The drastic impact of Eddington-limit induced mass ejections on massive star populations

This study introduces a physically motivated mass-loss prescription for Eddington-limit induced ejections in 1D stellar evolution codes, demonstrating that incorporating this mechanism into single and binary models successfully reconciles theoretical predictions with observed massive star populations in the Magellanic Clouds, including their luminosity limits, evolutionary stages, and binary fractions.

The Gist

Imagine massive stars as the rock stars of the universe. They are born huge, burn incredibly bright, and die young, but their lives are short and violent. They are the engines that shape galaxies, creating the heavy elements that make up planets and even us.

However, for a long time, astronomers have been trying to write the "biography" of these stars using computer models, and the story didn't add up. The models predicted that these stars should get so big and bright that they would explode or turn into giant red monsters (Red Supergiants) that we simply don't see in the sky. It was like a movie script predicting a dragon that no one has ever photographed.

This paper, written by a team of astronomers, fixes that story by introducing a new, dramatic plot twist: The Eddington Limit Mass Ejection.

Here is the breakdown of their discovery in simple terms:

1. The Problem: The "Overweight" Stars

In the old computer models, massive stars were like people eating too much cake. As they burned their fuel, they got so hot and bright that their own light pushed against their own gravity. The models predicted they would puff up into enormous, glowing red giants that were far too bright to exist.

But when astronomers looked at the sky (specifically in the Large and Small Magellanic Clouds, which are our cosmic neighbors), they saw a hard limit. There is a "ceiling" to how bright and big a star can get. This is called the Humphreys-Davidson Limit. No stars exist above this line. The old models were failing because they didn't know how the stars stopped themselves from getting too big.

2. The Solution: The "Cosmic Haircut"

The authors realized that when a massive star gets too close to its "Eddington Limit" (the point where its light is so strong it threatens to blow the star apart), it doesn't just sit there. It goes into a panic mode.

Think of a massive star like a balloon being blown up too fast.

• Old Theory: The balloon just keeps getting bigger until it pops or becomes a giant, floppy mess.
• New Theory: When the balloon gets too tight, it suddenly snaps a piece off! It violently ejects a huge chunk of its outer skin (its atmosphere) to relieve the pressure.

The authors created a new rule for their computer models: When a star inflates too much, it triggers a "mass ejection." It sheds its outer layers in a massive burst, similar to a Luminous Blue Variable (LBV) eruption. This is like the star giving itself a sudden, drastic haircut to stay within the size limits.

3. The Results: The Story Finally Fits

When they ran the simulations with this new "haircut" rule, the results were amazing. The computer models suddenly matched the real sky perfectly:

• No More Ghost Stars: The models stopped producing those impossible, super-bright Red Giants that don't exist.
• The Wolf-Rayet Mystery Solved: There are some very faint, hot stars called Wolf-Rayet stars that were a mystery. Old models said they should only exist if they were in a binary system (two stars dancing together). But astronomers found them alone. The new model showed that a single star can strip its own skin off via these mass ejections, becoming a Wolf-Rayet star all by itself.
• The Right Numbers: The models now predict the exact number of O-stars, Red Supergiants, and Wolf-Rayet stars we actually see in the Magellanic Clouds.

4. The Role of Binaries (The Dance Partners)

The paper also looked at stars that have a partner (binary systems). Usually, two stars can strip each other's skin off by stealing mass. The authors found that while this "dance" is important, it's not the only way things happen. Their new "self-haircut" rule explains a huge chunk of the population that was previously unexplained.

The Big Picture

This paper is like fixing a broken map. Before, the map of the universe showed mountains where there were only flat plains. By adding the rule that "stars shed their skin when they get too hot," the map now matches the terrain.

In a nutshell: Massive stars are wild, unstable creatures. When they get too big for their britches, they don't just explode; they violently shed their outer layers to stay safe. Once we programmed this behavior into our computers, the universe finally made sense.

Technical Summary

Here is a detailed technical summary of the paper "The drastic impact of Eddington-limit induced mass ejections on massive star populations" by Pauli et al. (2026).

1. Problem Statement

Current 1D stellar evolution models face significant discrepancies when predicting the populations of massive stars ($M_{\text{init}} \gtrsim 8 M_\odot$), particularly in low-metallicity environments like the Large and Small Magellanic Clouds (LMC and SMC). Key issues include:

• The Humphreys-Davidson (HD) Limit Violation: Standard models predict the existence of extremely luminous Red Supergiants (RSGs) and Yellow Supergiants (YSGs) beyond the empirical HD limit ($\log L/L_\odot \gtrsim 5.5$), where no stars are observed.
• The "Faint" Wolf-Rayet (WR) Problem: Models struggle to explain the existence of the least luminous, apparently single WR stars in low-metallicity galaxies. Standard wind prescriptions predict that single stars cannot strip their hydrogen envelopes efficiently enough at low metallicity to become WR stars; thus, these stars are expected to be binary products, yet observations suggest many are single.
• Luminous Blue Variable (LBV) Uncertainty: The physics of LBV eruptions (mass ejections near the Eddington limit) is poorly understood and often ignored or oversimplified in evolutionary codes, leading to inaccurate mass-loss histories.

2. Methodology

The authors employed the MESA (Modules for Experiments in Stellar Astrophysics) code (v24.08.1) to simulate stellar populations at LMC ($Z = 0.5 Z_\odot$) and SMC ($Z = 0.14 Z_\odot$) metallicities.

A. New Mass-Loss Prescription

The core contribution is a new, empirically calibrated prescription for Eddington-limit induced mass ejections.

• Trigger Mechanism: The model monitors the inflation of the stellar envelope. The "core" radius ($R_{\text{core}}$) is approximated as the location where gas pressure drops to 15% of total pressure. The inflation factor is defined as $R/R_{\text{core}}$.
• Thresholds: Mass ejections are triggered when the inflation exceeds a threshold dependent on surface hydrogen abundance ($X_H$):
  • For hydrogen-rich stars ($X_H \geq 0.2$): Threshold is $R/R_{\text{core}} \approx 2$.
  • For hydrogen-poor stars ($X_H < 0.2$): The threshold scales linearly down to $R/R_{\text{core}} \approx 1.1$ for helium stars.
• Mass-Loss Rate: Once the threshold is crossed, the mass-loss rate is increased to a critical value ($\dot{M}_{\text{eject}}$) derived from Petrovic et al. (2006), designed to prevent further inflation. The rate ramps up linearly between 95% and 100% of the threshold to avoid numerical instabilities.
• Wind Recipes: The models use updated empirical recipes for hot stars (Pauli et al. 2025) and RSGs (Yang et al. 2023), switching to WR recipes when appropriate.

B. Population Synthesis

• Single Stars: Grids of 33 models per metallicity covering $10 - 400 M_\odot$ with a Kroupa IMF and constant Star Formation Rate (SFR).
• Binary Stars: The single-star grids were combined with pre-computed binary evolution grids (Marchant 2016; Pauli et al. 2022) to account for mass transfer and mergers.
• Variations: The authors tested different physical setups, including varying the RSG mass-loss recipe (Nieuwenhuijzen & de Jager 1990 vs. Yang et al. 2023) and semiconvective mixing efficiency ($\alpha_{sc}$).

3. Key Contributions

• Implementation of a Physical LBV Proxy: The paper provides a computationally efficient 1D method to approximate the effects of LBV eruptions and Eddington-limit instabilities without requiring expensive 3D hydrodynamic simulations.
• Calibration via Inflation: Instead of arbitrary mass-loss boosts, the mass ejection is tied to the physical state of envelope inflation, calibrated against the observed HD limit and WR luminosity distributions.
• Resolution of the Faint WR Paradox: The study demonstrates that Eddington-limit induced mass ejections allow single stars to shed their hydrogen envelopes even at low metallicity, explaining the existence of faint, single WR stars without relying solely on binary interactions.

4. Results

A. Single-Star Population (Model Eject)

• HD Limit Compliance: The new prescription successfully prevents stars from residing beyond the HD limit for significant durations. Massive stars ($M_{\text{init}} \gtrsim 50 M_\odot$) undergo mass ejections during core hydrogen burning, preventing them from becoming super-luminous RSGs.
• WR Population: The models reproduce the observed number of WR stars in the LMC and SMC. Crucially, they explain the faintest single WR stars in the SMC and the WO star in IC 1613, which previous models could not.
• Luminosity Distributions: The synthetic populations match the observed luminosity distributions of O-stars, RSGs, and WR subtypes (WN, WC/WO) in the LMC and SMC.
• Discrepancies:
  • There is a slight overproduction of very massive stars (VMS) and bright H-free WN stars, likely due to assumptions in the Initial Mass Function (IMF) or Star Formation History (SFH).
  • The models underpredict the number of Blue (BSG) and Yellow Supergiants (YSG), a known issue potentially linked to binary mergers or mixing physics.

B. Binary Population (Model Binary)

• Binary Fraction: The synthetic binary population predicts a $\sim70\%$ binary fraction for O-stars and a $\sim40\%$ binary fraction for WR stars, consistent with observations.
• Formation Channels: While binary interactions (mass transfer) remain a primary channel for forming H-free WR stars, the study finds that self-stripping via Eddington-limit induced mass ejections in non-interacting binaries and mergers accounts for a large fraction of the WR population.
• Cool Supergiants: The inclusion of binaries slightly increases the RSG count (due to merger products) but does not resolve the deficit of BSGs/YSGs, suggesting these may be merger products not fully captured by single-star approximations.

C. Sensitivity Analysis

• RSG Mass Loss: Using weaker RSG mass-loss rates (Nieuwenhuijzen & de Jager 1990) fails to strip envelopes of intermediate-mass stars ($25-40 M_\odot$), leading to an overproduction of RSGs and a failure to produce faint WR stars. This confirms the need for strong, Z-independent RSG winds.
• Semiconvection: Increasing semiconvective mixing efficiency ($\alpha_{sc}=10$) produces blue loops (BSG/YSG) but fails to match the observed luminosity distribution and still struggles to produce WR stars.

5. Significance

This work resolves a long-standing tension between theoretical stellar evolution models and observed massive star populations. By introducing a physically motivated, inflation-triggered mass ejection mechanism, the authors demonstrate that:

1. The HD limit is a natural consequence of Eddington-limit instabilities, not an arbitrary observational cutoff.
2. Single-star evolution is sufficient to explain the existence of faint, hydrogen-poor WR stars in low-metallicity environments, challenging the paradigm that these must be binary products.
3. Binary interactions and self-stripping are complementary mechanisms; both are necessary to fully reproduce the observed diversity of massive star populations.

The study underscores the necessity of including Eddington-limit effects in 1D stellar evolution codes to accurately model the feedback, chemical enrichment, and final fates of massive stars in the Universe.