A quantum computer would surpass even today's supercomputers by being able to crack encryption schemes or simulate quantum physics in far less time. The key to its power is the quantum bit, or qubit, which is not limited to representing 0 or 1. Qubits exist in fuzzy states that are both 0 and 1, and can be combined to represent many numbers at once [see photo, " "]. As a result, a quantum computer would be like a massively parallel computer array, whose power grows exponentially with each additional qubit.
The most promising technology for constructing an ultrapowerful quantum computer is the ion trap, a nest of electrodes that holds ions in midair. If a quantum computer is ever to be scaled up and manufactured, however, ion traps must be built from semiconductors, the way computer chips are. Researchers have now built the first such ion-trap chips. Linking multiple chips may allow research groups to manipulate much larger numbers of ions and demonstrate rudimentary components of a quantum computer over the next few years.
One way to construct a scalable quantum computer that could handle thousands of qubits rather than just a few would be to turn electrical currents into qubits. But this approach has lagged, because the currents degrade fairly rapidly. Ions, by contrast, are extremely stable and have been used to demonstrate the fundamentals of quantum computation. But ion traps haven't been very scalable, because they had to be assembled one by one. "When you have thousands of electrodes, you can't hand-assemble it; the alignment is too hard," says ion trapper Christopher Monroe of the University of Michigan, in Ann Arbor. "You really need to make it on a multilayered chip."
Monroe's group has taken one of the first steps toward such a device. As reported in the January issue of Nature Physics, his team fashioned a smaller, squatter version of a conventional trap by alternating layers of gallium arsenide and aluminum-gallium-arsenide. The result was an insulating layer sandwiched between two electrode layers. The researchers carved a channel down the chip and etched the flanking regions into individual diving boardshaped pieces. They were able to load a single cadmium ion into the gap between these pieces. To cool and probe the ion, they shone a laser down the central channel.
When not being cooled, the trapped ion quickly absorbed a lot of energy from the surrounding device, causing it to "boil" out of its trap. Adding another ion also knocked the first one out. "It wasn't the best-behaving trap I've seen," says Monroe. The electrodes overhang the insulating layer by a short distance, which may cause them to vibrate at just the right frequency to agitate the ion, he hypothesizes.
A potential disadvantage of stacked-electrode designs is the wiring, which has to be crammed in from the side. A more planar trap, in which the electrodes lie on the same surface, would allow the electronics controlling each electrode to be plugged in from underneath, requiring fewer wires from the side.
To scale up, "you've got to control a lot of potentials on a lot of electrodes, and probably the only reasonable way to do that is to have the electronics on board," says David Wineland of the National Institute of Standards and Technology, in Boulder, Colo. Wineland's group has fabricated a planar trap that has held up to six magnesium ions in a neat row. In this setup, four gold electrodes lie on a quartz substrate and the ions hover above them. A laser plays across the surface.
Last year the U.S. intelligence community's Advanced Research and Development Activity, a low-profile operation based in Hanover, Md., began funding competing but complementary groups at Lucent Technologies Bell Laboratories and Sandia National Laboratories to build ion chips that would serve as common research platforms. They will serve the world's ion trap groups, including newcomers such as the one at the Massachusetts Institute of Technology, in Cambridge. Both labs have expertise in fabricating micrometer and submicrometer semiconductor devices: Lucent's group has built planar traps from silicon, while Sandia's is constructing planar and nonplanar chips.
A major benefit of chip-based ion traps is that they could be combined into large arrays. The goal is to create a big switchyard in which hundreds or thousands of ions are constantly being shuttled around in small groups, controlled by conventional electronics. In the next few years, researchers predict, ion traps will likely be scaled up from a handful of ions to 50 or 100.