Belling that cat: Oz boffins pass entanglement test
Two-cubit operations possible thanks to working single-atom transistor in silicon
Entanglement is easy to generate, but if you want to prove you have entanglement - and that's one of the many holy grails of quantum computing - you have to pass what's called "Bell's inequality test". Doing it in silicon is even better, and that combination has the corks popping at the University of New South Wales.
Their work, announced here, was based on using a single-atom transistor (phosphorus in ordinary silicon), and as the university says, the Bell score of 2.70 is the highest recorded. Since entanglement is regarded as proven for any score over 2.0, the team, led by Professor Andrea Morello, regards their work as a big deal.
Bell's inequality was proposed in 1964 as a test of whether entanglement exists (and at the same time excluding the “hidden variables” explanation of entanglement proposed by Einstein-Podolsky-Rosen to explain the phenomenon).
Speaking to The Register this afternoon, Professor Morello said:
"The electron in the phosphorus atom has a spin (that's one qubit), and the nucleus has a spin (the other qubit), so you get two qubits. What we've done is to entangle the electron with the nucleus, and use the Bell test, which is … is spectacularly unforgiving: as soon as there is something slightly wrong in your setup, your system will no longer be able to pass the test.
"The Bell test is spectacularly unforgiving"
Without satisfying the conditions of the test, he said, "you might have created entanglement in there, you can't prove it. That's why passing the Bell Test is a big deal: you have proven that throughout the entire sequence of the operation … everything is perfect. The overall quality of [our] two quantum bits is as high as it's ever gotten."
The high-quality entanglement means that as well as the four states you can represent in two conventional bits (00, 01, 10 and 11), the group can add qubits' superpositions to what can be represented in silicon.
With superpositions and entanglement stored with high fidelity (96 per cent at the moment), you can move onto the challenge of using the qubits, and because it's in relatively standard silicon, the qubits are accessible to manipulation.
While some of the exotic needs of quantum operations remain - the device needs to operate at cryogenic temperatures to reduce noise, for example - that's far less of a problem than the challenge of creating a quantum chip out of common materials and using relatively standard techniques.
Silicon chips work well at cryonic temperatures, so operating a large quantum chip in the chiller doesn't require anything exotic, Professor Morello explained.
Phosphorus is already a ubiquitous doping material in silicon chips: the big change is that instead of a lot of Phosphorus atoms, this device uses one. It's placed using an ion implanter technique developed at the University of Melbourne, in which "you ionise the Phosphorus, and shoot it into the silicon", Professor Morello said.
Their work has been published in Nature Nanotechnology, and there's a pre-press version at Arxiv here. ®