Ronald Walsworth, a lecturer at Harvard, led the effort, which is part of a larger program attempting to harness quantum mechanics to create uncrackable codes, photonic quantum memories and eventually, blazingly fast quantum supercomputers.

"We are one step closer to the first practical application of quantum information processing. Today we can nondestructively transfer information from light to the spin state of atoms, then transfer it back out into the same light beam, while preserving all the original phase information," Walsworth said.

Over the next five years, Walsworth's group plans to supply the missing pieces to a secure quantum system. Parallel research groups worldwide investigating rival materials include fellow Harvard professor Lene Hau, who is investigating Bose-Einstein condensates, and Texas A&M associate professor Phillip Hemmer, who is investigating rare-earth-doped insulators.

The first step, demonstrated last year by all three groups, was the ability to store a laser-encoded signal in the spin states of atoms and then nondestructively read them back out. In the current step, reported recently by Walsworth's group, the signal maintains phase coherence.

Computer components

The last step will be to demonstrate that information encoded in quantum states can be stored and retrieved. At that point, all the components for quantum computers will be in place.

One aspect that will be an automatic fallout of total quantum information-processing systems will be absolute security. That will result from the basic property of all quantum systems: Any reading of data will alter the data, permanently invalidating it. While a quantum computer is processing data that has been entered into the system, no one can know what the actual information content of a specific operation might be. It is only when the resulting quantum states are read out — which constitutes an observation of their states — that information content can be known. Thus, any attempt to discover the values of intermediate results would become an additional quantum operation that would invalidate the computed results for any subsequent calculations.

Those unusual properties, which current information encoding cannot mimic, result from the nature of a bit of quantum information, or qubit, which is simply some property of the quantum state of an elementary particle.

For example, a qubit could be stored as the spin of an electron. Because of the quantum nature of such media, a qubit of information is not strictly in either a logical "0" or "1" value but instead is in a somewhat mysterious "superposition of states" that is a combination of the two values. While a quantum computer is operating on the qubits, their actual information content — the phase information that describes the exact superposition of states they are in at that moment — is an inherently unknown quantity.

The recent Harvard experiment shows that it is possible to store qubits in a memory and retrieve them later, without having to observe, and therefore invalidate, their values.

Quantum phenomena abound in nature. Both waves (light/lasers) and particles (atoms/electrons) harbor quantum states. Those states are being harnessed by various research groups for communication and computations using qubits. Walsworth's group transferred quantum states in a laser beam to the atomic "spins" of rubidium atoms. The Rb atoms were able to store the spin state of a signal for about a thousandth of a second before having to be "refreshed" like a DRAM. "It takes about a millionth of a second to store the whole signal in the rubidium atoms, where it will stay for about a thousandth of a second, which is quite a long time for a computer memory," Walsworth said.

Besides demonstrating that the phase coherence was maintained, Walsworth's group also demonstrated that a global function could be performed on the signal, resulting in predictable "computations" on the stored values after they were read back out. Walsworth used a magnetic field to perform a "numerical operation" on the stored spin values on the Rb atoms. When the signal was read out, its phase was predictably shifted by the magnetic "computation" performed on it while stored. "We can alter the form of the quantum information without knowing what it is — that is what's nice about it. We can alter it in such a way that serves our purposes and then read out the results," he said.

The next milestone will be to demonstrate that quantum information is preserved by the store/retrieval process, creating essentially a quantum "repeater." Researchers hope to achieve that step within the next couple of years, while aiming at long-range quantum communications links in five years.

A quantum repeater will be an optical element for long-distance secure communications of qubit streams. Because qubits cannot be decoded without altering the original quantum information, a quantum repeater would enable a communications network with uncrackable encryption to span any given distance by adding repeaters.

Practical possibilities

"In about five years, we hope to have the first practical application of quantum information processing, which is brief but ultrasecure messaging," Walsworth said. "Because the information is quantum-mechanical, if it is measured in any way before it gets to you — if anybody tries to eavesdrop — the information will be destroyed, and you will not receive it. So if you get the information, then you can be sure it was not observed by anyone else." That has attracted the attention — and funding — of the Defense Advanced Research Projects Agency and the National Security Agency.

In the basic experiment, two laser beams are projected into a quantum medium consisting of rubidium atoms. In other experiments, fellow researcher Hau has used sodium atoms, and Texas A&M's Hemmer has used rare-earth doped insulators.

One beam is the signal laser that contains the information to be stored, and the other serves as a fixed reference. The interference states caused by the reference laser are transferred to the spin values of the atoms. To reconstruct the original, the reference laser beam is turned on, and the photons that hit the atoms pick up the differential signal, in effect turning the reference beam back into the original signal beam.

A somewhat indirect method of quantum-state transfer was required, because if only one laser beam illuminated the rubidium medium, the photons would simply be absorbed without transferring any quantum information. By adjusting a second control laser, a condition of "electromagnetically induced transparency" prevents the signal beam from being absorbed. As the leading edge of the signal laser enters the Rb vapors, it spatially compresses. As the atomic spins of the molecules are flipped, the signal and the spins form a coupled excitation called a polariton.

The signal becomes translated into atomic spins by gradually reducing the intensity of the control laser, thus slowing the velocity of the polariton until all the information comes to a halt, stored away in the form of the excited spins of the Rb atoms.

The system of rubidium atoms is in a coherent matter state that mirrors the coherent state of the laser light. By "tuning" the reference beam, the two coherent states instantaneously merge, allowing the quantum states encoded in the laser light to transfer smoothly to the quantum states of the rubidium medium. The process can be reversed by turning the reference beam back on.