It is hard to assign a definite date to mark the beginning of the field of quantum computing. Some may trace it back to the famous paper by Richard Feynman from the early 1980s, others may prefer to start with Peter Shor’s 1994 quantum algorithm for factorization or earlier algorithms developed in the early 1990s. But when it comes to the first conception of how to actually build a quantum computer, one result stands out. On 15 May 1995, a paper by Ignacio Cirac and Peter Zoller was published in Physical Reviews Letters. It showed that atomic ions, trapped in a vacuum and cooled using lasers, could be used as qubits with which elementary quantum logic operations could be performed. The Cirac–Zoller paper turned quantum computing from a bold theoretical idea into an experimental race to build an actual device.

At the International Conference on Atomic Physics in 1994 physicists shared the excitement about Shor’s freshly published algorithm and the potential offered by a quantum computer. Artur Ekert, one of the pioneers of quantum cryptography, challenged atomic physicists: how can we build one? The first candidate qubits were photons, but to perform two-qubit operations, essential for quantum computation, qubits need to interact, and photons are notoriously non-interacting. Cirac and Zoller proposed a different route: ions. Ion trapping and cooling were well-known technologies, allowing the ions to be isolated from the environment and their motion frozen out. The ion’s internal energy levels could be used to encode the two states of the qubit. These could be manipulated and read out by exciting the atomic transitions using lasers. All these ingredients promised a clean, controllable set of qubits ready for performing quantum operations on. “It was a wonderful time that lasted about 3 months and we really enjoyed seeing how different pieces were coming together,” Cirac and Zoller recall. But one crucial ingredient was missing: a two-qubit gate, more precisely, a controlled-NOT gate, which inverts the state of one qubit depending on the state of another (control) qubit. To realize such a gate, a way to transfer information between qubits was needed.

Credit: Image courtesy of Samuel Hile, Sussex University

“After several attempts, we came up with an idea to do that using the ‘phonon bus’, which is a set of degrees of freedom shared by all qubits, something that is used now with other platforms as well,” Cirac and Zoller remember. Using lasers, the internal energy levels of the ions can be coupled to the external motional degrees of freedom (the phonon bus) common to ions trapped together. In this way, the information could be communicated between different ions, and conditional operations such as the controlled-NOT gate could be realized.

By the end of 1995 researchers at NIST had already put Cirac and Zoller’s ideas into practice and demonstrated a two-qubit controlled-NOT gate using beryllium ions. From there the technology advanced quickly, and today the state of the art is hundreds of gates and tens of ions.

Cirac and Zoller’s idea to use a common degree of freedom as a data bus to enable two-qubit operations is simple and elegant, and for this reason it was adopted in other platforms. But the devil is in the detail. In practice, trapped ions are not perfectly isolated from their environment. All sorts of noise sources have been identified and combatted. The trapping and manipulation of tens of ions is not an easy feat and today there is an entire field dedicated to the development of sophisticated ion trapping technologies (see the Technical Review in this Issue).

In 1995, Cirac and Zoller came up with the blueprint for an ion trap quantum computer, and circuit-based quantum computer architecture in general. Twenty-five years on, despite tremendous progress, engineering challenges remain, but first-generation practical quantum computers seem tangible.