A fully realized quantum computer would far outperform even the most awe-inspiring supercomputer of today. This power stems from quantum bits, or qubits (typically subatomic particles), that exhibit quantum mechanical phenomena.
Compelling qubit behaviors include superposition, the ability to exist in multiple states simultaneously, and entanglement, a quantum state that defies current logic: entangled particles remain connected and interrelated, despite being physically separate.
Quantum computing is expected to fuel breakthroughs in fields from artificial intelligence (AI) and engineering to pharmaceuticals.
However, we are not there yet.
Due to their sensitivity, hundreds of thousands, perhaps even millions of qubits are required to produce low-error quantum set-ups that would revolutionize computing. Today, even the most sophisticated systems, such as Google's Sycamore and IBM's Raleigh, have just a few tens of qubits.
Following are three new approaches being pursued that could edge us closer to an actual quantum reality.
Hot qubits for cooler computing.
Andrew Dzurak and Henry Yang of Australia's University of New South Wales (UNSW), Sydney, are focusing on quantum technologies that integrate on silicon chips using qubits, which are a simple modification of the transistors used for conventional computing. They have been investigating the impact of temperature on scalability.
Superconducting qubits, the type being explored by companies like Google and IBM, need to operate at extremely low temperatures (close to absolute zero, typically 0.1 degrees Kelvin or less) because quantum coherence breaks down at high temperatures in solid-state systems. Large-scale multi-million-dollar dilution refrigerators keep current systems cold.
Explains Dzurak, "As you add extra quantum bits, and you also add the electronics to manipulate them, it becomes increasingly hard to keep a system cold."
If temperature control is difficult with tens of qubits, it becomes almost unfathomable with millions of qubits. The solution, say Dzurak and Yang, is "hot qubits."
Along with collaborators from Canada's Université de Sherbrooke, Finland's Aalto University, Japan's Keio University, and the Canadian Institute for Advanced Research, the researchers produced a proof-of-concept quantum processor unit cell using silicon qubits that operates at 1.5 Kelvin —15 times hotter than most other qubits.
The cell, made up of two qubits confined in a pair of quantum dots embedded in silicon, is based on experimental work carried out by Yang.
Typically, an electron reservoir is used to initialize and read the qubits' spin information. However, this process depends on low temperatures to keep the electrons from heating up and disturbing the energy distribution.
In the new protocol, quantum dots are isolated from the electron reservoir, and the initializing and reading of the qubits takes place via tunneling, a quantum process state that allows electrons to move through the barrier between the quantum dots. This negates the need for ultra-low temperature control, which is why "we can do the qubit operations at higher temperature, because we are able to now initialize and read out the spin at higher temperature," says Yang.
In practical terms, says Dzurak, this equates to cheaper, simpler cooling systems. "You can now imagine having millions of these silicon qubits on a chip and integrated with the control electronics, so it really solves this key issue of scalability up to millions of qubits."
Faster operations with giant atomic ions.
Speed also impacts scalability. As in classical computing, logic gates and circuits are bedrocks of quantum computing. To be useful, qubits need to implement gate operations at high speed, but weak interactions slow the process down.
Researchers at the Universities of Nottingham and Stockholm have developed an approach that uses trapped Rydberg ions to create faster quantum gate operations. The system combines the strong dipolar interaction of Rydberg atoms with the quantum benefits of trapped ions— ions suspended in a trap— which have low error rates and can be accurately controlled.
The trapped Rydberg ions are generated by using a laser to excite normal trapped ions from a ground state into high-lying Rydberg states. "Here the Rydberg atom is huge, their radius is thousands of times bigger than ground state, so they interact strongly," explains Weibin Li, a professor at Nottingham's School of Physics and Astronomy.
This strong dipolar Rydberg interaction supports the speeding up of trapped ion entangling operations, allowing the team to demonstrate a sub-microsecond entangling gate between two ions— around 100 times faster than a typical trapped ion system.
A Goldilocks ion for error reduction.
As we move into the Noisy Intermediate-Scale Quantum (NISQ) era and see the development of 50+ qubit computers, errors in the preparation and measurement of qubits remain a major challenge.
"In general, the field is striving to build a gate-based quantum computer where error correction can be implemented at or below the fault-tolerant threshold, i.e., applying error correction improves accuracy of the computation," says Justin Christensen, a researcher at UCLA's Hudson Lab.
Currently, there is a trade-off between number of errors and number of ancilla, or extra, qubits required to perform fault-tolerant error correction: the lower the number of errors, the fewer ancilla qubits are required.
Trapped-ion qubits have demonstrated low error rates and are below the minimum fault-tolerant threshold. However, the number of ancilla qubits required for fault-tolerant error correction remains prohibitive.
As Christensen explains, "Current experiments can control somewhere between 10 and 50 qubits, while fault-tolerant error correction would require many thousands of qubits."
The new UCLA qubit, 133Ba+, is hosted in a laser-cooled, radioactive barium ion. The so-called "Goldilocks ion" has a unique electronic structure that allowed the team to achieve a qubit preparation and measurement error rate of about 0.03%.
"This work demonstrated the lowest state preparation and measurement error of any qubit on any platform by approximately a factor of two, as well as high-fidelity single-qubit gates," says Christensen.
The work will be especially beneficial in the near term as fault-tolerant error correction is not feasible, Christensen adds. "As we continue to demonstrate improved operational fidelities of 133Ba+, we can move towards implementing and improving the total fidelity of quantum computation."
Karen Emslie is a location-independent freelance journalist and essayist.