The world of computing has held guarded optimism about one day having quantum computing capability. Recently, two teams of Australian scientists propelled the dream a great leap forward by developing the first silicon-based quantum qubits, or bits, the basic information-unit in quantum computing. The accomplishments are described in two recent publications in the journal Nature Nanotechnology.

“We have demonstrated that with silicon qubit we can have the accuracy needed to build a real quantum computer,” said Andrew Dzurak of the University of New South Wales, who is co-author on both papers. “That’s the first time this has been done in silicon.”

Previous work with phosphorous atoms as qubits achieved 50 percent accuracy when used in silicon. One of the new methods built upon these findings but with silicon atom qubits in silicon.

“In natural silicon each atom also has its own spin which affects the phosphorous atom, which is why the accuracy was only 50 per cent,” said Dzurak. “We solved the problem by removing all the silicon 29 isotopes that have magnetic spin leaving only silicon 28, which has no magnetic spin to influence the phosphorous, giving us an accuracy of 99.99 per cent.”

The second method involved turning a silicon transistor into an “artificial atom” qubit. As such, the accuracy of the data storage reached 99.6 percent. Transistors produce binary code, zeros and ones, by flowing electrons through an electronic gate that can be turned on or off.

“What we’ve done is make a silicon transistor with just one electron trapped in that transistor,” explained Dzurak. “This lets us use exactly the same sort of transistor that we use in computer chips and operate it as a qubit, opening the potential to mass-produce this technology using the same sort of equipment used for chip manufacturing.”

In addition, with the silicon chips, the two teams of scientists also improved coherence time, or the time over which a quantum system retains data.

“In solid-state systems these times are typically measured in nano or micro seconds before the information gets lost,” said Dzurak. “We are getting a 30-second coherence time, which on the time-scale of doing calculations is an eternity.”

Dzurak added that silicon is preferred in the industry because of its long coherence times while retaining accuracy.

“We can go better, these are only our first experiments,” says Dzurak.