---
title: 'Cliffhanger EMRI: Threshold Dynamics'
url: https://www.emergentmind.com/topics/cliffhanger-emri
type: topic
---

# Cliffhanger EMRI: Threshold Dynamics

Searching arXiv for recent papers on cliffhanger EMRIs and related EMRI waveform modeling.
A cliffhanger EMRI is an extreme-mass-ratio inspiral situated at a dynamical threshold between direct plunge and long-lived inspiral. In current usage, the term is not completely uniform. In low-mass massive-black-hole and intermediate-mass-black-hole systems, it denotes a “failed plunge” in which a compact object on a nearly radial orbit emits enough gravitational-wave energy at periapse to transition into a true EMRI rather than plunge directly [2304.13062]. In post-Newtonian Monte Carlo studies with local relaxation, the same phenomenon appears as a nonzero large-\(a_0\) plateau in the EMRI fraction \(S(a_0)\) and can account for a substantial fraction of EMRIs around \(M_\bullet\lesssim3\times10^5\,{\rm M}_\odot\) [2409.09122]. In waveform modeling, “cliffhangers” also refers to transient resonances that generate step-like glitches in individual \(h_{\ell m}(t)\) modes [1201.5715]. More recent electromagnetic and AGN-disk studies use the label for candidate EMRI/IMRI systems identified through quasiperiodic ultrafast outflows or for binaries hovering near a torque-balance threshold between inspiral and ejection [2410.12090; 2411.16070].

## 1. Terminological scope and core usages

The literature attaches “cliffhanger” to several closely related threshold phenomena. All usages share the idea that the system remains near a separatrix: plunge versus inspiral, smooth adiabatic evolution versus transient resonant departure, or inspiral versus ejection.

| Usage | Defining feature | Source |
|---|---|---|
| Failed-plunge EMRI formation | A plunge has an \(\mathcal O(1)\) probability of failing and transitioning into a novel “cliffhanger” EMRI for low-mass black holes | [2304.13062] |
| Large-\(a_0\) EMRI plateau | \(S(a_0)\gg0\) for \(a_0\gg a_{\rm c}\), reaching values as high as \(0.6\) for \(M_\bullet=10^4\,{\rm M}_\odot\) | [2409.09122] |
| Relativistic loss-cone “hanging” | Orbits “hang” at the relativistic loss-cone boundary after resonant relaxation is quenched by GR precession | [1702.00597] |
| Waveform resonance glitch | A transient feature, step-like jump, in the amplitude or phase of a single \((\ell,m)\) component near frequency commensurability | [1201.5715] |
| Electromagnetic or gas-disk candidate state | Quasiperiodic ultrafast outflows or a metastable balance between hydrodynamical torques and relativistic GW losses | [2410.12090], [2411.16070] |

This suggests that “cliffhanger EMRI” is best understood as a family of threshold behaviors rather than a single invariant definition. The most technically developed meaning is the failed-plunge formation channel in low-mass nuclei, but the term has broadened into waveform phenomenology and multimessenger candidate identification.

## 2. Relativistic loss-cone background

The conceptual precursor to the modern failed-plunge picture is the relativistic loss-cone analysis of EMRIs in a stellar background. In that framework, EMRIs are governed by a competition between resonant relaxation, Schwarzschild precession, and gravitational-wave dissipation. A leading-order expression for the resonant-relaxation timescale is
\[
T_{\rm RR}(a)\approx \frac{M_\bullet^2}{N(a)\,m^2}\,\frac{P(a)^2}{T_c(a)}\,,
\]
while the 1PN precession timescale is
\[
T_{\rm prec}(a,e)=\frac{c^2\,a\,(1-e^2)\,P(a)}{3\,G\,M_\bullet}\,,
\]
and the 2.5PN inspiral timescale is
\[
T_{\rm GW}(a,e)=\frac{5}{64}\,\frac{c^5\,a^4}{G^3\,M_\bullet^2\,m}\,
\frac{(1-e^2)^{7/2}}{1+\tfrac{73}{24}e^2+\tfrac{37}{96}e^4}\,.
\]
For typical Milky Way–like cusps, the hierarchy is
\[
T_{\rm RR}\;\gtrsim\;T_{\rm prec}\;\gg\;T_{\rm GW}\,,
\]
which creates a narrow window in which EMRIs can decouple from background torques and complete the inspiral [1702.00597].

The key structure is the adiabatic-invariance line \(j_{\rm AI}(a)\), defined by the condition that relativistic precession outruns the coherence of the stellar torques. Below this line, fast Schwarzschild precession suppresses resonant relaxation. The plunge boundary is
\[
j_{\bullet}(a)=\sqrt{\frac{16\,r_g}{a}}\,,
\qquad
r_g=\frac{G\,M_\bullet}{c^2}\,.
\]
EMRIs can form if
\[
j_{\rm AI}(a_{\rm GW})\gtrsim j_{\bullet}(a_{\rm GW})\,.
\]
In this regime, orbits diffuse toward the relativistic loss-cone boundary, background torques are quenched, and GW emission takes over before plunge. The paper explicitly describes such trajectories as stars that “hang” at the relativistic loss-cone boundary—hence “cliffhanger” EMRIs. The same analysis yields \(a_{\rm GW}\simeq0.03\,r_h\,(\ln Q)^{-4/5}\) and a steady-state branching ratio \(R_{\rm EMRI}/R_{\rm plunge}\sim a_{\rm GW}/r_h\sim10^{-2}\), identifying a small but persistent EMRI-producing phase-space region [1702.00597].

This relativistic-loss-cone picture is not identical to the later failed-plunge channel, but it provides the dynamical logic for why inspirals can survive near an apparently terminal boundary.

## 3. Failed plunges and the low-mass black-hole cliffhanger channel

The modern cliffhanger mechanism was formulated for compact objects on highly eccentric orbits around intermediate-mass black holes. The orbit evolves under two competing processes: GW dissipation and angular-momentum relaxation. The relevant timescales were written as
\[
t_{\rm GW}(a,q)
= \frac{3\cdot2^{7/2}\,85}{c^5\,G^3\,M_\bullet^2\,m}\,a^{1/2}\,q^{7/2},
\qquad q=a(1-e)\,,
\]
and
\[
t_{\rm AM}(r,e)
= \frac{2\,k\,\sigma^3(r)}{G^2\,n(r)\,\langle m^2\rangle\,\ln\Lambda}\,(1-e)\,,
\qquad
\sigma^2(r)=\frac{GM_\bullet}{(1+\gamma)\,r}\,.
\]
The classical EMRI/plunge separatrix is defined by \(t_{\rm GW}=f\,t_{\rm AM}\) at \(q_{\rm IBCO}=8\,R_g\), producing a critical semimajor axis \(a_c\) [2304.13062].

The new effect occurs for \(M_\bullet\lesssim M_c\), where the last, nearly radial plunge orbit can emit so much GW energy in a single periapse passage that the post-burst orbit satisfies \(a_{\rm new}<a_c\). The paper describes this as a compact object switching “from a would-be plunge into a long, GW-dominated inspiral.” Equating the one-orbit GW energy loss to the binding energy at \(a_c\) yields
\[
M_c
=\Bigl[\frac{85\pi\,m\,c^2\,a_c}{6\sqrt2\,G\,x_{\rm IBCO}^{7/2}}\Bigr]^{1/2}
\approx10^{4.9}\,M_\odot,
\qquad x_{\rm IBCO}=8\,.
\]
This is the defining failed-plunge cliffhanger mechanism [2304.13062].

The transition is encoded in the capture fraction \(S(a)\). Classically, \(S(a)\to1\) for \(a\ll a_c\) and \(S(a)\to0\) for \(a\gg a_c\). With cliffhangers, however, for \(M_\bullet<M_c\),
\[
S(a)\longrightarrow S_\infty>0
\qquad (a\gg a_c)\,,
\]
and
\[
P_{\rm cliff}(m_\star,a,M_\bullet)\equiv S(a\,|\,m_\star,M_\bullet)-\Theta(a_c-a)\,.
\]
Asymptotically, \(P_{\rm cliff}\simeq S_\infty(m_\star,M_\bullet)\), and \(P_{\rm cliff}\) rises from zero for \(M_\bullet>M_c\) to an \(\mathcal O(0.1\!-\!1)\) plateau at \(M_\bullet\ll M_c\). Monte Carlo results summarized in the paper show that \(S_\infty\) grows with compact-object mass: for a \(10\,M_\odot\) black hole, \(S_\infty\sim0.2\!-\!0.4\) at \(M_\bullet\sim10^4\,M_\odot\); for \(1.4\,M_\odot\) neutron stars, \(S_\infty\approx0\); and for \(50\,M_\odot\) black holes, \(S_\infty\gtrsim0.5\). Flatter cusps increase \(a_c\) and \(M_c\), while prograde Kerr orbits with \(x_{\rm IBCO}\simeq2\) instead of \(8\) raise \(M_c\) by orders of magnitude, potentially extending the cliffhanger regime to \(\sim10^6\!-\!10^7\,M_\odot\) [2304.13062].

The rate implications are correspondingly large. Because \(S(a)\) no longer falls to zero at \(a\gg a_c\), the range \(a_c<a<r_{\rm inf}\) contributes an extra \(\Delta\dot N_{\rm EMRI}\sim S_\infty\,\dot N_{\rm pl}\), and the total cosmic EMRI rate can be boosted by up to \(\sim10\times\) compared to the classical estimate. The same study states that if dwarf galaxies down to \(M_\bullet\sim10^4\,M_\odot\) host IMBHs at even moderate occupation fractions \((f_{\rm occ}\gtrsim0.5)\), the volumetric EMRI rate could rise from \(\sim1\) to \(\sim10\) Gpc\(^{-3}\)yr\(^{-1}\), yielding tens of LISA detections per year. The cliffhanger channel therefore modifies not only the plunge/EMRI branching ratio but the expected cosmological EMRI inventory [2304.13062].

## 4. Post-Newtonian Monte Carlo confirmation and refinement

A later study replaced orbit-averaged diffusion with a local, on-the-fly treatment of two-body relaxation coupled to post-Newtonian orbital evolution. The diagnostic quantity is
\[
S(a_0)=\frac{N_{\rm EMRI}(a_0)}{N_{\rm EMRI}(a_0)+N_{\rm DP}(a_0)}\,,
\]
the fraction of captures at initial semimajor axis \(a_0\) that become true EMRIs rather than direct plunges. Earlier work had suggested a single critical scale
\[
a_{\rm c}\simeq (2\!-\!5)\times10^{-2}\,R_{\rm inf},
\qquad
R_{\rm inf}\equiv \frac{GM_\bullet}{\sigma^2}\,,
\]
with \(S\approx1\) for \(a_0\ll a_{\rm c}\) and \(S\approx0\) for \(a_0\gg a_{\rm c}\). The new simulations solve
\[
H=H_0+\frac{1}{c^2}H_1+\frac{1}{c^4}H_2+\frac{1}{c^5}H_{2.5}+\cdots
\]
up to 2.5PN order with an adaptive Bulirsch–Stoer integrator, while applying continuous local Gaussian velocity kicks determined by instantaneous diffusion coefficients [2409.09122].

The principal result is that cliffhangers are not only confirmed but are more numerous than previously expected. They start to appear for \(M_\bullet\lesssim3\times10^5\,{\rm M}_\odot\) and can account for up to \(55\%\) of the overall EMRIs formed. For the lightest case \(M_\bullet=10^4\,{\rm M}_\odot\), the large-\(a_0\) plateau reaches \(\max S(a_0)\approx0.6\). The study gives a dynamical picture of a compact object scattered onto a nearly radial orbit that just misses the loss-cone radius
\[
R_{\rm lc}=\zeta\,R_g,\qquad
\zeta=
\begin{cases}
8 & (N)\\
6.45 & (1{\rm PN})\\
5.6 & (2{\rm PN}\ {\rm or}\ 2.5{\rm PN})
\end{cases},
\]
and then emits a large GW burst at pericenter in one passage. In the \(e\to1\) limit, the cliffhanger region in the \((1-e,a)\) plane is bounded by
\[
a_{\rm lc}(e)=\frac{\zeta\,R_g}{1-e}
\;<\;a\;<\;
a_{\rm cliff}(e)\,,
\]
and simple algebra gives an upper mass scale for cliffhangers,
\[
M_\bullet\lesssim1.8\times10^7\,\zeta^{-2.38}
\bigl(m_\bullet/10M_\odot\bigr)^{0.68}
\sim3\times10^5\,M_\odot\,,
\]
in excellent agreement with the simulations [2409.09122].

The numerical refinements matter. In pure Newtonian gravity with orbit-averaged Peters losses, the classical step-function picture is recovered with \(a_{\rm c}\simeq5\times10^{-3}R_{\rm inf}\). Turning on 2.5PN evolution pushes \(a_{\rm c}\to10^{-2}R_{\rm inf}\) for all \(M_\bullet\). The paper attributes this to two effects: PN pericenter precession allows stable orbits to reach \(\sim5.6R_g\) rather than \(8R_g\) before plunging, and the PN orbital shape further suppresses direct plunges at moderate periapses. Continuous local kicks also enhance EMRI formation in the full-loss-cone regime relative to orbit-averaged kicks at apocenter, which systematically underpredict \(S(a_0)\) for \(a_0\gg a_{\rm c}\), especially at low \(M_\bullet\) [2409.09122].

The resulting \(S(a_0)\) curves are explicitly nonclassical. For \(M_\bullet=3\times10^5\,M_\odot\), \(S\simeq1\) for \(a_0\lesssim10^{-2}\), falls to \(S\sim0.2\) at \(a_0\sim0.1\), and approaches \(S_\infty\simeq0.1\) at larger \(a_0\). For \(M_\bullet=10^5\,M_\odot\), \(S(a_0)\) plateaus at \(S_\infty\approx0.3\). For \(M_\bullet=10^4\,M_\odot\), \(S_\infty\approx0.6\). In a fiducial \(M_\bullet=3\times10^5\,M_\odot\) cluster, the revised rate estimate changes from
\[
\dot N_{\rm EMRI}^{\rm(cl)}\approx5\times10^{-8}\,{\rm yr}^{-1}
\]
to
\[
\dot N_{\rm EMRI}\approx7\times10^{-8}\,{\rm yr}^{-1},
\]
a \(\sim35\%\) increase. The authors emphasize that the total detection yield may increase by tens of percent and that cliffhanger EMRIs enter LISA with richer eccentricity distributions than in standard Fokker–Planck models [2409.09122].

## 5. Waveform-level cliffhangers in the Chimera scheme

A distinct but related usage arises in the Chimera waveform model for generic EMRIs. Chimera is a hybrid kludge that combines a multipolar, post-Minkowskian expansion for the far-zone metric perturbation and local self-force prescription, a post-Newtonian expansion for computing the multipole moments, and a black-hole perturbation-theory expansion in which the orbit is represented as a sequence of self-adjusting Kerr geodesics [1201.5715].

The dynamics begin with the MiSaTaQuWa point-particle equation of motion,
\[
\frac{d^2 z^\alpha}{d\tau^2}+\Gamma^\alpha_{\mu\nu}u^\mu u^\nu
=\frac{1}{m}F^\alpha,
\]
with
\[
F^\alpha
=-\frac12\,m\,(g^{\alpha\lambda}+u^\alpha u^\lambda)\,
u^\mu u^\nu\bigl(2\nabla_\mu h_{\nu\lambda}-\nabla_\lambda h_{\mu\nu}\bigr)\,.
\]
The trajectory is built from Kerr geodesic fragments joined by the osculating-elements method, with \((E,L_z,Q)\) evolving according to
\[
\frac{dE}{d\tau}=-\zeta^{(t)}_\alpha a^\alpha,\qquad
\frac{dL_z}{d\tau}=\zeta^{(\phi)}_\alpha a^\alpha,\qquad
\frac{dQ}{d\tau}=2\,\xi_{\alpha\beta}\,u^\alpha a^\beta\,.
\]
A coordinate map converts Boyer–Lindquist geodesics to harmonic coordinates for waveform generation, and the far-zone signal is obtained in TT gauge and then decomposed as
\[
h_+(t)+i\,h_\times(t)=\sum_{\ell,m}h_{\ell m}(t)\;{}_{-2}Y_{\ell m}(\theta,\phi)\,.
\]
This construction is explicitly local in time [1201.5715].

That locality is crucial for transient resonances. Because the Chimera self-force depends on the instantaneous worldline rather than on an averaged flux, the orbital elements acquire short-term oscillations at the fundamental frequencies \(\Omega_{r,\theta,\phi}\). When
\[
k\,\Omega_r+m\,\Omega_\theta+n\,\Omega_\phi=0
\]
is crossed, the two-time-scale expansion underlying adiabatic averaged schemes breaks down and one predicts a small secular jump \(\Delta{\cal O}\) in each orbital constant. The paper states that these show up automatically as sharp departures from the smooth drift in \(E(t)\), \(L_z(t)\), and \(Q(t)\). In the waveform, a resonance induces a phase-kink and amplitude glitch in the affected modes \(h_{\ell m}(t)\). Writing
\[
h_{\ell m}(t)\simeq A_{\ell m}(t)e^{-i\Phi_{\ell m}(t)}\,,
\]
the mode amplitude acquires an \({\cal O}(q)\) step and the phase an \({\cal O}(q)\) phase offset over a short timescale \(\sim1/\sqrt q\). The paper identifies the cliffhanger EMRI signature as a transient feature, a step-like jump, in the amplitude or phase of a single \((\ell,m)\) component localized to the moment of frequency commensurability [1201.5715].

This waveform usage differs from the failed-plunge channel: it concerns resonance structure during an inspiral rather than the birth of an inspiral from a failed plunge. The common element is again a sharply localized threshold event that leaves a persistent imprint.

## 6. Electromagnetic candidates and AGN-disk cliffhanger states

Electromagnetic work has extended the terminology into multimessenger candidate identification. In the quasiperiodic ultrafast outflow scenario, a secondary perturber on a moderately inclined orbit around an SMBH repeatedly plunges through the inner accretion flow, lifts gas above the disk midplane, and drives transient soft X-ray absorption troughs. The recurrence interval is set by the orbital period. For ASASSN-20qc, the observed QPOut period \(T\simeq8.3\) days is identified with \(P_{\rm orb}\), and for \(M_1\approx3\times10^7\,M_\odot\) this gives
\[
a\simeq79\,r_g\times
\left(\frac{M_1}{10^{7.5}M_\odot}\right)^{-2/3}
\left(\frac{T}{8\,{\rm days}}\right)^{2/3},
\]
that is, a few \(\times10^{13}\) cm or \(\sim100\,r_g\). The Hill-radius condition yields \(m_2\gtrsim{\rm few}\times10^2\,M_\odot\), while a statistical argument based on \(\tau_{\rm gw}\gtrsim10^4\) yr yields \(m_2\lesssim10^4\,M_\odot\), so the mass ratio is \(\sim10^{-5}-10^{-3}\), consistent with an IMBH perturber at \(\sim80\!-\!100\,r_g\). At \(a\approx100\,r_g\), \(f_{\rm orb}\approx1.4\times10^{-6}\) Hz and \(f_{\rm GW}\approx2f_{\rm orb}\sim3\times10^{-6}\) Hz, just below the LISA band; as the orbit decays, the source can sweep upward into the \(10^{-4}-1\) Hz LISA window. The proposed observational strategy is systematic soft X-ray monitoring, high-cadence campaigns to measure \(N_H(t)\) and \(\tau(E,t)\), deeper X-ray spectroscopy and polarimetry, and the maintenance of localization and ephemeris for future low-frequency GW observations [2410.12090].

A different extension appears in AGN-disk dynamics. Hydrodynamical simulations and relativistic three-body calculations find that a gap-opening IMBH can drive a surrounding stellar-mass black hole to migrate synchronously until \(\sim10\) Schwarzschild radii from the central SMBH. In the gas-dominated regime,
\[
\frac{da}{dt}
=
\frac{2a}{L}\,[\Gamma_I(a)+\Gamma_{II}(a)+\Gamma_{\rm tide}(a)]
+\left(\frac{da}{dt}\right)_{\rm GW},
\qquad
\frac{de}{dt}
=
-\frac{e}{t_{\rm damp}}
+\left(\frac{de}{dt}\right)_{\rm GW},
\]
with Peters’ terms governing the relativistic GW contribution. Once gas torques weaken and the system enters the GW-dominated regime, the stellar-mass black hole can be either captured or kicked out by the IMBH, producing either two subsequent IMRIs or an EMRI followed by an IMRI. The paper defines the “cliffhanger” EMRI as the threshold between stable inspiral and ejection: when gas plus tidal torques exactly counteract GW-driven orbital decay, the semimajor axis lingers near \(a_{\rm balance}\) with \(da/dt\approx0\), numerically at \(a_{\rm balance}\sim(20\!-\!10)\,R_S\). In this state the inspiral time becomes very long, \(t_{\rm insp}\sim10^4\!-\!10^6\) yr, so perturbations such as \(\delta\Gamma\sim10\%\) from disk turbulence or weak third-body passages can tip the system toward ejection or renewed inspiral. The same study reports \(P_{\rm capture}\simeq10\!-\!30\%\) and \(P_{\rm eject}\simeq70\!-\!90\%\) per close passage for representative parameters [2411.16070].

These multimessenger and AGN-disk usages are broader than the failed-plunge loss-cone channel. They nonetheless preserve the defining structure of a system poised near a transition surface, where a modest perturbation determines whether the evolution terminates, stalls, or becomes a long-lived GW source.

## 7. Significance, open distinctions, and implications for LISA

The principal significance of cliffhanger EMRIs is that they break the classical dichotomy between EMRIs and direct plunges. The older picture assigned true EMRIs to \(a\ll a_c\) and plunges to \(a\gg a_c\). The failed-plunge mechanism and the PN Monte Carlo results both show that, for sufficiently small \(M_\bullet\), this dichotomy fails: \(S(a)\) approaches a finite plateau at large semimajor axis, and a substantial fraction of nominal plunges become long, GW-dominated inspirals instead [2304.13062; 2409.09122].

This has several consequences. First, it changes rate estimates. Depending on the occupation fraction of low-mass black holes, the volumetric EMRI rate can be boosted by up to \(\sim10\times\), while in specific steady-state cluster models the revision may be more modest, such as the \(\sim35\%\) increase found for \(M_\bullet=3\times10^5\,M_\odot\) [2304.13062; 2409.09122]. Second, it changes source properties. Cliffhanger EMRIs can enter the LISA band with high eccentricities and periapses \(r_p\sim10\,R_g\), implying a broader parameter distribution than in standard orbit-averaged treatments [2409.09122]. Third, it sharpens the importance of relativistic dynamics. Both the relativistic-loss-cone picture and the PN Monte Carlo calculations show that GR precession and burst-like periapse emission are not small corrections to a Newtonian capture problem; they restructure the phase-space boundaries that determine EMRI survival [1702.00597; 2409.09122].

A recurrent misconception is that “cliffhanger EMRI” refers to one and only one mechanism. The literature does not support that simplification. In the loss-cone channel it means a failed plunge; in Chimera it means a transient resonance imprint in \(h_{\ell m}(t)\); in AGN and X-ray work it can denote a candidate source or a metastable torque-balance state [1201.5715; 2410.12090; 2411.16070]. A plausible implication is that future usage will remain context-dependent unless the community standardizes the term. For LISA data analysis and source-population modeling, the most consequential meaning remains the low-\(M_\bullet\) failed-plunge channel, because it directly alters event rates, eccentricity distributions, and the partition between direct plunges and long-lived inspirals [2304.13062; 2409.09122].

Source: https://www.emergentmind.com/topics/cliffhanger-emri