Five years ago, teams of physicists at Harvard University caused quite a sensation when they demonstrated that light pulses could be drastically slowed down and even brought to a standstill, then reactivated at will and sent on their merry way. Commentators were quick to predict stunning new applications in communications and in optical and quantum computing.

The enthusiasm quickly evaporated, however, when it sank in that the experiments at Harvard had required enormously complex laser apparatus that could fill a room.

Now, though, separate groups in the United States and Europe say that they have built and successfully tested more compact, rugged, and efficient means of delaying the pulses. Their work seems to clear the way for the kinds of applications foreseen by the Harvard pioneers, including not just those in optical switching and quantum communications but also others in network synchronization, radar, and even computer memory.

Of course, you can slow a light beam by directing it through glass or any other material with a relatively high index of refraction. And a dark piece of paper will stop a beam quite dependably. But by absorbing the photons, the paper destroys the beam irretrievably. What the Harvard researchers had found was a way of slowing or stopping light pulses without destroying their constituent photons and then re-creating the pulses utterly unchanged.

Lene Vestergaard Hau, a Danish physicist at Harvard, was the first to stop light. What she had done, in effect, was imprint the information carried by photons into spin patterns in clouds of atomic gases—“parking” the pulses in a gaseous medium, as she put it—and then reconstitute the pulses as desired, in a technique somewhat reminiscent of holography. Any information carried by the beam would remain perfectly intact.

Hau’s close competitor at Harvard, Mikhail Lukin, anticipated using this stop-light technology as a means of transporting quantum states from one part of a computer to another, an essential process in any large computer based on quantum principles.

There are nearer-term possibilities, too: a buffer for a router, for example, in which an optical delay line might keep one train of light pulses briefly on hold, allowing another train to pass through the router. Phased-array radars, commonly used in the military, could also benefit. In a phased-array radar, many small antennas transmit pulses that are delayed electronically in a systematic way to create a narrow beam that can be steered by changing the delays to the individual antennas.

But producing and controlling these delays electronically is costly. It might be cheaper to devise a system in which electronic input is converted to optical signals, delayed in a tunable system, and then reconverted into electronic signals that are fed to microwave signal amplifiers and individual antennas in the correct phase.

In the new work, the European and U.S. groups are slowing light pulses in optical fibers rather than in atomic gases, by up to several nanoseconds. They’re taking advantage of a phenomenon known as stimulated Brillouin scattering, which involves using sound waves to change the refractive index in a material. When incoming light waves encounter the changed refractive index, they scatter and slow down as some of the light is reflected back into the fiber and interferes with the incoming beam.

Both groups—a team led by Luc Thévenaz at the Swiss Federal Institute of Technology, in Lausanne, and the other led by Alexander Gaeta at Cornell University, in Ithaca, N.Y.—were able to send data pulses with wavelengths of roughly 1550 nanometers through one end of spooled optical fibers. The fibers ranged in length from several hundred meters to a few kilometers, simulating real-world conditions.

Using a pump beam with a slightly different frequency from the data beam, the teams generated sound waves in the fiber. The sound wave scatters the control beam, lowering its frequency to that of the data beam. Both beams interfere constructively, slowing the pulse down.

The team led by Gaeta reported delaying 15-nanosecond-long pulses by more than 25 ns. The Lausanne team reported similar results, delaying pulses by up to 30 ns [see photo, “Taking Pulse”]. To be sure, those delay times of barely more than a pulse length are still too short for data to actually be represented. “To be useful, this effect should be capable of delaying the pulse by at least a few pulse lengths,” comments Harvard’s Lukin.

Another limit, especially for broadband applications, is the maximum frequency of the delayed pulses achieved in the experiments, which was only 35 megahertz. But that problem seems solvable: both groups recently reported success in increasing the bandwidth by modulating the control beam, giving it a bandwidth of several hundred megahertz. That additional bandwidth increased the bandwidth of the slowed pulses, too. “There is no real limit for the extension of the bandwidth—we can extend it up to many tenths of a gigahertz,” says Thévenaz.

The first real-world applications may not be that distant, says Daniel Gauthier of Duke University, in Durham, N.C., who participated in the Gaeta group’s research. One application he sees right away is a pulse regenerator. Its use would restore pulse trains that have been distorted by traveling over long distances through optical fibers and are out of sync with the system clock, which enables the system to determine where meaningful data strings start. “You need to resynchronize the data pulse stream with the system clock, and for that you need one-pulse-width adjustment,” says Gauthier.