Quantum Hardware

Continued from page 1

But most ion traps are difficult to fabricate, consisting of a ceramic insulator and gold contacts for conducting an electrical current. In contrast, Monroe's team built their chip out of insulating layers of alloys of aluminum, gallium, and arsenide, with semiconducting layers of gallium and arsenide -- all easy to deposit on a chip using a conventional process called molecular beam epitaxy. They then etched a hole in the chip and "fashioned a set of cantilevered electrodes protruding over the hole," says Monroe.

The chip is placed in a vacuum, which then gets injected with a vapor of cadmium ions. When the appropriate voltages are applied to the electrodes, a cadmium ion with a free electron becomes trapped, floating between the cantilevers above the etched hole. In order to actually use the atom's free electron for computation, Monroe explains, the ion must be probed by a laser beam that reads the electron's spin state. The challenge "was just a matter of getting the right combination of parameters," such as electrode voltages and laser wavelengths ("and the phases of the moon," Monroe jokes).

This first ion-trap chip builds on a quantum computing roadmap that Monroe, David Kielpinski at MIT, and David J. Wineland at NIST published in Nature in 2002. Their plan envisioned a large-scale quantum computer using multiple ion traps on a chip.

"The trick is to take [a single ion trap] that is reasonably well understood and make more of them," says physicist Bruce Kane at the University of Maryland, who's working on a silicon-based quantum computer.

Putting multiple traps on one chip presents difficulties, Kane says, because as the number of traps increase, it becomes more difficult for a laser to read the state of an individual electron without interfering with the state of other electrons.

Despite such challenges, though, Carl Williams, coordinator of the NIST Quantum Information Program, says the Michigan research is "another step forward" in quantum computing.

"They've used state-of-the-art fabrication techniques to design much more complicated traps so that they can actually build a quantum computer," he says. "And that's an important result."

For quantum computers to be truly useful in cryptography, engineers will need to build in roughly 10,000 qubits, the number required to factor a 100-digit number, explains Kane.

As the number of traps increases, however, other potentially useful applications for quantum chips may appear before then. Says Kane: "I think most workers in the field would say, 'Yeah, these goals are necessary -- but we don't understand fundamentally what could happen' in the meantime.".