🔭 astrophysics

Stellar masses and mass ratios for Gaia open cluster members

This paper presents a fast, simulation-based inference framework that utilizes Gaia DR3 parallaxes and multi-band photometry to derive precise stellar masses and binary mass ratios for open cluster members, revealing that high-mass-ratio binary fractions correlate strongly with cluster age and weakly with metallicity.

Original authors: Sagar Malhotra, Alfred Castro-Ginard, Friedrich Anders, Carme Jordi, Judit Donada, Xavier Luri, Lola Balaguer-Núñez, Songmei Qin, Yueyue Jiang, Andrija Župić

Published 2026-02-04

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Sagar Malhotra, Alfred Castro-Ginard, Friedrich Anders, Carme Jordi, Judit Donada, Xavier Luri, Lola Balaguer-Núñez, Songmei Qin, Yueyue Jiang, Andrija Župić

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the night sky not as a collection of lonely stars, but as a bustling city where many stars live in pairs or groups, holding hands so tightly that even our most powerful telescopes see them as a single, glowing dot. These are called unresolved binaries.

For astronomers, trying to figure out how heavy these "double stars" are is like trying to guess the weight of two people hugging in a foggy room just by looking at the shadow they cast. If you assume it's just one person, you get the math wrong. This mistake throws off our understanding of how star clusters (the "neighborhoods" where stars are born) evolve and move.

This paper introduces a new, super-fast "detective tool" to solve this mystery for 42 star clusters in our cosmic neighborhood. Here is how they did it, explained simply:

1. The Problem: The "Hug" Effect

In a star cluster, stars are born at the same time and have similar chemical makeup. However, many of them are actually two stars orbiting each other. Because they are so close, they look like one bright star.

The Mistake: If you treat a double star as a single star, you think it is heavier and brighter than it really is.
The Consequence: This messes up the calculation of the cluster's total mass and how it changes over time.

2. The Solution: A "Cosmic Simulator" (SBI)

The authors didn't just look at the stars; they built a massive virtual reality simulation of what star clusters should look like.

The Training: They created a digital universe with about 2 million fake stars, mixing in single stars and binary pairs with different weights and ages. They taught a computer (using a technique called Simulation-Based Inference) to recognize the difference between a single star and a "hugging" pair by looking at how bright they are in different colors of light (from blue to infrared).
The Trick: Just like a human learns to spot a fake painting by seeing thousands of real ones, the computer learned to spot binary stars by studying millions of simulated ones.

3. The "Group Hug" Strategy (Iterative Fitting)

The method is clever because it doesn't look at stars one by one in isolation. It treats the whole cluster as a team.

The Loop: The computer makes a guess about the cluster's age and distance. Then, it checks if the stars fit that guess. If the stars look too old or too young for the guess, the computer adjusts the guess and tries again.
The Result: It keeps refining its guess (like tuning a radio until the static clears) until it finds the perfect age, distance, and dust level for the whole cluster. Once the cluster's "personality" is understood, it can accurately weigh every single star inside it.

4. The "Infrared Glasses" Advantage

The team used data from three different telescopes: Gaia (visible light), 2MASS, and WISE (infrared).

The Analogy: Looking at a star cluster with only visible light is like trying to see a dark room with a flashlight. Adding infrared data is like putting on night-vision goggles.
The Benefit: This allowed them to spot "lighter" binary pairs that were previously invisible. They found that they could reliably detect binary pairs where the smaller star is at least 20% to 50% the size of the bigger one (depending on how far away the cluster is).

5. What They Found

After applying this method to 42 clusters (containing over 27,000 stars), they created a new "ID card" for every star, listing its true mass and whether it has a companion.

The Age Factor: They found that older clusters have more binary stars (specifically those with partners of similar size). It's like an old neighborhood where the single people have moved away, leaving behind the couples who stick together.
The Metal Factor: There is a hint that clusters with fewer heavy elements (lower "metallicity") might have slightly more binaries, but this needs more study.
The Mass Factor: For stars heavier than our Sun, the number of binaries in these clusters matches what we see in the rest of the galaxy. However, for very small, faint stars, the data was a bit messy, likely because our "virtual models" of small stars aren't perfect yet.

Summary

Think of this paper as the release of a new, high-definition map for star clusters. Instead of seeing a blurry crowd of single dots, astronomers can now see the individual pairs, know exactly how heavy they are, and understand how the "family dynamics" of these star clusters change as they get older. This helps us understand the history of our galaxy with much greater precision.

Technical Summary: Stellar Masses and Mass Ratios for Gaia Open Cluster Members

Problem Statement
Unresolved binary stars in open clusters (OCs) introduce significant biases in stellar and cluster mass estimates, complicating the study of cluster dynamics and evolution. Traditional methods for characterizing these systems often rely on time-consuming spectroscopic surveys or eclipsing binary detection, which are not scalable to large samples. Furthermore, standard isochrone fitting often fails to account for the scatter caused by binaries in color-magnitude diagrams (CMDs), leading to inaccurate determinations of cluster parameters (age, distance, extinction) and individual stellar masses. While the Gaia mission has revolutionized astrometry and photometry, deriving precise individual masses and mass-ratios ( $q$ ) for unresolved binaries in a self-consistent manner across large cluster samples remains a challenge.

Methodology
The authors present a fast, robust framework based on Simulation-Based Inference (SBI), specifically utilizing Neural Posterior Estimation (NPE), to jointly derive stellar masses, mass-ratios, and optimal cluster parameters. The pipeline operates as follows:

Data Integration: The method utilizes multi-band photometry from Gaia, 2MASS, and WISE, combined with Gaia DR3 parallaxes for member stars of 42 well-populated OCs. Membership lists are drawn from Hunt & Reffert (2024).
Simulation-Based Inference (SBI): A neural network is trained on a simulated dataset of $\sim$ 2 million stars generated using PARSEC isochrones. The simulation incorporates realistic observational uncertainties (photometry and parallax) and physical priors (Initial Mass Function, metallicity, age, distance, extinction, and mass-ratio). The network learns to approximate the posterior distribution of astrophysical parameters (primary mass $M_A$ , mass-ratio $q$ , metallicity, age, distance modulus, and extinction) directly from observed data, bypassing the need for an explicit likelihood function.
Iterative Cluster Parameter Estimation: To break degeneracies between individual stellar parameters and global cluster properties, the authors employ an iterative loop. They assign weights to SBI posterior samples based on Gaussian priors for cluster age, distance, metallicity, and extinction. The cluster parameters are updated iteratively by selecting specific subsets of stars (e.g., Main Sequence Turn-Off stars) best suited for estimating specific parameters, converging after approximately 45 iterations. Spectroscopic metallicities from the literature are held fixed to reduce degeneracy.
Mass-Ratio Priors: The framework incorporates empirical findings that binary fraction increases with primary stellar mass. A prior is applied to the joint $M_A$ - $q$ posterior, weighting samples based on the probability of a star being in a binary system. A lower mass-ratio limit ( $q_{lim}$ ) is defined based on the minimum stellar mass in the simulations ( $0.1 M_\odot$ ), below which systems are treated as single stars.
Validation: The method is validated on simulated clusters with controlled binary properties and Gaia-like uncertainties, demonstrating that the inclusion of infrared photometry significantly improves the detection of low mass-ratio binaries.

Key Contributions and Results

Catalogue Generation: Applying the method to 42 OCs, the authors produced a catalogue of stellar masses and mass-ratios for 27,201 stars. The typical uncertainties achieved are 0.01 $M_\odot$ for the primary mass and 0.08 for the mass-ratio ( $q$ ).
Detection Thresholds: The study establishes that the minimum mass-ratio threshold for reliably identifying unresolved binaries ( $q_{thresh}$ ) depends on cluster properties and available photometry. With multi-band photometry (including IR), $q_{thresh}$ typically lies below 0.5, whereas using Gaia photometry alone raises the threshold to $\sim$ 0.55–0.56.
Case Studies (NGC 2360 and M67): Detailed analysis of these archetype clusters reveals:
- NGC 2360: Shows a monotonic increase in binary fraction for stars heavier than $\sim0.6 M_\odot$ , consistent with field star trends.
- M67: Exhibits dips in binary fraction at specific masses, likely due to gaps in the Main Sequence.
- Both clusters show higher binary fractions in their cores, consistent with mass segregation.
Population Trends: Analysis of the 42 clusters reveals:
- A strong positive correlation between the high mass-ratio binary fraction ( $q \ge 0.6$ ) and cluster age.
- A weak anti-correlation between binary fraction and cluster metallicity.
- A weak anti-correlation with distance (attributed to the inability of Gaia to resolve close binaries at larger distances).
- The variation of binary fraction with stellar mass in OCs aligns with the observed dependence for field stars heavier than $\sim0.6 M_\odot$ .
Comparison with Literature: The derived stellar masses agree well with previous studies (e.g., Hunt & Reffert 2024; Khalatyan et al. 2024; Almeida et al. 2023). The derived binary fractions are generally slightly higher (by a factor of $\sim$ 1.4) than studies relying solely on optical Gaia photometry, attributed to the improved detection capability of the multi-band approach.

Significance
The authors claim that this work paves the way for future population-level investigations of multiplicity statistics and precision stellar masses in extended samples of open clusters. By providing a fast, scalable framework that simultaneously infers individual stellar parameters and global cluster properties, the method addresses the critical need to properly treat unresolved binaries in dynamical and evolutionary studies. The inclusion of infrared photometry is highlighted as a key factor in lowering the detection threshold for low mass-ratio binaries, offering a more complete picture of the binary population in the solar neighborhood. The paper emphasizes that while the method is model-dependent, it offers a self-consistent approach to deriving precise parameters that were previously difficult to obtain for large samples.