Dynamical Ejection Scenario (DES) Overview

• Dynamical Ejection Scenario (DES) is a prompt ejection process where stars or matter are expelled on the system’s intrinsic dynamical timescale rather than via delayed mechanisms.
• In young clusters, close few-body interactions can eject massive stars with efficiencies up to 50%, altering mass functions and binary properties.
• In compact-object mergers, DES produces fast, anisotropic ejecta with distinct mass, velocity, and composition profiles that impact r-process nucleosynthesis and transient signals.

Searching arXiv for recent and core papers on “Dynamical Ejection Scenario” across its main astrophysical uses. Taken together, the usages represented in the literature suggest that the Dynamical Ejection Scenario (DES) is a family of prompt ejection channels in which stars or matter are expelled on the intrinsic dynamical timescale of a system rather than by a later secular process. In its classical stellar-dynamical sense, DES refers to close few-body encounters in dense young clusters that eject massive stars, binaries, and higher-order multiples. In compact-object merger work, the same term is used in relation to the prompt tidal- and shock-driven “dynamical ejecta” launched during coalescence, as distinct from neutrino-driven or viscous post-merger outflows. In binary-asteroid impact studies, the analogous problem is the dynamical fate of impact ejecta in a weak-gravity binary environment, where escape, reaccretion, temporary trapping, and resonant return are all possible (Oh et al., 2016, Sekiguchi et al., 2015).

1. Terminology and scope

In the classical runaway-star literature, DES is contrasted with the binary supernova scenario (BSS): a star is expelled from its natal cluster by a close encounter with a binary or another system, rather than by the supernova disruption of a binary. In merger studies, the same contrast is recast as prompt dynamical ejecta versus slower post-merger winds. In kinetic-impact studies, the relevant distinction is between prompt launch conditions and the later dynamical sorting of ejecta by gravity and radiation pressure (Sekiguchi et al., 2015, Kyutoku et al., 2017).

The acronym is not universal. In “A Magellanic Origin of the DES Dwarfs” the label “DES” refers to the Dark Energy Survey, not to a dynamical ejection scenario, and the paper is explicit that this is a different usage (Jethwa et al., 2016). This ambiguity is important because the same three letters occur in stellar dynamics, compact-object mergers, and survey astronomy with unrelated meanings.

A useful common structure nonetheless recurs across the supplied literature. DES is typically identified by four features: prompt launch, strong sensitivity to initial geometry, anisotropic ejecta, and a sharp distinction from later channels such as supernova kicks, neutrino-driven winds, viscous outflows, or long-term drag-dominated clearing. This suggests that “DES” functions less as a single mechanism than as a timing-and-driver classification.

2. Classical stellar DES in young clusters

In young-cluster dynamics, DES denotes the expulsion of massive stars through close few-body interactions before any supernova occurs. Direct $N$-body calculations for clusters with $M_{\rm ecl}=10^{3.5}\,M_\odot$ show that the initial density is the most influential parameter for the ejection fraction of massive systems: clusters with initial half-mass radius $0.1$ pc can eject up to 50% of their O-star systems on average, and clusters with $0.3$ pc can eject up to 30%. Most models show that the average ejection fraction decreases with decreasing stellar mass, the mass function of ejected stars becomes top-heavy, the retained cluster population becomes top-light, and the multiplicity fraction of ejected massive stars can be as high as 60%; massive high-order multiple systems can also be dynamically ejected (Oh et al., 2016).

The energetics of the classical mechanism are binary-driven. In the same study, the characteristic scaling for the ejection speed from a single-binary encounter is

$v_{\mathrm{ej} \propto \left(\frac{m}{a}\right)^{1/2},$

so tighter and more massive binaries act as more effective slingshots. The dependence on primordial mass segregation and binary demographics is therefore strong. Ordered pairing or a uniform massive-star mass-ratio distribution produces more efficient O-star ejection than random pairing, while compact clusters are much more effective than $r_{\rm h}=0.8$ pc systems, which eject hardly any OB stars by $3$ Myr (Oh et al., 2016).

A specialized version of the stellar DES was proposed by Kroupa and collaborators for the Orion Nebula Cluster: repeated all-OB-star ejection from a very compact, primordially mass-segregated core would temporarily quench feedback and permit renewed star formation, producing multiple stellar populations. Direct $N$-body tests indicate that standard ONC-like Plummer models are not efficient enough, but if $3$–$4$ OB stars are born with a maximum separation of about $M_{\rm ecl}=10^{3.5}\,M_\odot$0 pc, all OB stars have a high chance—about 50–70%—to escape from the center and not return within $M_{\rm ecl}=10^{3.5}\,M_\odot$1 Myr (Wang et al., 2018). This is a much narrower and more restrictive DES than the generic runaway-star channel.

Observationally, the Small Magellanic Cloud field-OB population provides one of the clearest tests of the classical scenario. Using RIOTS4 plus Gaia DR2, one study concluded that DES dominates over BSS by a factor of $M_{\rm ecl}=10^{3.5}\,M_\odot$2–3 in the SMC field runaway population, and that two-step ejections—dynamically ejected binaries later reaccelerated by supernovae—likely dominate the BSS-like runaway subset (Jones et al., 2020). An updated Gaia DR3 analysis lowered the median transverse velocity from 39 to 29 km s$M_{\rm ecl}=10^{3.5}\,M_\odot$3, but still found that among runaways the observed DES/BSS ratio is about 1.5, with a preferred model value of $M_{\rm ecl}=10^{3.5}\,M_\odot$4; non-compact binaries remain DES tracers, whereas OBe stars and HMXBs are more consistent with BSS (Phillips et al., 2024). These SMC studies are significant because they show that improved astrometry weakened the absolute velocity scale without overturning the qualitative case for DES dominance among runaways.

3. Runaway and hypervelocity stars as constraints

A central discriminant in the hypervelocity-star literature is the “arrival time,” the interval between stellar formation and ejection: $M_{\rm ecl}=10^{3.5}\,M_\odot$5 For disk runaway channels tied to massive stars—binary supernova disruption or dynamical interactions among massive stars in dense young clusters—the required ejection must occur within roughly $M_{\rm ecl}=10^{3.5}\,M_\odot$6–$M_{\rm ecl}=10^{3.5}\,M_\odot$7 yr, or at most about $M_{\rm ecl}=10^{3.5}\,M_\odot$8 Myr. By contrast, Galactic-center black-hole ejection allows much longer delays because stars may take $M_{\rm ecl}=10^{3.5}\,M_\odot$9–$0.1$0 Gyr to diffuse into the black hole’s loss cone (Brown et al., 2012).

HVS5 provides the clearest anti-DES case in the supplied literature. Keck HIRES spectroscopy shows that HVS5 is a rapidly rotating $0.1$1 main-sequence B star. Its age is

$0.1$2

its Galactic-center flight time is

$0.1$3

and its arrival time is therefore

$0.1$4

The paper is explicit that ejection during the first $0.1$5 yr of its lifetime is ruled out at the 3-$0.1$6 level. Flight times from elsewhere in the Galactic disk are similar, so a classical massive-star DES does not solve the timing problem (Brown et al., 2012).

The same study contrasts HVS5 with two unbound or marginally unbound disk runaways, HD 271791 and HIP 60350, whose arrival times are $0.1$7 Myr and $0.1$8 Myr, respectively. Those values are consistent with DES or binary-supernova ejection. The distinction is therefore not “bound versus unbound,” but whether the timing is compatible with a young massive-cluster environment. This addresses a common misconception: not all unbound B stars are products of the same ejection channel (Brown et al., 2012).

A broader Galactic-center constraint comes from a DESI search for slower, bound GC ejecta accumulated over the last $0.1$9 Gyr. That analysis looked for stars with high metallicity and small vertical angular momentum $0.3$0, whose distribution should peak at zero for isotropic GC launch. It found no statistically significant signal and inferred an upper limit on the average GC stellar ejection rate of

$0.3$1

over the past $0.3$2 Gyr (Verberne et al., 24 Jun 2025). This does not revive classical cluster DES for HVSs; rather, it constrains alternative dynamical ejection models centered on Sgr A$0.3$3, including Hills-type ejection and past massive-black-hole mergers.

4. Dynamical ejecta in binary neutron star mergers

In binary neutron star mergers, DES denotes matter expelled during the merger itself, on millisecond timescales, by tidal torques and shock heating rather than by later remnant-disk outflows. General-relativistic merger studies commonly identify dynamical ejecta with criteria such as

$0.3$4

or related Bernoulli-type conditions, and then characterize the asymptotic ejecta speed and composition (Sekiguchi et al., 2015, Fujibayashi et al., 2022).

Bauswein, Goriely, and Janka emphasized that relativistic DES in NS mergers is interface-dominated and shock-heated, rather than mainly cold tidal-tail stripping as in many Newtonian calculations. For $0.3$5–$0.3$6 binaries, they found ejecta masses of order $0.3$7–$0.3$8; for $0.3$9–$v_{\mathrm{ej} \propto \left(\frac{m}{a}\right)^{1/2},$0 systems, $v_{\mathrm{ej} \propto \left(\frac{m}{a}\right)^{1/2},$1–$v_{\mathrm{ej} \propto \left(\frac{m}{a}\right)^{1/2},$2; and characteristic velocities $v_{\mathrm{ej} \propto \left(\frac{m}{a}\right)^{1/2},$3–$v_{\mathrm{ej} \propto \left(\frac{m}{a}\right)^{1/2},$4. They also found that the ejecta robustly produce a nearly solar $v_{\mathrm{ej} \propto \left(\frac{m}{a}\right)^{1/2},$5-process pattern above $v_{\mathrm{ej} \propto \left(\frac{m}{a}\right)^{1/2},$6, with a stable $v_{\mathrm{ej} \propto \left(\frac{m}{a}\right)^{1/2},$7 ratio of about $v_{\mathrm{ej} \propto \left(\frac{m}{a}\right)^{1/2},$8–$v_{\mathrm{ej} \propto \left(\frac{m}{a}\right)^{1/2},$9 (Bauswein et al., 2013).

A full radiation-hydrodynamics study of equal-mass $r_{\rm h}=0.8$0 mergers with the SFHo, DD2, and TM1 equations of state showed a strong EOS dependence. Only for the soft EOS SFHo does the ejecta mass exceed $r_{\rm h}=0.8$1; the final mean electron fractions are $r_{\rm h}=0.8$2 for SFHo, $r_{\rm h}=0.8$3 for DD2, and $r_{\rm h}=0.8$4 for TM1, with a broad $r_{\rm h}=0.8$5 distribution extending roughly from $r_{\rm h}=0.8$6 to $r_{\rm h}=0.8$7. The study argues that shock heating and positron capture are the main causes of this broad $r_{\rm h}=0.8$8 distribution, while neutrino heating adds a smaller upward shift and a modest neutrino-driven component of order $r_{\rm h}=0.8$9 (Sekiguchi et al., 2015).

For mergers that leave only a short-lived massive neutron star before collapse, the dynamical ejecta remain the decisive heavy-element channel but become more sensitive to binary asymmetry. In SFHo models with $3$0, the dynamical ejecta masses are $3$1, $3$2, $3$3, and $3$4 as the mass ratio decreases from $3$5 to $3$6, while the mean dynamical-ejecta $3$7 drops from $3$8 to $3$9. The more massive $N$0, $N$1 case has $N$2 and $N$3. The paper’s central conclusion is that more asymmetric binaries make the DES component more tidal, colder, and more neutron rich, while post-merger ejecta compensate by supplying lighter $N$4-process nuclei (Fujibayashi et al., 2022).

The fast tail of BNS DES is not single-component. Full-GR Lagrangian simulations with SPHINCS_BSSN resolve two mechanisms for ejecta with $N$5: about 30% are “sprayed out” from the shear interface along the orbital plane, while the remaining $N$6 come from remnant “bounce back” after strong relativistic compression. The bounce component is more isotropic and faster by about $N$7, reaching resolved velocities up to $N$8, and can catch up with slower spray ejecta. Even a prompt-collapse case retains a fast tail of similar character, while slower ejecta are swallowed by the black hole (Rosswog et al., 2024). This two-component decomposition materially changes DES-based interpretations of shock breakout and late synchrotron afterglow.

5. Dynamical ejecta in black hole–neutron star mergers

In black hole–neutron star mergers, DES is even more explicitly tidal. Numerical-relativity studies find that the dynamical ejecta are driven primarily by tidal torque, are much more anisotropic than in BNS mergers, and are concentrated around the orbital plane with a half opening angle of 10–20 deg, often sweeping out only half of the plane. The ejecta mass can be as large as $N$9, with characteristic velocity $3$0–$3$1, and the remnant black hole–disk system can receive a kick of $3$2 km s$3$3 from ejecta linear momentum (Kyutoku et al., 2015).

Precessing NSBH simulations with a hot, composition-dependent DD2 EOS refine this picture. Disrupting systems produce

$3$4

with mean ejecta speeds of $3$5 for $3$6 black holes and $3$7 for the $3$8 disrupting case. The ejecta form a crescent with azimuthal opening angle $3$9 and vertical opening angle $4$0, and remain extremely neutron rich, $4$1 (Foucart et al., 2016). A major result of that study is that while ejecta mass is broadly consistent with earlier fitting formulae, the ejecta velocity is materially altered by the microphysical EOS treatment of cold, low-density neutron-rich matter.

The role of neutrino irradiation in this NSBH DES has been tested with full GR neutrino-radiation-hydrodynamics. For $4$2, $4$3 systems with SFHo, DD2, and TM1 EOS, the prompt dynamical ejecta masses at $4$4 ms after merger are approximately $4$5, $4$6, and $4$7, respectively, with velocities $4$8–$4$9 at $M_{\rm ecl}=10^{3.5}\,M_\odot$00 ms and asymptotic values $M_{\rm ecl}=10^{3.5}\,M_\odot$01–$M_{\rm ecl}=10^{3.5}\,M_\odot$02. The ejecta are equatorial, cold, low entropy, and have mean $M_{\rm ecl}=10^{3.5}\,M_\odot$03 between $M_{\rm ecl}=10^{3.5}\,M_\odot$04 and $M_{\rm ecl}=10^{3.5}\,M_\odot$05 (Kyutoku et al., 2017).

The same study shows that neutrino irradiation does not significantly modify the DES ejecta. The estimated absorption timescale to change the ejecta electron fraction is

$M_{\rm ecl}=10^{3.5}\,M_\odot$06

for $M_{\rm ecl}=10^{3.5}\,M_\odot$07 ms and $M_{\rm ecl}=10^{3.5}\,M_\odot$08 erg s$M_{\rm ecl}=10^{3.5}\,M_\odot$09, which is at least an order of magnitude longer than the expansion time. Direct reruns without neutrino absorption leave the ejecta mass unchanged within much less than numerical error and alter the mean $M_{\rm ecl}=10^{3.5}\,M_\odot$10 only minimally. Over the simulated interval, the neutrino-driven wind from the remnant disk is negligible compared with the prompt dynamical component (Kyutoku et al., 2017). In this regime, DES is therefore not merely present but dominant.

6. Impact ejecta in binary asteroids

In binary-asteroid impact studies, DES concerns the post-impact dynamical evolution of ejecta in a weak-gravity binary, especially the partition into escape, reaccretion, temporary orbiting, and long-lived survivors. A methodological study built a workflow that combines cratering scaling laws, SPH impact simulations, and large-$M_{\rm ecl}=10^{3.5}\,M_\odot$11 PKDGRAV integrations to determine what escapes, what falls back, and what lingers near the target. In SPH-derived initial conditions, that framework produced high-speed ejecta escaping at low angles of inclination and very slowly moving ejecta lofting off the surface at higher inclination angles, some of which re-impact; the authors explicitly noted that the realism of this slow lofted component remained under investigation (Schwartz et al., 2016).

A more complete dynamical model for the Didymos/Didymoon system showed that the binary environment fundamentally restructures the fate problem. For a full-scale AIDA-like simulation, the post-impact cloud after two weeks was partitioned into 71.1% escaped, 23.2% accreted, and 5.7% orbiting, and large debris on polar orbits had a survival advantage over smaller ejecta and low-latitude ejecta. Solar radiation pressure was found to be efficient in cleaning dust-size ejecta (Yu et al., 2016). This already implies that binary DES is strongly anisotropic and size dependent.

The follow-up study made the classification explicit by defining seven fate classes: early escape, late escape, early accretion on the primary, early reaccretion on the secondary, late accretion on the primary, late reaccretion on the secondary, and surviving orbit. Two broad dynamical mechanisms dominate. First, ejecta launched into mean-motion resonance with Didymoon produce long-term quasi-periodic showers back onto the secondary over at least a couple of weeks. Second, non-resonant ejecta produce a rapid and high re-accretion flux that occurs only once, because such particles leave the system unless they collide during their first encounter (Yu et al., 2017).

The speed interval

$M_{\rm ecl}=10^{3.5}\,M_\odot$12

is the dynamically rich regime in which temporary trapping, delayed reaccretion, and long-lived orbiting are all possible. All particles with $M_{\rm ecl}=10^{3.5}\,M_\odot$13 cm s$M_{\rm ecl}=10^{3.5}\,M_\odot$14 are early secondary fallback, while no particle with $M_{\rm ecl}=10^{3.5}\,M_\odot$15 cm s$M_{\rm ecl}=10^{3.5}\,M_\odot$16 reaccretes onto Didymoon. The launch site matters nearly as much as the speed: leading-side launches increasingly favor early escape, whereas leeward and inward-facing sites favor reaccretion and regolith-producing impacts (Yu et al., 2017).

A particularly important binary-specific result is the resonance-band approximation

$M_{\rm ecl}=10^{3.5}\,M_\odot$17

which identifies launch-site bands that feed $M_{\rm ecl}=10^{3.5}\,M_\odot$18 mean-motion resonances with Didymoon (Yu et al., 2017). In effect, binary-asteroid DES is not just an escape-speed problem. It is an orbital-dynamics problem in which companion swing-bys, resonant return, and solar-radiation-pressure sorting all materially control the later ejecta cloud. A plausible implication is that, for weak-gravity binaries, the low-speed tail of the ejecta distribution may dominate the physically interesting local consequences even when most of the total ejecta mass escapes.