Nonreciprocal Quantum Entanglement

• Nonreciprocal entanglement is defined as quantum entanglement that varies with direction, where reversing control variables like drive or magnetic field yields distinct entangled states.
• The phenomenon is achieved through mechanisms such as the Sagnac effect, Kerr nonlinearity, and Barnett shifts, leading to either unequal or unidirectional entanglement in both continuous- and discrete-variable systems.
• Experimental studies report directional contrast ratios and robustness metrics, indicating that nonreciprocal entanglement can be optimized for applications in quantum routers, isolators, and backaction-immune sensors.

Searching arXiv for papers on nonreciprocal entanglement to support the encyclopedia entry. Nonreciprocal entanglement is direction-dependent quantum entanglement: reversing a propagation direction, drive orientation, spin direction, magnetic-field orientation, or coupling phase changes the generated entanglement, and in ideal cases entanglement is present for one direction but absent for the opposite. In current arXiv literature, the concept appears in continuous-variable Gaussian platforms such as spinning optomechanical, magnomechanical, and exciton-optomechanical systems, as well as in discrete-variable settings including superconducting qubits and topological or plasmonic waveguides. Across these settings, the central feature is the same: reciprocity breaking modifies the effective dynamical generator asymmetrically, so that the steady or transient entangled state depends on direction (Xu et al., 14 May 2026, Jiao et al., 2020, Ren et al., 2024).

1. Definition and conceptual boundaries

In the relevant literature, nonreciprocity does not merely mean unequal transmission. It refers specifically to unequal entanglement under reversal of the control that defines forward and backward operation. In spinning resonators, the comparison is typically between CW and CCW driving, or left and right injection; in Kerr-based magnonic systems it is between opposite magnetic-field orientations; in Barnett-effect systems it is between positive and negative rotation-induced magnon shifts; and in waveguide or cavity-mediated qubit systems it is between opposite propagation directions or opposite initial excitations (Xu et al., 14 May 2026, Chabar et al., 13 Jul 2025, Chen et al., 2023, Ren et al., 2024).

The literature distinguishes several limiting cases. One is asymmetric entanglement, where both directions remain entangled but with unequal magnitudes. Another is one-way or ideal nonreciprocity, where the bidirectional contrast reaches unity and entanglement exists only in one direction. This ideal limit is explicitly discussed for Kerr-controlled cavity-magnon optomechanics, squeezed-magnon magnomechanics, Barnett-effect magnomechanics with an optical parametric amplifier, and several superconducting-qubit schemes (Chen et al., 2023, Imara et al., 9 Aug 2025, Chabar et al., 29 Sep 2025, Ren et al., 12 May 2026).

A recurrent misconception is that classical nonreciprocity and quantum nonreciprocity are interchangeable. Two recent results argue against that identification. In spinning optomechanics, nonreciprocal quantum entanglement can occur “even in the absence of any classical nonreciprocity” (Jiao et al., 2020). In an asymmetric Fabry–Perot cavity, the degrees of classical and quantum nonreciprocities “do not exhibit positive correlation as expected” (Wu et al., 5 Jun 2026). Nonreciprocal entanglement is therefore a distinct quantum property rather than a trivial by-product of directional transmission.

2. State-space formulations and entanglement measures

Most continuous-variable realizations use linearized quantum Langevin dynamics. A standard construction writes each operator as a steady value plus a fluctuation, introduces quadratures, and collects them into a fluctuation vector $u$. The dynamics is then written as

$\dot u(t)=A\,u(t)+n(t),$

where $A$ is the drift matrix and $n(t)$ is the noise vector. For Gaussian steady states, the covariance matrix

$$V_{ij}=\frac12\langle u_i(\infty)u_j(\infty)+u_j(\infty)u_i(\infty)\rangle$$

is obtained from the Lyapunov equation

$A\,V+V\,A^T + D = 0,$

with $D$ the diffusion matrix fixed by damping and thermal occupancies (Xu et al., 14 May 2026).

Bipartite entanglement is usually quantified by logarithmic negativity. Given a $4\times4$ covariance submatrix $V^{(\mu\nu)}$ for a chosen bipartition, one computes the smallest symplectic eigenvalue $\eta^-$ of the partially transposed block and evaluates

$\dot u(t)=A\,u(t)+n(t),$0

Variants of this formula recur across cavity magnomechanics, cavity-magnon optomechanics, molecular optomechanics, and asymmetric Fabry–Perot systems (Xu et al., 14 May 2026, Chen et al., 2023, Berinyuy et al., 23 Jan 2025, Wu et al., 5 Jun 2026). Some works extend the analysis to Gaussian EPR steering, using Rényi-2 entropies, or to genuine tripartite entanglement via the minimum residual contangle (Chabar et al., 13 Jul 2025, Ahmed et al., 2024, Lin et al., 2 Mar 2026).

Discrete-variable platforms instead use master equations and concurrence. In general 3D nonreciprocal photonic environments, the reduced two-qubit density matrix obeys a Lindblad equation whose coherent and dissipative rates are determined by the dyadic Green function. Entanglement is then quantified by Wootters’ concurrence

$\dot u(t)=A\,u(t)+n(t),$1

with $\dot u(t)=A\,u(t)+n(t),$2 the eigenvalues of $\dot u(t)=A\,u(t)+n(t),$3 in decreasing order (Gangaraj et al., 2017). The same measure is used for DC-current-biased graphene plasmons and for superconducting-qubit schemes based on coherent–dissipative interference or loss-induced phases (Berres et al., 2022, Ren et al., 2024, Ren et al., 12 May 2026).

3. Mechanisms that break reciprocity

Mechanism	Control variable	Representative consequence
Sagnac–Fizeau shift	$\dot u(t)=A\,u(t)+n(t),$4 set by rotation $\dot u(t)=A\,u(t)+n(t),$5 and drive direction	Detuning shifts differ for CW/CCW or left/right driving (Xu et al., 14 May 2026, Berinyuy et al., 23 Jan 2025)
Kerr, Barnett, squeezed-magnon, OPA effects	Sign of $\dot u(t)=A\,u(t)+n(t),$6, sign of $\dot u(t)=A\,u(t)+n(t),$7, squeezing phase $\dot u(t)=A\,u(t)+n(t),$8, OPA gain/phase	Magnon frequency shifts, pair terms, or damping shifts become direction dependent (Chen et al., 2023, Chabar et al., 13 Jul 2025, Imara et al., 9 Aug 2025)
Chiral, topological, and interference-based coupling	$\dot u(t)=A\,u(t)+n(t),$9, loss phases, qubit spacing, gap Chern number	Forward and backward effective couplings become unequal or strictly one-way (Fan et al., 2024, Ren et al., 12 May 2026, Gangaraj et al., 2017)

The Sagnac mechanism is the most widely used in spinning resonators. In a spinning whispering-gallery cavity, CW versus CCW drive flips the sign of $A$0, shifting the effective cavity detuning $A$1, the intracavity photon number $A$2, and the linearized couplings. In the two-YIG-sphere spinning cavity-magnon system, this asymmetry enters the drift matrix together with the Kerr-shifted magnon detunings, so that the covariance matrix and logarithmic negativity become direction dependent (Xu et al., 14 May 2026). In molecular optomechanics, the same sign flip can move the system between red-detuned and blue-detuned operation, thereby turning the two-mode-squeezing pathway on or off (Berinyuy et al., 23 Jan 2025).

Kerr and Barnett mechanisms act primarily through direction-sensitive magnon detunings. In cavity-magnon optomechanics, reversing the magnetic-field direction flips the sign of the Kerr coefficient $A$3, and therefore the signs of both the magnon frequency shift $A$4 and the two-magnon coefficient $A$5 (Chen et al., 2023). In Barnett-effect magnomechanics, reversing the rotation changes $A$6, which changes the effective magnon detuning and breaks time-reversal symmetry of the drift matrix (Chabar et al., 13 Jul 2025). In squeezed-magnon proposals, the phase $A$7 controls $A$8 and $A$9, so that $n(t)$0 interchanges the effective detunings and dampings of the magnon quadratures (Imara et al., 9 Aug 2025). In spinning exciton-optomechanics, an embedded optical parametric amplifier contributes additional detuning and damping shifts through $n(t)$1 and $n(t)$2, thereby regulating the Sagnac-induced asymmetry (Lin et al., 2 Mar 2026).

A different class of mechanisms relies on asymmetric couplings rather than rotating frames. In a torus-shaped cavity magnomechanical system, chiral cavity-magnon coupling with $n(t)$3 means that driving one circulating mode produces a large $n(t)$4 and a strong enhanced magnomechanical rate $n(t)$5, while driving the opposite mode can leave the interaction negligible (Fan et al., 2024). In superconducting circuits, one can combine coherent and dissipative couplings so that their interference is directional; a fully nonreciprocal interaction is obtained when the qubit separation is $n(t)$6 and $n(t)$7 (Ren et al., 2024). In a distinct superconducting scheme, two lossy auxiliary resonators generate effective couplings

$n(t)$8

so that $n(t)$9 whenever the coherent and loss-induced phase differences do not satisfy reciprocal conditions (Ren et al., 12 May 2026).

Topological and plasmonic platforms implement reciprocity breaking at the Green-function level. In 3D photonic topological insulators and Weyl semimetals, a nonreciprocal or unidirectional surface plasmon polariton exists in the bandgap, often tied to a nonzero gap Chern number. Because only one propagation channel is available, the collective rates become intrinsically asymmetric, and entanglement inherits that asymmetry (Gangaraj et al., 2017, Ukhtary et al., 2022). This suggests a unifying interpretation: nonreciprocal entanglement appears whenever the asymmetric control acts directly on the drift matrix, on the effective non-Hermitian coupling, or on the photonic Green function that mediates the interaction.

4. Principal physical platforms

The most heavily developed family is magnonic and magnomechanical. In a spinning whispering-gallery cavity coupled to two YIG spheres, magnon Kerr nonlinearity and the Sagnac effect together generate steady nonreciprocal magnon–magnon entanglement. For experimentally realistic parameters $$V_{ij}=\frac12\langle u_i(\infty)u_j(\infty)+u_j(\infty)u_i(\infty)\rangle$$0, $$V_{ij}=\frac12\langle u_i(\infty)u_j(\infty)+u_j(\infty)u_i(\infty)\rangle$$1, $$V_{ij}=\frac12\langle u_i(\infty)u_j(\infty)+u_j(\infty)u_i(\infty)\rangle$$2, and $$V_{ij}=\frac12\langle u_i(\infty)u_j(\infty)+u_j(\infty)u_i(\infty)\rangle$$3, one finds $$V_{ij}=\frac12\langle u_i(\infty)u_j(\infty)+u_j(\infty)u_i(\infty)\rangle$$4; the entanglement grows for small $$V_{ij}=\frac12\langle u_i(\infty)u_j(\infty)+u_j(\infty)u_i(\infty)\rangle$$5, peaks near $$V_{ij}=\frac12\langle u_i(\infty)u_j(\infty)+u_j(\infty)u_i(\infty)\rangle$$6, and vanishes near $$V_{ij}=\frac12\langle u_i(\infty)u_j(\infty)+u_j(\infty)u_i(\infty)\rangle$$7, although one direction can tolerate $$V_{ij}=\frac12\langle u_i(\infty)u_j(\infty)+u_j(\infty)u_i(\infty)\rangle$$8 while the opposite direction loses entanglement around $$V_{ij}=\frac12\langle u_i(\infty)u_j(\infty)+u_j(\infty)u_i(\infty)\rangle$$9 (Xu et al., 14 May 2026). In Kerr-modified spinning cavity magnomechanics, photon–phonon entanglement can be preserved at $A\,V+V\,A^T + D = 0,$0 K, with the strongest enhancement occurring when $A\,V+V\,A^T + D = 0,$1 (Chen et al., 2023). In cavity-magnon optomechanics using Kerr control alone, bipartite and tripartite entanglements can be switched between reciprocal and ideal one-way regimes by tuning the magnon detuning, the two-magnon coefficient, and the bath temperature (Chen et al., 2023). Multipartite extensions in two-cavity magnomechanics explicitly report peak bipartite negativities such as $A\,V+V\,A^T + D = 0,$2 for $A\,V+V\,A^T + D = 0,$3 at $A\,V+V\,A^T + D = 0,$4 versus $A\,V+V\,A^T + D = 0,$5 at $A\,V+V\,A^T + D = 0,$6, along with a tripartite contrast measure $A\,V+V\,A^T + D = 0,$7 (Ahmed et al., 2024).

Barnett-controlled variants introduce additional tunability. In a hybrid magnomechanical–optomechanical system with coherent feedback, increasing the beam-splitter reflectivity $A\,V+V\,A^T + D = 0,$8 reduces the effective optical damping and boosts both entanglement and steering, while the Barnett shift $A\,V+V\,A^T + D = 0,$9 magnifies directional asymmetry; under optimized conditions entanglement can persist up to $D$0–$D$1 (Chabar et al., 13 Jul 2025). In Barnett magnomechanics with an optical parametric amplifier, ideal one-way operation is reported for $D$2, $D$3, $D$4, $D$5, and $D$6, with $D$7–$D$8 and $D$9 (Chabar et al., 29 Sep 2025). Squeezed-magnon proposals attain ideal nonreciprocity by phase-controlled asymmetric magnon detunings and dampings, with $4\times4$0 in optimized windows (Imara et al., 9 Aug 2025).

Spinning optomechanical systems realize the same idea outside magnonics. In cascaded spinning optomechanical resonators, remote entanglement between two mechanical oscillators can be generated only from one input direction, and for frequency-mismatched oscillators the revival factor

$4\times4$1

remains $4\times4$2 even for $4\times4$3 (Jiao et al., 2022). In spinning molecular optomechanics, positive photon–vibration and vibration–vibration entanglement persists even at $4\times4$4 K in the optimized CCW/blue-sideband configuration, and the vibration–vibration channel benefits from increasing the number of molecules, with a threshold $4\times4$5 for entering the stable, entangled regime (Berinyuy et al., 23 Jan 2025). In a spinning exciton-optomechanical system with an OPA, the OPA enhances photon–exciton and tripartite entanglement but weakens photon–phonon and exciton–phonon entanglement; the photon–exciton channel remains appreciable up to $4\times4$6 K and persists to $4\times4$7 K for $4\times4$8 and $4\times4$9 K for $V^{(\mu\nu)}$0 (Lin et al., 2 Mar 2026). In an asymmetric Fabry–Perot cavity, a stable working point at $V^{(\mu\nu)}$1, $V^{(\mu\nu)}$2, $V^{(\mu\nu)}$3, $V^{(\mu\nu)}$4, $V^{(\mu\nu)}$5, and $V^{(\mu\nu)}$6 gives $V^{(\mu\nu)}$7, $V^{(\mu\nu)}$8, and $V^{(\mu\nu)}$9, with forward entanglement surviving up to $\eta^-$0 K (Wu et al., 5 Jun 2026).

Discrete-variable nonreciprocal entanglement is now established in qubit and emitter systems as well. In a biased photonic topological insulator, a unidirectional surface plasmon polariton gives defect-immune entanglement dynamics, while in a Weyl semimetal the nonreciprocal SPP leads to the analytic concurrence $\eta^-$1, which peaks at $\eta^-$2 (Gangaraj et al., 2017, Ukhtary et al., 2022). DC-current-biased graphene plasmons increase the maximum concurrence from $\eta^-$3 to $\eta^-$4 as the drift reaches $\eta^-$5, and maintain $\eta^-$6 out to $\eta^-$7 (Berres et al., 2022). In superconducting-qubit platforms, one scheme stabilizes a nonreciprocal steady-state entangled state by driving the appropriate qubit (Ren et al., 2024), while another uses interference between two lossy resonator-mediated paths to realize a transient concurrence up to $\eta^-$8 for one initial excitation and $\eta^-$9 for the reversed one (Ren et al., 12 May 2026).

5. Performance, robustness, and metrics of ideality

A standard way to quantify directionality is the bidirectional contrast ratio. In spinning molecular optomechanics,

$\dot u(t)=A\,u(t)+n(t),$00

with $\dot u(t)=A\,u(t)+n(t),$01 corresponding to ideal nonreciprocity (Berinyuy et al., 23 Jan 2025). Closely related definitions are used for Kerr-controlled entanglement,

$\dot u(t)=A\,u(t)+n(t),$02

for microwave–optical optomagnomechanics,

$\dot u(t)=A\,u(t)+n(t),$03

and for Barnett-shift control,

$\dot u(t)=A\,u(t)+n(t),$04

In all these formulations, the unit value marks one-way entanglement (Chen et al., 2023, Liu et al., 2024, Chabar et al., 29 Sep 2025).

Thermal robustness varies strongly by platform. Magnon–magnon entanglement in spinning cavity-magnon systems survives to about $\dot u(t)=A\,u(t)+n(t),$05 under realistic parameters (Xu et al., 14 May 2026). Kerr-modified spinning cavity magnomechanics reaches $\dot u(t)=A\,u(t)+n(t),$06 K for photon–phonon entanglement (Chen et al., 2023). Barnett-plus-feedback magnomechanics remains entangled up to $\dot u(t)=A\,u(t)+n(t),$07–$\dot u(t)=A\,u(t)+n(t),$08 (Chabar et al., 13 Jul 2025). Asymmetric Fabry–Perot cavities extend forward optomechanical entanglement to $\dot u(t)=A\,u(t)+n(t),$09 K (Wu et al., 5 Jun 2026). Spinning exciton-optomechanics with an OPA pushes photon–exciton nonreciprocal entanglement to room temperature and beyond (Lin et al., 2 Mar 2026), while spinning molecular optomechanics exhibits positive entanglement even at $\dot u(t)=A\,u(t)+n(t),$10 K (Berinyuy et al., 23 Jan 2025). The temperature scale is therefore not intrinsic to nonreciprocity itself; it is set by the specific entangling channel and by whether the asymmetry enhances a noise-resilient two-mode-squeezing process.

The literature also reports that nonreciprocity and robustness do not always move together in a trivial way. In cavity-magnon optomechanics, increasing the environmental temperature can under some conditions enhance the contrast ratio until the entanglement itself is suppressed at very high $\dot u(t)=A\,u(t)+n(t),$11 (Chen et al., 2023). In spinning exciton-optomechanics, the OPA improves the thermal resilience of photon–exciton entanglement but worsens the temperature dependence of phonon-involving bipartite and tripartite entanglement (Lin et al., 2 Mar 2026). In the asymmetric Fabry–Perot system, large classical nonreciprocity does not guarantee large quantum nonreciprocity (Wu et al., 5 Jun 2026). These results rule out a single monotonic robustness principle.

6. Functions, related phenomena, and unresolved issues

The proposed functions of nonreciprocal entanglement are consistently operational. Magnon-based spinning cavities are presented as building blocks for nonreciprocal quantum routers, isolators, or circulators in hybrid magnonic networks (Xu et al., 14 May 2026). Kerr-controlled cavity-magnon optomechanics points toward routers, circulators, and entanglement-diodes (Chen et al., 2023). Chiral cavity magnomechanics explicitly implements channel multiplexing quantum teleportation from a microwave field to a solid-state magnon mode, with a reported coherent-state teleportation fidelity $\dot u(t)=A\,u(t)+n(t),$12 (Fan et al., 2024). Loss-engineered superconducting circuits realize an “entanglement diode” for modular networks (Ren et al., 12 May 2026), while Barnett-plus-OPA magnomechanics is proposed as a route to nonreciprocal single-phonon devices (Chabar et al., 29 Sep 2025). Spinning optomechanics connects directional entanglement to backaction-immune quantum sensors and noise-tolerant quantum processors (Jiao et al., 2020).

Nonreciprocal entanglement is also tightly connected to nearby notions of directional quantum correlation. One magnomechanical–optomechanical study simultaneously demonstrates nonreciprocal entanglement, one-way steering, two-way steering, and multipartite entanglement, with coherent feedback enhancing both steering and entanglement (Chabar et al., 13 Jul 2025). In the exciton-optomechanical setting, the OPA enhances photon–exciton bipartite entanglement and tripartite entanglement while suppressing photon–phonon and exciton–phonon channels (Lin et al., 2 Mar 2026). Multipartite contrast ratios have likewise been formulated for two-cavity magnomechanics, where the tripartite contrast is defined directly from the residual contangle (Ahmed et al., 2024). Nonreciprocal entanglement is therefore not an isolated resource but part of a broader directional hierarchy of Gaussian and non-Gaussian correlations.

Several unresolved issues are already identified in the literature. In Barnett-effect magnomechanics with an OPA, stated open questions include full quantum modeling of mechanical decoherence at higher $\dot u(t)=A\,u(t)+n(t),$13, the impact of Kerr and multi-magnon processes on one-way operation, extension to multipartite nonreciprocal entanglement, and integration with traveling-wave microwave guides for on-chip routing (Chabar et al., 29 Sep 2025). A plausible implication is that the next stage of the subject will be comparative rather than merely platform-specific: which asymmetry source—rotation, nonlinearity, topology, dissipation, or engineered interference—most efficiently converts a given coupling graph into a robust directional entangling resource.