Physicists Trim the Noise from Quantum Computing

The deep realm of the very small is hardly a pristine place. This is especially true when electric charge is involved, which is maybe not such a big deal in dealing with the enormous bundles of particles involved in classical charge-based information and computing schemes—where information is encoded as bits via "high" and "low" electric currents—but noise becomes a significant barrier in quantum computing.

Quantum computing is the ever-looming digital future in which information is stashed in the states of single particles or very small particle collections, in contrast to the gushing particle pipelines of electrical current in use now. A quantum dot, for instance, might be just a single excited electron confined to a small depression (or "cage"). This is precarious, in some large part because of omnipresent electrical background interference.

A group of physicists based at the University of Wisconsin and Sandia National Laboratories has devised a new alternative to semiconductor charge-qubit manipulation, a very high-speed method of switching qubits around that mostly avoids the "charge noise" barrier that comes with manipulating qubits with the DC voltage pulses of prior experiments. Here, microwaves are used instead, offering a way of juggling qubits without the interference; quantum information can be both stored and switched at very high-speeds while also maintaining very high-fidelity, which is a necessary capability.

"The noise in the experiment appears to be dominated by charge noise," Mark Eriksson, lead author of a new paper in Nature Nanotechnology describing the microwave pulse method, told me. "That is, the electric field from the environment near the qubit fluctuates randomly in time. Microwave manipulation helps avoid the worst effects of this charge noise, because microwave manipulation of a charge qubit can be performed entirely at the operating point that is most well protected from charge noise."​

Information in electrically-controlled quantum information schemes involves the spatial manipulation of a particle using charge as a sort of probe that bounces electrons between adjacent sites or "holes" in a semiconductor, a qubit implementation known as a "charge qubit." A voltage pulse can do the job, but only within a certain "sweet spot," an operating point beyond which the qubits will undergo dephasing, which is like a kind of quantum decay. A dephased particle is no longer in a superposition of states, resulting in information loss and, indeed, low fidelity. Really, it's just not a qubit anymore.

It'd be nice if we could hew to that sweet spot, but there's a catch in that DC gating—gating being how how operations like AND and OR and XOR are implemented and, thus, how information is actually processed—requires some deviation from that sweet spot, causing information (coherence) to be squeezed out of the system like water from a washcloth, resulting in unacceptably low fidelities. A catch-22. Enter microwaves.

"The reason microwaves add this [lower noise] capability is that the phase of the microwave pulses can be controlled in the experiment, and changing the phase changes the axis about which the qubit rotation occurs," Eriksson explained. "In contrast, DC voltage pulses manipulate the charge qubit by switching the qubit between two or more operating points; only one of those points can be the optimal point, so dc voltage pulses necessarily cause the qubit to experience a bit more of the detrimental effects of the charge noise."

It turns out that there's a whole catalog of different possible qubit forms. In addition to shuttling electrons around cage to cage as in with charge qubits, it's also possible to store quantum information via other properties, like spin or light polarization or even time.

But making any them work in the world inside of actual, even everyday machines is its own profound challenge, requiring both quantum computers that can be built with reasonable materials, like silicon, the material behind Eriksson's work, and that be reasonably robust. Quantum information that withers in the presence of a bit of background interference doesn't have much of a place in the computing future.

An open-access preprint version of Eriksson's study is available at arXiv.