Entanglement Harvesting in Quantum Fields

• Entanglement harvesting is a process where localized quantum detectors extract vacuum correlations to generate entanglement, even when they are spacelike separated.
• The protocol employs Unruh–DeWitt detector models with smooth Gaussian switching functions to optimize the interplay between local excitation noise and nonlocal entangling terms.
• Extensions to curved spacetimes and multiparty setups reveal that detector geometry, switching details, and background topology crucially influence the efficiency and feasibility of entanglement extraction.

Searching arXiv for recent and foundational papers on entanglement harvesting to support the encyclopedia entry. Entanglement harvesting is the process in which localized quantum systems, usually modeled as particle detectors, become entangled by locally coupling to a quantum field, even when the detectors remain spacelike separated during the interaction. The detectors start uncorrelated, the field is typically in its vacuum, and after a finite-time local interaction one traces out the field and finds entanglement in the detector sector. In this sense, vacuum correlations of quantum fields are operationally accessed and converted into usable entanglement. In relativistic quantum information, the standard model is the Unruh–DeWitt detector, which idealizes an atom or qubit with two energy levels linearly coupled to a scalar field (Mendez-Avalos et al., 2022).

1. Conceptual framework and detector models

The standard harvesting protocol begins with two or more localized detectors prepared in product ground states and a quantum field in the vacuum. Each detector couples locally to the field for a finite time through a switching function, after which the field is traced out. The resulting reduced detector state can contain bipartite or multipartite entanglement, depending on the number of detectors and the geometry of the interaction region (Mendez-Avalos et al., 2022).

In the usual Unruh–DeWitt setting, a detector is a two-level system with ground state $\ket{0}$, excited state $\ket{1}$, and energy gap $\Omega$, coupled locally and linearly to a scalar field. For a detector $D$, the monopole moment is

$$\hat \mu_D(\tau_D) = e^{i\Omega_D \tau_D}\ket{1}_D\bra{0}_D + e^{-i\Omega_D \tau_D}\ket{0}_D\bra{1}_D,$$

and the interaction is modulated by a switching function $\chi_D$ along the detector worldline $x_D(\tau_D)$ (Mendez-Avalos et al., 2022). A common choice is Gaussian switching,

$\chi(\tau)=e^{-\tau^2/(2\sigma^2)},$

with $\sigma$ the interaction timescale. Smooth switching is described as adiabatic in the three-detector analysis and reduces UV artifacts relative to sharp or instantaneous switching (Mendez-Avalos et al., 2022).

The reduced detector density matrix is determined by the field’s Wightman function $W(x,x')=\langle 0|\phi(x)\phi(x')|0\rangle$. In the two-detector case, the key quantities are local excitation probabilities $\ket{1}$0, a coherence term $\ket{1}$1, and a nonlocal term $\ket{1}$2; in the three-detector case these generalize to $\ket{1}$3, $\ket{1}$4, and $\ket{1}$5 for all detector pairs (Henderson et al., 2017, Mendez-Avalos et al., 2022). The term $\ket{1}$6 or $\ket{1}$7 is the quantity most directly tied to harvested entanglement, since it competes against local noise from $\ket{1}$8.

Several extensions of the basic detector model appear in the literature. One line replaces classical detector trajectories by quantum-delocalized center-of-mass degrees of freedom, yielding a generalized Unruh-deWitt detector model for matter systems such as atoms, ions, or molecules; in that setting, coherent spreading of the center-of-mass wave function suppresses harvesting (Stritzelberger et al., 2020). Another line replaces scalar-field harvesting by harvesting from the quantum gravitational field itself, deriving a detector-like coupling to the curvature component $\ket{1}$9 in linearized quantum gravity and showing that spacelike separated probes can harvest entanglement from graviton vacuum fluctuations (Perche et al., 2022). A further generalization couples pointlike Unruh-DeWitt detectors not to a free field but to scalar primary operators in general $\Omega$0-dimensional conformal field theories, extending harvesting beyond free fields to interacting conformal theories and arbitrary spatial dimensions (Wurtz et al., 6 Feb 2026).

2. Reduced states, entanglement measures, and computational methods

To leading nontrivial order in perturbation theory, the reduced detector state typically takes an $\Omega$1-state form. For two detectors, in the basis $\Omega$2, the state is

$\Omega$3

with $\Omega$4 the local transition probabilities and $\Omega$5 the nonlocal entangling term (Henderson et al., 2017). In the three-detector case, the reduced state has a structured “X+coherence” form in the computational basis $\Omega$6, with entries determined by $\Omega$7, $\Omega$8, and $\Omega$9 (Mendez-Avalos et al., 2022).

For two-qubit states, concurrence and negativity are the standard entanglement measures used in the supplied literature. In the BTZ black-hole analysis, concurrence is

$D$0

so harvesting occurs precisely when $D$1 (Henderson et al., 2017). In other work, negativity is used instead; for the two-detector state in a superposed spacetime, negativity and concurrence reduce to

$D$2

for identical detectors (Chakraborty et al., 2024). For three detectors, negativity is used both for one-vs-rest bipartitions such as $D$3 and for pairwise reductions such as $D$4, and these negativities are then assembled into the $D$5-tangle,

$D$6

with

$D$7

and cyclic permutations (Mendez-Avalos et al., 2022).

For mixed three-qubit detector states, the $D$8-tangle is explicitly treated as a lower bound to genuine tripartite entanglement rather than as the full convex-roof Coffman–Kundu–Wootters residual (Mendez-Avalos et al., 2022). This distinction matters because detector states are mixed after tracing out the field. The same mixed-state issue also motivates alternative operator-algebraic formulations: a modular-operator treatment constructs modular conjugation operators for Unruh-DeWitt type qubits interacting with a scalar field, and the harvested entanglement is directly quantified by an expectation value involving the modular conjugation operator $D$9 (Chatterjee, 17 Aug 2025).

A major technical development concerns the evaluation of the entangling term $$\hat \mu_D(\tau_D) = e^{i\Omega_D \tau_D}\ket{1}_D\bra{0}_D + e^{-i\Omega_D \tau_D}\ket{0}_D\bra{1}_D,$$0 or $$\hat \mu_D(\tau_D) = e^{i\Omega_D \tau_D}\ket{1}_D\bra{0}_D + e^{-i\Omega_D \tau_D}\ket{0}_D\bra{1}_D,$$1 when the Wightman function is only known through a mode expansion. By exploiting the conjugate symmetry of the Wightman function, one may split the entangling integral into parts dependent on the commutator and anti-commutator of the field. In cases where the commutator vanishes, such as spacelike separation or timelike separation if the strong Huygens principle holds, the entangling term can be expressed in terms of the simpler $$\hat \mu_D(\tau_D) = e^{i\Omega_D \tau_D}\ket{1}_D\bra{0}_D + e^{-i\Omega_D \tau_D}\ket{0}_D\bra{1}_D,$$2 correlators. For the vacuum state, this can be translated into a simple Fourier transform and thus a single sum over modes, simplifying the procurement of closed expressions (Ng et al., 2018). This suggests that non-signalling and signalling contributions to harvested entanglement can be cleanly separated in favorable regimes.

3. Geometry, switching, and kinematic control of harvesting

The amount of harvested entanglement depends strongly on detector separation, interaction duration, energy gap, and global or local spacetime structure. In the simplest flat-space setting with smooth finite-duration switching, harvesting is strongest when nonlocal correlations encoded in $$\hat \mu_D(\tau_D) = e^{i\Omega_D \tau_D}\ket{1}_D\bra{0}_D + e^{-i\Omega_D \tau_D}\ket{0}_D\bra{1}_D,$$3 overcome local excitation noise $$\hat \mu_D(\tau_D) = e^{i\Omega_D \tau_D}\ket{1}_D\bra{0}_D + e^{-i\Omega_D \tau_D}\ket{0}_D\bra{1}_D,$$4, and it typically decreases with growing separation and increasing detector transition energy (Barman et al., 2023).

Gaussian switching is repeatedly used across the supplied works. In the three-detector Minkowski study, adiabatic Gaussian switching is contrasted with instantaneous single-kick models; smooth switching allows entanglement harvesting even for spacelike separations and reduces UV artifacts (Mendez-Avalos et al., 2022). In curved backgrounds, Gaussian switching continues to set the relevant interaction time scale $$\hat \mu_D(\tau_D) = e^{i\Omega_D \tau_D}\ket{1}_D\bra{0}_D + e^{-i\Omega_D \tau_D}\ket{0}_D\bra{1}_D,$$5, and many results are expressed in terms of ratios such as $$\hat \mu_D(\tau_D) = e^{i\Omega_D \tau_D}\ket{1}_D\bra{0}_D + e^{-i\Omega_D \tau_D}\ket{0}_D\bra{1}_D,$$6, $$\hat \mu_D(\tau_D) = e^{i\Omega_D \tau_D}\ket{1}_D\bra{0}_D + e^{-i\Omega_D \tau_D}\ket{0}_D\bra{1}_D,$$7, $$\hat \mu_D(\tau_D) = e^{i\Omega_D \tau_D}\ket{1}_D\bra{0}_D + e^{-i\Omega_D \tau_D}\ket{0}_D\bra{1}_D,$$8, and $$\hat \mu_D(\tau_D) = e^{i\Omega_D \tau_D}\ket{1}_D\bra{0}_D + e^{-i\Omega_D \tau_D}\ket{0}_D\bra{1}_D,$$9 (Henderson et al., 2017, Liu et al., 2021, Barman et al., 2023).

Geometry can alter harvesting qualitatively. In the presence of a perfectly reflecting plane boundary, the boundary generally degrades harvested entanglement when detectors are very close to the boundary. However, when the distance between detectors and the boundary becomes comparable to the interaction duration parameter, the amount of the harvested entanglement approaches a peak that can exceed the value without a boundary. At the same time, the parameter space of detector separation and acceleration allowing entanglement harvesting is enlarged by the boundary (Liu et al., 2021). This supports the description of the boundary as playing a double-edged role: degrading harvested entanglement in general while enlarging the entanglement-harvesting-achievable parameter space (Liu et al., 2021).

Topological defects can also reshape the vacuum correlations that detectors probe. In cosmic string spacetime, which is locally flat but globally conical, the presence of the string can either aid or hinder harvesting for detectors on the same side of the string, depending on the distance between the detectors and the string and on the alignment. For detectors on opposite sides of the string, more entanglement is harvested than in flat spacetime without a string, and the harvesting-achievable range is enlarged in that configuration (Ji et al., 2024). This contrasts with the reflecting-boundary case, where the boundary invariably extends the harvesting-achievable range (Ji et al., 2024).

The nature of the background can matter even when it is not classical. In a quantum superposition of two quotient Minkowski spaces with different compactification lengths, the superposed nature of spacetime induces interference effects that can significantly enhance entanglement for both twisted and untwisted fields. The concurrence reaches its maximum when the final spacetime superposition state matches the initial spacetime state, and for twisted fields the no-entanglement region differs significantly from that in classical Minkowski space or a single quotient Minkowski space (Chakraborty et al., 2024). This suggests that entanglement harvesting can serve as an operational probe of quantum superpositions of spacetime topologies.

Background dynamics also matter. In linearized gravitational-wave burst backgrounds, harvesting is possible for Gaussian, sech-squared, Heaviside step, and tanh burst profiles. In all cases it decreases with increasing detector distance and transition energy. The presence or absence of gravitational-wave memory changes the low-gap behavior qualitatively: with memory, harvesting keeps increasing with decreasing transition energy, while without memory it tends to reach finite values (Barman et al., 2023).

4. Curved spacetime, horizons, and topological sectors

Black-hole backgrounds provide some of the sharpest demonstrations that harvesting is sensitive to spacetime geometry. In the vicinity of a BTZ black hole, two static detectors outside the horizon experience strong inhibition of entanglement harvesting as they approach the horizon, even when their proper separation is held fixed. The harvested entanglement rapidly falls to zero near the horizon because gravitational redshift suppresses the nonlocal term $\chi_D$0, while Hawking radiation increases the local noise $\chi_D$1 through the local temperature

$\chi_D$2

This leads to a sudden-death region near the horizon where no entanglement can be harvested (Henderson et al., 2017).

Rotation qualitatively modifies this picture. In the rotating BTZ background, concurrence between co-rotating detectors is amplified relative to large-distance values, most strongly for near-extremal small mass black holes. Rotation can allow harvesting at large spacelike detector separations and reduces the size of the entanglement shadow, the region near the black hole from which entanglement cannot be extracted (Robbins et al., 2020). This amplification is tied to how the rotating BTZ Wightman function reorganizes the balance between the nonlocal term $\chi_D$3 and the local noise $\chi_D$4 (Robbins et al., 2020).

Hidden topology behind horizons can also be visible to harvesting protocols. In the comparison between the BTZ black hole and its $\chi_D$5 geon counterpart, the classically hidden topology of the geon changes the amount of entanglement harvested from the exterior vacuum for sufficiently small mass. Small-gap detectors harvest more entanglement in BTZ, but as the energy gap increases, more entanglement is harvested in the geon spacetime. The crossover depends on black-hole mass, and the entanglement shadow is larger in the geon spacetime for small gaps but larger in BTZ for large gaps (Henderson et al., 2022). A plausible implication is that harvesting protocols can probe topological sectors that are classically hidden from external observers.

Cosmological or horizon-like backgrounds are not the only curved cases of interest. The developer-supplied material also notes that black-hole inhibition is argued to be generic to horizons because it combines Hawking radiation with redshift suppression of nonlocal correlations (Henderson et al., 2017). This suggests that harvesting near horizons can diagnose the competition between nonlocal field correlations and local thermalization.

More generally, harvesting has now been extended from free scalar fields in fixed backgrounds to the quantum gravitational field itself. In linearized quantum gravity, two spacelike separated probes can harvest entanglement from the gravitational vacuum, with detector couplings expressed in terms of the curvature component $\chi_D$6. The resulting negativity obeys the familiar structure $\chi_D$7, but the spin-2 nature of gravity imposes strong angular-momentum selection effects and the magnitudes are extremely small for atomic probes (Perche et al., 2022).

5. Multipartite, pure-state, and nonstandard harvesting regimes

The two-detector protocol is only the simplest instance of harvesting. With three Unruh-DeWitt detectors adiabatically interacting with a massless scalar field in Minkowski space, genuine tripartite entanglement can be harvested. The study of linear, equilateral triangular, and scalene triangular configurations shows that more entanglement can be extracted in the linear configuration than in the equilateral one under the same parameters (Mendez-Avalos et al., 2022). The geometry determines the pattern of pairwise distances $\chi_D$8, and therefore the arguments of the Wightman integrals controlling $\chi_D$9 and $x_D(\tau_D)$0.

A central multipartite result is that no bipartite entanglement is required to harvest tripartite entanglement. There are parameter regimes in which pairwise negativities involving one detector vanish while the $x_D(\tau_D)$1-tangle remains positive because one-vs-rest negativities such as $x_D(\tau_D)$2 and $x_D(\tau_D)$3 remain nonzero (Mendez-Avalos et al., 2022). The same analysis shows that tripartite entanglement can be harvested at larger spacelike separations than those allowing bipartite entanglement, indicating that the vacuum correlations being operationally accessed are genuinely multipartite rather than merely pairwise (Mendez-Avalos et al., 2022).

At small detector separations in the three-detector problem, bipartite correlations can become larger than tripartite ones, leading to an apparent violation of the Coffman-Kundu-Wootters inequality when the $x_D(\tau_D)$4-tangle is computed directly from mixed-state negativities. The analysis shows that this is not a perturbative artifact but instead arises because the detector state is mixed after tracing out the field; the convex-roof quantity appearing in rigorous mixed-state monogamy inequalities is larger than the directly computed negativity (Mendez-Avalos et al., 2022). Thus, the observed negative $x_D(\tau_D)$5-tangle is interpreted not as a breakdown of monogamy but as a limitation of using mixed-state negativity in a pure-state monogamy formula.

Another nonstandard regime is pure-state harvesting. Standard UDW harvesting protocols typically extract tiny and fragile mixed-state entanglement because the detectors remain entangled with the field. A different protocol constructs partner modes $x_D(\tau_D)$6 within a discretized scalar field vacuum and then swaps their entanglement to external devices $x_D(\tau_D)$7 through bilinear couplings,

$x_D(\tau_D)$8

This protocol extracts pure entanglement rather than mixed detector-field entanglement and can, in principle, extract an arbitrarily large amount of entanglement even with a finite UV cutoff and a finite number of field modes (Trevison et al., 2018). The mechanism relies on constructing spatially overlapped partner modes and tuning a parameter $x_D(\tau_D)$9 so that the entanglement entropy

$\chi(\tau)=e^{-\tau^2/(2\sigma^2)},$0

can become arbitrarily large (Trevison et al., 2018).

This pure-state protocol comes with an explicit energy–entanglement trade-off. In the three-oscillator example analyzed there, the harvested entanglement diverges logarithmically as $\chi(\tau)=e^{-\tau^2/(2\sigma^2)},$1, while the energy cost of the swap diverges as $\chi(\tau)=e^{-\tau^2/(2\sigma^2)},$2 (Trevison et al., 2018). Infinite entanglement extraction therefore requires infinite energy, even though the underlying field system has a finite UV cutoff (Trevison et al., 2018).

6. Limitations, experimental relevance, and current directions

Several limitations recur throughout the literature. Most explicit calculations are perturbative in a small detector-field coupling, often keeping only $\chi(\tau)=e^{-\tau^2/(2\sigma^2)},$3 terms in the reduced density matrix and entanglement monotones (Mendez-Avalos et al., 2022, Henderson et al., 2017, Chakraborty et al., 2024). Smooth switching helps control UV behavior, but pointlike detectors can still require care, particularly in topological backgrounds or at coincident-image loci, where spatial smearing would regulate otherwise singular behavior (Mendez-Avalos et al., 2022, Ji et al., 2024).

The mixed-state nature of ordinary harvesting is both a limitation and a source of structure. Because detectors remain entangled with the field, extracted entanglement is usually small and fragile (Trevison et al., 2018). This has motivated both pure-state harvesting protocols based on mode engineering (Trevison et al., 2018) and operator-algebraic reformulations in terms of modular reflection and curvature of entanglement (Chatterjee, 17 Aug 2025).

The physical realizability of harvesting depends strongly on the detector model. For coherently delocalized matter systems such as atoms, ions, or molecules, harvesting is Gaussian suppressed in the initial center-of-mass delocalization of the detectors, and harvesting in media is significantly suppressed when the wave propagation speed is much smaller than the vacuum speed of light (Stritzelberger et al., 2020). The paper further concludes that spatial smearing profiles used to model finite atomic size are not suited to model center-of-mass delocalization (Stritzelberger et al., 2020). This indicates that realistic matter-wave implementations must treat motional and internal degrees of freedom separately.

In gravitational settings, harvesting has been proposed as an operational witness of quantum gravity. Harvesting from the gravitational vacuum in linearized quantum gravity produces extremely small negativities for atomic probes, but the coupling scales with detector mass, suggesting that more massive mesoscopic systems may be more favorable in the future (Perche et al., 2022). This suggests that harvesting could eventually complement gravity-mediated entanglement proposals by probing entanglement already present in the gravitational vacuum itself.

At the level of formal theory, the modular-operator approach introduces a notion of curvature of entanglement, defined as the second derivative of an entanglement monotone viewed as a functional of $\chi(\tau)=e^{-\tau^2/(2\sigma^2)},$4 and $\chi(\tau)=e^{-\tau^2/(2\sigma^2)},$5. At modular self-duality points in a Markovian two-qubit model, the curvature of entanglement coincides in magnitude with the quantum Fisher information (Chatterjee, 17 Aug 2025). This suggests a connection between harvesting, modular structure, and information geometry.

Recent conformal-field-theory work extends harvesting to general interacting CFTs in arbitrary spatial dimensions, with pointlike Unruh-DeWitt detectors coupled to scalar primary operators. Increasing the operator scaling dimension suppresses both negativity and mutual information, reflecting the faster decay of correlations, and in holographic CFTs bulk effective field theory enables a separation between field-harvested and communication-mediated entanglement (Wurtz et al., 6 Feb 2026). This suggests that harvesting may become a useful operational probe not only of spacetime geometry but also of operator content and large-$\chi(\tau)=e^{-\tau^2/(2\sigma^2)},$6 structure in strongly coupled quantum field theories.