Asteroseismology

Why a star is a resonator

Strike a bell and it rings, because the geometry of the bell singles out a discrete set of standing-wave frequencies — its eigenmodes — and ambient stochastic forcing exciles them. A star does the same thing on a vastly larger scale. Its convective envelope is a turbulent reservoir of mechanical energy; that energy leaks broadband into the spectrum of normal modes that the star's stratification permits. Some modes get amplified, others damp; the ones that survive are observable as periodic brightness or velocity fluctuations of the surface.

The key distinction from a bell is that a star is not uniform. Sound speed and density both vary by orders of magnitude from core to surface, and so do gravity, composition, and rotation rate. Each oscillation mode samples a different region: shallow modes carry information about the envelope, deep modes about the core. Measure enough modes, invert the eigenvalue problem, and you reconstruct the run of the relevant physical quantities along the radius. This is exactly the strategy that seismologists use to map the Earth's interior — the field is called asteroseismology by direct analogy.

Two families of waves: p-modes and g-modes

The eigenmodes of a non-rotating, non-magnetic star fall into two well-separated families, distinguished by which restoring force dominates.

Pressure modes (p-modes) are acoustic waves. The restoring force is gas pressure; the dispersion relation reduces to the sound wave equation in the WKB limit. p-modes propagate where the wave frequency exceeds the local Lamb frequency, which essentially means anywhere with sufficient compressibility — primarily the convective envelope. The Sun's famous "five-minute oscillation" is the integrated signature of millions of p-modes in the envelope. Periods are short, of order minutes for the Sun and similar for other cool stars; amplitudes at the surface are a few cm/s in velocity and a few parts per million in brightness.

Gravity modes (g-modes) are buoyancy waves. Displace a fluid parcel vertically in a stably stratified layer and Archimedes' principle pulls it back; let it overshoot and it oscillates at the Brunt-Väisälä frequency. g-modes propagate only where the medium is convectively stable, i.e. in the radiative interior. Periods are far longer — minutes to days — because the buoyancy frequency is set by the much weaker entropy gradient rather than the sound speed. The convective envelope of a cool star acts as an evanescent layer for g-modes: their amplitude is exponentially attenuated by the time they reach the surface, which is the principal reason why solar g-modes have eluded definitive detection despite half a century of effort.

In between sit the mixed modes of subgiants and red giants: oscillations that have g-mode character in the contracted helium core and p-mode character in the puffed-up envelope. These are particularly powerful because a single observed frequency probes both regions, dramatically improving the inversion.

Labelling modes: n, l, m

The spatial dependence of every linear mode of a spherical body separates into a radial wave function times a spherical harmonic Y_l^m(θ, φ). Each mode is therefore tagged by three quantum-number-like integers:

• n — the radial order, counting the number of nodal surfaces from centre to surface. Large n means many wiggles in radius.
• l — the angular degree, counting the number of nodal great circles on the surface. l = 0 is purely radial ("breathing"); l = 1 is dipole; l = 2 quadrupole, and so on.
• m — the azimuthal order, with |m| ≤ l, counting how many of the nodal lines are meridians. Without rotation, modes with the same n, l but different m are degenerate; rotation breaks the degeneracy and splits each l-multiplet into 2l + 1 sublines, exactly as the Zeeman effect splits atomic levels.

From Earth we only ever see disk-integrated light or velocity, so high-l modes are almost completely cancelled by averaging. For Sun-as-a-star observations and for Kepler photometry of other stars, only l = 0, 1, 2, and occasionally 3 are practically detectable.

The asteroseismic scaling relations

Solar-like oscillators — F, G, K dwarfs and red giants — all show p-mode comb structures with two distinctive scales: the large frequency separation Δν, which is the average spacing between consecutive radial orders of the same l, and the frequency of maximum power ν_max, where the envelope of mode amplitudes peaks. Both turn out to scale cleanly with global stellar parameters.

Δν   ∝ √(M / R³)                  large separation ∝ √mean density
ν_max ∝ M / (R² √T_eff)           Brown 1991, Kjeldsen & Bedding 1995

Δν follows from the asymptotic theory of high-order acoustic modes: the sound-travel time across a star is essentially the diameter divided by the average sound speed, and for an ideal gas the average sound speed scales as √T̄ — which, by hydrostatic equilibrium, scales as √(M/R). Putting these together gives Δν ∝ √(M/R³), i.e. the square root of the mean density. The ν_max relation is more empirical but justified by the argument that mode excitation peaks at the acoustic cutoff frequency of the photosphere.

Solving the two relations simultaneously gives M and R independently, with the Sun as the calibrator:

M / M☉ ≈ (ν_max / ν_max,☉)³ (Δν / Δν,☉)⁻⁴ (T_eff / T_eff,☉)^(3/2)
R / R☉ ≈ (ν_max / ν_max,☉)   (Δν / Δν,☉)⁻²  (T_eff / T_eff,☉)^(1/2)

where ν_max,☉ ≈ 3090 µHz, Δν☉ ≈ 135 µHz, T_eff,☉ = 5777 K. For a Kepler red giant with ν_max ≈ 30 µHz, Δν ≈ 4 µHz, T_eff ≈ 4800 K, the relations return roughly M ≈ 1.4 M☉ and R ≈ 11 R☉ — typical for a red clump star — with formal uncertainties of a few percent.

The small separation as an age clock

Where Δν captures the global density, the small frequency separation δν₀₂ — the gap between an l = 0 mode and the nearest l = 2 mode of one higher radial order — probes the sound-speed gradient at the very core, where the l = 2 mode has its turning point but l = 0 does not. On the main sequence, δν₀₂ decreases monotonically as the central hydrogen fraction falls: a hot, fully ionised hydrogen-burning core has a different sound-speed profile than a partly helium-converted one. Plotted against Δν, evolutionary tracks of different mass sweep out a fan in the (Δν, δν₀₂) plane known as the Christensen-Dalsgaard or C-D diagram. Locate a star on that diagram and you read off both its mass and its evolutionary age — directly, from frequencies alone, without isochrone fitting.

Helioseismology — the Sun in 3D

The five-minute oscillation was discovered by Robert Leighton, Robert Noyes and George Simon in 1962, using a magnetograph at Mt Wilson to look for Doppler shifts on the solar photosphere. For more than a decade nobody knew what they were; Roger Ulrich, Henry Hill and others gradually established that they are trapped p-modes — global oscillations of the whole Sun, not surface phenomena. Once that was clear, the floodgates opened.

Era	Instrument	Result
1962	Mt Wilson magnetograph	Discovery of 5-min oscillation
1970s	BiSON (ground network)	Identified as global p-modes
1980s	GONG (ground network)	Resolved (n, l, m) mode spectrum
1996+	SOHO / MDI (L1)	Inverted sound-speed and rotation profile
2010+	SDO / HMI	Continuous Doppler imaging at 4096²

The principal results of helioseismology are textbook:

• Base of the convection zone at 0.713 R☉. A sharp discontinuity in the sound-speed gradient marks the transition from radiative interior to convective envelope, fixed by inversions to within 0.001 R☉.
• Differential rotation in the envelope, solid-body in the core. The convection zone rotates with latitude (equator faster than poles), but the radiative interior below 0.7 R☉ rotates as a rigid body. The thin shear layer between them is the tachocline, the conjectured seat of the solar dynamo.
• Helium abundance Y_surf ≈ 0.249. The signature of helium's second ionisation zone shows up as a small dip in Δν versus n; fitting it gives the surface helium fraction with high accuracy.
• Solar neutrino vindication. The 1990s helioseismic sound-speed profile agreed with standard solar-model predictions to better than 0.5 percent, ruling out astrophysical solutions to the missing solar neutrino problem and pushing the resolution onto neutrino physics — confirmed by SNO and Super-Kamiokande in 2001-2002.

Kepler and the red-giant revolution

Solar-like oscillations of other stars are hard. The mode amplitudes are micromagnitudes; the lifetimes are days; you need uninterrupted, photometrically stable, high-cadence light curves for weeks. None of those existed at scale before Kepler. The mission's prime goal was transit photometry of Sun-like dwarfs, but a happy by-product of staring at 150,000 stars for 4 years was that, for the brighter red giants, the asteroseismic mode comb sat well above the noise.

By the end of the prime mission, the Kepler Asteroseismic Science Consortium had catalogued Δν and ν_max for over 16,000 red giants. The Sun, suddenly, was a sample of one in a sea of thousands. Eclipsing-binary red giants validated the scaling-relation masses and radii to within 5 percent, and grid-based modelling using the full mode pattern improved that further. Population-level results followed:

• Red-clump vs. RGB discrimination. Helium-burning clump stars and hydrogen-shell-burning red giant branch stars have nearly identical (Δν, ν_max) but different period spacings of their gravity-dominated mixed modes — Kepler resolved this and gave an unambiguous evolutionary tag for every giant.
• Core rotation of red giants. Rotational splitting of mixed modes revealed that giant cores rotate roughly 10× faster than envelopes, but far slower than angular-momentum-conserving contraction would predict — implying efficient internal angular-momentum transport that is still not fully explained.
• Galactic archaeology. Asteroseismic ages combined with Gaia distances and APOGEE/GALAH abundances produced the first detailed age-metallicity-kinematics maps of the Milky Way thick and thin disks, resolving long-standing degeneracies in stellar population studies.

Classical pulsators beyond the solar-like regime

Solar-like oscillations are stochastic — driven by convection and damped by it. They are not the only kind of stellar pulsation. The HR diagram is dotted with instability strips where coherent, large-amplitude pulsations are self-excited by the κ-mechanism (opacity bumps in partially ionised hydrogen or helium drive heat into the pulsation cycle).

Class	HR-diagram location	Periods	Mode types	Diagnostic
Classical Cepheids	F-G supergiants, instab. strip	1 – 100 d	Radial p-modes	P-L relation (distances)
RR Lyrae	Horizontal branch, instab. strip	0.2 – 1 d	Radial, sometimes l = 1	Halo distances
δ Scuti	A-F dwarfs / subgiants	0.5 – 6 h	Low-order p and mixed	Core mixing
γ Doradus	F dwarfs, just cooler than δ Sct	0.3 – 3 d	High-order g-modes	Near-core rotation
β Cephei	O-B main sequence	2 – 8 h	Low-order p and g	Massive-star interiors
SPB (slowly pulsating B)	B dwarfs	0.5 – 5 d	High-order g-modes	Core overshooting
White dwarf (ZZ Cet, DBV, GW Vir)	WD cooling track	1 – 30 min	g-modes	WD interior, crystallisation

γ Doradus stars are particularly valuable because their high-order g-modes form a near-uniform period spacing — the gravity-mode analogue of the p-mode Δν — and any deviation from that spacing encodes the sharp gradient at the convective core boundary. Combined δ Scuti / γ Doradus hybrids therefore probe both the radiative outer envelope (via p-modes) and the convective core (via g-modes) in the same star.

Worked example — masses for a Kepler red giant

Suppose Kepler delivers the following measurements for a red giant: Δν = 4.0 µHz, ν_max = 30 µHz, T_eff = 4800 K. The scaling relations give

M / M☉ = (30 / 3090)³ × (4.0 / 135)⁻⁴ × (4800 / 5777)^(3/2)
       = (9.71 × 10⁻³)³ × (2.96 × 10⁻²)⁻⁴ × 0.762
       ≈ 9.16 × 10⁻⁷ × 1.30 × 10⁶ × 0.762
       ≈ 0.91
M ≈ 0.91 M☉

R / R☉ = (30 / 3090) × (4.0 / 135)⁻² × (4800 / 5777)^(1/2)
       = 9.71 × 10⁻³ × 1138 × 0.913
       ≈ 10.1
R ≈ 10.1 R☉

So the star is a red giant of about a solar mass, swollen to ten solar radii. With a measured period spacing of the gravity-dominated mixed modes, one could further distinguish whether it sits on the first ascent of the RGB or the helium-burning red clump — settling a question that has historically required colour-magnitude diagram membership or sparse spectroscopic surface-gravity measurements.

Common pitfalls

• Treating the scaling relations as exact. Δν ∝ √ρ̄ rests on an asymptotic argument that breaks down for evolved giants where the envelope is no longer near-isothermal. Corrections of a few percent must be applied at metal-poor and at high-luminosity end; modern work uses model-calibrated reference values rather than the solar value.
• Confusing the small separation δν₀₂ with the rotational splitting. Both produce close pairs of modes near each Δν peak. δν₀₂ separates l = 0 and l = 2 of consecutive n; rotational splitting separates m components of one (n, l). The first is set by interior structure, the second by rotation rate. Failing to disentangle them gives spurious masses.
• Assuming the Sun is calibrated. The "surface effect" — a systematic offset between observed and computed high-frequency p-modes — is a few µHz in the Sun and depends on near-surface convection that is not yet fully captured by 1D stellar models. Asteroseismic analyses correct for it empirically or via 3D hydro simulations.
• Ignoring background. The granulation background in cool stars rises as a power law toward low frequency and can swamp ν_max if the analysis pipeline does not fit it correctly. Misfit backgrounds shift ν_max systematically.
• Forgetting that g-modes don't make it through convection. Solar g-modes are predicted at amplitude orders of magnitude below their excitation level because the convective envelope acts as an evanescent layer. Any claim of solar g-mode detection has to either invoke unusually efficient transmission or has been disputed.

PLATO and what comes next

ESA's PLATO mission, launched in 2026, was conceived as the asteroseismic successor to Kepler. Its 26 small telescopes plus 2 fast cameras stare at large fields of bright stars at high cadence, with photometric stability matched to the few-ppm amplitudes of solar-like oscillations. Over a planned decade, PLATO will scan large fractions of the sky and deliver high-quality light curves for several million stars — hundreds of thousands of which will be solar-like oscillators with measurable Δν and ν_max, giving asteroseismic masses, radii and ages.

The science driver is not asteroseismology in isolation. PLATO's primary deliverable is Earth-analogue exoplanets, and the host-star characterisation by asteroseismology is what turns transit-and-radial-velocity observations into planet masses, radii, and ages. Asteroseismology, having transformed stellar physics, is now the rate-limiting step in characterising the exoplanets we find. After PLATO, the same dataset will support galactic archaeology at unprecedented scale, the calibration of gyrochronology, the mapping of stellar rotation and magnetic activity across the HR diagram, and — quite possibly — the long-elusive detection of solar g-modes.