🔭 astrophysics

Asteroseismic ages for 17,000 stars in Kepler, K2 and TESS

This study presents a homogeneous analysis of over 17,000 red giant stars using asteroseismic data from Kepler, K2, and TESS combined with Gaia, APOGEE, and GALAH observations to infer robust stellar ages and parameters, while identifying unreliable measurements and exploring Galactic archaeology trends across different stellar populations.

Original authors: Emma Willett, Andrea Miglio, Saniya Khan, Yvonne Elsworth, Benoît Mosser, Karsten Brogaard, Giada Casali, Cristina Chiappini, Valeria Grisoni, Amalie Stokholm, Diego Bossini, William J. Chaplin

Published 2026-05-29

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Emma Willett, Andrea Miglio, Saniya Khan, Yvonne Elsworth, Benoît Mosser, Karsten Brogaard, Giada Casali, Cristina Chiappini, Valeria Grisoni, Amalie Stokholm, Diego Bossini, William J. Chaplin

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 Milky Way galaxy as a massive, ancient library. For a long time, astronomers could see the books (stars) on the shelves, but they didn't know how old they were or where they were born. Without knowing the age of the books, it's hard to understand the story of how the library was built and how it changed over billions of years.

This paper is like a team of librarians who have finally figured out how to date over 17,000 of these "books" (specifically, red giant stars, which are old, swollen stars nearing the end of their lives). They did this by combining data from three different space telescopes—Kepler, K2, and TESS—with precise maps from the Gaia satellite and chemical fingerprints from ground-based telescopes (APOGEE and GALAH).

Here is a simple breakdown of what they did and what they found:

1. The "Heartbeat" of the Stars (Asteroseismology)

Stars don't just sit still; they pulse and vibrate like giant, glowing drums. These vibrations create ripples in the light we see from them.

The Analogy: Think of a bell. If you hit a small, thin bell, it rings at a high pitch. If you hit a large, thick bell, it rings at a low pitch. By listening to the "ringing" (vibrations) of these stars, the team could calculate their size, mass, and most importantly, their age.
The Challenge: The K2 mission (one of the telescopes used) only watched each patch of sky for a short time (about 80 days). It's like trying to identify a song by listening to only 10 seconds of it. Sometimes, the team couldn't be sure if the "song" was real or just noise. They developed a special filter to spot and remove the stars where the "heartbeat" measurement was unreliable, ensuring they didn't date the wrong stars.

2. Building a Better Map

Once they knew the age and mass of the stars, they combined this with data from the Gaia satellite, which tells us exactly where the stars are and how they are moving.

The Result: They created a massive catalog of 17,000 stars with known ages, masses, and orbits. This is like giving every book in the library a label that says: "Born 8 billion years ago, originally from the North Wing, now drifting near the entrance."

3. What the Library Tells Us (Key Findings)

By looking at these dated stars, the team uncovered several stories about the history of our galaxy:

The "Old" vs. "Young" Neighborhoods: They found that the galaxy has different "neighborhoods" based on chemical makeup.
- The "High-Alpha" Group: These are the oldest stars (mostly over 6–10 billion years old). They live in the "thick disc" and the "halo" (the outer, fuzzy edges of the galaxy). They are like the original settlers who built the foundation.
- The "Low-Alpha" Group: These are younger stars (mostly under 6 billion years old). They live in the "thin disc" (the flat, main part of the galaxy where the Sun is). They are like the newer residents who moved in later.
The Drifters: The team discovered that stars don't stay where they were born. Over billions of years, they drift outward or inward, and up and down. The older stars have had more time to drift, so they are found all over the place, even far from the center. The younger stars are still mostly hanging out near where they were born.
The "Ex-Situ" Visitors: Some stars in the sample have chemical fingerprints that suggest they weren't born in the Milky Way at all. They were likely "stolen" from smaller galaxies that crashed into ours billions of years ago. The team found that these visitors are slightly younger (about 8.6 billion years) than the native old stars, fitting the story that they arrived during a major galactic merger.
The "Chemical Clocks": The team checked if the ratio of certain chemicals in the stars (like Carbon to Nitrogen) matched their calculated ages. It did! This confirms that their dating method is working correctly, even for very low-mass stars that other surveys haven't studied much yet.

4. Why This Matters

This paper isn't just about listing numbers; it's about providing a reliable training set.

The Analogy: Imagine trying to teach a computer to guess the age of a star just by looking at a picture. To do that, the computer needs to study thousands of examples where the age is already known and verified.
This paper provides that verified dataset. It gives machine learning algorithms a "textbook" of 17,000 stars with known ages, which will help astronomers date millions more stars in the future, even those where we can't measure the "heartbeat" directly.

In short: The authors used the "vibrations" of 17,000 stars to build a detailed timeline of our galaxy's history, separating the ancient settlers from the newer residents, identifying visitors from other galaxies, and creating a master key to help future scientists unlock the ages of stars across the entire universe.

Technical Summary: Asteroseismic Ages for 17,000 Stars in Kepler, K2, and TESS

Problem and Motivation
The availability of precise asteroseismic constraints for tens of thousands of red giant (RG) stars has enabled robust age estimates, a critical component for understanding stellar evolution and the formation history of the Milky Way (MW). While previous studies have established ages for specific fields or populations, there is a need for large, homogeneous samples spanning different Galactic domains (disc, thick disc, and halo) to address open questions in Galactic archaeology. This work addresses the challenge of deriving reliable stellar parameters—specifically mass and age—for a vast sample of RGs by combining data from multiple photometric missions (Kepler, K2, TESS) with astrometric (Gaia DR3) and spectroscopic (APOGEE DR17, GALAH DR3) constraints. A primary motivation is to characterize the reliability of these parameters, particularly given the varying observation durations and data qualities across missions, to support both direct astrophysical analysis and the training of machine-learning age estimators.

Methodology
The study compiles a sample of over 17,000 RGs by cross-matching asteroseismic catalogues from Kepler (Yu et al. 2018), K2 (using the EVEREST pipeline and BHM/COR analysis methods), and TESS (Southern Continuous Viewing Zone) with Gaia DR3 astrometry and spectroscopic data from APOGEE DR17 and GALAH DR3.

Stellar parameters (mass $M_*$ , radius $R_*$ , age $\tau$ , and distance $D$ ) are inferred using the Bayesian code PARAM. The analysis employs a reference grid of MESA models (G2) including microscopic diffusion and a linear helium-metallicity relation. The study rigorously tests the sensitivity of results to different observational constraints:

Asteroseismic Constraints: Global parameters $\nu_{\text{max}}$ (frequency of maximum power) and $\Delta\nu$ (average large frequency separation) are used alongside effective temperature ( $T_{\text{eff}}$ ) and metallicity ([Fe/H]).
Luminosity Constraints: Alternatively, the study utilizes Gaia-derived luminosities ( $L$ ) combined with $\nu_{\text{max}}$ and $T_{\text{eff}}$ .
Quality Control and Flagging:
- $\Delta\nu$ Reliability: The authors identify stars with unreliable $\Delta\nu$ measurements, particularly in short-duration K2 campaigns, by comparing masses derived from $\Delta\nu$ versus those derived from $L$ . Stars showing significant discrepancies (indicating bias in $\Delta\nu$ ) are flagged and removed.
- Parallax Zero-Point: Different corrections for Gaia parallax zero-points are applied (Lindegren et al. 2021 for Kepler/TESS; a constant 17 $\mu$ as for K2) to assess sensitivity.
- Prior Sensitivity: Stars where the inferred age is dominated by the prior (specifically the 20 Gyr upper limit) are flagged.
- Evolutionary State: While evolutionary states (RGB vs. Core-He Burning) are used for Kepler where available, K2 and TESS samples are treated conservatively without state-specific priors to avoid bias.

Orbital parameters (guiding radius $R_g$ , maximum vertical excursion $Z_{\text{max}}$ , eccentricity $e$ , etc.) are calculated using the galpy package and the MWPotential2014 potential, utilizing covariance matrices from Gaia and spectroscopic radial velocities.

Key Contributions

Homogeneous Catalogue: The paper presents a unified catalogue of over 17,000 RGs with homogeneously determined stellar and orbital parameters.
Reliability Characterization: It provides a detailed assessment of the reliability of asteroseismic ages, specifically highlighting the impact of short observation durations (K2) on $\Delta\nu$ measurements and the sensitivity of luminosity-based ages to $T_{\text{eff}$ scales and parallax zero-points.
Quality Flags: The authors introduce specific quality flags to identify stars with unreliable $\Delta\nu$ or those whose ages are heavily influenced by priors, offering a "cleaned" subset for Galactic archaeology studies.
Extended Parameter Space: The K2 sample extends known mass-[C/N] ratio relationships to lower masses, probing older stellar populations not well-sampled by Kepler or TESS.

Results

Kinematics and Chemistry: The study maps the samples onto kinematic planes (Toomre diagram, $E$ vs. $L_z$ ) and chemical spaces ([ $\alpha$ /M] vs. [Fe/H]). It confirms that the TESS sample is confined to the thin disc, Kepler extends into the thick disc, and K2 provides superior coverage of the halo and inner Galaxy.
Age Trends: Analysis of $R_g$ and $Z_{\text{max}}$ as a function of age reveals that older stars exhibit wider distributions in guiding radius and vertical excursion, consistent with radial migration and vertical heating.
Population Differences: High- $\alpha$ populations peak at $\approx$ 10 Gyr, while low- $\alpha$ populations peak at 6–7 Gyr. The study identifies "ex situ" stars (chemically selected via [Mg/Mn] vs. [Al/Fe]) which are, on average, slightly younger than in situ high- $\alpha$ stars.
Internal Mixing: The [C/N] ratio vs. mass relationship is confirmed across all missions, with increased scatter at low masses attributed to initial abundance variations in old, metal-poor halo stars.
Chemical Clocks: Using GALAH data, the study validates chemical clocks (e.g., Y/Mg, Ba/Mg) against asteroseismic ages, demonstrating their utility for inferring ages in populations lacking direct asteroseismic constraints.

Significance
The authors state that these catalogues are instrumental for addressing key questions in Galactic archaeology and stellar evolution. Specifically, the data allows for time-resolved studies of different stellar populations across the Milky Way. The K2 dataset, in particular, is highlighted for its spatial range, which has already been utilized in subsequent works to study radial metallicity gradients, cerium enrichment, and young $\alpha$ -rich stars. Furthermore, the robust, homogeneously determined ages serve as high-quality training sets for machine-learning techniques, which are essential for scaling age determinations to the vast populations expected from future missions like PLATO and the Nancy Grace Roman Telescope. The paper emphasizes that while the results are robust, the identified limitations (such as $\Delta\nu$ biases in K2 and $T_{\text{eff}}$ sensitivity in luminosity-based ages) must be carefully considered when applying these data to specific astrophysical problems.

1. The "Heartbeat" of the Stars (Asteroseismology)

2. Building a Better Map

3. What the Library Tells Us (Key Findings)

4. Why This Matters

More like this