Unruh Thermalization in Accelerated Detectors

• Unruh Thermalization is the process in which an accelerated detector, weakly coupled to a quantum field, asymptotically relaxes to a Gibbs state at the Unruh temperature.
• The phenomenon is modeled using an open-system framework with a GKSL master equation, where the detector’s transition rates satisfy detailed balance via the KMS condition.
• Key factors such as detector-field coupling, spacetime dimensions, field mass, and switching effects determine the rate and universality of the thermalization process.

Searching arXiv for recent and foundational work on Unruh thermalization and closely related detector thermalization results. First, I’ll gather papers directly addressing Unruh detector thermalization, asymptotic Gibbs states, and related generalizations. Unruh thermalization is the late-time relaxation of the reduced state of an accelerated particle detector to a Gibbs state at the Unruh temperature $T_U=a/(2\pi)$ when the field state, restricted to the detector’s local time evolution, satisfies the KMS condition. In the standard setting, a uniformly accelerated Unruh–DeWitt detector is treated as an open quantum system weakly coupled to a quantum field initially in its vacuum state; the resulting reduced dynamics are governed by a GKSL master equation whose jump rates are determined by the detector response functions. The central result is that detailed balance fixes the asymptotic state to $\rho\propto e^{-H_D/T_U}$, while the approach to equilibrium depends on the detector–field coupling, spacetime dimension, field mass, smearing, switching, and the observable used to probe the field (Arrechea et al., 2021, Moustos, 2018, Perche, 2021).

1. Open-system formulation of accelerated detectors

A standard model is a two-level detector with energy gap $\omega$, states $|g\rangle,|e\rangle$, and interaction Hamiltonian

$H_I(\tau)=\lambda\,m(\tau)\,\phi^n[x(\tau)],$

with $\lambda\ll1$ and

$$m(\tau)= e^{+i\omega\tau}\sigma_+ + e^{-i\omega\tau}\sigma_- .$$

For $n>1$ one normal-orders $\phi^n$ to remove tadpole-type divergences. The reduced dynamics are obtained to $O(\lambda^2)$ in the Born–Markov approximation: at $\rho\propto e^{-H_D/T_U}$0 the field remains in its initial state, and coarse-graining in proper time on scales much larger than the field correlation time makes the detector evolution local in $\rho\propto e^{-H_D/T_U}$1 (Arrechea et al., 2021).

The population equations are

$\rho\propto e^{-H_D/T_U}$2

$\rho\propto e^{-H_D/T_U}$3

with

$\rho\propto e^{-H_D/T_U}$4

where $\rho\propto e^{-H_D/T_U}$5 is the pulled-back $\rho\propto e^{-H_D/T_U}$6-point Wightman function (Arrechea et al., 2021).

Equivalently, one writes the Schrödinger-picture master equation in Lindblad form,

$\rho\propto e^{-H_D/T_U}$7

with $\rho\propto e^{-H_D/T_U}$8, $\rho\propto e^{-H_D/T_U}$9 a Lamb-shift term, $\omega$0, and $\omega$1 (Arrechea et al., 2021). A closely related parametrization uses $\omega$2 and a Kossakowski matrix $\omega$3 in the Bloch-vector equation $\omega$4 (Wang et al., 6 Sep 2025). In both formulations, Unruh thermalization is an open-system statement about the detector’s reduced density matrix, not merely a statement about a single transition probability.

2. Response functions, KMS structure, and dimensional dependence

The response function is the object that encodes the coupling, the field correlations, and the thermal character of the accelerated vacuum. For a real scalar field in $\omega$5 spacetime dimensions with linear coupling $\omega$6,

$\omega$7

By stationarity and the KMS condition with $\omega$8,

$\omega$9

which is the detailed-balance relation responsible for thermalization at $|g\rangle,|e\rangle$0 (Arrechea et al., 2021).

For free fields, closed forms can be given for both a genuine thermal bath and an accelerated detector in vacuum. In a static thermal bath,

$|g\rangle,|e\rangle$1

with

$|g\rangle,|e\rangle$2

For a uniformly accelerated detector in vacuum,

$|g\rangle,|e\rangle$3

where $|g\rangle,|e\rangle$4 is a polynomial of degree $|g\rangle,|e\rangle$5; for massless fields $|g\rangle,|e\rangle$6 in even $|g\rangle,|e\rangle$7, and $|g\rangle,|e\rangle$8 in odd $|g\rangle,|e\rangle$9 (Arrechea et al., 2021).

The KMS property can also be stated directly at the level of the pulled-back Wightman function. For a uniformly accelerated trajectory in four-dimensional Minkowski space,

$H_I(\tau)=\lambda\,m(\tau)\,\phi^n[x(\tau)],$0

and contour integration gives

$H_I(\tau)=\lambda\,m(\tau)\,\phi^n[x(\tau)],$1

so the detector thermalizes at

$H_I(\tau)=\lambda\,m(\tau)\,\phi^n[x(\tau)],$2

(Garay et al., 2016).

A generalization beyond pointlike scalar couplings is available for smeared detectors coupled locally to any operator in a quantum field theory in curved spacetimes. If the field state is $H_I(\tau)=\lambda\,m(\tau)\,\phi^n[x(\tau)],$3-KMS with respect to the detector’s local time evolution, then adiabatic long-time interactions yield

$H_I(\tau)=\lambda\,m(\tau)\,\phi^n[x(\tau)],$4

and, under mild dynamical assumptions, the reduced detector state approaches $H_I(\tau)=\lambda\,m(\tau)\,\phi^n[x(\tau)],$5 (Perche, 2021). This operator-independence is a strong formulation of the robustness of Unruh thermalization.

3. Emergence of the Gibbs state and the rate of equilibration

Setting

$H_I(\tau)=\lambda\,m(\tau)\,\phi^n[x(\tau)],$6

the excited-state population satisfies

$H_I(\tau)=\lambda\,m(\tau)\,\phi^n[x(\tau)],$7

with solution

$H_I(\tau)=\lambda\,m(\tau)\,\phi^n[x(\tau)],$8

Hence,

$H_I(\tau)=\lambda\,m(\tau)\,\phi^n[x(\tau)],$9

and detailed balance implies

$\lambda\ll1$0

Therefore the asymptotic state is

$\lambda\ll1$1

namely a Gibbs state at the Unruh temperature (Arrechea et al., 2021).

The convergence rate is governed by

$\lambda\ll1$2

For massless $\lambda\ll1$3, one finds $\lambda\ll1$4 up to polynomials. In odd dimensions, $\lambda\ll1$5 contains a Fermi-Dirac factor $\lambda\ll1$6 modulated by a polynomial, so $\lambda\ll1$7 is suppressed at large $\lambda\ll1$8 by that polynomial. For a massive field with $\lambda\ll1$9,

$$m(\tau)= e^{+i\omega\tau}\sigma_+ + e^{-i\omega\tau}\sigma_- .$$0

so the thermalization time $$m(\tau)= e^{+i\omega\tau}\sigma_+ + e^{-i\omega\tau}\sigma_- .$$1 grows exponentially with $$m(\tau)= e^{+i\omega\tau}\sigma_+ + e^{-i\omega\tau}\sigma_- .$$2 (Arrechea et al., 2021).

Open-system analyses of derivative couplings and electromagnetic couplings sharpen the distinction between transient response and asymptotic state. The early-time transition rate may depend strongly on the type of interaction and may not follow a Planck distribution, but the late-time asymptotic state is always thermal at the Unruh temperature; the permanent characterization is the Gibbs steady state rather than the early-time rate alone (Moustos, 2018). This same asymptotic logic is what underlies the statement that the field vacuum acts as a genuine thermal bath for the detector.

4. Statistics inversion, genuine thermal baths, and anti-Unruh behavior

A distinctive feature of Unruh response functions is the “statistics inversion” that appears for a linearly coupled detector in odd spacetime dimensions. In odd $$m(\tau)= e^{+i\omega\tau}\sigma_+ + e^{-i\omega\tau}\sigma_- .$$3,

$$m(\tau)= e^{+i\omega\tau}\sigma_+ + e^{-i\omega\tau}\sigma_- .$$4

where $$m(\tau)= e^{+i\omega\tau}\sigma_+ + e^{-i\omega\tau}\sigma_- .$$5 is an even polynomial of degree $$m(\tau)= e^{+i\omega\tau}\sigma_+ + e^{-i\omega\tau}\sigma_- .$$6. The denominator is Fermi-Dirac-like, and the jump rates become

$$m(\tau)= e^{+i\omega\tau}\sigma_+ + e^{-i\omega\tau}\sigma_- .$$7

Nevertheless,

$$m(\tau)= e^{+i\omega\tau}\sigma_+ + e^{-i\omega\tau}\sigma_- .$$8

so the asymptotic Gibbs state is unchanged (Arrechea et al., 2021). The inversion is therefore a statement about the structure of $$m(\tau)= e^{+i\omega\tau}\sigma_+ + e^{-i\omega\tau}\sigma_- .$$9, not about a failure of thermalization.

The comparison with an inertial detector in a genuine thermal bath is exact only in restricted cases. In $n>1$0 with $n>1$1, the accelerated detector in vacuum and the inertial detector in a thermal bath have exactly the same relaxation rate, coherences, and equilibration. In all other cases considered in the open-system treatment—$n>1$2, $n>1$3, self-interactions, or $n>1$4—the response functions differ,

$n>1$5

so the precise path to equilibrium, including the decay constant and Lamb shift, differs even though both situations yield the same final Gibbs state at $n>1$6 (Arrechea et al., 2021).

This inequivalence is also visible in the anti-Unruh literature. For accelerated detectors coupled to a KMS state, one may have a regime in which the excitation rate decreases as the KMS temperature increases:

$n>1$7

A stronger notion uses the effective EDR-temperature

$n>1$8

with “strong” anti-Unruh corresponding to $n>1$9 (Garay et al., 2016). The effect is characteristic of accelerated detectors and cannot appear for inertially moving detectors in a thermal bath. If the commutator function $\phi^n$0 is independent of $\phi^n$1, then $\phi^n$2 always, so there is no anti-Unruh. If $\phi^n$3 does depend on $\phi^n$4, as for accelerated detectors coupled to the vacuum of a massive scalar in $\phi^n$5 or $\phi^n$6 dimensions, both weak and strong anti-Unruh can occur for certain ranges of parameters (Garay et al., 2016). The standard infinite-time Unruh effect remains an exact thermalization result, but the route to equilibrium is more delicate than the phrase “accelerated detector = thermal bath” suggests.

5. Generalizations, universality, and domain of validity

The modern literature extends Unruh thermalization well beyond a pointlike qubit linearly coupled to a scalar field. For smeared detectors probing arbitrary local operators, the long-time thermalization result follows from KMS plus adiabatic switching, provided the detector remains small relative to the acceleration and curvature scales. The explicit small-smearing conditions are

$\phi^n$7

ensuring that $\phi^n$8 remains an approximate proper-time flow for all smearing points (Perche, 2021).

Spatial extension does not obstruct thermalization. For a uniformly accelerated two-spin system with direct spin–spin coupling $\phi^n$9, the unique zero mode of the Markov generator is

$O(\lambda^2)$0

and in a Rindler frame each part of the extended system “feels” a local Unruh temperature $O(\lambda^2)$1 satisfying the Tolman-law condition $O(\lambda^2)$2 (Lima et al., 2018). This addresses objections based on spatially extended probes by showing that the vacuum state does induce thermalization of an accelerated extended system.

Non-perturbative cavity calculations likewise support universality. For a uniformly accelerated harmonic-oscillator detector in optical cavities, if the switching is smooth enough, the detector thermalizes to the Unruh temperature regardless of Dirichlet, Neumann, or periodic boundary conditions, and regardless of minor variations in the detector–field coupling (Brenna et al., 2013). The asymptotic reduced covariance takes the form

$O(\lambda^2)$3

with near-thermality when $O(\lambda^2)$4 (Brenna et al., 2013).

There are, however, important limitations to simplistic thermal criteria. For higher-dimensional detectors, detailed balance of individual transition channels is not sufficient to establish full Gibbs thermalization. Qudit models have multiple gaps and possible selection rules, and initial coherences need not vanish in a naive second-order Dyson treatment; the more faithful operational test is convergence of the full reduced state to $O(\lambda^2)$5 (Lima et al., 2023). A further limitation concerns what thermalizes: for a detector with a dynamical center of mass, the internal two-level system satisfies detailed balance at $O(\lambda^2)$6, but the center-of-mass degree of freedom does not. The fluctuation–dissipation theorem fails for the center of mass because momentum recoil spoils the strict KMS periodicity of its correlators, so the center of mass is not in thermal equilibrium with the Unruh bath, while the internal level can be (Stargen et al., 2 Sep 2025).

6. Nonequilibrium extensions and recent developments

Recent work has shifted attention from the existence of the Gibbs limit to the geometry and kinetics of the approach toward it. In a Bloch-vector formulation, the detector state follows a one-way spiral in the Bloch ball toward

$O(\lambda^2)$7

with $O(\lambda^2)$8 fixed by the KMS relation $O(\lambda^2)$9. Quantum heat, coherence, Uhlmann fidelity, and the quantum Fisher information define thermodynamic and information-geometric diagnostics of the trajectory (Wang et al., 6 Sep 2025). Within this framework, a quantum Mpemba-like effect appears: for a two-temperature protocol, one finds $\rho\propto e^{-H_D/T_U}$00 for all $\rho\propto e^{-H_D/T_U}$01, and the maximum fidelity difference $\rho\propto e^{-H_D/T_U}$02 distinguishes Unruh thermalization from classical thermal-bath-driven thermalization of an inertial detector (Wang et al., 6 Sep 2025).

Many-body generalizations can depart qualitatively from mono-exponential relaxation. For many non-interacting, equally accelerated atoms in the regime $\rho\propto e^{-H_D/T_U}$03, the Liouvillian acquires quasi-integrals of motion

$\rho\propto e^{-H_D/T_U}$04

and the system first relaxes to a prethermal generalized Gibbs ensemble

$\rho\propto e^{-H_D/T_U}$05

before drifting to full thermal equilibrium. The prethermal plateau persists for

$\rho\propto e^{-H_D/T_U}$06

and is accompanied by a Dicke superradiance-type radiation burst rather than simple mono-exponential decay (Saha et al., 6 Sep 2025). This establishes that single-detector Unruh thermalization is not the generic late-time story for accelerated many-body systems.

Non-uniform and experimentally motivated trajectories also modify the canonical picture. In an alternating-sign-acceleration cavity setup, the detector can “forget” the cavity structure and thermalize to a temperature proportional to its acceleration when $\rho\propto e^{-H_D/T_U}$07, $\rho\propto e^{-H_D/T_U}$08, and $\rho\propto e^{-H_D/T_U}$09, but the continuum Unruh slope $\rho\propto e^{-H_D/T_U}$10 is replaced by $\rho\propto e^{-H_D/T_U}$11 in that finite-cavity, slow-motion regime (Vriend et al., 2020). For oscillatory trajectories, the late-time state is a non-equilibrium squeezed-thermal state, and the mean effective temperature is trajectory dependent and generally smaller than the naive $\rho\propto e^{-H_D/T_U}$12 at high accelerations (Doukas et al., 2013). These results do not negate Unruh thermalization; rather, they delimit the precise regime in which the eternal uniformly accelerated detector provides the correct asymptotic template.

In this broader sense, Unruh thermalization is both robust and sharply circumscribed. It is robust because KMS structure, long interaction times, and weak-coupling open-system dynamics drive a wide class of accelerated detectors to Gibbs equilibrium at $\rho\propto e^{-H_D/T_U}$13. It is circumscribed because the exact response function, the transient rates, the presence of statistics inversion, the possibility of anti-Unruh behavior, the role of detector extension or multiple transition channels, and the distinction between internal and center-of-mass degrees of freedom all determine how, and in some cases whether, the canonical thermalization narrative applies (Arrechea et al., 2021, Garay et al., 2016, Perche, 2021).