Scientists have developed what they say is the first complete design for a silicon quantum microchip. It portends the possibility of quantum computers being produced entirely using conventional silicon technology.
Classical computers switch transistors either on or off to symbolize data as ones and zeroes. In contrast, quantum computers use quantum bits or qubits that, because of the nature of quantum mechanics, can be in a state of superposition where they are both 1 and 0 simultaneously.
Superposition lets a qubit perform two calculations at once. If two qubits are quantum-mechanically linked, or entangled, they can help perform 22 or four calculations simultaneously; three qubits, 23 or eight calculations; and so on. In principle, a quantum computer with 300 qubits could perform more calculations in an instant than there are atoms in the visible universe.
Research teams around the globe are exploring a variety of different ways to build a useful quantum computer. There are at least five major quantum computing approaches being explored: silicon spin qubits, ion traps, superconducting loops, diamond vacancies, and topological qubits.
One challenge all quantum computing strategies face is that qubits are highly vulnerable to disruption from heat and other noise. To overcome this fragility, quantum computers need to employ error correction codes to protect them from interference. For example, the information from a single “logical” qubit can spread to several highly entangled “physical” qubits to reduce the chances that any environmental disturbance will tamper with the information in question.
The demands of quantum error correction codes suggest that quantum computers will likely need up to millions of qubits in order to solve complex problems and prove useful. Scaling up existing quantum computing strategies to such large numbers of qubits has proven a daunting challenge.
Now, a group of scientists in Australia has developed a design they say could integrate millions of silicon spin qubits onto microchips. Moreover, “all of the components can be manufactured using a conventional silicon chip manufacturing plant using standard CMOS materials,” says Andrew Dzurak, director of the Australian National Fabrication Facility at the University of New South Wales in Sydney. Dzurak was senior author of the paper detailing the group’s work that was published today in the online version of the journal Nature.
One key advance in this chip blueprint is an architecture of conventional electronic components that can manipulate and read out data from millions of qubits, Dzurak says. Most other quantum computation designs have focused only on the qubit-level architecture and have ignored the conventional electronic components, or else have included the control electronics on separate chips, he adds. “This is the first that integrates everything on one chip,” Dzurak says.
Another key feature of this blueprint is a new kind of error correction code designed specifically for silicon spin qubits, Dzurak says. “Basically, it would allow 1 million qubits to be contained on a chip that is the same size as a conventional microprocessor,” he says.
One of the first steps needed to implement this blueprint is to design the transistors used to select the qubits, Dzurak says. Another is designing the precise component layouts and feature sizes to minimize cross-talk between the components, he adds. “All stages will involve the design and development of the specific fabrication processes for each of the components of the architecture, from the qubit gate electrodes to the connecting lines,” says Dzurak.
The researchers currently have US $63.7 million in funding to develop a 10-qubit prototype silicon quantum integrated circuit by 2022. The funds are part of a deal reached by the University of New South Wales, Australian telecommunications company Telstra, the Commonwealth Bank of Australia and the Australian and New South Wales governments. In August, these partners launched Australia's first quantum computing company, Silicon Quantum Computing Pty Ltd, to advance the development and commercialization of the research team's technologies.
“We would not aim for a million-qubit chip on day one,” Dzurak says. “We would start with a more modest goal—say a 10-qubit device, then move to 100 qubits, then gradually progress to 1 million, in the same way that Moore's Law has held for conventional microprocessors.”