“The silicon-based qubit can also be manipulated with 10 times greater precision than qubits based on GaAS, which has been widely used to make qubits in the past,” explains team leader Lieven Vandersypen.

A quantum computer will be able to solve certain computational problems – such as factoring large numbers or simulating complex chemical reactions as well as the interactions between large numbers of fundamental particles – exponentially faster and more efficiently than a current classical computer can. The building block of a quantum computer is a quantum bit (or qubit), which is the quantum analogue of the conventional binary digit (bit). In principle, powerful quantum computers can be built from a collection of qubits.

For a qubit based on an electron spin, for example, these states would be "spin up" and "spin down", with one state representing a logical "1" and the other "0". Each qubit can be in a superposition of two quantum states at the same time and N qubits could be quantum-mechanically entangled to represent 2N values simultaneously. This would lead to the parallel processing of information on a massive scale that is simply not possible with conventional computers.

Bit errors

However, quantum computers are extremely fragile, and a computation can be easily destroyed by "bit errors" in the qubit state that occur when external noise in the environment affects the values of the qubits. While it is proving very difficult to create practical qubits that are robust enough to eliminate such errors, an alternative approach is to accept that errors will occur and to try to correct for them as the quantum calculation progresses.

Now, Vandersypen and colleagues have made an important advance here by studying the qubit performance of an electron spin in an Si/SiGe quantum dot (a semiconductor structure just tens of nm across) and looking into the dominant error mechanisms in it. The researchers have found that they can electrically control the Si/SiGe-based qubit with sufficient accuracy so that remaining errors could, in principle, be corrected using known protocols

To use the electron spin as a qubit, they trapped one single electron in the quantum dot. The trap is based on the different material properties of Si and SiGe, explains Vandersypen, and on small electric fields from tiny metal electrodes on the substrate surface.

Electrically controlling quit spin

“Normally, spins are controlled with oscillating magnetic fields (like what is done, for instance, in MRI equipment in hospitals),” he says. “We controlled the spin of our qubit electrically. We place a small magnet next to the quantum dot and apply an oscillating electric field via one of the electrodes that forms the quantum dot. The oscillating electric field wiggles the electron position back and forth, and thanks to the magnetic field varying with position, the electron experiences an oscillating magnetic field, which allows us to control the spin.”

When placed in an arbitrary state, the qubit stays in that state for a microsecond, which is about 100 times longer than for a GaAs qubit, he tells nanotechweb.org. “With additional control signals to further stabilize the qubit state, we can extend this to 0.4 ms. What is more, we have quantified the probability of errors when we try to manipulate the state, and find that this probability is only about 1%.”

A collaboration with Intel

A 1% error probability during qubit manipulation is at the threshold of what is required to make a large-scale quantum computer, adds team member Susan Coppersmith. “The relatively long lifetime of the qubit memory of 0.4 ms also facilitates scaling up, proving that silicon is in fact a very attractive material for implementing electron spin qubits and we expect that it will generate a great deal of interest.”

Indeed, as if to prove this point, Intel Corporation entered the quantum computing arena last year and started a long-term collaboration with Vandersypen’s team.

The team, reporting its work in PNAS doi: 10.1073/pnas.1603251113, says that it is now busy looking into how best to couple several spin qubits together. “Moreover, we must develop the technology to scale up to a large number of (nearly) identical qubits – and this is an area where players like Intel can make a difference, since their high-quality nano-fabrication processes are optimized to be extremely reproducible,” says Coppersmith.