Post-Merger Echoes in Astrophysics

• Post-merger echoes are delayed, structured signals that capture the after-coalescence dynamics in both compact-object and galaxy mergers.
• In gravitational-wave studies, echoes arise from cavity mechanisms where partially trapped radiation yields a sequence of weaker, delayed pulses.
• In galaxy evolution, echoes manifest as prolonged features such as enhanced star formation, AGN triggering, and tidal structures, indicating post-coalescence evolution.

“Post-merger echoes” denotes delayed, structured aftereffects that persist after coalescence and are studied in at least two distinct astrophysical settings. In compact-object mergers, the term refers to late-time gravitational-wave structures that follow the primary merger–ringdown signal when radiation is repeatedly reflected or trapped between an outer potential barrier and an inner reflecting region, producing a delayed sequence of pulses or a comb of spectral resonances (Conklin et al., 2017). In galaxy evolution, the phrase is usefully extended to the long-lived post-coalescence response of galaxies—enhanced star formation, rapid quenching, AGN triggering, and faint tidal features—that persists for $0.1$–$1.8$ Gyr or longer after the stellar bodies have merged (Ferreira et al., 2024).

1. Gravitational-wave usage and physical definition

In the gravitational-wave literature, echoes are late-time, quasi-periodic repetitions of the signal that appear after the main merger or ringdown burst (Pani et al., 2018). The basic mechanism is a cavity: perturbations are partially trapped between an outer scattering region, typically the photon-sphere or angular-momentum barrier, and an inner reflecting surface or structure, so that each partial leakage to infinity produces a delayed, weaker pulse (Abedi et al., 2018). A characteristic timescale is the echo delay, with a corresponding frequency scale $f_{\rm echo} \sim \pi/\tau_{\rm echo}$ or, in some model-agnostic treatments, $f_{\rm echo}=1/\Delta t_{\rm echo}$ (Pani et al., 2018).

This usage is not unique to one class of remnant. For black-hole-like remnants, echoes are associated with a partially reflecting membrane or wall just outside the would-be horizon, often motivated by near-horizon quantum structure (Abedi et al., 2018). For horizonless ultracompact objects, the cavity can be produced without any horizon-scale quantum ingredient: the photon sphere acts as the outer barrier, while the regular stellar interior or center provides the inner reflective region (Pani et al., 2018). A necessary condition in that case is $R<3M$, so that the remnant possesses a photon sphere outside its surface (Mannarelli et al., 2018).

The same late-time structures can also arise in dynamical or rotating classical general-relativistic settings without exotic compact objects. In binary black-hole mergers, “post-merger chirps” or secondary post-merger chirps are additional time-frequency peaks correlated with the evolving geometry of the common horizon, not with repeated reflections from a near-horizon cavity (Bustillo et al., 2019). This distinction matters because post-merger structure in the data is not, by itself, sufficient to imply horizon-scale new physics. A plausible implication is that “post-merger echoes” is best treated as a family resemblance term rather than a single mechanism.

2. Cavity mechanisms, compactness, and time-delay scales

For compact-object echoes, the canonical picture is wave propagation in an effective cavity. In a Kerr or Schwarzschild exterior, the angular-momentum barrier near the light ring partially transmits and partially reflects gravitational perturbations (Conklin et al., 2017). If the inward-propagating part is reflected by a surface or inner wall rather than being absorbed at a classical horizon, repeated bounces generate a train of delayed signals (Maggio et al., 2019).

For black-hole-like exotic compact objects, the echo delay is logarithmically enhanced because the inner boundary lies at a very small proper distance from the horizon. A representative estimate is

$\Delta t_{\rm echo} \simeq 8 M \log M$

in Planck units in the Schwarzschild-like case, while for a spinning remnant one may write

$$\Delta t_{\rm echo} \simeq \frac{4 G M_{\rm BH}}{c^3} \left(1+\frac{1}{\sqrt{1-a^2}}\right)\ln\!\left(\frac{M_{\rm BH}}{M_{\rm Planck}}\right),$$

with $f_{\rm echo}\equiv \Delta t_{\rm echo}^{-1}$ (Abedi et al., 2018). This scaling places the fundamental frequency in the tens-of-Hz range for a $2.6$–$2.7\,M_\odot$ remnant, which motivated searches around $1.8$0–$1.8$1 Hz for GW170817 (Abedi et al., 2018).

For ultracompact stars, the same delayed behavior can emerge in classical GR if the radius satisfies $1.8$2 and approaches the Buchdahl bound $1.8$3 (Pani et al., 2018). In the constant-density toy model, the echo timescale is

$1.8$4

and, near the Buchdahl limit, behaves as

$1.8$5

so the divergence is a power law rather than logarithmic (Pani et al., 2018). For $1.8$6, the low-frequency $1.8$7 Hz claim for GW170817 requires $1.8$8, meaning a radius extremely close to $1.8$9 (Pani et al., 2018).

This compactness requirement immediately separates models. Causal strange-star equations of state can cross the photon-sphere line $f_{\rm echo} \sim \pi/\tau_{\rm echo}$0, but even with $f_{\rm echo} \sim \pi/\tau_{\rm echo}$1 they do not approach $f_{\rm echo} \sim \pi/\tau_{\rm echo}$2 closely enough to yield echoes at tens of Hz; instead their echo frequencies are of order $f_{\rm echo} \sim \pi/\tau_{\rm echo}$3–$f_{\rm echo} \sim \pi/\tau_{\rm echo}$4 kHz in the most compact stable configurations (Mannarelli et al., 2018). A common misconception is therefore that any ultracompact star naturally explains low-frequency post-merger echoes. The literature summarized here indicates the opposite: low-frequency echoes require either near-horizon structures or stellar configurations in strong tension with current neutron-star models (Pani et al., 2018).

3. Spectral structure, greybody factors, and analytic templates

The time-domain echo train has a frequency-domain counterpart: a comb of nearly evenly spaced spectral resonances (Conklin et al., 2017). In a broad class of models, the cavity transfer function is written as a geometric series in the product of the barrier reflectivity, the wall reflectivity, and the round-trip phase. This leads to a sequence of resonances with spacing set approximately by the inverse delay time and widths set by the leakage through the barrier (Maggio et al., 2019).

For spinning remnants, an analytic low-frequency template can be constructed in terms of the Kerr barrier coefficients and a complex reflectivity $f_{\rm echo} \sim \pi/\tau_{\rm echo}$5. A representative transfer function is

$f_{\rm echo} \sim \pi/\tau_{\rm echo}$6

where $f_{\rm echo} \sim \pi/\tau_{\rm echo}$7, $f_{\rm echo} \sim \pi/\tau_{\rm echo}$8 encodes the compactness, and $f_{\rm echo} \sim \pi/\tau_{\rm echo}$9 and $f_{\rm echo}=1/\Delta t_{\rm echo}$0 describe the photon-sphere barrier (Maggio et al., 2019). In this framework the barrier acts as a spin- and frequency-dependent high-pass filter, so successive echoes become progressively lower-frequency, and a major fraction of the energy is contained in low-frequency resonances corresponding to the quasi-normal modes of the remnant (Maggio et al., 2019).

A more recent line of work argues that greybody factors are more robust observables than QNM spectra for post-merger signals (Rosato, 26 May 2025). In that formulation, the perturbation equation is

$f_{\rm echo}=1/\Delta t_{\rm echo}$1

with reflection and transmission encoded by

$f_{\rm echo}=1/\Delta t_{\rm echo}$2

The proposed advantage is that greybody factors are stable under small perturbations of the potential, whereas QNM frequencies are spectrally unstable (Rosato, 26 May 2025). In the wormhole or horizonless-ultracompact-object case, oscillatory structure in $f_{\rm echo}=1/\Delta t_{\rm echo}$3 Fourier-transforms into multiple late-time peaks, providing a natural explanation for echoes in the time-domain response (Rosato, 26 May 2025).

An analytically distinct, but conceptually related, surrogate description writes the spectrum as a sum of Lorentzian lines,

$f_{\rm echo}=1/\Delta t_{\rm echo}$4

with line positions $f_{\rm echo}=1/\Delta t_{\rm echo}$5, widths $f_{\rm echo}=1/\Delta t_{\rm echo}$6, amplitudes $f_{\rm echo}=1/\Delta t_{\rm echo}$7, and phases $f_{\rm echo}=1/\Delta t_{\rm echo}$8 calibrated from the underlying transfer function (Conklin et al., 2021). In that model, the wall reflectivity is Boltzmann-like,

$f_{\rm echo}=1/\Delta t_{\rm echo}$9

and the non-uniform resonance spacing is emphasized as observationally important because constant-spacing combs can incur severe overlap loss (Conklin et al., 2021).

4. Searches, candidate events, and current controversy

The most widely discussed candidate was GW170817. A model-agnostic cross-correlation search reported a peak at $R<3M$0 Hz, around $R<3M$1 s after the merger, with a stated false alarm probability $R<3M$2, corresponding to $R<3M$3 within the chosen time-frequency window (Abedi et al., 2018). The same work interpreted the signal as consistent with a $R<3M$4–$R<3M$5 remnant with spin $R<3M$6–$R<3M$7, and noted agreement with an electromagnetic estimate of the collapse time $R<3M$8 s (Abedi et al., 2018).

That interpretation is not unique. One analysis argued that echoes at $R<3M$9 Hz could be reproduced by an incompressible ultracompact star with mass $\Delta t_{\rm echo} \simeq 8 M \log M$0 and radius very close to the Buchdahl limit, though this would be in tension with all current neutron-star models (Pani et al., 2018). Another showed that causal strange stars cannot explain a $\Delta t_{\rm echo} \simeq 8 M \log M$1 Hz signal, because their corresponding echo frequencies lie in the tens of kHz (Mannarelli et al., 2018). A further body of work emphasized that low-frequency echoes could also arise from quantum black holes with Boltzmann reflectivity, with the first $\Delta t_{\rm echo} \simeq 8 M \log M$2 echoes decaying inversely with time and later echoes decaying exponentially (Wang et al., 2019).

Claims are controversial beyond GW170817. A frequency-domain comb search reported signals with $\Delta t_{\rm echo} \simeq 8 M \log M$3-values of order $\Delta t_{\rm echo} \simeq 8 M \log M$4 or smaller in several LIGO/Virgo events, including GW151226, GW170104, GW170608, GW170814, and GW170817, and found echo delays broadly consistent with a simple truncated-Kerr model (Conklin et al., 2017). For GW190521, a Lorentzian/Boltzmann surrogate search inferred a fractional energy in post-merger echoes

$\Delta t_{\rm echo} \simeq 8 M \log M$5

with the uncertainty representing the 90% credible region, and characterized the evidence as moderate rather than definitive (Conklin et al., 2021).

At the same time, the observational status remains unsettled. Several papers explicitly note that echo claims in binary black-hole events are controversial, that different pipelines give different significances, and that independent confirmation with additional events and detectors is required (Pani et al., 2018). A useful caution is that standard GR can generate complex post-merger structure of its own. Unequal-mass, spinning, and precessing binary black holes can show secondary post-merger chirps in the time-frequency domain, correlated with horizon geometry, without any reflective near-horizon wall (Henshaw et al., 23 May 2025). This suggests that some apparent “echo-like” features may instead be higher-mode, spin-driven, or horizon-dynamical structure within classical GR.

5. Interval structure and evolving cavities

A standard simplifying assumption is that successive echoes are equally spaced in time. One paper argues that this need not hold if the post-merger object is dynamical rather than stationary (Wang et al., 2018). In the example of a wormhole whose throat slowly pinches off and approaches a black hole, the cavity length in tortoise coordinate increases with time, so the delay between successive echoes also increases (Wang et al., 2018).

For a Morris–Thorne wormhole modeled by gluing two Schwarzschild spacetimes at

$\Delta t_{\rm echo} \simeq 8 M \log M$6

the barrier separation is approximately

$\Delta t_{\rm echo} \simeq 8 M \log M$7

and the echo interval is

$\Delta t_{\rm echo} \simeq 8 M \log M$8

As $\Delta t_{\rm echo} \simeq 8 M \log M$9 decreases, later echoes are more widely separated (Wang et al., 2018). A direct implication is that equal-interval templates are not generally sufficient for all horizonless or evolving near-horizon scenarios.

This time dependence provides a conceptual bridge between compact-object echoes and other post-merger phenomena. In both cases, the observable late-time structure reflects not only the existence of a remnant, but also its internal dynamical evolution after coalescence. The difference is that, in the compact-object case, the relevant evolution is the changing cavity geometry or reflectivity, whereas in galaxy mergers it is the time-dependent response of star formation, AGN fueling, and tidal debris.

6. Galaxy-merger post-merger echoes

In galaxy evolution, post-merger echoes are long-lived signatures that remain after the stellar bodies have coalesced. These include enhanced star formation, rapid post-starburst quenching, AGN triggering, and faint tidal structures (Ferreira et al., 2024). The recent UNIONS-based studies use the MUMMI deep-learning framework to classify galaxies as pairs or post-mergers and to assign time since coalescence $$\Delta t_{\rm echo} \simeq \frac{4 G M_{\rm BH}}{c^3} \left(1+\frac{1}{\sqrt{1-a^2}}\right)\ln\!\left(\frac{M_{\rm BH}}{M_{\rm Planck}}\right),$$0 in four bins extending to $$\Delta t_{\rm echo} \simeq \frac{4 G M_{\rm BH}}{c^3} \left(1+\frac{1}{\sqrt{1-a^2}}\right)\ln\!\left(\frac{M_{\rm BH}}{M_{\rm Planck}}\right),$$1 Gyr (Ferreira et al., 2024).

The star-formation response peaks around coalescence and remains elevated well into the post-merger phase. Using a sample of $$\Delta t_{\rm echo} \simeq \frac{4 G M_{\rm BH}}{c^3} \left(1+\frac{1}{\sqrt{1-a^2}}\right)\ln\!\left(\frac{M_{\rm BH}}{M_{\rm Planck}}\right),$$2 star-forming post-mergers with $$\Delta t_{\rm echo} \simeq \frac{4 G M_{\rm BH}}{c^3} \left(1+\frac{1}{\sqrt{1-a^2}}\right)\ln\!\left(\frac{M_{\rm BH}}{M_{\rm Planck}}\right),$$3 at $$\Delta t_{\rm echo} \simeq \frac{4 G M_{\rm BH}}{c^3} \left(1+\frac{1}{\sqrt{1-a^2}}\right)\ln\!\left(\frac{M_{\rm BH}}{M_{\rm Planck}}\right),$$4, one study found that mergers enhance star formation by, on average, up to a factor of two; that this enhancement peaks within $$\Delta t_{\rm echo} \simeq \frac{4 G M_{\rm BH}}{c^3} \left(1+\frac{1}{\sqrt{1-a^2}}\right)\ln\!\left(\frac{M_{\rm BH}}{M_{\rm Planck}}\right),$$5 Myr of coalescence; that enhancements continue for up to $$\Delta t_{\rm echo} \simeq \frac{4 G M_{\rm BH}}{c^3} \left(1+\frac{1}{\sqrt{1-a^2}}\right)\ln\!\left(\frac{M_{\rm BH}}{M_{\rm Planck}}\right),$$6 Gyr after coalescence; and that merger-induced star formation contributes $$\Delta t_{\rm echo} \simeq \frac{4 G M_{\rm BH}}{c^3} \left(1+\frac{1}{\sqrt{1-a^2}}\right)\ln\!\left(\frac{M_{\rm BH}}{M_{\rm Planck}}\right),$$7 of excess stellar mass per event, corresponding to $$\Delta t_{\rm echo} \simeq \frac{4 G M_{\rm BH}}{c^3} \left(1+\frac{1}{\sqrt{1-a^2}}\right)\ln\!\left(\frac{M_{\rm BH}}{M_{\rm Planck}}\right),$$8–$$\Delta t_{\rm echo} \simeq \frac{4 G M_{\rm BH}}{c^3} \left(1+\frac{1}{\sqrt{1-a^2}}\right)\ln\!\left(\frac{M_{\rm BH}}{M_{\rm Planck}}\right),$$9 of the final stellar mass (Ferreira et al., 2024). Most of that in-situ stellar mass growth occurs after coalescence, not before (Ferreira et al., 2024).

A complementary study examined rapid quenching using post-starburst diagnostics in $f_{\rm echo}\equiv \Delta t_{\rm echo}^{-1}$0 MUMMI-identified post-mergers. It found that the post-coalescence population evolves from one dominated by star-forming and starbursting galaxies at $f_{\rm echo}\equiv \Delta t_{\rm echo}^{-1}$1 Gyr to one dominated by quenched galaxies by $f_{\rm echo}\equiv \Delta t_{\rm echo}^{-1}$2 Gyr (Ellison et al., 2024). The excess of post-starbursts peaks at $f_{\rm echo}\equiv \Delta t_{\rm echo}^{-1}$3 Gyr: PCA-selected PSBs are more common than in controls by $f_{\rm echo}\equiv \Delta t_{\rm echo}^{-1}$4, while classically selected E+A PSBs are more common by $f_{\rm echo}\equiv \Delta t_{\rm echo}^{-1}$5 in that same interval (Ellison et al., 2024). The same study found that the majority of PSBs are linked to mergers, with a total merger fraction of $f_{\rm echo}\equiv \Delta t_{\rm echo}^{-1}$6 for E+A systems and $f_{\rm echo}\equiv \Delta t_{\rm echo}^{-1}$7 for PCA-selected PSBs under the fiducial classification (Ellison et al., 2024).

AGN triggering shows a parallel but distinct post-merger echo. Using the same time-resolved merger sequence, another study found that the excess of AGN—identified via mid-IR colors, narrow emission lines, and broad emission lines—peaks immediately after coalescence, in the bin $f_{\rm echo}\equiv \Delta t_{\rm echo}^{-1}$8 Gyr (Ellison et al., 2024). The excess persists long after coalescence: both mid-IR selected AGN and broad-line AGN remain more common than in matched controls even at $f_{\rm echo}\equiv \Delta t_{\rm echo}^{-1}$9 Gyr (Ellison et al., 2024). The excess is larger for more luminous and bolometrically dominant AGN, and the deficit of broad-line AGN in the pre-merger phase turning into an excess in post-mergers is interpreted as evolution in the covering fraction of nuclear obscuring material (Ellison et al., 2024). This suggests a sequence in which tidally triggered inflows initially increase nuclear obscuration before AGN feedback clears at least some of that material after coalescence (Ellison et al., 2024).

Morphological echoes persist even longer than the spectroscopic ones. Numerical simulations of equal-mass disk mergers showed that the merger-feature time depends strongly on image depth: for a shallow surface-brightness limit of $2.6$0 mag arcsec$2.6$1, the merger-feature time is on average $2.6$2 times the final coalescence time, whereas for a deeper limit of $2.6$3 mag arcsec$2.6$4 it is a factor of two longer (Ji et al., 2014). The same work found that tidal forces in a cluster potential strip post-merger features and reduce the merger-feature time (Ji et al., 2014). A useful interpretation is that galaxy-merger echoes form a hierarchy: AGN and starbursts peak promptly, post-starburst signatures peak after a delay of a few $2.6$5 yr, and very faint morphological debris can remain visible for several Gyr.

7. Conceptual synthesis and common misconceptions

Across both compact-object and galaxy-merger literatures, post-merger echoes are delayed structures that preserve information about the remnant after coalescence. In compact-object mergers, the controlling variables are cavity size, barrier transmission, reflectivity, compactness, and spin (Maggio et al., 2019). In galaxy mergers, the controlling variables are gas inflow, obscuration, stellar-population aging, and the phase-mixing and stripping of tidal debris (Ellison et al., 2024).

One misconception is that all post-merger structure in gravitational-wave data implies exotic near-horizon physics. The literature reviewed here does not support that. Secondary post-merger chirps can be generated by higher-order modes, spin, precession, and the evolving geometry of the common horizon in standard GR (Henshaw et al., 23 May 2025). Another misconception is that the term should be restricted to gravitational waves. The time-resolved UNIONS studies demonstrate that post-merger echo language also captures a well-defined sequence in galaxy evolution: starburst, AGN triggering, rapid quenching, and long-lived tidal signatures after coalescence (Ferreira et al., 2024).

Taken together, these results suggest that post-merger echoes are best understood as delayed observables of remnant structure. In one domain they probe strong-field gravity, horizon-scale microphysics, and ultracompact-matter models; in the other they probe gas inflow, feedback, quenching, and morphological relaxation. The common logic is the same: coalescence is not the end of the event, but the start of an after-response whose timing, spectral content, and persistence encode the physical nature of the remnant.