🔭 astrophysics

Asteroseismically Inferred Ages of 132,000 Red Giants with TESS

This study utilizes TESS oscillation data and custom MESA stellar evolution models to derive asteroseismic ages for approximately 132,000 red giants with an average uncertainty of 23%, providing a vital resource for future galactic archaeology research.

Original authors: Artemis Theano Theodoridis, Leslie Morales, Jamie Tayar

Published 2026-02-16

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Artemis Theano Theodoridis, Leslie Morales, Jamie Tayar

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 giant, bustling city that has been growing and changing for billions of years. To understand its history—where the neighborhoods formed, how the buildings aged, and which parts were added later—we need to know the ages of the stars living there.

However, stars don't wear birth certificates. For decades, astronomers had to guess a star's age by looking at how bright it is and what color it shines, but this was like trying to guess a person's age just by looking at their skin tone; it's often inaccurate and depends heavily on which "rulebook" (model) you use.

This paper is a massive breakthrough in solving that problem. Here is the story of how the authors did it, explained simply:

1. The New "City Census" (TESS and Gaia)

The authors used two powerful space telescopes as their data sources:

TESS (Transiting Exoplanet Survey Satellite): This is like a giant, all-sky security camera that watches almost the entire sky. It found 158,000 red giant stars (stars that are old, puffy, and nearing the end of their lives) that are "humming" or vibrating.
Gaia: This is like a super-precise GPS for stars, telling us exactly how far away they are and how big they are.

2. Listening to the "Heartbeat" (Asteroseismology)

The secret sauce here is Asteroseismology.

The Analogy: Imagine a bell. If you strike a small, light bell, it rings at a high pitch. If you strike a huge, heavy bell, it rings at a low pitch.
The Science: Stars vibrate like bells. By listening to the specific "notes" (frequencies) a star hums, astronomers can weigh it with incredible precision. This is called Asteroseismology.
The Breakthrough: In the past, we could only "hear" a few thousand stars. With TESS, they could "hear" over 150,000. Because they know the mass (weight) of the star from its vibration, they can calculate its age much more accurately than before. It's like knowing a person's exact weight and height allows you to guess their age much better than just looking at their face.

3. Building a Custom "Time Machine" (Stellar Models)

To turn that mass into an age, you need a simulation.

The Problem: Astronomers have used different "rulebooks" (models) for decades, and they often disagreed on the age of the same star.
The Solution: The authors built their own custom "Time Machine" using a super-computer code called MESA. They didn't just use a generic rulebook; they tailored it specifically for the types of stars they found (red giants with specific weights and chemical compositions).
The Result: They ran simulations for thousands of different star types to create a massive grid. This grid acts like a lookup table: "If a star weighs X and has Y amount of iron, it is Z years old."

4. The Big Reveal: 132,000 New Ages

After cross-referencing the TESS vibrations, Gaia distances, and their custom computer models, they successfully calculated the ages for 132,794 stars.

The Accuracy: They are about 23% accurate on average. In the world of stellar archaeology, this is a huge leap forward. Previously, for red giants, age guesses could be off by 80% if the mass wasn't known. Now, with the mass known, the error dropped to roughly 23%.
The Map: They found that these stars follow similar age patterns to what we saw in smaller, older studies (like those from the Kepler mission), but now we have a much bigger, more diverse map of the galaxy.

5. Why This Matters (Galactic Archaeology)

Think of the Milky Way as a layered cake.

The core is the oldest part.
The thin disk (where our Sun lives) is younger.
The halo is a mix of old stars and stars that were "stolen" from other small galaxies that crashed into ours.

By having the ages of 132,000 stars spread across the sky, astronomers can finally start to:

Reconstruct the timeline of how our galaxy was built.
Figure out when different "neighborhoods" (like the thick disk or the halo) formed.
Understand how the galaxy ate smaller galaxies over time.

The Bottom Line

This paper is like finding a massive, dusty library of birth certificates for 132,000 stars that no one knew existed before. By listening to their "heartbeats" and using a custom-built computer model, the authors have given us a much clearer picture of the Milky Way's history. They have handed this data to the entire scientific community, allowing anyone to dig deeper into the story of our cosmic home.

In short: They turned a blurry, guesswork-heavy map of the galaxy into a high-definition, time-stamped history book.

1. Problem Statement

Accurate stellar ages are fundamental to Galactic archeology, yet determining them for large populations faces two primary challenges:

Limited Generalizability: Previous precise age determinations were restricted to small sky areas (e.g., the Kepler field), limiting their applicability to the entire Galaxy.
Model Dependence and Uncertainty: Stellar ages are not directly measurable and rely heavily on stellar evolution models. For Red Giant Branch (RGB) stars, age uncertainties can reach up to 80% when mass is unknown, as models of different masses exhibit similar temperatures on the RGB. This leads to significant discrepancies between studies depending on the physics assumptions (e.g., convective overshoot, mixing length) and model grids used.

The authors aim to overcome these limitations by leveraging the all-sky coverage of the Transiting Exoplanet Survey Satellite (TESS) to provide a massive, homogeneous sample of red giants with seismically constrained masses, thereby reducing model-dependent systematics.

2. Methodology

Data Collection and Cross-Matching

Seismology: The study utilizes a catalog of 158,000 oscillating red giants identified in TESS data (Years 1 and 2) by M. Hon et al. (2021) using deep learning on 2D power spectra.
Astrometry & Radii: Radii ( $R$ ) are derived from Gaia EDR3 parallaxes using the isoclassify package.
Spectroscopy: Effective temperatures ( $T_{\text{eff}}$ ) and metallicities ([M/H]) are estimated using XGBoost machine learning models trained on Gaia XP spectra and calibrated against APOGEE DR17.
Sample Selection: After cross-matching TESS, Gaia, and XGBoost data, the final sample consists of 134,901 red giants.

Seismic Mass and Surface Gravity Calculation

Masses ( $M$ ) and seismic surface gravities ( $\log g_{\text{seis}}$ ) are calculated using the standard asteroseismic scaling relations:
$\frac{M}{M_\odot} = \left( \frac{\nu_{\text{max}}}{\nu_{\text{max},\odot}} \right)^3 \left( \frac{R}{R_\odot} \right)^2 \left( \frac{T_{\text{eff}}}{T_{\text{eff},\odot}} \right)^{-1.5}$
Where $\nu_{\text{max}}$ is the frequency of maximum power. The authors adopt a conservative uncertainty of ~6% for $\nu_{\text{max}}$ based on cross-validation with independent pipelines.

Stellar Evolution Modeling (MESA)

Grid Construction: The authors constructed a custom grid of 619 stellar models using MESA (Modules for Experiments in Stellar Astrophysics).
Parameters: The grid covers masses from 0.7 to 3.0 $M_\odot$ and initial metallicities ([Fe/H]) from -2.0 to +0.5.
Physics: The models include updated physics tailored to the sample, such as the Henyey mixing length scheme ( $\alpha_{\text{MLT}} = 1.94$ ), OPAL/Ferguson opacities, and specific nuclear reaction networks. Non-standard processes like rotation and binary interactions are excluded to maintain consistency with standard large-sample analysis, though their impact is accounted for in the error budget.
Solar Calibration: The grid is calibrated to match solar temperature, luminosity, and age (4.57 Gyr).

Age Interpolation and Mass Loss

Interpolation: The open-source package kiauhoku is used to interpolate stellar parameters (Mass, Metallicity, Equivalent Evolutionary Point) from the MESA grid to the observed stars. A four-point cubic interpolation scheme is used where possible to capture non-linear age-mass relationships.
Evolutionary State Classification: To distinguish between RGB stars and Red Clump (RC) stars, the authors use a temperature difference metric ( $\Delta T_{\text{eff}}$ ). Stars are classified based on whether their observed $T_{\text{eff}}$ matches the model prediction for an RGB or RC phase. A cutoff of 100 K is used to separate the populations.
Mass Loss Handling: Three scenarios are computed to address mass loss prior to the RC phase:
1. No Mass Loss: Assumes current mass = birth mass (RGB assumption).
2. Empirical Mass Loss: Uses a metallicity-dependent empirical prescription derived from $\alpha$ -enhanced RGB vs. RC star differences.
3. Theoretical Mass Loss: Uses Reimers/Blöcker wind schemes from the MESA tracks.
  The final adopted ages use the empirically calibrated mass loss (Option 2) combined with a theoretical mass dependence.

3. Key Contributions

Massive Dataset: The paper provides asteroseismic ages for ~132,794 red giants, increasing the known sample of seismically dated stars by an order of magnitude compared to previous Kepler-based studies.
Reduced Systematics: By constraining stellar mass directly via asteroseismology, the authors reduce age uncertainties from ~80% (typical for RGB stars without mass constraints) to an average of 23%.
Custom MESA Grid: A tailored modeling grid covering the specific mass and metallicity ranges of the TESS sample, ensuring high fidelity for the target population.
Open Data: The full catalog, including seismic masses, metallicities, evolutionary states, and ages under different assumptions, is made publicly available (Zenodo) for Galactic archeology research.

4. Results

Age Distribution: The age distributions for RGB and RC stars in the TESS sample show strong agreement with Kepler (APOKASC-3) data, validating the methodology.
Galactic Structure: The TESS sample reveals a higher fraction of younger stars compared to Kepler, attributed to TESS's coverage of the Galactic plane (where younger stars are more common) versus Kepler's focus on the thick disk/halo.
Uncertainties:
- Average mass uncertainty: 12%.
- Average age uncertainty: 23% (including a 10% systematic floor for model-to-model variations).
Outliers: Approximately 4,045 stars (3%) have ages exceeding the age of the Universe (13.8 Gyr). These are likely due to measurement errors, binary interactions, or model convergence issues, particularly at low metallicities and high masses.
Striping Effect: The age-metallicity diagram exhibits a "striping" pattern caused by interpolation weights biasing toward lower-mass tracks and gaps in the model grid where models failed to converge (typically at high metallicity/old ages).

5. Significance

This work represents a paradigm shift in Galactic archeology by providing a homogeneous, all-sky dataset of stellar ages.

Validation of TESS Seismology: It confirms that TESS data, when combined with machine learning and Gaia astrometry, can produce asteroseismic results comparable to the Kepler gold standard.
Future Applications: The dataset serves as a critical training set for machine learning models aiming to estimate ages for even larger, non-oscillating populations.
Galactic Evolution: The ages enable detailed tracing of the Milky Way's formation history, including the evolution of the thin/thick disks, the halo, and accreted dwarf galaxies, with significantly reduced model-dependent biases.

The authors conclude that while challenges remain regarding binary interactions and model physics at extreme parameters, this dataset provides a robust foundation for understanding the chronological evolution of the Milky Way.