Unconditionally teleported quantum gates between remote solid-state qubit registers
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Nature Communications |
| Published | 2026-05-26 |
| Authors | Mariagrazia Iuliano, N. Demetriou, H. Benjamin van Ommen, Constantijn Karels, Tim H. Taminiau |
| DOI | 10.1038/s41467-026-72818-6 |
The authors performed a teleported Controlled-NOT gate between carbon-13 nuclear-spin qubits in two diamond nodes connected by 2 m of optical fiber. All intermediate measurement outcomes were accepted without post-selection or data filtering, and the remote Bell-state fidelity was 63(4)%. A heralded four-qubit GHZ state across the nodes reached 64(4)% measured fidelity.
Unconditional gate logic
Section titled “Unconditional gate logic”Unconditional operation applies to the gate logic after remote entanglement has been heralded. The experiment still detects photons to establish that entanglement, so it does not remove the probabilistic stage of network startup. Once the link is ready, every mid-circuit measurement outcome is accepted and processed through real-time feed-forward. A wrong measurement assignment can still select the wrong operation, so unconditional acceptance does not mean error-free gates.
Each node couples an NV-center electron spin to one carbon-13 nuclear spin. The electron spin serves as the communication qubit, while the nuclear spin serves as the data qubit. Alice and Bob operate in separate cryostats at 3.9 K and 3.4 K, respectively. Initialization and single-shot communication-spin readout use spin-selective optical transitions. The mid-circuit communication-spin readout is nondestructive, which limits extra data-qubit dephasing. Local gates inside each two-qubit register supply the logic required by the teleported operation.
Node architecture and control conditions
Section titled “Node architecture and control conditions”Alice and Bob use external magnetic fields of 189 mT and 31 mT, respectively. Electron-spin operations use microwave pulses delivered through on-chip gold striplines. Alice drives the carbon-13 qubit with phase-controlled radio-frequency pulses interleaved with dynamical decoupling, a method called DDRF. Bob uses tailored dynamical decoupling sequences for the same nuclear-spin role. Radio-frequency control lies in the MHz range, while microwave control lies in the GHz range.
Red 637 nm and yellow 575 nm lasers provide optical control through one combined excitation path. DC Stark tuning adjusts the emitted optical frequencies at each node. Charge repumping uses a high-power 575 nm pulse on the neutral NV zero-phonon line to counteract ionization. The paper notes that high-field Stark tuning also depended on favorable strain conditions.
The average initialization fidelity, corrected for known communication-spin tomography errors, was 85(1)% for Alice and 96(1)% for Bob. The six test states covered the positive and negative X, Y, and Z Bloch-sphere axes. Identified error sources included communication-spin pulse errors, laser leakage, state-mapping errors, and imperfect decoupling from the nuclear-spin bath.
Remote entanglement under repeated attempts
Section titled “Remote entanglement under repeated attempts”One remote-entanglement attempt lasts 8.392 μs. Each node emits a photon toward a midpoint 50:50 in-fiber beamsplitter connected to superconducting nanowire single-photon detectors. A detector click heralds electron-spin entanglement in the protocol. The optical phase difference at the beamsplitter is actively stabilized before entanglement generation.
The measured entangled-state fidelities were 77(2)% for Ψ plus and 76(2)% for Ψ minus. The average simulated fidelity was 79%, with an average photon-state parameter α of 0.045 between the setups.
The carbon-13 data qubits have intrinsic coherence times of tens of milliseconds in the current devices, but entanglement attempts cause faster decay. During an attempt, the communication-spin state cannot be tracked perfectly, so the nuclear-spin phase develops stochastic errors. Alice calibrates the mean phase per attempt and updates the RF local oscillator. Bob applies calibrated XY8 dynamical-decoupling sequences compiled into a control lookup table.
The dephasing metric N1/e counts attempts after which the fidelity contrast falls by 1/e. The fitted values were 391(31) attempts for Alice and 479(19) attempts for Bob. The authors set a 50-attempt timeout before reinitializing the data qubit, balancing experiment rate against fidelity. This timeout shows that accepting every gate branch does not yet provide fully deterministic network operation.
Four-qubit GHZ benchmark
Section titled “Four-qubit GHZ benchmark”The same protocol creates a four-qubit Greenberger-Horne-Zeilinger state across the two nodes. Initialization, remote entanglement, rephasing, local operations, feed-forward, and mid-circuit readout all contribute to the state. The measured fidelity was 64(4)%, compared with 66% from simulations using measured parameters. The reported GHZ and gate fidelities use corrections for known tomography errors.
Because the result exceeded 0.5, the authors used it as evidence of genuine four-partite entanglement. The state was generated without post-selection. The authors state that, to their knowledge, it was the largest heralded GHZ state across optically connected solid-state network nodes.
Remote-entanglement imperfections and data-qubit initialization were the main limits. Incorrect assignment of communication-spin outcomes occurred for about 5% of Alice’s and 9% of Bob’s measured 1 outcomes. Those assignment errors reduced the observed fidelity by about 7%, leading the authors to estimate an actual GHZ fidelity of about 71%. Incorrect assignment therefore affects both feed-forward and the rephasing sequence compiled for data-qubit readout.
Teleported Controlled-NOT performance
Section titled “Teleported Controlled-NOT performance”The nonlocal Controlled-NOT was first tested with the two data qubits prepared in computational basis states. The four measured two-qubit fidelities were above 70% on average and agreed with the simulated truth table.
The authors then prepared Alice’s data qubit in the X state and Bob’s data qubit in the 1 state. The ideal gate transforms this input into the Ψ plus Bell state. Measured XX, YY, and ZZ correlators gave a Bell-state fidelity of 63(4)%, compared with 65% in simulation. This result shows that the remote gate preserved quantum coherence and generated entanglement between the data qubits.
The experiment accepted every mid-circuit outcome. A simulation restricted to the 00 communication-qubit outcome predicts 90% average truth-table fidelity and 76% Bell-state fidelity. The difference indicates that better communication-spin readout would materially improve the gate.
Scaling requirements and prior context
Section titled “Scaling requirements and prior context”Full deterministic operation would remove the 50-attempt timeout and the associated data-qubit reinitialization. The authors state that active link efficiency must exceed one. This requires data-qubit coherence under networking to exceed the time needed to generate one or more entangled states.
The authors cite prior work using weakly coupled carbon-13 spins and pairs of nearby carbon-13 spins as routes to much longer coherence under network activity. They also suggest dynamical-decoupling pulses on the data qubit to extend coherence in the current platform. These are proposed extensions, not measured improvements in this experiment.
For the entanglement rate, the authors point to cavity enhancement and more efficient communication qubits. They identify the integrated DDRF method as an enabler for multi-qubit control and larger resource states. Earlier photonic gate-teleportation experiments relied on post-selection. Unconditional operation had been reported for superconducting qubits in one cryogenic system, a segmented ion trap, and nearby trapped-ion systems. This experiment extends that result to solid-state qubits across an optical link.
The NV platform had previously enabled heralded entanglement over 10 km using 25 km of deployed fiber and basic protocols on a three-node network. For device engineering, the priorities are stable optical transitions, long nuclear-spin coherence during network activity, accurate phase tracking, and communication-spin readout with low assignment error. Indistinguishable photon frequencies enable the link, but the data show that coherence and readout also limit gate performance.
Materials and support
Section titled “Materials and support”For this quantum-node application, 6CCVD’s diamond materials catalog lists single-crystal diamond grown by MPCVD and Au metallization. The paper requires NV-center diamond with optical access, microwave and RF control, and gold striplines. These device requirements should be reviewed through engineering support before selecting a material.
View Original Abstract
Quantum networks connecting quantum processing nodes via photonic links enable distributed and modular quantum computation. In this framework, quantum gates between remote qubits can be realized using quantum teleportation protocols. The essential requirements for such non-local gates are remote entanglement, local quantum logic within each processor, and classical communication between nodes to perform operations based on measurement outcomes. Here, we demonstrate an unconditional Controlled-NOT quantum gate between remote diamond-based qubit devices. The control and target qubits are Carbon-13 nuclear spins, while NV electron spins enable local logic, readout, and remote entanglement generation. We benchmark the system by creating a Greenberger-Horne-Zeilinger state, showing genuine 4-partite entanglement shared between nodes. Using deterministic logic, single-shot readout, and real-time feed-forward, we implement non-local gates without post-selection. These results demonstrate a key capability for solid-state quantum networks, enabling exploration of distributed quantum computing and testing of complex network protocols on full-stack systems.
High Concentration Quantum Grade Single Crystal Diamond
View Specs →Products
View Specs →
Single Crystal Cvd Diamond Plate
View Specs →References we cover
Section titled “References we cover”- Spin-photon correlations from a Purcell-enhanced diamond nitrogen-vacancy center coupled to an open microcavity
- Improved Electron-Nuclear Quantum Gates for Spin Sensing and Control
- Quantum networks based on color centers in diamond
- Realization of a multinode quantum network of remote solid-state qubits