Globular cluster dating

Globular clusters as cosmic chronometers

Globular clusters are dense, spherical collections of typically 100,000 to over one million stars, bound by mutual gravity and orbiting in the halos and bulges of galaxies. In the Milky Way, approximately 150 globular clusters are known, and spectroscopic analysis of their stars reveals very low abundances of elements heavier than helium (low metallicity), consistent with formation early in galactic history when the interstellar medium had been enriched by only a few generations of supernovae.^{1, 10} Because the stars in a globular cluster formed essentially simultaneously from the same gas cloud, a cluster provides a natural experiment: a coeval, chemically homogeneous population of stars of different masses, all at the same distance, evolving together. The age of the oldest globular clusters therefore sets a firm lower limit on the age of the universe.^{1, 2}

The color-magnitude diagram and main-sequence turnoff

The principal tool for dating globular clusters is the color-magnitude diagram (CMD), the observational analog of the Hertzsprung-Russell diagram. When the apparent magnitudes of a cluster’s stars are plotted against their colors (a proxy for surface temperature), the stars trace a characteristic pattern: a main sequence of hydrogen-burning stars, a subgiant branch, a red giant branch, a horizontal branch, and in some cases an asymptotic giant branch.⁶

The key feature for age determination is the main-sequence turnoff (MSTO) — the point at which the main sequence bends toward the subgiant branch. Stars at the turnoff are just exhausting hydrogen in their cores and beginning to evolve off the main sequence. Because the rate of hydrogen burning depends strongly on stellar mass (roughly as M^3.5 for main-sequence luminosity), higher-mass stars exhaust their fuel faster and turn off the main sequence sooner.^{1, 6} In a young cluster, the turnoff occurs at high luminosity (massive, blue stars are still on the main sequence). In an old cluster, only low-mass, faint stars remain on the main sequence, and the turnoff occurs at correspondingly lower luminosities. The luminosity of the turnoff is therefore a direct clock: it measures the time elapsed since the cluster formed.^{1, 2}

Isochrone fitting

To convert the observed turnoff luminosity into an age, astronomers compare the observed CMD with theoretical isochrones — curves that predict the positions of stars of different masses in the CMD at a given age and metallicity, calculated from stellar evolution models.^{3, 6} The best-fitting isochrone yields the cluster age. This process, called isochrone fitting, requires three primary inputs: the cluster’s distance (to convert apparent magnitudes to absolute magnitudes), its metallicity (which affects the opacity and hence the structure of the stars), and a set of stellar evolution models that predict the theoretical isochrones.^{1, 3}

Modern isochrone fitting incorporates sophisticated stellar physics including nuclear reaction rates, radiative and convective energy transport, equation of state for stellar matter, opacity tables, and atmospheric models for converting internal stellar properties to observable colors and magnitudes.^{3, 6}

Ages of the oldest clusters

Applying the MSTO method to the most metal-poor globular clusters consistently yields ages of 11–13 Gyr. Among the best-studied ancient clusters, 47 Tucanae has a derived age of approximately 11.5–12.5 Gyr, M92 yields 12–14 Gyr, and M3 gives approximately 11.5–12.5 Gyr, depending on the adopted distance scale and stellar models.^{2, 5, 9}

A comprehensive study by Marin-Franch and colleagues (2009) used homogeneous ACS/HST photometry of 64 globular clusters and derived ages for 55 of them using a differential MSTO method that minimizes sensitivity to distance and reddening errors. They found that the oldest clusters have ages of approximately 12.8 Gyr (with a spread of about 1 Gyr among the most metal-poor systems), while more metal-rich clusters tend to be 1–2 Gyr younger.⁷ These results are consistent with a picture in which the Galaxy’s halo formed over a period of 1–2 Gyr beginning approximately 12–13 Gyr ago.

Relationship to cosmological age estimates

The globular cluster ages provide a critical cross-check on the age of the universe derived from cosmological observations. The Planck satellite’s analysis of the cosmic microwave background yields an age of 13.80 ± 0.02 Gyr under the standard Lambda-CDM cosmological model.⁸ For consistency, the oldest globular clusters must be younger than the universe — and they are, with typical best-estimate ages of 12–13 Gyr leaving roughly 0.5–1.5 Gyr for the formation of the first stars and the assembly of the clusters themselves, a timeline that agrees with theoretical predictions for early structure formation.^{2, 5}

This concordance was not always the case. Through much of the 20th century, globular cluster ages appeared older than the universe implied by the Hubble constant, creating the “cosmic age problem.” The resolution came from improved distance measurements (which reduced cluster ages) and the discovery of accelerating expansion (which increased the cosmological age), bringing the two estimates into agreement.^{2, 4}

Sources of uncertainty

The dominant systematic uncertainties in globular cluster ages arise from several sources. Distance is the single largest contributor: a 10 percent error in distance translates to approximately a 2 Gyr error in age, because the inferred absolute luminosity of the turnoff changes with the square of the distance.^{1, 4} Historically, distances to globular clusters relied on RR Lyrae variables, horizontal branch fitting, and main-sequence fitting to nearby subdwarfs with parallaxes from Hipparcos. Gaia parallaxes have dramatically improved subdwarf distances and are progressively reducing this source of error.^{4, 12}

Helium abundance is another significant uncertainty. Globular clusters show evidence of multiple stellar populations with different helium abundances, and helium-enriched populations evolve faster (yielding younger apparent ages for a given turnoff luminosity). The discovery that essentially all globular clusters host multiple populations with helium spreads of up to ΔY = 0.10–0.15 has complicated the simple picture of a single-age, single-composition cluster.^{10, 11}

Convective overshooting — the extent to which material in convective regions penetrates beyond the classical boundary — affects the size of the convective core and hence the main-sequence lifetime. Greater overshooting extends the main-sequence lifetime and yields older ages for a given turnoff luminosity. The treatment of overshooting varies among stellar evolution codes and remains a source of systematic scatter in derived ages.^{3, 6}

Relative age ranking and formation history

Beyond absolute age determination, the differential comparison of globular cluster CMDs reveals a systematic pattern of relative ages that constrains the formation history of the Milky Way. De Angeli and colleagues developed a parameter-free method for measuring relative ages by comparing the colour difference between the main-sequence turnoff and the base of the red giant branch, a quantity that is sensitive to age but relatively insensitive to distance and reddening uncertainties.¹⁸ Their analysis, together with similar studies, demonstrates that the most metal-poor clusters in the outer halo are uniformly old (within approximately 1 Gyr of each other), while more metal-rich clusters associated with the inner halo and thick disk show a wider age spread, with some being 2–3 Gyr younger.^{7, 18}

This age-metallicity relationship is consistent with hierarchical galaxy formation models in which the outer halo was assembled first from primordial gas clouds, while the inner regions continued to accrete enriched material and form clusters over a longer period. The age spread among the more metal-rich clusters may also reflect the accretion of satellite galaxies, each with its own globular cluster system formed at different times.^{10, 18}

The horizontal-branch morphology of globular clusters — the distribution of stars along the horizontal branch, ranging from blue to red — has been used as an independent age indicator. At a given metallicity, older clusters tend to have bluer horizontal branches because their lower-mass stars have had more time to evolve. This “second parameter” effect (metallicity being the first) provided early evidence for an age spread among globular clusters before high-precision turnoff photometry became available.¹⁷

The Gaia era

The Gaia mission has brought dramatic improvements to globular cluster distance determinations and, by extension, to absolute age estimates. Gaia parallaxes for field subdwarf stars — the local calibrators used in main-sequence fitting — are an order of magnitude more precise than those from Hipparcos, substantially reducing the dominant source of uncertainty in the distance-age chain. O’Malley and colleagues used Gaia EDR3 subdwarf parallaxes to derive ages for a sample of well-studied globular clusters, obtaining 12.2 ± 0.8 Gyr for M92 and 11.7 ± 0.8 Gyr for 47 Tucanae.¹⁶ These results are consistent with earlier estimates but carry smaller uncertainties, progressively tightening the concordance between stellar and cosmological ages.

Complementary methods

Globular cluster ages derived from the MSTO can be cross-checked using independent methods. White dwarf cooling ages measure the elapsed time since the oldest white dwarfs in a cluster formed by fitting the faint end of the cluster’s white dwarf cooling sequence to cooling models. Hansen and colleagues applied this method to M4, deriving an age of 12.7 ± 0.7 Gyr, in excellent agreement with the MSTO-based age.¹³ Nucleocosmochronology — using the abundance ratios of radioactive isotopes such as thorium-232 and uranium-238 in the oldest metal-poor stars — provides yet another independent age constraint, yielding ages broadly consistent with those from the MSTO and white dwarf methods.⁵ Additionally, the baryon density measured independently from primordial deuterium abundances constrains the cosmological parameters used to calibrate isochrone ages, tightening the connection between cluster dating and Big Bang nucleosynthesis.¹⁴

The convergence of these independent dating techniques on a consistent age range of approximately 11–13 Gyr for the oldest stellar populations, younger than the 13.8 Gyr CMB-derived age of the universe, provides one of the most robust consistency checks in modern astrophysics and a cornerstone of the cosmic concordance between cosmological and stellar age estimates. The techniques developed for globular clusters have also been applied to old open clusters, where CMD morphology and white dwarf cooling sequences provide additional calibration points.^{5, 8, 15}