Gravitationally Mediated Entanglement

• Gravitationally mediated entanglement is the generation of nonseparable quantum correlations between masses via gravitational interaction alone, evidenced by branch-dependent phase differences.
• The protocol exploits spatial superposition of masses and precise phase measurements to differentiate quantum gravitational effects from classical gravitational channels.
• Entanglement witnesses and interference tests certify nonclassical gravitational effects while highlighting stringent experimental constraints on noise and decoherence.

Searching arXiv for recent and foundational work on gravitationally mediated entanglement. Gravitationally mediated entanglement is the generation of nonseparable correlations between two quantum systems through gravitational interaction alone. In the canonical tabletop scenario, two neutral masses are each prepared in a spatial superposition, interact only via the Newtonian potential, and acquire branch-dependent phases such as $\delta\phi = G m^2 t/(\hbar d)$; after recombination, the resulting state is entangled if the relative phase is nontrivial (Marletto et al., 2017). The subject occupies a central place in contemporary quantum-gravity phenomenology because the appearance of entanglement is incompatible with a mediator that is merely a classical communication channel, yet the precise inference from an observed signal to “quantum gravity” depends on how the mediator and its dynamics are modelled (Christodoulou et al., 2022).

1. Information-theoretic criterion and the classical-mediator no-go

The modern discussion was crystallized by the claim that any system capable of mediating entanglement between two quantum systems must itself admit non-commuting observables (Marletto et al., 2017). In the gravity setting, the standard quantum-information statement invoked is that Local Operations and Classical Communication cannot generate entanglement. In the formulation recalled for the gravitational case, if Alice’s and Bob’s only communication is through the gravitational field, then a strictly classical gravitational channel cannot create entanglement between their masses (Christodoulou et al., 2022).

Subsequent work has made this claim mathematically broader. A classical mediator has been modelled as a commutative unital C*-algebra, encompassing continuous and infinite-dimensional classical systems as well as the finite classical mediators used in earlier arguments. In that framework, if the joint evolution consists of any sequence of local channels acting only on $\mathcal A\otimes\mathcal G$ or $\mathcal G\otimes\mathcal B$, then the reduced state on $\mathcal A\otimes\mathcal B$ remains separable after any finite or countably infinite sequence of rounds; accordingly, any entanglement measure that vanishes on separable states remains zero throughout (Ludescher et al., 17 Jul 2025).

A distinct classical-gravity argument has been formulated in Newton–Cartan terms. There, a purely classical Newtonian-limit gravitational field leads to an interaction Hamiltonian that splits into a sum of one-body operators, so the total time-evolution operator factorizes as $U_1(t)\otimes U_2(t)$ and no off-diagonal entangling terms are created. On this view, if a screened gravitationally induced entanglement experiment were nevertheless to show entanglement, then either gravity was not a purely classical mediator, or some additional interaction supplied the effect (Schneider et al., 24 Nov 2025).

2. Canonical two-mass protocol and Newtonian phase accumulation

The standard protocol uses two test masses in spatial superposition. In one spatial dimension, the Newtonian two-body Hamiltonian is

$$H=\frac{p_1^2}{2m_1}+\frac{p_2^2}{2m_2}-\frac{Gm_1m_2}{|x_1-x_2|},$$

with the four interferometric branches labelled $\ket{LL},\ket{LR},\ket{RL},\ket{RR}$ (Marchese et al., 2024). For identical masses, one usually takes $r_{LL}=r_{RR}=d$, $r_{LR}=d+\Delta x$, and $r_{RL}=d-\Delta x$, so that the branch phases are set by $\mathcal A\otimes\mathcal G$0 and the evolved two-mass state can be written, up to an overall phase, as

$\mathcal A\otimes\mathcal G$1

which is entangled whenever it is not a product state (Marchese et al., 2024).

For this pure two-qubit state, the concurrence is

$\mathcal A\otimes\mathcal G$2

This expression makes explicit that entanglement is controlled by branch-dependent Newtonian energy differences rather than by the absolute interaction energy (Marchese et al., 2024). In the simpler branch-phase discussion often used in the original proposals, the essential quantity is the phase accumulated in the branch with the smallest separation,

$\mathcal A\otimes\mathcal G$3

which already captures why recombination of the branches can produce a nonclassical joint state (Christodoulou et al., 2022).

A closely related implementation maps the orbital phase onto internal spins. In the Stern–Gerlach proposal, each mass carries an embedded spin and the hold stage produces a spin-only state

$\mathcal A\otimes\mathcal G$4

so entanglement can be certified by spin correlations rather than direct orbital tomography (Bose et al., 2017).

3. Mediator degrees of freedom, relational observables, and geometric interpretation

One interpretive line emphasizes that the mediator need not be thought of as a collection of independent field modes attached to a preferred background split. In a relational or “quantum reference” frame, one can choose coordinates so that one mass is always at $\mathcal A\otimes\mathcal G$5 and the other at $\mathcal A\otimes\mathcal G$6; the only geometric variable left is the relational distance

$\mathcal A\otimes\mathcal G$7

In this description, the total state is written as a superposition over the four branches with a metric state $\mathcal A\otimes\mathcal G$8, showing that only the single scalar degree of freedom $\mathcal A\otimes\mathcal G$9 of the metric is entangled with the two masses. The distinction between “longitudinal” and “transverse” gravitational degrees of freedom is then a frame-dependent mathematical decomposition rather than a physically distinct choice of mediator (Christodoulou et al., 2022).

A path-integral treatment of a unified “path” and “oscillator” protocol reaches a closely related conclusion in linearised quantum gravity. Two masses prepared in superpositions of Gaussian center-of-mass states evolve according to an effective interaction obtained by integrating the gravitational field on shell, and the entanglement that appears is traced to a coherent sum over distinct classical field configurations $\mathcal G\otimes\mathcal B$0. In that formulation, both the discrete-path regime and the delocalized-oscillator regime are interpreted as manifestations of a quantum superposition of geometries rather than of a single classical spacetime (Bengyat et al., 2023).

A more radical reformulation goes further and removes spacetime distance from the primitive description altogether. In a conformally invariant twistor framework, the entangling object is a bilocal phase functional $\mathcal G\otimes\mathcal B$1 defined directly on twistor data, with no prior notion of metric distance. The familiar Newtonian phase $\mathcal G\otimes\mathcal B$2 arises only after conformal invariance is broken by introducing the infinity twistor, which selects a specific spacetime representation. This suggests that, in that framework, the entangling channel is primary and the Newtonian $\mathcal G\otimes\mathcal B$3 description is contingent on a particular geometric representation (Williams, 5 Feb 2026).

Within linearised quantum gravity, another field-theoretic analysis argues that there is no clear operational distinction between entanglement mediated by the Newtonian field and entanglement mediated by on-shell gravitons under the relevant protocols. In that sense, the observation of Newtonian-limit gravitational entanglement may be viewed as implying graviton entanglement as well (Danielson et al., 2021).

4. Interpretive disputes: LOCC, Newtonian dynamics, and what an observation would establish

The central controversy concerns not the abstract LOCC statement itself, but what counts as a classical gravitational model for the protocol. A recent analysis showed that, in the same two-mass interferometer setup, one can generate the same amount of entanglement using classical time evolution given by Newton’s laws of motion. Using Liouville evolution for the Wigner function with a quadratic fit or second-order Taylor expansion of the Newtonian potential, the reduced-state purity and concurrence match the Schrödinger prediction up to negligible corrections on the experimental timescale; for the quadratic fit, the purity coincides with the quantum result to within $\mathcal G\otimes\mathcal B$4 over $\mathcal G\otimes\mathcal B$5 (Marchese et al., 2024).

The implication drawn there is limited but sharp. Observation of gravity-mediated entanglement rules out local classical channels in which gravity is only measurement–feedback or a purely classical field in the LOCC sense, but entanglement generation alone does not certify that gravitons or $\mathcal G\otimes\mathcal B$6 terms in the Moyal expansion are operative. On that account, genuinely nonclassical dynamics would have to be established by probing effects proportional to $\mathcal G\otimes\mathcal B$7, such as deviations from Liouville evolution beyond a second-order potential fit (Marchese et al., 2024).

This dispute is closely tied to the use of effective Hamiltonians. A Newton–Cartan critique argues that the familiar “naive Newtonian” bipartite Hamiltonian

$\mathcal G\otimes\mathcal B$8

can indeed entangle the two masses, but only because it is a direct two-body coupling rather than a three-system model of “mass 1 + mass 2 + mediator.” In that reading, the confusion resides in interpreting applications of a Hamiltonian formalism rather than in the LOCC theorem itself (Schneider et al., 24 Nov 2025).

An experimentally motivated electromagnetic analogy sharpens the same issue. In a single-ion implementation of a gravity-mediated-entanglement analogue, the authors conclude that without considering the light-crossing time, the protocol does not distinguish a quantum-field-theoretic description from a quantum-controlled classical field. In their language, to rule out the latter one must demonstrate entanglement production on the scale $\mathcal G\otimes\mathcal B$9, rather than only in the effectively instantaneous regime $\mathcal A\otimes\mathcal B$0 (Bian et al., 2023).

5. Protocol families and generalizations

A large family of protocols now falls under the heading of gravitationally mediated entanglement. They differ mainly in which degree of freedom is superposed and in how the entangling phase is encoded.

Platform	Characteristic interaction or phase	Citation
Spatially superposed masses	$\mathcal A\otimes\mathcal B$1	(Marletto et al., 2017)
Path/oscillator unified protocol	Entanglement from superposition of geometries	(Bengyat et al., 2023)
Rotational-energy superpositions	$\mathcal A\otimes\mathcal B$2	(Higgins et al., 2024)
Light beams or photons	Branch-dependent photonic gravitational phase	(Aimet et al., 2022)
Constrained mechanical dynamics	Same short-time phase with $\mathcal A\otimes\mathcal B$3 corrections	(Williams, 1 May 2026)
BEC phonons	Effective pair-creation term $\mathcal A\otimes\mathcal B$4	(Sen et al., 22 Apr 2026)

The oscillator and path proposals are formally unified by allowing each particle to occupy a superposition of two Gaussian packets of width $\mathcal A\otimes\mathcal B$5 separated by $\mathcal A\otimes\mathcal B$6. In the limit $\mathcal A\otimes\mathcal B$7, one recovers the usual path-protocol scaling $\mathcal A\otimes\mathcal B$8; in the limit $\mathcal A\otimes\mathcal B$9, one obtains the delocalized-oscillator scaling

$U_1(t)\otimes U_2(t)$0

with the same conceptual conclusion in both regimes (Bengyat et al., 2023).

A genuinely relativistic proposal replaces positional superposition by a superposition of rotational energies. Two fixed rotors are put into qubit states $U_1(t)\otimes U_2(t)$1 and $U_1(t)\otimes U_2(t)$2, where $U_1(t)\otimes U_2(t)$3 carries additional rotational energy $U_1(t)\otimes U_2(t)$4 and therefore additional gravitational mass through mass–energy equivalence. The entangling term is

$U_1(t)\otimes U_2(t)$5

giving the phase

$U_1(t)\otimes U_2(t)$6

Because this interaction vanishes in the limit $U_1(t)\otimes U_2(t)$7, it isolates a relativistic effect absent from the usual Newtonian path protocol (Higgins et al., 2024).

Photonic variants bifurcate into laboratory and cosmological proposals. For two counter-propagating light pulses in adjacent interferometers, the entangling phase is estimated as

$U_1(t)\otimes U_2(t)$8

and laboratory-scale parameters imply photon numbers of order $U_1(t)\otimes U_2(t)$9 for $$H=\frac{p_1^2}{2m_1}+\frac{p_2^2}{2m_2}-\frac{Gm_1m_2}{|x_1-x_2|},$$0 (Aimet et al., 2022). In an FLRW background, an analogous two-photon calculation gives extremely small present-day phases, with concurrence and logarithmic negativity of order $$H=\frac{p_1^2}{2m_1}+\frac{p_2^2}{2m_2}-\frac{Gm_1m_2}{|x_1-x_2|},$$1 for typical cosmic-photon scenarios (Zhang, 26 Nov 2025).

Mechanical and many-body extensions have been proposed as well. Under constrained dynamics, short-time effectively inertial motion reproduces the free-fall entangling phase, with relative correction $$H=\frac{p_1^2}{2m_1}+\frac{p_2^2}{2m_2}-\frac{Gm_1m_2}{|x_1-x_2|},$$2 and visibility changes bounded by $$H=\frac{p_1^2}{2m_1}+\frac{p_2^2}{2m_2}-\frac{Gm_1m_2}{|x_1-x_2|},$$3 in the parameter regime analysed for carbon-nanotube pendula (Williams, 1 May 2026). In two Bose–Einstein condensates, linearized quantum gravity induces an effective interaction between phonon modes, yielding concurrence

$$H=\frac{p_1^2}{2m_1}+\frac{p_2^2}{2m_2}-\frac{Gm_1m_2}{|x_1-x_2|},$$4

which is reported to be substantially larger than in standard two-mass QGEM at very small separations, though with a faster fall-off as distance increases (Sen et al., 22 Apr 2026).

6. Entanglement witnesses, resource conversion, and experimental constraints

The original experimental proposals emphasized indirect certification through internal degrees of freedom. In the spin-embedded Stern–Gerlach protocol, a witness built from spin correlations satisfies $$H=\frac{p_1^2}{2m_1}+\frac{p_2^2}{2m_2}-\frac{Gm_1m_2}{|x_1-x_2|},$$5 for all separable states, while the proposed parameters $$H=\frac{p_1^2}{2m_1}+\frac{p_2^2}{2m_2}-\frac{Gm_1m_2}{|x_1-x_2|},$$6, $$H=\frac{p_1^2}{2m_1}+\frac{p_2^2}{2m_2}-\frac{Gm_1m_2}{|x_1-x_2|},$$7, $$H=\frac{p_1^2}{2m_1}+\frac{p_2^2}{2m_2}-\frac{Gm_1m_2}{|x_1-x_2|},$$8, and $$H=\frac{p_1^2}{2m_1}+\frac{p_2^2}{2m_2}-\frac{Gm_1m_2}{|x_1-x_2|},$$9 give $\ket{LL},\ket{LR},\ket{RL},\ket{RR}$0, or equivalently $\ket{LL},\ket{LR},\ket{RL},\ket{RR}$1, thereby certifying entanglement (Bose et al., 2017). A simpler conceptual version dispenses with explicit witness construction and states only that, once a nontrivial branch phase is acquired, one may reveal the entanglement through a Bell test on ancillary spins carried along the paths (Christodoulou et al., 2022).

A resource-theoretic reformulation interprets the same dynamics as coherent conversion of local coherence into bipartite entanglement. For equal superposition inputs, the reduced one-particle state has local coherence

$\ket{LL},\ket{LR},\ket{RL},\ket{RR}$2

while the bipartite negativity is

$\ket{LL},\ket{LR},\ket{RL},\ket{RR}$3

so that

$\ket{LL},\ket{LR},\ket{RL},\ket{RR}$4

In this formulation, initial local coherence is necessary for entanglement generation, and maximal entanglement requires initial maximal coherence (Dolatkhah et al., 10 Feb 2026).

The dominant obstacle to observation is not the formal entangling mechanism but the extreme sensitivity to non-gravitational phases and technical noise. In shielded setups, residual Casimir and magnetic-dipole interactions can imprint phases that overwhelm the gravitational one, and run-to-run positional or orientational fluctuations convert those phases into effective decoherence. The condition for retaining discernible entanglement is $\ket{LL},\ket{LR},\ket{RL},\ket{RR}$5, with tolerable fluctuations reported as $\ket{LL},\ket{LR},\ket{RL},\ket{RR}$6 and $\ket{LL},\ket{LR},\ket{RL},\ket{RR}$7 for silica, and $\ket{LL},\ket{LR},\ket{RL},\ket{RR}$8 and $\ket{LL},\ket{LR},\ket{RL},\ket{RR}$9 for superconducting lead (Bulling et al., 24 Apr 2026). The same analysis finds that shield vibrational modes can create persistent particle–shield correlations and can even mediate particle–particle entanglement that mimics a gravitational signal, making shield cooling and geometry optimization essential (Bulling et al., 24 Apr 2026).

Optomechanical amplification proposals illustrate the same tension between signal growth and noise. Modulating the optomechanical coupling changes the entangling phase from linear to cubic-in-time growth and can reduce the entanglement timescale from $r_{LL}=r_{RR}=d$0 to $r_{LL}=r_{RR}=d$1–$r_{LL}=r_{RR}=d$2 in the parameter sets analysed. However, the thermal-noise constraint remains unchanged in form: the fundamental bound is

$r_{LL}=r_{RR}=d$3

and optimistic parameter choices still require $r_{LL}=r_{RR}=d$4–$r_{LL}=r_{RR}=d$5, far beyond current technology (Plato et al., 2022).

Gravitationally mediated entanglement is therefore both a concrete experimental target and a conceptual probe of what it means for gravity to be quantum. Across its Newtonian, relativistic, field-theoretic, and algebraic formulations, the central issue remains the same: whether the observed nonlocal phase can be attributed to a mediator with genuinely noncommuting degrees of freedom, or whether the protocol has only constrained a narrower class of classical descriptions. The strength of the subject lies precisely in making that question experimentally sharp.