Evaporation (Stellar Dynamics)

Evaporation in stellar dynamics is the secular, continuous escape of stars from a gravitationally bound stellar system—such as an open cluster, globular cluster, or nuclear star cluster—driven by cumulative two-body gravitational scattering that repopulates the high-velocity tail of the stellar speed distribution above the cluster's escape velocity (v_esc). First formulated quantitatively by Viktor Ambartsumian (1938) and Lyman Spitzer (1940), stellar evaporation is the gravitational analogue of the kinetic evaporation of high-speed molecules from a liquid droplet or planetary atmosphere. Whereas violent relaxation and phase mixing govern collisionless systems on a short dynamical crossing time, collisional star clusters evolve over tens to hundreds of half-mass relaxation times (t_rh) by shedding lower-mass halo stars across their outer King profile tidal boundary while their self-gravitating cores contract toward the gravothermal catastrophe.

In an isolated, self-gravitating cluster of total mass M and N stars with mean stellar mass m̄ = M / N in steady-state virial theorem equilibrium, the total kinetic energy K and gravitational potential energy W satisfy 2K + W = 0, so the mean-square three-dimensional stellar velocity ⟨v²⟩ is related to the mass-weighted mean gravitational potential ⟨Φ⟩ by:

⟨v²⟩ = 2K / M = −W / M = −½ ⟨Φ⟩

Meanwhile, a star situated at position x becomes gravitationally unbound from the cluster if its speed v reaches or exceeds the local escape speed v_esc(x), where its total specific orbital energy ε = ½ v² + Φ(x) equals zero:

v_esc²(x) = −2 Φ(x)

Taking the mass-weighted average of the squared escape speed across the entire cluster yields the fundamental Ambartsumian–Spitzer relation:

⟨v_esc²⟩ = −2 ⟨Φ⟩ = 4 ⟨v²⟩ ⇒ v_esc,rms = 2 v_rms

Thus, in any isolated virialized star cluster, the root-mean-square escape velocity is exactly twice the root-mean-square thermal velocity of the cluster stars (v_esc,rms = 2 v_rms).

As cumulative small-angle two-body gravitational encounters thermalize the stellar velocities on the half-mass relaxation timescale t_rh, they continuously drive the speed distribution f(v) toward a three-dimensional Maxwell–Boltzmann distribution with one-dimensional dispersion σ = v_rms / 3^1/2:

f_MB(v) dv = (4π)⁻¹ (2π σ²)^−3/2 4π v² exp(−v² / 2σ²) dv

Integrating the Maxwellian distribution over all speeds exceeding the RMS escape threshold v ≥ 2 v_rms = 12^1/2 σ ≈ 3.464 σ determines the fraction ξ_e of stars residing in the unbound high-velocity tail:

ξ_e = ∫_2vrms^∞ f_MB(v) dv = erfc(6^1/2) + 2 (6 / π)^1/2 exp(−6) ≈ 0.00053 + 0.00685 = 0.00738 (≈ 0.74%)

Assuming that two-body encounters replenish this 0.74% high-velocity tail once every half-mass relaxation time t_rh (as unbound stars escape across the cluster within a single crossing time t_cross ≪ t_rh), the classical Ambartsumian–Spitzer evaporation rate for an isolated single-mass cluster is:

dN / dt = −ξ_e N / t_rh ≈ −N / (136 t_rh)

which predicts a finite cluster evaporation lifetime of t_evap ≈ 100 to 140 t_rh.

In 1960, Michel Hénon pointed out a physical subtlety in the classical Ambartsumian–Spitzer calculation for an isolated cluster: because two-body gravitational encounters in a large-N system are dominated by distant, small-angle deflections (diffusive Fokker–Planck random walks in energy space) rather than close, large-angle two-body collisions, a star whose binding energy gradually diffuses upward toward ε = 0 does not immediately acquire v > v_esc in the core. Instead, it migrates onto an elongated, loosely bound halo orbit where the local density is low and two-body scattering virtually ceases, building an extended outer halo before escaping. Consequently, in a strictly isolated cluster without an external tidal field, diffusive evaporation is roughly an order of magnitude slower (ξ_e ≈ 0.00085 in Hénon's isolated homologous model) than the Maxwellian tail estimate.

In real galaxies such as the Milky Way, however, star clusters are embedded in the external gravitational potential Φ_gal(R) of the host galaxy. For a cluster of mass M_c orbiting at galactocentric radius R_G where the galaxy's angular speed is Ω_G and Oort constants define the local shear, the effective potential in the corotating frame possesses saddle points at the inner (L₁) and outer (L₂) Lagrange points located at the Jacobi (tidal) radius:

r_J = [ G M_c / (4 Ω_G² − κ_G²) ]^1/3 ≈ ( M_c / 3 M_gal )^1/3 R_G

This external tidal boundary accelerates stellar evaporation in three ways:

1. Depression of the Escape Energy Threshold: Instead of needing to climb to zero energy at r → ∞, any star scattered above the lowered Jacobi potential barrier Φ(r_J) becomes unbound, raising the diffusive escape rate by nearly two orders of magnitude to ξ_e ≈ 0.05 (so t_evap ≈ 20 t_rh).
2. Anisotropic Escape via Lagrange Funnels and Küpper Epicyclic Tails: Stars with energies slightly above the Jacobi threshold (E > E_J) cannot cross the zero-velocity surface everywhere; they must find one of the two narrow necks around the L₁ (inward toward the Galactic Center) and L₂ (outward toward the anticenter) Lagrange points. Once they exit L₁ or L₂ at low relative speeds (Δv ≈ 1–3 km s⁻¹), the Coriolis force and galactic differential shear stretch them into leading and trailing tidal tails exhibiting periodic epicyclic overdensities (Küpper et al., 2008).
3. Gravitational Tidal Shocks (Disk and Bulge Shocking): Whenever a globular cluster on an inclined eccentric orbit plunges through the dense Galactic disk midplane or passes pericenter near the Galactic bulge, the rapid time-varying tidal force (lasting Δt ≪ t_orb for outer halo stars) injects kinetic energy ΔE ∝ r² into outer cluster stars (Ostriker, Spitzer, & Chevalier, 1972), unbinding outer shells and accelerating core collapse.

Because two-body relaxation drives mass segregation toward kinetic energy equipartition (m₁⟨v₁²⟩ ≈ m₂⟨v₂²⟩), low-mass red dwarfs (m ≈ 0.1–0.4 M_☉) acquire higher thermal velocities than solar-mass turnoff stars (m ≈ 0.8 M_☉) or heavy compact remnants (1.2–10 M_☉) and are preferentially scattered into the outer halo near the tidal boundary r_J. Consequently, evaporation strips low-mass stars at a far higher rate than higher-mass stars. Deep Hubble Space Telescope and James Webb Space Telescope photometry of Milky Way globular clusters (such as Palomar 5, NGC 6397, and M4) and old open clusters (such as the Hyades and M67) confirms that highly evolved, tidally stripped clusters exhibit strongly flattened or inverted present-day stellar mass functions (dN/dm ∝ m^−α with α ≤ 0.5, compared to a canonical Kroupa/Salpeter initial slope of α ≈ 2.3), having lost 50% to 90% of their birth low-mass M-dwarf populations into the Galactic field halo and disk.

1. Spitzer, L. (1940). "The stability of isolated clusters." Monthly Notices of the Royal Astronomical Society, 100(5), 396–410. https://doi.org/10.1093/mnras/100.5.396
2. Hénon, M. (1961). "Sur l'évolution dynamique des amas globulaires." Annales d'Astrophysique, 24, 369–419. https://doi.org/10.1051/0004-6361/201117031
3. Gieles, M., Heggie, D. C., & Zhao, H. (2011). "The life cycle of star clusters in a tidal field." Monthly Notices of the Royal Astronomical Society, 413(4), 2509–2524. https://doi.org/10.1111/j.1365-2966.2011.18320.x
4. Baumgardt, H., & Makino, J. (2003). "Dynamical evolution of star clusters in tidal fields." Monthly Notices of the Royal Astronomical Society, 340(1), 227–246. https://doi.org/10.1046/j.1365-8711.2003.06286.x