Chance discovery brings quantum computing using standard microchips a step closer
An accidental innovation has given a dark-horse approach to quantum computing a boost. For decades, scientists have dreamed of using atomic nuclei embedded in silicon—the familiar stuff of microchips—as quantum bits, or qubits, in a superpowerful quantum computer, manipulating them with magnetic fields. Now, researchers in Australia have stumbled across a way to control such a nucleus with more-manageable electric fields, raising the prospect of controlling the qubits in much the same way as transistors in an ordinary microchip.
“That’s incredibly important,” says Thaddeus Ladd, a research physicist at HRL Laboratories LLC., a private research company. “This could potentially change the game for nuclear qubits in silicon.”
An ordinary computer flips bits from 1 to 0 and back again. A quantum computer employs qubits that can be set to 0, 1, or, thanks to the bizarre rules of quantum mechanics, 0 and 1 at the same time. This enables a quantum computer to crunch a huge number of inputs simultaneously, which is one reason why a big one should able to solve certain types of complex problems that would swamp any conventional computer. Last year, researchers with Google claimed their small quantum computer performed an abstruse calculation that would have taken conventional supercomputers millennia.
Still, it’s far from clear which of many types of qubit is best. Each of the 53 qubits in Google’s machine consists of a small circuit of superconducting metal that has two settings or “states” with different energies. By tickling it with microwaves, each circuit can be set to one state, the other, or both. However, the qubits must run at near absolute zero temperatures in a bulky contraption called a dilution refrigerator, which is about the size of a phone booth. Google researchers think they can fit about 1000 qubits in one dilution fridge. A full-fledged quantum computer with millions of qubits might then require thousands of interconnected dilution refrigerators.
In 1998, Bruce Kane, a condensed matter physicist now at the University of Maryland, College Park, thought up a more compact technology with individual phosphorus atoms embedded in silicon and placed in a magnetic field. Each phosphorus nucleus spins like a top, and with an oscillating magnetic field, researchers could coax it to spin in the same direction as the field, the opposite way, or in both ways at once, making it a qubit. The delicate spin states would last for nearly 1 second, and in principle, researchers could put millions of qubits on a single silicon microchip.
But this technology lags behind superconducting qubits. One difficulty comes in addressing a single nucleus with a magnetic field because the field tends to spill over and mess with neighboring nuclei. Now, Andrea Morello, a quantum engineer at the University of New South Wales (UNSW), Sydney, and colleagues have found a way to control such a nucleus with a more manageable electric field, they report today in Nature.
Morello and colleagues studied an antimony nucleus embedded in silicon. The larger antimony nucleus has higher spin than phosphorus. So, in a magnetic field, it has not just two basic states but eight, ranging from pointing in the same direction as the field to pointing in the opposite direction.
In addition, the distribution of electric charge within the nucleus isn’t uniform, with more charge around the poles than the equator. That uneven charge distribution gives experimenters another handle on the nucleus in addition to its spin and magnetism. They can grab it with an oscillating electric field and controllably ease it from one spin state to another or into combinations of any two. All it takes is applying an electric field of the right frequency with a simple electrode, the researchers report.
The researchers discovered the effect by accident, Morello says. For reasons that have nothing to do with quantum computing, they had wanted to study how the antimony nucleus embedded in a silicon chip would react to jolts of the oscillating magnetic field generated by a wire on the chip. But the wire melted and broke, turning the current-carrying wire into a charge-collecting electrode that instead generated an oscillating electric field.
The discovery was even more complicated. For the oscillating electric field to create an effect, the nucleus first had to sit in an uneven static electric field. Luckily for the researchers, that uneven field arose naturally from distortion in the silicon caused by aluminum leads on its surface that had contracted as the chip was cooled to its operating temperature near absolute zero.
It took a month for researchers to figure out what was going on, says Serwan Asaad, a postdoc at UNSW. That’s “longer than I would like to admit to a reporter,” he says. Morello says theorists predicted in 1958 that an oscillating electric field could flip a nucleus, but no one had observed it.
The embedded antimony nucleus isn’t exactly a qubit because it has eight basic states, notes Gavin Morley, a quantum physicist at the University of Warwick. But that’s OK, he says, because eight states are actually equivalent to three two-state qubits. The all-electric technique could be used to control other nuclei, he says, but not phosphorus, because it has a uniform charge distribution.
It’s too early to say whether nuclei in silicon will succeed as qubits, Kane says, as researchers struggle to assemble more than two of them in a device. Still, the advance highlights the allure of silicon technology to build a compact quantum computer. “That’s what keeps our little field funded,” Kane says. If a silicon-based quantum computer is possible, it may be inevitable, Morley says. “There’s a saying that if something can be done in silicon, it will be done in silicon, just because the industry is so big.”