New neutral-atom qubit offers advantages for quantum computing

13 Oct 2022 Ali Lezeik

Near-deterministic loading: The neutral-atom qubit setup at JILA. (Courtesy: Kaufman lab)

Two US-based research teams have developed quantum information processors that use neutral ytterbium (Yb) atoms as qubits – the first time this atomic species has been employed for this purpose. Trapping 100 Yb atoms in a 10 × 10 array, the researchers showed that they could perform entangled two-qubit gate operations on them, paving the way towards quantum computers based on this choice of qubit.

In principle, qubits can be any quantum system capable of transporting information through a so-called quantum register, which houses qubits in the same way as a classical register houses bits in group of 8, 16, 32 and 64. Previously, however, all neutral atom qubits were based on alkali metals such as rubidium or caesium. Owing to their single valence electron, this group of atoms is highly controllable using advanced, well-understood techniques such as laser cooling and trapping.

In the latest experiments, independent teams led by Adam Kaufmann from JILA in Colorado and Jeff Thompson from Princeton University in New Jersey instead used the nuclear spin of a Yb isotope, Yb-171, as their choice of qubit. The rich internal structure of the “alkali earth-like” metal Yb offers numerous possibilities for cooling and trapping while also making it possible to create qubit systems that are robust to external perturbations. Yb-based qubits could therefore allow for more efficient gate operations, boosting the performance of quantum information processors.

Setting up an optical tweezer array

An important criterion for a high-fidelity quantum computer is to have as much control as possible over the way the quantum register is set up. In a technique the JILA team call “near-deterministic loading”, a gas of atoms is first cooled and prepared in a magneto-optical trap. The gas is then compressed to increase the atom density before the atoms are loaded into an optical potential formed by a 10 × 10 array of devices known as optical tweezers. The increase in density ensures that each of the 100 tweezer sites contains at least one atom.

The trapped atoms are then placed in a magnetic field, which divides them into separated groups determined by their magnetic substates. This allows the researchers to use an additional laser beam to “kick out” excess atoms from overloaded tweezer sites in order to isolate a single atom at each site. This sequence loaded a single atom in more than 90% of the array, and according to Aruku Senoo, a PhD student working on the JILA experiment, combining it with the well-developed tweezer rearrangement protocol should make it possible to scale qubit numbers.

Single-qubit gate operations

Once they prepared their qubits in the -½ magnetic substate of Yb-171, members of both teams were able to demonstrate single-qubit operations, initializing the qubits to the ½ state with a fidelity (a measure of the operation’s control) of 99.95%. Because this sequence exploits the magnetic substructure of Yb’s energy levels, Thompson thinks the operation’s maximum coherence time – that is, the qubit lifetime – of 3.7 seconds can be further increased by stabilizing the magnetic field used in the setup. Furthermore, the trapping mechanism depends on the polarization state of the light fields, so optimizing this further could make the trapping more efficient.

The biggest challenge both teams had to overcome was determining the final state of the qubit. A common way to do this is by fluorescence imaging – essentially, shining light on the atoms to excite a transition between atomic energy levels and then measuring the light they emit in response. However, picking the right wavelength for the imaging beam proved tricky. While the JILA team utilized a broad transition at 399 nm, the Princeton team decided to use a so-called “magic” wavelength that would leave the qubit state unchanged during imaging and reduce the loss of atoms. But since the energy levels of the Yb-171 isotope have not yet been mapped in detail, the Princeton team first had to find this magic wavelength.

“That spectroscopy took a month or two because we were cobbling together random lasers with low power that could sometimes only make one or two tweezers, but it was necessary as there was no precise theoretical prediction,” Thompson says.

Two-qubit entangled Rydberg states

According to Thompson, these experiments are “just the beginning of finding out what we can do with qubits in Yb-171”. One particular avenue of interest would be to develop scalable quantum computers based on entanglement mediated by highly-excited Rydberg states. The Princeton team demonstrated such an entangled state in Yb-171 for the first time. Using a sequence of light pulses, the corresponding entangled states, or Bell states, were generated with a fidelity around 85%.READ MORE

Trapped ytterbium ions could form backbone of a quantum internet, say researchers

Although the two-qubit gate fidelity demonstrated is below that demonstrated by ion or superconducting qubit platforms so far, Senoo says that a Yb-based qubit system has a promising path to building 1000-qubit arrays, whereas scaling up the number of trapped-ion or superconducting qubits even to the 100-qubit level is not very straightforward. Moreover, Rydberg state-mediated entanglement has the advantage of limiting crosstalk and undesired interactions in a many-qubit entangled system. Such interactions decrease the fidelity of qubit operations, as has been shown with trapped ions and superconducting qubits.

According to Thompson, neutral atoms are certainly having a moment now. Both teams are working towards quantum error correction to achieve a better two-qubit gate fidelity by utilizing other transitions in Yb-171. Their research is published in back-to-back papers in Phys Rev X.

Ali Lezeik is a PhD student contributor to Physics World. Ali is currently working on quantum optics at the Leibniz University Hannover. Find out more about our student contributor networks

from physicsworld.com                   15/11/2022