Features:

A SHOCK TO THE SYSTEM
by J. William Bell NCSA Science Editor

Few, if any, were incredulous of the claims made by John Joannopoulos and his team in May 2003. Joannopoulos is a physics professor at the Massachusetts Institute of Technology and has been a leading light in the photonics world for three decades. He wrote the book, two texts in fact, on his field.

But everyone was surprised by the team's findings that a shockwave passing through a photonic crystal could be used to modulate the frequency of light regardless of its intensity, narrow its bandwidth, and trap it for a time. No other known system is capable of these effects. New Scientist magazine put a fine point on it, saying "Claims of 'unexpected and stunning new physical phenomena' are rare in the abstract of a reputable scientific paper. But the latest report from Joannopoulos' group does not disappoint."

"The degree of control over light really is quite shocking," said Eli Yablonovitch, who developed the first photonic crystal while at Bell Communications Research in the late 1980s.

Even the team didn't fully anticipate its discovery, born of numerical simulations conducted in part on NCSA's SGI Origin2000 system and continuing on the center's Titan Linux cluster and IBM p690 supercomputer.

"The motivation was to try to find the simplest system that had the periodicity [that all photonic crystals exhibit], have a shock propagate through, and see what happens. These results, these new physical phenomena that came out, were an unexpected consequence of this thought experiment," says Evan Reed, a postdoc in the Joannopoulos lab and lead author on the Physical Review Letters article that announced the findings.

What Semiconductors Are To Electrons

Photonic crystals are made up of one or more materials, precisely laid atop one another to form a lattice of repeating, identical patterns. The different materials impact the light in different ways, various defects can be introduced to further manipulate the light, and the combinations of these effects can alter the resulting light's properties. The key features of photonic crystals are "band gaps." Band gaps block particular frequencies of light within the crystals, as a result of the crystals' lattice structures. Thus, photonic crystals can be used to stop some types of light while allowing others to pass through.

"In a nutshell," Joannopoulos wrote in a 2001 Nature article, "the idea is to design periodic structures that affect the behavior of photons in much the same way that crystalline semiconductors affect the properties of electrons."

Already, photonic crystals are used in optical communications networks and in directional antennas on cellular phones that thwart radiation headed toward your head. The ability to not only block certain frequencies but also alter those frequencies, narrow light's bandwidth, and hold the light offers more grand possibilities.

Reed comes up with two in short order. Today, telecommunications networks typically use light of two different frequencies to represent the ones and zeros of a binary data transmission. Modulating the frequency of that light-- and modulating it as quickly as possible--is thus crucial. The modulations are currently done via electronic processing, and the data rate is limited by this processing. Shocked photonic crystals, however, would allow for all-optical systems, circumventing the slow down that accompanies conversion back and forth through the electronic frequency modulators in contemporary networks.

Solar power might also someday benefit from shocked photonic crystals' bandwidth narrowing. "The sun emits a very broad electromagnetic spectrum, and you want to harness as much of that spectrum as possible," Reed explains. "Well, the way you do that now is with solar cells with layers of many different materials, and each material is good at harnessing a certain narrow regime….If you have a technology that takes a big chunk of that bandwidth and just compresses it down to a single frequency, that can make a big difference in efficiently capturing this energy."

Can't Start It Like A Car, Can't Stop It With A Gun

The Joannopoulos team is currently collaborating with researchers at Lawrence Livermore National Laboratory and Los Alamos National Laboratory to see if their numerical models can be reproduced experimentally. Their approach? Bullets or high-intensity lasers are fired at test crystals to induce shockwaves, and the team examines the light that passes through the crystals in the instant between the compression due to the impact and the destruction by the same. It's not exactly a feasible means in the applied world, but the tests will confirm the computational results.

The team imagines microelectromechanical machines or acoustic systems will eventually do the shocking. "Think of much less violent systems that won't destroy the crystal," says Reed. "These will be repeatable and practicable."

While the folks at Livermore are on the shooting range, the Joannopoulos team stays in the machine room, conducting other photonic crystal studies. Some focus solely on the light. There's no need to model the crystal when they already know, through previous studies of their own and by others, how it will react. Those conditions can just be plugged into the Maxwell's equations that govern the light's behavior.

Their studies of Cerenkov radiation, for example, are conducted with this mindset. Cerenkov radiation--which team member Chiyan Luo likens to a sonic boom--is the light projected forward when a charged particle travels through a material at a speed greater than the speed of light in the material. In a January 2003 Science article, the team reported that Cerenkov radiation projects backward in some photonic crystals.

Other studies, meanwhile, center on the crystals themselves, examining new designs and new means of modeling those designs.

Without Preconceived Notions

Joannopoulos and his team currently use systems from throughout the National Science Foundation's Partnerships for Advanced Computational Infrastructure and TeraGrid programs. In addition to NCSA's p690 and Titan cluster, the team uses the TCS1 system at the Pittsburgh Supercomputing Center. The Pittsburgh Supercomputing Center is a partner in the National Science Foundation's TeraGrid project, which includes NCSA.

This year, they're on track to expend about 650,000 hours of computing time. Last year, they came in at just under 500,000 hours and also used time at the San Diego Supercomputer Center, another TeraGrid partner. "I probably shouldn't admit this," Joannopoulos says, "but we run projects wherever we think we're going to get the fastest turnaround times on a given day."

Those big numbers stem from nearly 30 years of numerical simulation, dating back to the earliest Cray machines. "I probably shouldn't admit this, either, because it dates me," Joannopoulos says, "but I've been at this longer than NCSA has."

Supercomputing offers an incentive that's worth the investment of a career, however. Using ab initio methods--working from the most basic theoretical principles, like Maxwell's equations--means coming at the problem without any unintentional biases, Joannopoulos explains. You set the rules of engagement, you provide the parameters, and then you get out of the way. Physics takes over.

An experiment can't perfectly capture theory. There's always a smudge on the lens, the scale is only precise to so many significant figures, or there's something equally confounding that you can't account for. But, according to Joannopoulos, the Maxwell's equations that the team uses for its photonics simulations are so straightforward and the computing power is so abundant that you can look at the light's properties and behavior to "whatever degree of accuracy you want." This is in contrast to most computational models where "the model comes with preconceived notions--about how bonding is handled or how defects are handled. When dealing with electrons, for example, you only have a very approximate idea of how those electrons are interacting."

That's why "experiment still gets the upper hand [in most materials studies]." But with ab initio studies of photons, he says, "We know precisely what we put into it and precisely how those things are going to interact. We're way ahead of the game."

Funding statement This research is supported by the National Science Foundation, the Department of Defense/Office of Naval Research Multidisciplinary Research Program of the University Research Initiative, the Department of Energy, and the Materials Research Institute and the Energetic Materials Center at Lawrence Livermore National Laboratory.

For further information: http://ab-initio.mit.edu/.

Team members Peter Bermel David Chan Kerwyn Huang Mihai Ibanescu John Joannopoulos Steven Johnson Elefterios Lidorikis Chiyan Luo Michelle Povinelli Evan Reed David Roundy Marin Soljacic