Galaxy Mergers and Interactions

Two galaxies drifting toward one another are not two billiard balls. Each is an extended cloud of billions of stars and a still larger halo of dark matter, humming with internal orbital motion. When they pass close, the encounter cannot be scored on the two-body orbit alone, because that orbit does not keep its energy: tidal torques and a gravitational drag called dynamical friction bleed orbital kinetic energy steadily into the internal random motions of the stars. The encounter is inelastic — and that single fact, energy leaking from the orbit into heat, is what lets galaxies that would merely scatter as point masses instead spiral together and coalesce. This article assembles the analytic machinery of that coalescence: the drag law from a massive body plowing a wake through a sea of lighter stars, the merger clock from balancing that drag against a satellite's angular momentum, and the survival-or-disruption verdict from the impulse approximation for fast fly-bys. Two dimensionless ratios organize everything: the encounter speed measured against the internal dispersion, v/σ, which decides whether a passage merges or merely harasses; and the mass ratio M₁/M₂, which decides whether the remnant is a rebuilt spheroid or a thickened survivor. The frequency of such events, integrated over cosmic time, is the engine of hierarchical structure formation — today's massive galaxies are the accreted debris of countless smaller progenitors, and the same physics that sinks a satellite also funnels gas to feed starbursts and quasars, strips dwarfs into tidal streams, and drives supermassive black holes toward their own gravitational-wave coalescence.

Key takeaways

A galaxy encounter is inelastic: tidal torques and dynamical friction drain orbital energy into internal random motion, so extended systems merge where point masses would only scatter.
Dynamical friction is a gravitational wake — a massive body focuses background stars into a trailing overdensity that pulls it back; the drag grows as M² and falls as 1/v².
Chandrasekhar's formula gives the merger clock t_df ≈ 1.17 r_i² v_c / (G M lnΛ): heavier satellites sink faster, so major mergers finish first and set the ordering of hierarchical growth.
Whether a passage merges is set by v/σ: slow near-parabolic encounters (groups) raise tidal tails and coalesce; fast fly-bys (clusters) deliver only a weak impulsive kick, ΔE ∝ M_p²/(b⁴v²), and galaxies survive.
Major disk mergers destroy ordered rotation and re-virialize to 2K + W = 0 as dispersion-supported ellipticals; minor mergers strip satellites beyond their Jacobi radius and thicken disks.
Key idea: an encounter between extended stellar systems is inelastic — tidal forces and dynamical friction bleed orbital energy into internal random motion, so galaxies that pass slowly enough relative to their internal dispersion inevitably decay and merge rather than scatter freely.

Where this lives today

Mergers are now watched, simulated, and heard. Large cosmological hydrodynamic simulations — IllustrisTNG, EAGLE, FIRE — resolve mergers galaxy by galaxy, with Chandrasekhar's friction acting on live particles, and reproduce the full pageant from first passage to relaxed remnant, matching observed elliptical scaling relations and merger-rate statistics. In our own Galaxy the Gaia mission has turned the abstraction into archaeology: the velocity-space debris of the Gaia-Sausage/Enceladus event records a major accretion some 8–10 billion years ago, while the Sagittarius stream is a dwarf caught mid-disruption, its leading and trailing arms tracing the dark halo exactly as the Jacobi-radius picture predicts. At the endpoint of major mergers the two central supermassive black holes sink by dynamical friction toward a bound binary; pulsar timing arrays are now detecting the stochastic gravitational-wave hum from the whole population of such binaries, and the planned LISA observatory will hear individual coalescences. The open "final-parsec problem" — whether these binaries actually merge or stall — is a direct descendant of the same drag physics.

The physical picture

Picture a heavy body driving through a crowd. Everyone it passes is tugged toward its track and, arriving late, piles up just behind it — a trailing overdensity, a gravitational wake — that pulls the intruder backward. This is dynamical friction: the body is decelerated not by contact but by the collective gravity of the crowd it has stirred up. Reverse the logic and the scalings fall out. A heavier intruder gathers a denser wake, so the drag grows with its mass; the wake's pull back on the intruder is again proportional to its mass, so the retarding force scales as M². And a faster intruder outruns the pile-up before it can fully form, so the drag weakens as it speeds up. The first figure freezes this picture and attaches the law to it.

Figure

A massive body of mass M moving through a sea of lighter stars gravitationally focuses them into a trailing overdensity, the wake, whose backward pull is the drag F_df ≈ 4π G² M² ρ lnΛ / v² — growing as M² and falling as 1/v²

Whether a given passage ends in a merger is a race between two clocks: the encounter clock, roughly the pericentre-passage time, and the internal clock, a stellar orbital period inside each galaxy. When the encounter is slow — the relative speed v small compared with the internal dispersion σ — the stars respond resonantly over many internal periods, orbital energy drains, tidal tails are raised, and the galaxies merge within a few dynamical times. When it is fast, v ≫ σ, each galaxy receives only a brief impulsive kick and sails on largely intact. Distinct from this yes-or-no question is how fast a bound, already-captured satellite then spirals in: that is set by the drag law and depends steeply on the satellite's mass. The second figure turns that dependence into the sinking time, showing why massive satellites merge within a Hubble time while dwarfs survive far longer.

Figure

Dynamical friction sinking time t_df versus satellite mass for a circular orbit at r_i = 50 kpc in a v_c = 220 km/s halo: a 10¹⁰ M☉ satellite merges in ≈ 3.6 Gyr, well inside a Hubble time, while a 10⁸ M☉ dwarf needs ≈ 170 Gyr and survives — massive satellites sink first

Tides themselves are just differential gravity. During a close passage the near side of a disk is pulled harder than the far side, and the mismatch stretches the galaxy: a long tidal tail is flung off the outer, anti-companion side while a bridge reaches toward the companion — the antennae seen in interacting pairs, the visible fossil of angular momentum and energy handed from the orbit to the stars. When a merger finally completes it is inelastic: the orbital kinetic energy is converted into the random internal motion of the merged system, heating it, with some carried off by escaping stars and stripped debris. The remnant re-virializes into a hotter, rounder, dispersion-supported spheroid — the physical reason two spiral disks can merge into a single elliptical.

The core equations

The central result is Chandrasekhar's dynamical friction formula. A massive body moving through a Maxwellian background of lighter stars feels a deceleration that is the summed reaction of all the two-body deflections it induces, weighted so that only stars slower than the body contribute net drag.

\frac{\mathrm{d}\mathbf{v}_M}{\mathrm{d}t} = -\frac{4\pi G^2 M \rho \ln\Lambda}{v_M^3}\left[\mathrm{erf}(X) - \frac{2X}{\sqrt{\pi}}\,e^{-X^2}\right]\mathbf{v}_M \qquad (1)

where: $M$ — mass of the intruder (satellite); $\mathbf{v}_M$ — its velocity relative to the background; $\rho$ — density of the light background particles; $\ln\Lambda$ — Coulomb logarithm, $\Lambda \approx b_{\max}/b_{\min}$ ; $X \equiv v_M/(\sqrt{2}\,\sigma)$ — speed in units of the background dispersion; $\mathrm{erf}$ — error function. Reading it: the drag is antiparallel to the motion, scales as $M$ in the acceleration (so $M^2$ in the force), is independent of the star mass $m$ , and is throttled by the bracket, which vanishes as $v_M \to 0$ and saturates to $1$ at high speed. Taking that high-speed saturation gives the clean force law for a fast intruder and exposes the $1/v^2$ falloff directly.

F_{\mathrm{df}} \simeq \frac{4\pi G^2 M^2 \rho \ln\Lambda}{v_M^2} \qquad (v_M \gg \sigma) \qquad (2)

where: $F_{\mathrm{df}}$ — magnitude of the drag force; $M$ — intruder mass (note the $M^2$ ); $v_M$ — intruder speed. Reading it: faster intruders feel weaker friction, so an orbit decays slowest where it moves fastest. Balancing that drag against a satellite's angular momentum in a flat-rotation-curve host gives the merger clock — the time for a circular orbit to spiral in.

t_{\mathrm{df}} \approx \frac{1.17}{\ln\Lambda}\,\frac{r_i^2\, v_c}{G M} \qquad (3)

where: $t_{\mathrm{df}}$ — sinking time; $r_i$ — initial orbital radius; $v_c$ — host circular velocity (constant for a singular isothermal sphere); $\ln\Lambda \approx \ln(M_{\mathrm{host}}/M)$ . Reading it: the clock runs inversely with satellite mass, so heavy satellites merge quickly and light ones linger — the origin of hierarchical growth ordering. For a fast fly-by the perturber is a point flashing past on a straight line; integrating its transverse pull along that line gives each star a velocity kick.

\Delta \mathbf{v} = \frac{2 G M_p}{b\,v}\,\hat{\mathbf{b}} \qquad (4)

where: $M_p$ — perturber mass; $b$ — impact parameter; $v$ — relative speed; $\hat{\mathbf{b}}$ — unit vector toward closest approach. Reading it: the kick is uniform only to leading order — it is the variation across the galaxy that heats rather than merely displaces it. Summing the kinetic energy of the differential (tidal) part of those kicks over all stars gives the internal energy injected by a fast encounter.

\Delta E = \frac{4}{3}\,\frac{G^2 M_p^2}{b^4 v^2}\,M_{\mathrm{gal}}\,\langle r^2\rangle \qquad (5)

where: $\Delta E$ — internal energy deposited; $\langle r^2\rangle$ — mean-square stellar radius; $M_{\mathrm{gal}}$ — mass of the perturbed galaxy. Reading it: the steep $b^{-4}$ and $v^{-2}$ mean only close, slow passages disrupt — distant or fast ones barely warm the system. Finally, a satellite is stripped wherever the host tide overwhelms its self-gravity — outside the tidal (Jacobi) radius.

r_t \approx r_p\left(\frac{m}{(2+e)\,M(<r_p)}\right)^{1/3} \qquad (6)

where: $r_t$ — tidal radius (stars beyond it are stripped); $r_p$ — pericentric distance; $m$ — satellite mass within $r_t$ ; $M(<r_p)$ — host mass enclosed at pericentre; $e$ — orbital eccentricity. Reading it: the satellite is peeled from the outside in, its stripped stars fanning into leading and trailing streams that trace the host potential.

Derivation AChandrasekhar dynamical friction from two-body deflections

The goal is to build the drag law in equation (1) from the ground up: treat the intruder as scattering a sea of lighter stars one encounter at a time, sum the momentum it loses over all impact parameters, then average over the Maxwellian background so that only the stars it actually overtakes contribute. The payoff is to see exactly where each piece comes from — the $M^2$ force, the Coulomb logarithm, and the error-function bracket.

Step 1. A background star of mass $m$ passing at impact parameter $b$ with relative speed $v_{\mathrm{rel}}$ is deflected, gaining $\Delta v_{\parallel} = (2GM/b\,v_{\mathrm{rel}})\,[1+(b v_{\mathrm{rel}}^2/GM)^2]^{-1}$ toward the intruder's path. Why: by Newton's third law the reaction on the intruder is equal and opposite, so every star it focuses pulls it backward — the microscopic origin of the drag.

Step 2. The momentum handed to that star is $m\,\Delta v_{\parallel}$ , and by conservation the intruder recoils opposite to $\mathbf{v}_{\mathrm{rel}}$ . Why: the wake's amplitude is set by $M$ and the pull on the intruder is again set by $M$ , so two factors of $M$ appear even though a single star's mass $m$ drops out of the final acceleration.

Step 3. Encounters arrive at a rate $n\,v_{\mathrm{rel}}\,2\pi b\,\mathrm{d}b$ ; integrating the momentum transfer over $b$ produces $\int \mathrm{d}b/b = \ln(b_{\max}/b_{\min}) = \ln\Lambda$ , cut off below at the ninety-degree deflection radius $b_{\min}\sim GM/v^2$ and above at the system size $b_{\max}$ . Why: distant encounters are individually feeble but far more numerous, and their equal logarithmic weighting gives the drag its characteristic $\ln\Lambda$ factor of order a few.

Step 4. Averaging over an isotropic Maxwellian $f(v_a)\propto e^{-v_a^2/2\sigma^2}$ keeps only field stars slower than the intruder — faster ones cancel — leaving $\langle\dot{\mathbf v}_M\rangle \propto \int_0^{v_M} f(v_a)v_a^2\,\mathrm{d}v_a$ , which integrates to the bracket $[\mathrm{erf}(X)-\tfrac{2X}{\sqrt\pi}e^{-X^2}]$ with $X=v_M/(\sqrt2\sigma)$ . Why: this is the content of "you are only slowed by the crowd you outrun", and it makes the drag vanish for a very slow intruder and saturate for a very fast one.

Step 5. Assembling the pieces with $\rho=nm$ gives equation (1) exactly, confirming both the $M^2$ force scaling and the velocity bracket. Why: this reproduces Chandrasekhar's 1943 result and Binney and Tremaine's equation 8.7 from first principles. *Verified: matches Binney and Tremaine (2nd ed.) eq. 8.7 and Chandrasekhar (1943); the bracket evaluates to 0.428 at X = 1.*

Go deeper · Derivation B · 1 minOrbital decay time in a singular isothermal halo

Derivation BOrbital decay time in a singular isothermal halo

The goal is to convert the drag law into a concrete merger clock by following a circular-orbit satellite as it loses angular momentum to friction in a flat-rotation-curve host. A singular isothermal sphere has density $\rho(r)=v_c^2/(4\pi G r^2)$ , dispersion $\sigma=v_c/\sqrt2$ , and a circular satellite at $v_M=v_c$ , fixing $X=v_c/(\sqrt2\,\sigma)=1$ so the error-function bracket holds the constant value 0.428 throughout the orbit — the reason the calculation closes in elementary form.

Step 1. Evaluate the bracket once and for all: $\mathrm{erf}(1)-\tfrac{2}{\sqrt\pi}e^{-1}=0.8427-0.4151=0.428$ , a fixed number for any circular orbit in an isothermal background. Why: because $X$ stays at 1 as the satellite spirals in, this factor never changes, and the equation of motion has constant coefficients.

Step 2. Substitute the isothermal density into the drag force (2): $F_{\mathrm{df}}=0.428\times 4\pi G^2 M^2 \rho\ln\Lambda/v_c^2 = 0.428\,G M^2 \ln\Lambda/r^2$ , the $4\pi$ and $v_c^2$ cancelling. Why: the $1/r^2$ profile makes the drag torque tractable, turning the problem into a simple radial ODE.

Step 3. The angular momentum is $L=M v_c r$ with $v_c$ constant, so the drag torque gives $M v_c\,\mathrm{d}r/\mathrm{d}t = -0.428\,G M^2\ln\Lambda/r$ . Why: on a flat rotation curve the circular speed is fixed as the orbit shrinks, so all lost angular momentum shows up as shrinking radius.

Step 4. Separate and integrate, $\int_{r_i}^0 r\,\mathrm{d}r = -0.428\,(GM\ln\Lambda/v_c)\int_0^{t_{\mathrm{df}}}\mathrm{d}t$ , giving $\tfrac12 r_i^2 = 0.428\,(GM\ln\Lambda/v_c)\,t_{\mathrm{df}}$ and hence $t_{\mathrm{df}}=r_i^2 v_c/(0.856\,GM\ln\Lambda)=1.17\,r_i^2 v_c/(GM\ln\Lambda)$ . Why: the coefficient 1.17 is exactly $1/(2\times0.428)$ , so it traces directly back to the bracket value — a clean internal check.

Step 5. Read off the scaling: $t_{\mathrm{df}}\propto 1/M$ , so a satellite ten times heavier sinks ten times faster. Why: this inverse dependence orders hierarchical assembly — massive companions merge promptly while dwarfs orbit for many Hubble times. *Verified: reproduces Binney and Tremaine (2nd ed.) eq. 8.24; coefficient 1.17 = 1/(2 × 0.428).*

Go deeper · Derivation C · 1 minImpulsive heating and why fast fly-bys spare galaxies

Derivation CImpulsive heating and why fast fly-bys spare galaxies

The goal is to show that a fast encounter injects internal energy scaling as $M_p^2/(b^4 v^2)$ , so only slow, close passages can unbind a galaxy while fast cluster fly-bys leave it nearly untouched. Treat the galaxy as a frozen collection of stars at positions $\mathbf r_i$ — frozen because the perturber crosses in far less than an internal orbital time — give each the impulse of equation (4), and keep only the part that heats rather than translates.

Step 1. Each star receives $\Delta\mathbf v_i = (2GM_p/b_i v)\,\hat{\mathbf b}_i$ from the straight-line passage, where $b_i$ is that star's own impact parameter. Why: in the fast limit the perturber's trajectory is undeflected and every star simply integrates the transverse tug, so the kick is fully determined by geometry.

Step 2. Subtract the mean kick, $\delta\mathbf v_i = \Delta\mathbf v_i-\langle\Delta\mathbf v\rangle$ ; the mean merely displaces the galaxy's centre of mass and carries no internal energy. Why: only the spread of kicks across the galaxy — the tidal, differential part — can raise the internal energy, and expanding to first order in $r_i/b$ gives $\delta v\propto r_i/b^2$ .

Step 3. Sum the kinetic energy of those residual kicks, $\Delta E=\tfrac12\sum_i m_i|\delta\mathbf v_i|^2$ ; with the tidal gradient $\sim 2GM_p/(b^2 v)$ and an isotropic average one obtains $\Delta E=\tfrac{4}{3}G^2M_p^2 M_{\mathrm{gal}}\langle r^2\rangle/(b^4 v^2)$ . Why: the two powers of the tidal gradient give the $b^{-4}$ and $v^{-2}$ signatures, and the galaxy's extent enters through $\langle r^2\rangle$ .

Step 4. Interpret the scaling: at fixed $b$ , doubling $v$ cuts the heating fourfold, so fast cluster encounters deposit little energy and galaxies survive, while slow group encounters can deliver energy comparable to the binding energy and disrupt. Why: this single scaling law divides the sky — harassment in clusters, mergers in groups — and follows entirely from the frozen-star impulse picture. *Verified: consistent with Spitzer (1958) and Binney and Tremaine (2nd ed.) §8.2.*

Limiting cases

Outcomes are governed by the encounter speed relative to the internal dispersion, $v/\sigma$ , and by the mass ratio of the two systems.

Case	Result	Meaning
Fast fly-by	$v \gg \sigma$	Impulse approximation holds; internal energy rises only as $\Delta E \propto M_p^2/(b^4 v^2)$ and the galaxies do not merge. Typical in rich clusters, where repeated such kicks cause gradual harassment.
Slow / near-parabolic	$v \lesssim \sigma$	Tides act over many internal periods; strong resonant energy transfer raises tidal tails and drives merger within a few dynamical times. Typical in groups.
Major merger	$M_1/M_2 \sim 1$	Violent relaxation randomizes stellar orbits, disks are destroyed, and the remnant is a dispersion-supported, often triaxial spheroid — a classical elliptical.
Minor merger	$M_1/M_2 \gg 1$	The satellite sinks by dynamical friction and is tidally stripped, heating and thickening the host disk and building bulges and stellar halos without destroying the host.
Very slow intruder	$v_M \ll \sigma$	Bracket $\to \tfrac{4}{3\sqrt\pi}X^3$ , so $F_{\mathrm{df}}\propto v_M$ ; the drag becomes a simple linear resistance, like a body settling through a viscous fluid.

Worked examples

1 · Will a 10¹⁰ M☉ satellite merge within a Hubble time? (fully worked)

A satellite of M = 10¹⁰ M☉ is on a circular orbit at r_i = 50 kpc in a Milky-Way-like halo with flat rotation v_c = 220 km/s. Using G = 4.30×10⁻⁶ kpc (km/s)² M☉⁻¹, find the sinking time and decide its fate.

Step 1. Fix the Coulomb logarithm. The isothermal host mass inside the orbit is $M(<r_i)=v_c^2 r_i/G = (220)^2\times 50/(4.30\times10^{-6}) = 5.63\times10^{11}\,M_\odot$ , so $\ln\Lambda \approx \ln(M_{\mathrm{host}}/M) = \ln(56.3) = \mathbf{4.03}$ .

Step 2. Assemble the kinematic factor: $r_i^2 v_c/(GM) = (2500)(220)/(4.30\times10^{-6}\times10^{10}) = 12.79$ in kpc/(km/s), and $1\ \mathrm{kpc/(km/s)} = 0.978\ \mathrm{Gyr}$ .

Step 3. Apply equation (3): $t_{\mathrm{df}} = (1.17/4.03)\times 12.79 \times 0.978\ \mathrm{Gyr} = 0.290\times 12.51\ \mathrm{Gyr} = \mathbf{3.6\ Gyr}$ . Since this is far shorter than the ≈ 13.8 Gyr age of the Universe, the satellite spirals in and merges — this is the marked point on the second figure.

Step 4. Check the scaling. Because $t_{\mathrm{df}}\propto 1/M$ (only the weak logarithm varying), a ten-times-lighter 10⁹ M☉ dwarf gives $\ln\Lambda=6.33$ and $t_{\mathrm{df}}\approx \mathbf{23\ Gyr}$ — longer than a Hubble time, so it survives as a satellite. The heavy companion merges; the light one orbits on. t_df(10¹⁰) ≈ 3.6 Gyr merges; t_df(10⁹) ≈ 23 Gyr survives.

2 · Tidal radius of a stripped dwarf (fully worked)

A dwarf of m = 10⁸ M☉ reaches pericentre at r_p = 20 kpc on a near-circular orbit (e ≈ 0) in the same v_c = 220 km/s halo. Where is it stripped?

Step 1. Enclosed host mass at pericentre: $M(<r_p)=v_c^2 r_p/G = (220)^2\times 20/(4.30\times10^{-6}) = \mathbf{2.25\times10^{11}\,M_\odot}$ .

Step 2. Insert into equation (6) with $e=0$ : $r_t = r_p\,[\,m/(2M(<r_p))\,]^{1/3} = 20\,[\,10^8/(4.50\times10^{11})\,]^{1/3}\ \mathrm{kpc}$ .

Step 3. Evaluate the ratio and cube root: $10^8/4.50\times10^{11} = 2.22\times10^{-4}$ , cube root $0.0606$ , giving $r_t = 20\times 0.0606 = \mathbf{1.2\ kpc}$ . Stars orbiting beyond ≈ 1.2 kpc from the dwarf's centre are torn off at each pericentre and fanned into leading and trailing tidal streams that trace the Galactic potential — the morphology of the Sagittarius stream. r_t ≈ 1.2 kpc.

3 · Why a cluster fly-by barely heats a galaxy (estimate)

Two galaxies pass at cluster speed v = 1500 km/s, impact parameter b = 20 kpc, with M_p = 10¹¹ M☉ and a victim of M_gal = 10¹¹ M☉, ⟨r²⟩ ≈ (5 kpc)², dispersion σ ≈ 150 km/s. How badly is it heated?

Step 1. Specific heating from equation (5): $\Delta E/M_{\mathrm{gal}} = \tfrac{4}{3}G^2 M_p^2\langle r^2\rangle/(b^4 v^2) \approx \mathbf{17\ (km/s)^2}$ .

Step 2. Compare with the binding energy per unit mass, of order $\sigma^2 = (150)^2 = 2.25\times10^4\ (\mathrm{km/s})^2$ . The fractional heating is $17/22500 \approx 8\times10^{-4}$ — less than a tenth of a percent. The galaxy sails through essentially intact, and only the accumulation of many such kicks over a Hubble time (harassment) slowly transforms it. ΔE/|E_bind| ≈ 0.08% per fast passage.

Physics of the portrait

Dynamical friction — the quiet drag that makes galaxies sink and merge — is the invention of a single 1943 paper, and its author is one of the towering figures of twentieth-century astrophysics.

Origins · 1 minSubrahmanyan Chandrasekhar and how this was discovered

Subrahmanyan Chandrasekhar

1910–1995 · stellar dynamics and gravitational encounters

Subrahmanyan Chandrasekhar was a theorist of extraordinary range who worked each subject to exhaustion and then moved on. Born in Lahore in 1910 and educated at Cambridge, he is most famous for the Chandrasekhar limit — the maximum mass of a white dwarf, about 1.4 solar masses — which earned him a share of the 1983 Nobel Prize. But in the early 1940s he turned to stellar dynamics, and in a landmark 1943 paper he asked how a star moving through a stellar system is slowed by its interactions with all the others. His answer, dynamical friction, showed that a massive body deflects the field stars into a trailing overdensity whose collective pull decelerates it — a drag proportional to the body's mass and inversely to the square of its speed, mediated by the error-function bracket and Coulomb logarithm that still bear his name. The result sat largely dormant until the era of galaxy formation made it indispensable: it is the clock that times the decay of satellite orbits, the sinking of massive black holes to galactic centres, and the whole ordering of hierarchical assembly — robust enough to survive intact into the age of cosmological simulation, half a century after he wrote it down.

History

1943

— Chandrasekhar derives dynamical friction from two-body encounters

1958

— Spitzer analyzes the impulsive tidal disruption of star clusters

1972

— Toomre and Toomre reproduce bridges and tails in interacting galaxies

1977

— Toomre proposes that major mergers of disks build elliptical galaxies

1978

— White and Rees frame galaxy formation as hierarchical clustering

Going further

Texts. Binney and Tremaine, Galactic Dynamics (2nd ed.), Ch. 8 (Collisions and Encounters of Stellar Systems) — the definitive treatment of dynamical friction and impulsive encounters · Carroll and Ostlie, An Introduction to Modern Astrophysics (2nd ed.), Ch. 26 (Galactic Evolution) · Mo, van den Bosch and White, Galaxy Formation and Evolution, Ch. 12 (satellite dynamics, mergers) · Chandrasekhar (1943), ApJ 97, 255 (Dynamical Friction I) · Toomre and Toomre (1972), ApJ 178, 623 (Galactic Bridges and Tails) · Dodelson, Modern Cosmology, chapters on hierarchical growth and the halo model.

Provenance. All equations were cross-checked against Binney and Tremaine (2nd ed.) Ch. 8: the dynamical friction formula (eq. 8.7), the decay time (eq. 8.24), and the impulse approximation and tidal heating (§8.2). A verifier should confirm: (1) the error-function bracket evaluates to 0.428 at X = 1 using erf(1) = 0.8427; (2) the coefficient 1.17 = 1/(2 × 0.428) from the isothermal decay integral; (3) the impulse kick Δv = 2GM_p/(bv) from integrating the transverse force; (4) the worked merger clock t_df ≈ 3.6 Gyr for M = 10¹⁰ M☉, r_i = 50 kpc, v_c = 220 km/s, lnΛ = 4.0; (5) the Jacobi radius r_t ≈ 1.2 kpc for m = 10⁸ M☉ at r_p = 20 kpc. Qualitative merger phenomenology follows Carroll and Ostlie Ch. 26 and Toomre and Toomre (1972); hierarchical-growth context from Mo, van den Bosch and White, and Dodelson. No non-standard numerical values are asserted.

Builds on: Gravitational N-Body Problem · Collisionless Boltzmann Equation · Virial Theorem in Stellar Systems · Two-Body Relaxation Time · Galactic Potential Models · Gravitational Two-Body Problem. Leads to: Jeans Equations · Orbits in Axisymmetric and Triaxial Potentials · Action-Angle Variables.

Ask the tutor: Ask me to walk through the dynamical friction bracket, estimate the sinking time or tidal radius for any satellite you name, or show why a slow major merger of two disks re-virializes into an elliptical.

Why does an encounter between galaxies not conserve the orbital energy?

Because a galaxy is not a point — it is a huge reservoir of internal orbital motion. During a passage, tidal torques do work on the individual stars and dynamical friction transfers momentum from the orbit into a gravitational wake, so orbital kinetic energy is continuously converted into internal random motion. The total energy is of course conserved; what is not conserved is the energy in the two-body orbital degrees of freedom alone. That leakage is exactly why two galaxies on what looks like an unbound hyperbolic orbit can nonetheless fall back and merge.

What is the difference between a merger and an interaction?

An interaction is any close gravitational encounter — a fly-by that raises tidal tails, distorts disks, and exchanges some energy and angular momentum but leaves the two galaxies as separate systems, at least for now. A merger is the endpoint in which the two systems coalesce into a single bound remnant. Every merger begins as an interaction; whether an interaction becomes a merger depends on the ratio v/σ. A slow, near-parabolic passage (as in a galaxy group) loses enough orbital energy to friction and tides that the pair falls back and merges within a few orbits, while a fast passage (as in a cluster) survives as a repeated interaction.

Why do heavier satellites sink faster?

The drag force scales as M² — a heavier satellite builds a denser wake and is pulled back harder — so its deceleration scales as M, and the merger clock for a circular orbit becomes t_df ∝ 1/M: a satellite ten times more massive sinks in roughly a tenth of the time. Physically, a heavy companion carves out and drags a large overdensity, shedding angular momentum quickly, while a light dwarf barely disturbs the background and orbits for many Hubble times. This inverse dependence is what orders hierarchical growth: the big pieces merge first.

Why do fast fly-bys spare galaxies but slow ones destroy them?

The internal energy deposited scales as ΔE ∝ M_p²/(b⁴v²). The v⁻² factor makes each fast kick weak — a galaxy in a rich cluster, where relative speeds reach a thousand kilometres per second, is heated by well under a percent of its binding energy per passage. A slow encounter, by contrast, lets the tidal field act coherently over many internal orbital periods, so the transfer is resonant and can rival the binding energy, disrupting disks and driving coalescence. This is why mergers happen in groups while clusters instead grind galaxies down by the slow accumulation of many weak kicks — harassment.

Why does a major merger of two spiral disks make an elliptical?

Because it destroys the ordered rotation that defines a disk. In a slow major merger, violent relaxation — the rapid, collective scrambling of stellar orbits in a wildly time-varying potential — randomizes the stellar velocities. The ordered angular momentum of the progenitor disks is cancelled or expelled with tidal debris, and what remains is supported against gravity by random motion rather than rotation. When it re-virializes to 2K + W = 0 it is a hotter, rounder, dispersion-supported spheroid — a classical elliptical, often triaxial, sometimes retaining a kinematically decoupled core that remembers the progenitors' orbital angular momentum.

What sets the cutoffs in the Coulomb logarithm lnΛ?

It is lnΛ = ln(b_max/b_min), the ratio of the largest to smallest relevant impact parameters. The maximum b_max is roughly the host size or the satellite's orbital radius, beyond which the background is no longer a useful infinite homogeneous medium; the minimum b_min is the larger of the satellite's own size or the ninety-degree deflection radius GM/v², below which the point-mass treatment breaks down. For a merging satellite this reduces to lnΛ ≈ ln(M_host/M), of order a few — about 4 in the worked example. Because the drag depends only logarithmically on these choices, an order-of-magnitude estimate suffices.