Pump-Free Microwave-Optical Bell Pair Generation for Teleportation-Based Quantum Transduction

The paper in detail

A superconducting processor works with microwave photons at gigahertz frequencies, and optical fibres carry photons at hundreds of terahertz. Direct conversion typically uses a strong optical pump, whose light near the superconducting circuit adds heat and noise. The proposal avoids the pump: first create an entangled microwave–optical pair, then transfer quantum states by teleportation.

The optical half uses a spin in a cavity. An optical photon in an equal superposition of early and late time bins, \(\tfrac1{\sqrt2}(\ket E+\ket L)\), reflects off a cavity containing a spin in \(\tfrac1{\sqrt2}(\ket0+\ket1)\). Spin state \(\ket0\) couples to the cavity and reflects the photon with probability \(\mathcal R_{\max}\); \(\ket1\) does not reflect it. A spin \(\pi\)-pulse is applied between the two time bins. Keeping only events in which the photon returns gives \[\ket\Phi=\tfrac1{\sqrt2}\bigl(\ket E\ket1+\ket L\ket0\bigr).\] The spin is then mapped to a microwave photon by Purcell-enhanced emission into a microwave resonator, and a transmon heralds that photon’s arrival.

Four platforms are analysed: single colour centres in diamond (NV⁻ and SnV⁻), an ensemble of ⁸⁷Rb atoms, and a phonon-mediated variant in which the spin couples to a mechanical mode that couples piezoelectrically to the microwave resonator. The analysis assumes transmon relaxation time \(T_{1,t}=300\ \mu\mathrm s\) and measurement error \(\epsilon=0.01\). The predicted maximum fidelities and heralding rates are 0.92 and 3.0 kHz (NV⁻), 0.86 and 1.1 kHz (SnV⁻), 0.97 and 3.7 kHz (⁸⁷Rb), and 0.99 and 28.3 kHz (phonon-mediated). Rate and fidelity peak at different detection windows, so the two maxima are not reached at the same operating point.

Teleportation turns the pair into a transducer. A Bell measurement acts on the microwave input and the microwave half of the pair. Its two classical outcome bits select a Pauli correction on the optical photon, which then carries the input state. The paper describes this step qualitatively; its quantitative analysis concerns the entangled-pair source.

Transferable methods

1. Separate resource generation from state transfer

   The proposal first creates microwave–optical entanglement. A Bell measurement and a communicated correction then transfer the input state by teleportation.

   The teleportation identity. Let \(\ket{\beta_{jk}}=(I\otimes X^jZ^k)\ket{\Phi^+}\) be the four Bell states. For an input \(\ket\psi\) on system 1 and a pair \(\ket{\Phi^+}\) on systems 2 (microwave) and 3 (optical), \[\ket\psi_1\ket{\Phi^+}_{23}=\tfrac12\sum_{j,k\in\{0,1\}}\ket{\beta_{jk}}_{12}\otimes X^jZ^k\ket\psi_3\] up to signs on individual terms. Outcome \((j,k)\) therefore leaves the optical photon in \(X^jZ^k\ket\psi\), and applying \((X^jZ^k)^{-1}\) recovers \(\ket\psi\). The source’s state \(\tfrac1{\sqrt2}(\ket E\ket1+\ket L\ket0)\) is \(\ket{\Phi^+}\) up to a local relabelling. (Editorial; identity checked numerically.)

   Figures of merit. A heralded source is characterized by a rate and a conditional fidelity, \[R=\frac{P_{\mathrm{succ}}}{\tau+T_{\mathrm{exe}}},\qquad\mathcal F=\bra\Phi\rho_{\mathrm{herald}}\ket\Phi,\] where \(\tau\) is the detection window, \(T_{\mathrm{exe}}\) the rest of the protocol time, and \(P_{\mathrm{succ}}=\tfrac12\frac{C^2}{(1+C)^2}P_{\mathrm{mw}}\). For qubit teleportation with ideal Bell measurement and a pair of fidelity \(\mathcal F\), the average transfer fidelity is \(f=(2\mathcal F+1)/3\) (the teleportation channel is a Pauli channel with identity weight \(\mathcal F\)). Thus \(\mathcal F=0.92\) gives \(f\approx0.95\), above the classical limit \(2/3\). (Editorial.)

   How to reuse it

   Evaluate a heterogeneous interface first as a source of entangled pairs, with a rate \(R\) and a conditional fidelity \(\mathcal F\), and only then as a transducer via \(f=(2\mathcal F+1)/3\). This separates what the source must deliver from the cost of the Bell measurement, which is a demanding microwave operation in its own right and is not analysed here.

2. Use an intermediate spin as an interface

   A reflected optical photon becomes entangled with a spin, and cavity-enhanced emission then maps the spin state to a microwave photon. Timed control produces time-bin entanglement.

   The optical step. For a one-sided cavity at critical coupling on resonance, the empty cavity reflects nothing. A coupled spin with cooperativity \(C=g^2/(\kappa\gamma)\) gives amplitude \(r=1-\frac1{1+C}=\frac C{1+C}\). Hence \[\mathcal R_{\max}=\frac{C^2}{(1+C)^2},\qquad P_{\mathrm{opt}}=\tfrac12\mathcal R_{\max}.\] The \(\tfrac12\) is the fraction of time-bin and spin branches in which the photon meets the reflecting spin state. A large \(C\) is essential; \(C=3\) already loses 44% of the reflected intensity.

   The microwave step. Purcell emission at rate \(\Gamma_0=4g_{se}^2/\kappa_e\) for a single spin, or a collectively enhanced \(\Gamma\sim\sqrt Ng_{se}\) for an ensemble, gives emission probability \(p_{\mathrm{mw}}=1-e^{-\Gamma\tau/2}\) within the window \(\tau\). A transmon with error \(\epsilon\) heralds a click with probability \(P_{\mathrm{mw}}=p_{\mathrm{mw}}(1-\epsilon)+\epsilon(1-p_{\mathrm{mw}})\), where the second term counts false heralds. The heralded fidelity is \[\mathcal F_{\mathrm{herald}}=\frac{F_M}4\Bigl(1+e^{-\tau/T_{1,t}}+2e^{-\tau/2T_{1,t}}e^{-\tau/T_{2,s}}\Bigr),\qquad F_M=\frac{(1-\epsilon)p_{\mathrm{mw}}}{P_{\mathrm{mw}}}.\] Read it as the Bell overlap \(\tfrac12(\rho_{aa}+\rho_{bb}+2|\rho_{ab}|)\) with \(\rho_{aa}=\tfrac12\), \(\rho_{bb}=\tfrac12e^{-\tau/T_{1,t}}\) (the branch holding an excitation decays) and \(|\rho_{ab}|=\tfrac12e^{-\tau/2T_{1,t}}e^{-\tau/T_{2,s}}\) (coherence lost to relaxation and spin dephasing), scaled by the true-herald fraction \(F_M\).

   The trade-off. A longer \(\tau\) raises \(p_{\mathrm{mw}}\) and the true-herald fraction. It also lowers the fidelity through decoherence and lowers the rate through the denominator \(\tau+T_{\mathrm{exe}}\). Taking \(\tau\approx2/\Gamma\), the paper estimates \[R_{\max}\sim\frac12\frac{C^2}{(1+C)^2}\,\frac1{2/\Gamma+T_{\mathrm{exe}}}.\] The fidelity optimum lies elsewhere, so no single window maximizes both.

   How to reuse it

   Compare candidate mediators by three quantities: the cooperativity, through \(\mathcal R_{\max}=C^2/(1+C)^2\); the retrieval rate \(\Gamma\) relative to the coherence times \(T_{2,s}\) and \(T_{1,t}\); and the dead time \(T_{\mathrm{exe}}\). Then choose the window \(\tau\) for the rate–fidelity balance the application needs. Heralding discards failed attempts, but it does not remove false heralds (the \(\epsilon\) term) or decoherence within the window.

Source: Figures 1–2 and the main-text protocol; Appendices A–B. Based on the linked paper version. Derivations marked as editorial, and suggested uses in other problems, are editorial guidance rather than results of the paper.

Context & limitations

Why it matters. Offers a potential interface between microwave quantum processors and optical communication.

Limitations. These are proposed operating regimes, not an experimental network demonstration.

Description based on the full text of the linked arXiv version, read 2026-10-19; equations follow the paper’s notation.