Binary-Driven Mass Loss Mechanisms

• Binary-driven mass loss is the process where a companion star shapes the mass transfer through mechanisms like Roche-lobe overflow, common-envelope ejection, and Lagrange-point outflows.
• It involves redirecting stellar winds and triggering equatorial outflows that are quantified via hydrodynamic models and observational diagnostics in systems such as V1309 Sco.
• Studies show that the angular momentum carried away with the lost mass profoundly affects orbital evolution and mass ratio changes in various stellar populations.

Searching arXiv for recent and foundational papers on binary-driven mass loss, including Roche-lobe overflow, outer-Lagrange-point outflows, and reviews of binarity–mass-loss coupling. Binary-driven mass loss is the removal and redistribution of stellar material in which the presence of a companion star changes how mass is lost, how much of it is retained by the companion, and how much is expelled into the interstellar medium. In published usage, it includes binarity-influenced winds, Roche-lobe overflow, common-envelope-related outflows, and merger-driven ejection, and it is often tightly coupled to orbital angular-momentum loss and mass-ratio evolution rather than to single-star wind physics alone (Sana, 2022). In close interacting systems, the relevant channels range from $L_1$-mediated transfer with partial systemic loss to outer-Lagrange-point outflows, hot-spot-driven ejection, and circumbinary torus formation (Rensbergen et al., 2010).

1. Physical scope and astrophysical domain

Binary-driven mass loss is not a single mechanism but a family of interaction regimes. In detached systems, binarity can redirect stellar winds into non-spherical structures, create wind-collision regions, and modify observational diagnostics of $\dot M$. In closer systems, direct interaction strips envelopes through stable mass transfer, non-conservative Roche-lobe overflow, or common-envelope ejection. In contact binaries and systems approaching merger, mass can escape through the outer critical surface near $L_2$, carrying large specific angular momentum and feeding equatorial spirals or circumbinary disks (Sana, 2022).

The astrophysical relevance of this coupling is population-wide. For O-type stars, the overall multiplicity fraction is close to unity once wide companions are included; early B stars have $>60\%$ multiplicity. Using observed OB period distributions, integrating over periods shorter than a characteristic red-supergiant radius suggests that about two-thirds of OB binaries will interact before or during red-supergiant evolution (Sana, 2022). In the massive-star context, weaker clumping-corrected line-driven winds shift a substantial part of hydrogen-envelope removal to binary mass transfer and eruptive mass loss; the large majority of massive stars, roughly $2/3$–$3/4$, reside in binaries short-period enough to interact, while Sana et al. (2012) estimate that about $1/4$ will merge, about $1/3$ will have their H envelopes stripped, about $14\%$ will be spun up by accretion, and only about $1/4$ are effectively single (Smith, 2014).

The same logic extends beyond massive stars. In intermediate-separation binaries of order $\dot M$0 au, white dwarfs are inferred to be about $\dot M$1 less massive than their isolated counterparts, with a typical mass of about $\dot M$2; their progenitors likely lost $\dot M$3–$\dot M$4 of their mass, with binary interactions enhancing mass loss by an additional $\dot M$5 (Shahaf, 16 Jan 2025). This suggests that binary-driven mass loss is important not only in classical close-binary or merger channels, but also in a regime where interaction is moderate rather than catastrophic.

2. Principal mechanisms of mass expulsion

In semi-detached binaries, the most explicit formulation is Roche-lobe overflow with a time-dependent accretion efficiency. In the Brussels models of Algol progenitors, the donor’s RLOF rate $\dot M$6 is computed self-consistently, while the accretion rate onto the gainer is parameterized as

$\dot M$7

with $\dot M$8 for conservative transfer and $\dot M$9 for liberal evolution (Rensbergen et al., 2010). Mass loss to interstellar space occurs whenever the instantaneous transfer rate exceeds a critical rate set by the combined effects of gainer spin-up and hot-spot radiation. The key physical point is that $L_2$0 does not require the accretion luminosity alone to exceed the Eddington limit and does not require the gainer to be exactly at critical rotation; the combined spin-up plus radiation pressure matters. In that framework, binaries with a late B-type initial primary hardly lose any mass, whereas binaries with an early B-type initial primary evolve in a strongly non-conservative way (Rensbergen et al., 2010).

In contact binaries and binaries approaching merger, mass loss through the outer Lagrange point $L_2$1 becomes central. For V1309 Sco, Pejcha et al. interpret the pre-outburst evolution as runaway loss of mass and angular momentum from $L_2$2 over thousands of orbits, producing a long-lived equatorial “death spiral” outflow before final coalescence. In that system, the prolonged pre-dynamical phase lasted several years, reached $L_2$3–$L_2$4, and removed $L_2$5 before the final dynamical event (Pejcha et al., 2017). A related analysis of V1309 Sco models the slow rise to maximum as the expansion of the photosphere of a continuous optically thick outflow produced by dynamical mass loss, with the binary progressively buried by its own ejecta (Pejcha, 2013).

Hydrodynamic calculations of unstable RLOF just before common envelope yield a complementary picture. Across mass ratios $L_2$6–0.3, the mass expelled as the separation decreases from the Roche limit to the donor’s original radius is of order $L_2$7 of the accretor’s mass, assembling into a toroidal circumbinary distribution with approximately constant specific angular momentum due to momentum transport by spiral shocks (Macleod et al., 2020). The same pre-common-envelope phase therefore acts both as a mass-loss channel and as an initial-condition generator for the later dynamical inspiral.

The geometrical details near $L_2$8 are themselves non-trivial. Ballistic calculations show four broad outcomes for material launched from the vicinity of $L_2$9: unbound outflow, fallback followed by decretion-disk formation, collision with the binary surface, and self-intersecting loops that generate shocks near the binary (Hubova et al., 2019). A notable result is that even for initial velocities slower than corotation, some initial position offsets still produce unbound outflows, because time-dependent tidal torques, Coriolis forces, and initial conditions compete in a strongly structured way (Hubova et al., 2019).

A separate but related development is the unification of atmospheric underflow and $>60\%$0-stream overflow into one continuous onset-of-transfer formalism. The “Unified Rapid Mass Transfer” method computes the mass flux through the $>60\%$1 region by integrating over the full nozzle cross-section in the same 3D binary potential used to evolve the donor, applies over $>60\%$2, and can also be applied to hot donors (Ivanova et al., 2024). This suggests that the earliest stages of binary-driven mass loss are more sensitive to geometry and equation-of-state effects than older optically thin/optically thick switches imply.

3. Angular momentum loss and orbital evolution

The defining dynamical feature of binary-driven mass loss is that matter is usually lost with non-negligible specific orbital angular momentum. In Van Rensbergen et al.’s liberal evolution, the angular-momentum loss rate is written

$>60\%$3

with the lost matter assumed to leave from the hot spot on or near the gainer and to carry the gainer’s orbital specific angular momentum, so that

$>60\%$4

This makes angular-momentum loss strongest when the masses are comparable and less effective after mass-ratio reversal, when the donor has become much lighter (Rensbergen et al., 2010).

That distinction has direct orbital consequences. In conservative transfer from the more massive star to the less massive star, the orbit initially shrinks and later widens after mass-ratio reversal. In the liberal case, the gainer’s growth is limited, the fall in $>60\%$5 is moderated, and the orbit is altered by systemic loss of mass and angular momentum. This is one reason liberal evolution produces more high-$>60\%$6 systems than conservative evolution, even if it still fails to match the observed high-$>60\%$7 tail (Rensbergen et al., 2010).

Outer-Lagrange-point loss is dynamically more aggressive. In V1309 Sco, the observed period decay and the inferred mass-loss history are consistent with a scenario in which $>60\%$8 outflow extracts orbital angular momentum efficiently enough to drive merger (Pejcha et al., 2017). In the EMRI calculation of L2 leakage from a Roche-filling main-sequence star around a supermassive black hole, $>60\%$9 loss is likewise intrinsically non-conservative and decreases the evolution timescale of the emitted gravitational-wave signal by up to a few tens of per cent (Linial et al., 2017). Although the parameter regime is very different, the underlying point is the same: once mass escapes through an outer channel rather than remaining in an $2/3$0-fed exchange, the orbital evolution is governed by the specific angular momentum of the escaping matter.

A common misconception is that binary-driven mass loss necessarily widens the orbit because the system mass decreases. The literature instead shows that the sign and magnitude of $2/3$1 and $2/3$2 depend on where the mass is lost, how much angular momentum it carries, and whether mass-ratio reversal has already occurred. Mass loss from $2/3$3 or from a circumbinary torus can shrink the orbit efficiently, while loss carrying only the gainer’s orbital specific angular momentum can be dynamically milder (Rensbergen et al., 2010).

4. Observational manifestations

The observational phenomenology of binary-driven mass loss is correspondingly diverse. In detached massive binaries with strong winds, the wind-collision region is set by ram-pressure balance,

$2/3$4

and generates thermal X-rays, synchrotron radio emission, inverse Compton hard X-rays, and optical/IR/UV recombination lines (Sana, 2022). In adiabatic colliding-wind systems such as WR 140, the X-ray flux scales approximately as $2/3$5, where $2/3$6 is the stellar separation (Sana, 2022). In WC+O binaries, the same interaction region becomes a dust factory, producing pinwheel spirals whose pitch and brightness encode the wind speed, orbital period, and dust content.

R144 illustrates how binary-driven wind structure can become a precision diagnostic. Modeling of its heartbeat-like optical variability implies $2/3$7, $2/3$8–$2/3$9, and dynamical masses $3/4$0 and $3/4$1, with the photometric modulation attributed primarily to double wind eclipses and a phase-locked contribution from the wind-collision region rather than to tidal deformation (Sana, 2022). The broader implication is that binarity can convert wind diagnostics into direct measurements of mass loss and masses.

In semi-detached Algols, the population-level test is the joint distribution of orbital period and mass ratio. Liberal models that include Case B transfer and hot-spot/spin-up-driven systemic loss reproduce the observed period distribution of 303 Algols for $3/4$2 d, with Kolmogorov–Smirnov maximum difference $3/4$3 compared with $3/4$4 for conservative models (Rensbergen et al., 2010). Mass ratios remain problematic: conservative evolution gives only about $3/4$5 of Algols with $3/4$6, liberal evolution raises this to about $3/4$7, but the observed fraction is about $3/4$8, including about $3/4$9 with $1/4$0 versus only about $1/4$1 predicted (Rensbergen et al., 2010).

Approaching merger, the photometric signatures shift again. In V1309 Sco, the phased light curve evolved from a double-hump contact-binary pattern to a single-hump morphology and then disappeared as the optically thick outflow buried the system; the mean brightness first changed gradually and then rose by more than $1/4$2 mag over about eight months before the dynamical peak (Pejcha et al., 2017). A related interpretation of the slow-rise phase treats it as a continuous optically thick outflow whose expanding photosphere hides the binary and explains the $1/4$3-day rise to maximum (Pejcha, 2013). For AT2018bwo, the inferred pre-dynamical mass loss is concentrated in the orbital plane, and collision between slower equatorial mass loss and later faster ejecta powers a bright red transient (Pejcha, 2022).

5. Consequences for stellar and binary evolution

Binary-driven mass loss is one of the principal reasons that single-star evolutionary mappings fail in interacting populations. In massive stars, reduced line-driven wind rates imply that binary mass transfer and eruptive loss must shoulder much of the hydrogen-envelope stripping. This reassigns the origin of many stripped-envelope supernova progenitors, many Wolf–Rayet stars, and many classical Be stars from purely single-star channels to binary interaction (Smith, 2014). The lack of Be+main-sequence binaries and the presence of some Be+stripped companions strongly support a post-mass-transfer origin for most classical Be stars, while the much lower red-supergiant multiplicity fraction relative to OB progenitors implies a large hidden population of post-interaction products (Sana, 2022).

The same population logic appears in white-dwarf formation. In the intermediate-separation sample analyzed with the truncated-Pareto method, the inferred WD distribution peaks around $1/4$4–$1/4$5, with median $1/4$6 and mean $1/4$7, and the progenitors lost $1/4$8–$1/4$9 of their mass rather than the $1/3$0 typical of single-star-like IFMR determinations (Shahaf, 16 Jan 2025). A plausible implication is that tidally enhanced winds, wind Roche-lobe overflow, or grazing-envelope evolution can lower WD masses systematically without forcing the system into a short-period post-common-envelope state.

Cataclysmic variables provide a further example of binary-driven mass loss acting as an angular-momentum-loss mechanism rather than only as envelope stripping. In the nova context, recent hydrodynamical arguments motivate binary-driven mass loss in which a major fraction of the ejecta leaves near $1/3$1. Detailed calculations show that such BDML can provide extra angular-momentum loss below the period gap, produce a large range of mass transfer rates at a given orbital period in agreement with the observed secular distribution, avoid runaway mass transfer, allow white dwarfs to grow by $1/3$2, and cause both positive and negative orbital-period changes across nova eruptions (Tang et al., 2024). This suggests that systemic mass loss shaped by the binary potential can alter secular evolution even when the ejection event itself is brief.

6. Uncertainties, controversies, and modeling limits

Several of the central prescriptions remain poorly constrained. In the Algol hot-spot model, the factor $1/3$3 that measures how efficiently accretion power is converted into hot-spot radiation is calibrated from only 13 semi-detached systems and shows large error bars; the relation between $1/3$4 and total binary mass is therefore uncertain (Rensbergen et al., 2010). The geometry of the escaping matter is also simplified: loss is often assumed to carry only the orbital specific angular momentum of the gainer or of $1/3$5, while anisotropy, jets, and spin angular momentum of near-critically rotating accretors are neglected (Rensbergen et al., 2010).

The treatment of rapid mass transfer is another active fault line. The “Unified Rapid Mass Transfer” formalism finds that the revised underflow/overflow rates always differ from the optically thin and optically thick rates widely used in binary evolution calculations, and that the amount of mass lost before reaching the outer Lagrange equipotential can differ by more than an order of magnitude depending on the formalism adopted (Ivanova et al., 2024). A plausible implication is that some disagreements about common-envelope onset or survival are, at least in part, disagreements about how the onset of binary-driven mass loss is computed.

There is also a direct controversy over donor response and stability. Woods and Ivanova argue that for giants with deep convective envelopes, the local thermal timescale of the superadiabatic surface layer remains comparable to that of mass loss in most cases of “dynamical” mass loss, so the polytropic approximation gives much too strict a criterion for stability (Woods et al., 2011). This challenges the routine identification of convective-envelope donors with prompt catastrophic runaway and suggests that detailed evolutionary calculations are required to determine the effective $1/3$6 on rapid timescales.

Finally, the pre-dynamical phase of common-envelope or merger evolution is now difficult to ignore. In V1309 Sco, the gradual $1/3$7 outflow removed $1/3$8, comparable to the mass inferred for the later dynamical ejection, while broader arguments for AT2018bwo and related luminous red novae indicate prolonged mass loss amounting up to a few solar masses before the dynamical event (Pejcha et al., 2017). This shifts the conceptual center of gravity away from a purely explosive common-envelope picture toward one in which long-lived binary-driven outflows, equatorial density enhancement, and shock interaction are integral parts of the evolutionary pathway rather than precursors of negligible consequence.