Oxford Team Makes a Quantum Ghost Show Up in an Ion Computer

Physicists at the University of Oxford have made the Aharonov-Bohm effect appear inside a hybrid trapped-ion quantum computer, a setting where it had not been observed before, according to a paper in Nature Physics published on September 25, 2026 ^[2][3].

It is a laboratory demonstration of one of quantum mechanics' strangest predictions. It is also a bench test for a way of building quantum machines that mixes two kinds of quantum hardware, aimed at physics too hard for ordinary computers to calculate ^[2].

The ghost in the loop

Yakir Aharonov and David Bohm predicted the effect in 1959, and experiments with real electrons later confirmed it ^[3]. A charged particle can pick up a measurable change in its quantum phase by travelling around a region of magnetic flux, without ever entering the region where the magnetic field is. The particle feels a field it never touches.

The picture to hold is two waves. A quantum particle going around a loop behaves like waves taking both the left and right paths at once. If the paths arrive in step, the particle shows up. If they arrive out of step, crest against trough, they cancel.

That cancelling is what the Oxford team saw. In their setup, matter could tunnel, that is, hop, around a loop when no flux was present. When flux was present, the two paths interfered destructively and the tunnelling was suppressed ^[2][3]. The team reports this as an observation of the Z2 Aharonov-Bohm effect. Z2 refers to a field that can take only two values, the simplest kind of gauge field.

What the machine is made of

The device is a trapped-ion computer, with charged atoms held in place by electric fields. It uses two quantum ingredients at once. Qubits, stored in the internal electronic states of the ions, encode the gauge fields. The ions' vibrations, which behave as quantum oscillators, encode the matter ^[2][3].

The approach is described as hybrid because it pairs two-level qubits with oscillators, which have a ladder of many possible states. The paper says it combines digital operations for preparation, error suppression and measurement with analogue evolution under an engineered gauge-invariant Hamiltonian, the rule set that governs how the system changes in time ^[2].

The experiment went in two steps. First the team ran a single ion and showed that matter and gauge-field dynamics were correlated in a way consistent with Gauss's law on a single Z2 link. Gauss's law ties the field to the matter that sources it ^[2].

Then they used a two-ion crystal to build a loop from two qubits and two oscillators ^[2].

The two gauge-field qubits were prepared in an entangled state that represents a Z2 magnetic flux through the loop ^[2][3].

Why the field had to move

Sebastian Saner, the paper's lead author at Oxford, described the key step as encoding the flux in a gauge field that was itself dynamic, rather than making matter move through a fixed background ^[3]. In this machine, matter and field both evolve as parts of one quantum system, so their interplay can be studied together ^[2].

That distinction matters because the theories involved are hard to simulate classically. The paper says prior experiments on real-time Z2 lattice gauge dynamics mostly used one-dimensional geometries, where visons, a kind of quasiparticle, and topological effects are absent, or effective approaches that removed the matter degrees of freedom ^[2]. A loop is the smallest geometry in which those effects can appear at all.

The other authors include Oana Băzăvan at Oxford and Alejandro Bermúdez of the Instituto de Física Teórica in Madrid ^[2][3]. The team began developing the experiment in 2022, and the result reached print in 2026 ^[3].

Small by design

The experiment did not send real electrons around a physical magnet. It represented magnetic flux inside a simulated lattice gauge theory, a framework that places matter on grid points and fields on the links between them ^[2][3]. The loop is a building block, not a model of the Standard Model of particle physics.

It is also not evidence that the machine can yet beat classical computers at a practical task. The authors present the work as a path toward quantum simulations of such theories in higher dimensions, and do not claim a useful general-purpose quantum computer ^[2].

The prize for getting there is broad. The paper says lattice gauge theories connect to models in particle physics and condensed matter, including high-temperature superconductors, exotic magnetic materials and methods for protecting quantum computers from their own errors ^[2].

That is the quieter point of this result. Quantum computing is usually sold as a threat to encryption or a shortcut for chemistry. Here it is a machine for asking physicists' own questions, and the answer is a textbook ghost caught in an instrument built to hold it.

A parallel result

The Oxford group was not alone. A University of Maryland-led group published a related Nature Physics paper using a hybrid quantum system to simulate a Yukawa model, an interaction relevant to nuclear and particle physics ^[1][4]. The two groups developed their approaches independently and coordinated their submissions ^[1].

Two teams reaching for the same idea at the same time is usually a sign that a field has found its footing.

What is not yet public

The sources available do not give the number of measurement repetitions behind the loop result, the size of its uncertainties, or the main sources of error. They also do not say what system size or lattice geometry the Oxford team will try next, or on what timescale. Whether the hybrid approach can beat classical methods on a scientifically useful problem is, by the authors' own framing, a later question ^[2].

That question decides what this result is worth. The next test is how far these methods can scale to more links and higher dimensions while keeping control of the gauge rules and the real-time dynamics that give the simulations their value ^[2].