In a single ion held in an electromagnetic trap, the Oxford group stacks three kinds of quantum states that have no classical analog into a single quantum superposition, widening the toolkit available to future quantum computers, ultra precise sensors, and tests of quantum…
Oxford physicists have moved past asking whether they can put a quantum system in two contradictory states at once. A new experiment, newly reported in Nature Physics on 5 May 2026 and highlighted by the University of Oxford via ScienceDaily this month, demonstrates a method for stacking non-classical states to make new, even stranger ones. The capability widens the alphabet of quantum states available for building better quantum computers, ultra-precise sensors, and tests of quantum foundations.
The famous image for a quantum superposition is 'Schrödinger's cat,' the thought experiment's box that holds a cat simultaneously alive and dead until someone looks. In the lab, the 'cat' is a quantum oscillator, any system that can vibrate in a wave-like way, sitting in a superposition of two clearly different states. What the Oxford group has done is go beyond the standard cat state.
The team's paper entangles a single trapped ion's internal spin, a controllable quantum property acting as the control knob, with the ion's motion, the oscillator that holds the state. A mid-circuit measurement on the spin then projects the motion into a chosen superposition of nonclassical oscillator states, a process the authors describe as 'sculpting' the result. In a single ion, the team demonstrated three higher-order non-Gaussian states: squeezing, trisqueezing, and quadsqueezing. 'Non-Gaussian' is the technical name for states that go beyond what a simple harmonic oscillator can produce on its own. Such states are required for universal quantum computing with oscillators, so they sit on the field's shopping list rather than its trophy shelf.
A Nature Physics research briefing describing the result lays out what makes the demonstration durable. The reconstructed Wigner functions, which are maps of a quantum state in phase space, show sixfold rotational symmetry and Wigner-negative regions for the trisqueezed-state superposition. 'Wigner negativity' is the field's accepted fingerprint for a state with no classical analog. Producing it in a higher-order state on a single trapped ion is a capability expansion, not a record for its own sake.
The platform itself is straightforward for anyone who follows the field. A single ion sits in an electromagnetic trap; its internal spin acts as the control knob; the ion's motion acts as the oscillator. By tuning the strength of two spin-dependent interactions, the team can shape which non-Gaussian superposition ends up in the motion. The protocol extends the same group's earlier Z2 lattice gauge theory simulation, which used a similar approach to project the motion into a state useful for high-energy physics problems, and traces its theoretical foundation to a 2021 paper by Sutherland and Srinivas proposing universal hybrid quantum computing in trapped ions.
That theoretical foundation is what makes the result more than a one-off. According to the research briefing, the new approach is platform-agnostic across spin-oscillator systems and scales to higher nonlinear orders and to multiple oscillators. For builders in the continuous-variable quantum computing community, where information is encoded in the wave-like properties of oscillators rather than in discrete two-level 'qubits,' this is the part that matters: the result is a primitive, not a product.
The authors' stated forward-looking applications include error-resistant hybrid continuous-variable quantum computing, where richer state alphabets could support simpler error-correction strategies, and quantum sensing, where non-classical states can push past the noise floor of classical probes. The paper also points to foundational tests of the quantum-classical boundary. These are possibilities the authors name, not outcomes the paper demonstrates. A single trapped ion whose motion has been sculpted into a higher-order nonclassical state is not a scalable processor, and a May 2026 paper showing a state-engineering primitive does not by itself buy more resilient quantum computers next year.
What it does buy is a method. The shift the Oxford work represents is from 'can we make a cat state?' to 'can we build arbitrary non-Gaussian states on demand?' That second question is the one continuous-variable quantum computing, quantum error correction with oscillator states, and quantum sensing have all been waiting on. The answer in this paper is at least: yes, in a single ion, and on a protocol that extends.