Component masses in stellar and substellar binaries from Gaia astrometry and photometry

Imagine you are looking at a distant streetlight at night. If it's just one bulb, it's easy to guess how bright it is. But what if that single point of light is actually two bulbs stuck together, spinning around each other? And what if they are so far away that your eyes (or even a powerful telescope) can't separate them? You just see one glowing dot.

This is the challenge astronomers face with binary stars (two stars orbiting each other) and star-planet systems. The Gaia space telescope is like a super-precise camera that doesn't just take pictures; it tracks the tiny wobbles of these "single" dots as they dance around their common center of gravity.

Here is the problem: The wobble tells you how heavy the system is, but not how heavy each individual partner is. It's like seeing a seesaw wobble; you know the total weight of the two kids, but you don't know if it's a heavy adult and a light child, or two medium-sized kids.

The Big Idea: Weighing the Invisible

In this paper, Bailer-Jones and Kreidberg developed a clever new way to solve this puzzle using Gaia's data. They combined two types of clues:

The Dance (Astrometry): How much the system wobbles.
The Glow (Photometry): How bright the system is in three different colors of light.

The "Recipe" Analogy

Think of a star like a cake.

Mass is the size of the cake.
Brightness is how much light it emits.
Age and Ingredients (Metallicity) are like how long the cake has been baking or if you added extra sugar.

Usually, if you know the ingredients and the baking time, you can guess the size of the cake just by looking at how bright it is. But in a binary system, you have two cakes baked together in one pan. You can see the total light, but you don't know the ratio of the two cakes.

The authors created a mathematical recipe book (based on computer models of how stars evolve) that says: "If a star is this heavy and this old, it should shine this bright in these three colors."

They then ran a massive simulation (like a digital game of "Guess Who?") to find the only combination of two stars that fits both the wobble data and the total brightness data.

What Did They Find?

1. The "Equal Partner" Trap
One of the biggest headaches in finding exoplanets is false alarms. Sometimes, a system looks like it has a tiny planet because the wobble is small. But it might actually be two stars of almost equal weight spinning around each other.

The Old Way: You'd have to go to a different telescope, get a spectrum, and spend weeks confirming it.
The New Way: This method spots those "equal partner" pairs immediately. If the math says the two stars are nearly equal in mass, the "wobble" would be tiny, explaining why it looked like a planet. This helps clean up lists of potential exoplanets before astronomers waste time on follow-up.

2. How Accurate Is It?

For the Big Star (Primary): They can determine its mass with about 10–20% accuracy. That's like guessing a person's weight within 15 pounds. Pretty good!
For the Small Partner (Secondary): This is harder. The small partner could be a tiny star, a brown dwarf (a "failed star"), or a giant planet. The accuracy is lower (around 25% for the better cases), but they can still tell if it's a planet or a star.

3. The "Extra Clues" Test
The authors asked: "What if we add more data?"

Infrared Light: They added data from other telescopes that see heat (infrared). Result? Tiny improvement. It didn't change the weight estimates much, just made the "error bars" slightly smaller.
Spectroscopy (Speed): They added data about how fast the stars are moving toward/away from us. Result? Almost no change.
The Takeaway: You don't need expensive, time-consuming follow-up observations for every system. Gaia's basic data is often enough to get a solid answer.

Why This Matters

Imagine you have a giant list of 20,000 "wobbling stars." Before this paper, astronomers had to guess which ones were planets and which were just tricky star pairs. Now, they have a catalog of mass estimates for all of them.

For Planet Hunters: It's a filter. It helps them ignore the "fake" planet candidates (which are actually equal-mass stars) and focus on the real ones.
For Star Mappers: It gives us a better census of the "failed stars" (brown dwarfs) and giant planets hiding in our cosmic neighborhood.

The Bottom Line

The authors built a digital scale that can weigh two invisible partners just by watching them dance and measuring their combined glow. It's not perfect, but it's a massive leap forward, turning Gaia from a simple "wobble detector" into a powerful "mass measurer" for the universe.

1. Problem Statement

Determining the masses of stars and substellar objects (brown dwarfs, exoplanets) in unresolved binary systems is a fundamental challenge in astrophysics.

The Astrometric Limitation: For unresolved binaries, Gaia observes the motion of the system's photocentre (the flux-weighted center of light) rather than the barycentre (center of mass). The observed semi-major axis of the photocentre ( $a_p$ ) depends on both the mass ratio ( $M_1/M_2$ ) and the flux ratio ( $f_1/f_2$ ) of the components (Equation 2).
The Degeneracy: Without knowing the flux ratio, a single astrometric measurement cannot uniquely determine the individual masses. A system with a massive primary and a faint planet can produce the same photocentre wobble as a system with two stars of similar mass and brightness.
Current Gaps: Standard methods often assume the secondary contributes negligible light ( $f_2 \approx 0$ ), which fails for near-equal-mass binaries (a major source of false positives in exoplanet searches) or when the secondary is a brown dwarf. Existing solutions often require double-lined spectroscopic data (SB2) or radial velocity follow-up, which are unavailable for the vast majority of Gaia systems.

2. Methodology

The authors develop a probabilistic inference framework to break the mass-flux degeneracy using Gaia DR3 astrometry and three-band photometry ( $G, BP, RP$ ).

A. The Forward Model

Stellar Evolution Models: They utilize PARSEC (Bressan et al. 2012) isochrones to establish a relationship between stellar mass ( $M$ ), age ( $\tau$ ), metallicity ( $[M/H]$ ), and absolute magnitude ( $M_X$ ) in the Gaia bands.
Machine Learning Fit: A gradient-boosted tree model is trained on the PARSEC grid to predict absolute magnitudes as a function of mass, age, and metallicity.
Flux Calculation: The model predicts the flux for each component ( $f_1, f_2$ ) based on the inferred parameters. The sum of these fluxes is compared to the observed total flux.
Extrapolation: For substellar objects below the PARSEC grid limit ($0.09 M_\odot$), the authors linearly extrapolate the absolute magnitude as a function of log-mass, assuming no dependence on age or metallicity.

B. Likelihood and Inference

Parameters: The inference samples a 4-dimensional posterior probability density function (PDF) for the parameters: $\theta = (M_1, M_2, \tau, [M/H])$ .
Likelihood Terms:
1. Astrometric Likelihood: Based on the "astrometric mass function" (Equation 3), linking the measured $a_p^3/T^2$ to the mass and flux ratios.
2. Photometric Likelihood: Gaussian likelihoods comparing the observed total fluxes in $G, BP, RP$ bands against the sum of the model-predicted fluxes.
Priors: Broad, minimally informative priors are used for mass (log-uniform), age (constant star formation rate), and metallicity (linear increase).
Sampling: The posterior is sampled using Markov Chain Monte Carlo (MCMC) (emcee). The median of the posterior is taken as the point estimate, with 16th and 84th percentiles defining the $1\sigma$ uncertainties.

3. Key Contributions

Simultaneous Mass Estimation: The method uniquely determines the masses of both components in an unresolved binary without requiring the secondary to be spectroscopically detected or assumed to be dark.
Flux Ratio Handling: It explicitly models the unknown flux contribution of the secondary, allowing for the identification of near-equal-mass binaries that would otherwise be misidentified as single stars or planetary systems.
Catalogue Generation: The authors provide a catalogue of mass estimates for 20,334 systems (the "Orbital300" sample) within 300 pc of the Sun, selected from the Gaia DR3 Non-Single Star (NSS) catalogue.
Substellar Sensitivity: The method extends down to planetary-mass objects, identifying candidates where the secondary mass is constrained to be below the deuterium-burning limit.

4. Results

A. Mass Precision and Accuracy

Primary Stars: Masses can be determined with a precision of 10–20% (1 $\sigma$ posterior width) in 90% of cases.
Secondary Objects: Precision is lower due to the weaker astrometric signal from the secondary. However, 50% of secondaries are determined with better than 25% precision.
Comparison with Gaia DR3:
- Compared to Gaia Collaboration (2023a) estimates, the primary masses show a bias of -5.5% and a scatter (MAR) of 6.4%.
- Secondary masses show a larger bias (-38%) and scatter (39%) in the general sample, largely because the comparison set often assumes $f_2 \approx 0$ or uses less data.
- On a brighter, spectroscopic subset (AstroSpectro300), the agreement improves significantly (scatter ~14% for secondaries), validating the method even without spectroscopic input.

B. Impact of Additional Data

Infrared Photometry (2MASS/WISE): Adding $J, K_s, W1, W2$ bands improves the precision (narrows posteriors by ~7–34% for primaries) but changes the mean mass estimates by less than 4%.
Spectroscopic Orbits (SB1): Adding Gaia's spectroscopic orbital parameters changes mass estimates by less than 1% and improves precision only modestly (up to 40% for secondary upper errors). This suggests Gaia's epoch radial velocities are not precise enough to significantly refine masses beyond what astrometry + photometry provides for this sample.
Interstellar Extinction: Neglecting extinction has a negligible impact (<2% bias) for systems within 300 pc.

C. Exoplanet Candidates

The method identified 17 exoplanet candidates (where the 84th percentile of the secondary mass is $< 0.012 M_\odot$ ).
Cross-matching with follow-up studies (Stefánsson et al. 2025) revealed that some candidates were false positives (actually SB2 binaries), highlighting the difficulty of distinguishing planets from stellar companions purely via astrometry, but confirming the method's utility in filtering targets.

5. Significance and Limitations

Significance: This work demonstrates that high-precision masses for stars and substellar objects can be derived using only Gaia astrometry and photometry, without the need for expensive spectroscopic follow-up. This is crucial for characterizing the population of binary stars and brown dwarfs, and for cleaning exoplanet candidate lists by identifying false positives caused by equal-mass stellar binaries.
Limitations:
- Model Dependence: The accuracy relies heavily on the PARSEC stellar models. Discrepancies between models and real stars (especially for low-mass objects) are a dominant source of error.
- Low-Mass Extrapolation: The linear extrapolation for masses $< 0.09 M_\odot$ is a simplification, though it has minimal impact on the primary mass inference.
- Selection Effects: The sample is biased toward systems with detectable astrometric signals, which favors specific mass ratios and periods.

Conclusion

The paper presents a robust Bayesian framework that leverages the correlation between mass and luminosity to break the degeneracy in unresolved binary astrometry. By treating the flux ratio as a variable to be inferred alongside mass, age, and metallicity, the authors achieve statistically significant mass estimates for thousands of systems, paving the way for large-scale characterization of binary populations in future Gaia data releases (DR4).