Researchers at the University of Pennsylvania entangled four physical qubits inside a diamond at room temperature with a single control sequence, they reported in Nature Nanotechnology on September 14, 2026. The gate took 14.8 microseconds, 10 times faster than building the same state from sequences of two-qubit gates, the authors wrote.
Joseph Minnella, Mathieu Ouellet, Amelia Klein, and Lee Bassett of Penn’s Quantum Engineering Laboratory worked with a nitrogen-vacancy center in diamond. A nitrogen-vacancy center is a defect in which a nitrogen atom replaces a carbon atom next to a vacant site in the crystal lattice. The team used the center’s electron spin as the central qubit and three weakly coupled carbon-13 nuclear spins nearby as the other three.
The team made one entangling operation faster and more accurate inside a single four-qubit register, but the authors wrote that errors in preparing and measuring the qubits still limit the fidelity of the entangled states.
How the gate works
Entangled qubits have linked quantum states that cannot be described independently of one another. In nitrogen-vacancy registers, small groups of qubits around one defect, researchers usually build multi-qubit entangled states by entangling the electron with one nuclear spin at a time. The authors wrote that this sequential method is slow and prone to errors from crosstalk, unintended interactions between qubits.
The electron interacts continuously with the nuclear spins around it. The Penn team tuned the timing and repetition count of an XY8 dynamical decoupling sequence, a train of microwave pulses normally used to suppress unwanted interactions, so that one sequence entangled the electron with all three nuclear spins at once. The team used an entanglement metric framework that Evangelia Takou, Edwin Barnes, and Sophia Economou developed in theory papers published in 2023 and 2024.
The resulting four-qubit state is a Greenberger–Horne–Zeilinger, or GHZ, state, in which the electron and the three nuclear spins share one collective quantum state. The authors wrote that the parallel gate operates close to the fundamental speed limit set by the strength of the magnetic coupling between the electron and the nuclei. They also wrote that multipartite entanglement, shared among more than two qubits, is needed to run quantum algorithms, implement error correction, and achieve quantum-enhanced sensing.
Fidelity and its limits
The team estimated the parallel four-qubit gate’s fidelity at 0.92 ± 0.04, against 0.69 ± 0.03 for the sequential approach in the same register. Fidelity is a measure of how closely an operation matches its ideal version, with 1 meaning a perfect match.
The team also ran three-qubit parallel gates on every pair of nuclear spins. Those gates averaged a fidelity of 0.83 ± 0.09, against 0.78 ± 0.08 for sequential gates, and were almost twice as fast.
The researchers separated gate errors from those made in preparing and measuring the qubits by applying each gate repeatedly and measuring the state after every round. They wrote that preparation and measurement errors are the main limit on the fidelity of the entangled states and that readout is less efficient at room temperature.
To confirm the size of the entangled state, the researchers measured multiple quantum coherences, a technique in which the frequency of the readout signal indicates how many qubits share the state. The parallel gate produced the frequency expected for three entangled nuclear spins, while the sequential gate gave a shifted value of 2.6 instead of 3. The authors wrote that the method establishes the size of the entangled state but not its fidelity.
Scope and next steps
To estimate how widely the method applies, the researchers simulated registers with randomly placed nuclear spins. About 64% of the randomly generated registers contained at least one nuclear spin coupled to the electron more strongly than a cutoff the team set. The team evaluated the remaining 36% for parallel entanglement.
Among 500 such weakly coupled registers, gates entangling the electron with two or three nuclear spins in parallel were available in most configurations. The authors described their results as a lower bound on the gates available.
The authors wrote that the approach is readily applicable to other defects in diamond and to other host materials, such as silicon carbide and silicon. To cut room-temperature preparation and measurement errors, they pointed to repetitive readout at high magnetic fields, spin-to-charge conversion, and dynamical nuclear polarization, as well as to operating at cryogenic temperatures.
The authors wrote that the speedup creates new options for error correction and could make fault tolerance, or reliable computation despite errors, possible at room temperature. The team did not run error correction in the experiment. The authors simulated error correction only in the paper’s supplementary information.
The team first posted the work as a preprint, a paper shared before peer review, on arXiv in August 2025. The researchers archived the raw data behind the paper’s main figures on Zenodo, an open research repository. The U.S. National Science Foundation and Canada’s Natural Sciences and Engineering Research Council funded the work.
