Common Envelope Evolution

A star swallows its companion — and they survive it

Take two stars born close together, one a little heavier than the other. The heavier one burns through its hydrogen first, climbs the red-giant branch, and balloons outward by a factor of a hundred or more. At some point its swelling envelope reaches across the gap and the companion star is simply inside it — orbiting through the outer layers of a giant whose core it is now nested within. The two dense cores — the giant's degenerate core and the intact companion — are sharing a single, common envelope of gas. This is the configuration that gives common envelope evolution its name, and what happens next is one of the fastest, most consequential transformations in stellar astrophysics.

The shared envelope does not rotate in lockstep with the embedded cores. The companion plows through gas that is moving more slowly than it is, and that gas drags on it like an atmosphere drags on a satellite skimming the top of a planet. The drag — partly hydrodynamic, partly gravitational torque from the wake the companion raises — saps orbital energy and angular momentum. With less energy the cores fall closer together, which makes them orbit faster, which increases the drag, which removes energy faster still. The result is a runaway inspiral: a death spiral that takes the cores from a separation of an astronomical unit or more down to a small fraction of that in roughly a thousand years. The orbital energy released in the plunge is dumped into the envelope, and if there is enough of it, the envelope is heaved off into space, leaving behind a tight binary of two compact cores — or, if there is not enough, a single merged star.

Bohdan Paczyński articulated the idea in 1976 to explain a puzzle: we observe close binaries — cataclysmic variables, short-period white-dwarf pairs — whose components are smaller than the orbit that separates them, yet at least one of them was once a giant far larger than that orbit. They could not have formed where we find them. Common envelope evolution is the mechanism that takes a wide binary, briefly merges the two stars into a shared cocoon, and then ejects the cocoon to leave the survivors in an orbit no giant could ever have fit inside.

How the spiral-in works

The phase begins with unstable mass transfer. In a close binary, each star is bounded by its Roche lobe — the teardrop-shaped region inside which gas is gravitationally bound to that star. When the evolving primary fills its Roche lobe (see Roche-lobe overflow), gas starts streaming onto the companion through the inner Lagrange point L1. If the mass ratio is modest and the donor compact, this transfer can be stable and gentle. But when the donor is a giant with a deep convective envelope and is more massive than the accretor, transferring mass causes the donor to expand (or its Roche lobe to shrink) faster than the star can shrink back. The transfer becomes dynamically unstable and runs away.

The accretor cannot swallow material at that rate. Gas piles up, overflows the companion's Roche lobe and the outer Lagrange points, and within a few orbits the companion is buried inside the donor's envelope. Now the two cores orbit inside one gaseous body. The envelope retains some rotation but lags the orbiting cores, so each core sweeps through slower-moving gas. The drag force on a body of mass M moving at speed v through gas of density ρ is captured by gravitational dynamical friction, F ≈ 4π G² M² ρ / v² (the Bondi–Hoyle–Lyttleton form), and it always opposes the motion. Energy and angular momentum flow from the orbit into the envelope, and the cores fall inward.

As they fall, two things race each other. The deposited energy heats and inflates the envelope, working to unbind it. Meanwhile the shrinking orbit releases ever more energy, accelerating the deposition. If enough energy is liberated before the cores reach a separation where the giant's core would itself be disrupted, the envelope is ejected and the spiral-in halts at a tight, stable orbit. If not, the cores keep falling and merge. The whole contest is over in of order 1000 years — a dynamical-to-thermal timescale, vastly shorter than any nuclear-burning phase.

The α_CE energy formalism

Modeling the full three-dimensional hydrodynamics of a common envelope is brutally expensive, so population studies use a simple bookkeeping argument due to Webbink (1984) and van den Heuvel (1976). The released orbital energy must, with some efficiency, supply the energy needed to unbind the envelope:

α_CE · ( E_orb,final − E_orb,initial ) = E_bind

The left side is the change in orbital energy times an efficiency factor α_CE; the right side is the gravitational binding energy of the envelope that must be removed. Writing the orbital energy of two point masses as E_orb = −G M₁ M₂ / (2a), and parameterising the envelope binding energy with a structure constant λ, the standard expressions are:

E_bind   = G · M_donor · M_env / ( λ · R_donor )

α_CE · [ −G·M_core·M₂/(2 a_f)  +  G·M_donor·M₂/(2 a_i) ]  =  G·M_donor·M_env / (λ·R_donor)

Here M_donor is the giant's total mass before ejection, M_core its core mass, M_env = M_donor − M_core the envelope mass, M₂ the companion mass, R_donor the giant's radius at the onset of the common envelope, a_i and a_f the initial and final separations. Solving for the final separation a_f gives the post-common-envelope orbit. The dimensionless λ encodes how centrally concentrated the envelope is (deeply bound envelopes have small λ, around 0.1; loosely bound ones approach 1 or more once recombination energy is counted). The efficiency α_CE is bounded by physics — α_CE > 1 demands extra energy sources such as recombination of ionised H and He — and population studies typically adopt α_CE between about 0.2 and 1. Because α_CE and λ enter only as the product α_CE·λ, and because neither is well measured, this product is the single largest free parameter in binary population synthesis.

Worked example: a giant engulfing a 0.6 M☉ companion

Consider a 5 M☉ red giant with a 0.8 M☉ degenerate helium core, swollen to R_donor = 1 AU = 215 R☉, with a 0.6 M☉ main-sequence companion orbiting at a_i = 2 AU. The envelope mass is M_env = 5 − 0.8 = 4.2 M☉. Take λ = 0.5 and α_CE = 0.5. In solar units (G = 1, masses in M☉, lengths in R☉, energy in G M☉²/R☉), 1 AU = 215 R☉ and 2 AU = 430 R☉.

Envelope binding energy:
  E_bind = M_donor · M_env / (λ · R_donor)
         = (5 · 4.2) / (0.5 · 215)
         = 21 / 107.5  ≈  0.195   [G M☉²/R☉]

Initial orbital energy (donor still 5 M☉):
  E_i = − M_donor · M₂ / (2 a_i)
      = − (5 · 0.6) / (2 · 430)  ≈  −0.00349

Energy budget — solve α·(E_f − E_i) = E_bind for final orbit (core now 0.8 M☉):
  E_f = E_i + E_bind/α = −0.00349 + 0.195/0.5 = −0.00349 + 0.390 ...

  Note E_f must be NEGATIVE (a bound orbit). With these numbers
  the required binding energy FAR exceeds what the orbit can supply,
  so the cores would have to spiral to a_f → very small to release enough.

Set E_f = −M_core·M₂/(2 a_f) and solve:
  −(0.8·0.6)/(2 a_f) = −0.00349 + 0.195/0.5  →  RHS positive ⇒ no bound solution
  ⇒ envelope cannot be ejected at α=0.5, λ=0.5 ⇒ MERGER predicted.

The lesson is concrete: with this envelope and these efficiencies, the orbit simply cannot release enough energy to unbind 4.2 M☉ of envelope, so the model predicts a merger. Push α_CE or λ up (more efficient use of orbital energy, or a more loosely bound envelope), or start with a less massive envelope, and a surviving tight binary appears instead. Redo it with λ = 1, α_CE = 1, M_env = 1 M☉ (a less evolved donor): now E_bind = (1.8·1)/(1·215) ≈ 0.0084, and the cores need only spiral from 430 R☉ down to a_f ≈ M_core·M₂/(2(|E_i|+E_bind)) ≈ (0.48)/(2·0.0119) ≈ 20 R☉ — an orbit roughly 20× tighter than they started, ejecting the envelope and surviving. This razor's-edge sensitivity to α_CE·λ is exactly why predicted compact-binary formation rates carry large uncertainties.

Variants and regimes

Not every common envelope looks the same. The outcome depends on the donor's evolutionary state, the mass ratio, and how much energy is available:

• Single versus double common envelope. Many compact-binary channels require two common-envelope phases — one when the primary becomes a giant, another when the (initially secondary) star later evolves. Each phase shrinks the orbit, building up the extreme tightness of double white dwarfs and double neutron stars.
• Self-regulated (slow) spiral-in. If the envelope can radiate or convect away the deposited energy, the inspiral can stall into a quasi-static, self-regulated phase lasting far longer than the dynamical 1000 years, during which the system slowly tightens. Whether this happens versus a fast dynamical plunge is an active research question.
• Grazing envelope evolution. A regime proposed by Soker in which the companion skims the surface and helps drive a strong wind/outflow, removing the envelope without a full deep plunge — a softer cousin of the classical common envelope.
• Recombination-assisted ejection. As the ejected envelope cools and expands, ionised hydrogen and helium recombine, releasing latent energy that helps complete the unbinding — the leading way to justify effective α_CE values approaching or exceeding 1.
• Merger outcomes. When ejection fails, the cores coalesce, producing rapidly rotating, sometimes magnetic stars; luminous red novae such as V1309 Sco (2008) are observed in-progress mergers from terminal common envelopes.

Common envelope outcomes by progenitor

Donor / system	Typical core remnant	Companion	Post-CE product	Final fate
AGB star, low mass	CO white dwarf (~0.6 M☉)	Main-sequence star	Pre-cataclysmic binary	Cataclysmic variable / nova
Red giant + WD	He / CO white dwarf	White dwarf	Double white dwarf	Type Ia SN (double-degenerate)
Massive giant + NS	Neutron star	Helium star → NS	Double neutron star	Kilonova / GW merger
Massive giant + BH	Black hole	Helium star → BH	Binary black hole	GW150914-type merger
HB/RGB star	Hot subdwarf (sdB)	Low-mass dwarf/WD	Close sdB binary	Long-lived helium burner
AGB star + low-mass star	White dwarf + nebula	Main-sequence star	PN central binary	Bipolar planetary nebula
Insufficient α_CE·λ	(envelope not ejected)	(engulfed)	Merged single star	Luminous red nova → fast rotator

Why Type Ia supernovae need a common envelope

A Type Ia supernova is the thermonuclear detonation of a carbon-oxygen white dwarf that reaches a critical mass — at or below the Chandrasekhar limit of 1.4 M☉. Both leading progenitor channels lean on common envelope evolution to get there. In the double-degenerate channel, two white dwarfs must end up in a sub-AU orbit so that gravitational-wave emission can drive them to merge within a Hubble time; only common envelope evolution (often two episodes) can squeeze a pair of white dwarfs — each descended from a giant far larger than the final orbit — into such a tight pair. In the single-degenerate channel, a white dwarf must orbit a non-degenerate companion closely enough to accrete from it; again a common envelope sets up the tight post-CE binary that later becomes a mass-transferring system.

Because Type Ia supernovae are the standardisable candles that revealed the accelerating expansion of the universe, the uncertainties in common envelope physics feed directly into open questions about their progenitors and their delay-time distribution — the spread of times between star formation and explosion. A factor-of-a-few uncertainty in α_CE·λ translates into large uncertainties in predicted Type Ia rates, making the common envelope one of the most important unsolved problems standing between stellar evolution and precision cosmology.

Observational status

We never catch a common envelope in its dynamical 1000-year plunge — the odds are vanishingly small — but we see the system on both sides of it. Before: wide interacting binaries and symbiotic systems poised for unstable mass transfer. During (possibly): luminous red novae like V1309 Scorpii, whose pre-outburst light curve showed a contact binary with a shrinking period, are interpreted as common envelopes ending in merger. After: the post-common-envelope binaries are everywhere once you know to look — close double white dwarfs found by spectroscopic surveys, hot subdwarf binaries, cataclysmic variables, the central binaries of bipolar planetary nebulae, and the double-neutron-star and double-black-hole systems whose mergers LIGO and Virgo detect as gravitational waves. The shapes of bipolar and equatorial-ring planetary nebulae are themselves fossil evidence of envelopes ejected preferentially in the orbital plane during a common envelope. Three-dimensional hydrodynamic simulations (e.g., Ohlmann 2016; MESA/AREPO/FLASH studies) have begun to resolve the fast inspiral and confirm that the dynamical phase ejects only part of the envelope, leaving the rest to a slower process — a tension between theory and the energy formalism that remains unresolved.

Common pitfalls and misconceptions

• "The two stars merge into one." Sometimes — but the canonical, interesting outcome is the opposite: a tighter binary of two surviving cores. The envelope is what gets ejected, not necessarily the stars.
• "The companion accretes the envelope." The companion accretes very little. The defining physics is that it cannot swallow the inflow, which is exactly why a shared envelope forms in the first place. The envelope is ejected, not consumed.
• "It's just rapid Roche-lobe overflow." Roche-lobe overflow is the trigger, but the common envelope is a distinct phase: the companion is fully engulfed and the energetics are governed by drag inside a single envelope, not by a stream through L1.
• "α_CE is a measured constant." It is a fudge factor standing in for unresolved 3D hydrodynamics, degenerate with λ, and calibrated only loosely against observed post-CE binaries. Treat any single quoted value with skepticism.
• "The spiral-in is gradual, like gravitational-wave inspiral." No — it is a dynamical runaway of order 1000 years, set by drag in gas, not the slow general-relativistic decay (which takes over only much later for the resulting compact binary).
• "Common envelopes always eject the whole envelope cleanly." Simulations show the dynamical plunge often unbinds only part of the envelope; completing the ejection may require recombination energy or a longer self-regulated phase, a genuine open problem.