Gravity Darkening

A star that is hot at the top and cool around the middle

Take an ordinary star — roughly spherical, the same temperature all over — and start spinning it faster and faster. Two things happen. The equator bulges outward, because centrifugal force fights against gravity and the material there sags away from the center. And, less obviously, that same bulging equator goes cooler and dimmer while the poles, which sit on the rotation axis, stay hot and bright. Spin a star fast enough and you get a body that is several thousand kelvin hotter at its poles than around its waist — a flattened, two-tone star. This is gravity darkening, and for the fastest-spinning stars it is not a subtle correction: it sets the very shape, color and apparent brightness of the star.

The textbook fast rotators make the point. Regulus (Alpha Leonis) spins at about 86% of the velocity that would tear it apart, and its equatorial radius is roughly a third larger than its polar radius. Vega — long used as the photometric standard star — turned out to be a rapid rotator seen nearly pole-on, which is part of why its measured temperature seemed anomalous. Altair completes a rotation in under nine hours. Achernar is the most oblate star known, flattened to an axis ratio near 1.5. In every case the poles are the hot bright caps and the equator is the cool dim belt.

How it works: effective gravity and the von Zeipel theorem

The driving quantity is the effective gravity — the net inward acceleration felt by the gas at the surface once you subtract the outward centrifugal push. In the rotating frame, at a point whose distance from the spin axis is R_perp, the surface material feels

g_eff = g_grav − Ω² R_perp     (vector sum; centrifugal term ⟂ axis)

where Ω is the angular rotation rate. At the poles R_perp = 0, so the centrifugal term vanishes and g_eff equals the full gravitational value. At the equator R_perp is largest, the centrifugal term is largest and points directly outward against gravity, and the equator also sits at a larger radius (the bulge), so the gravitational term is weaker too. Both effects drive g_eff down at the equator. At the critical (breakup) rotation rate, the equatorial g_eff reaches zero — the star is on the verge of shedding material from its waist.

Now connect effective gravity to temperature. In 1924 Hugo von Zeipel proved a remarkable result for a star in hydrostatic and radiative equilibrium rotating as a barotrope: the emergent radiative flux at the surface is everywhere proportional to the local effective gravity,

F ∝ g_eff.

Because the Stefan–Boltzmann law relates the local flux to the local effective temperature by F = σ T_eff⁴, taking the fourth root gives the form quoted everywhere:

T_eff ∝ g_eff^β,   with β = 1/4 = 0.25  (radiative envelope).

So the cool, low-gravity equator radiates less and runs cooler; the high-gravity poles radiate more and run hotter. The exponent β = 0.25 is the classical von Zeipel exponent. For stars whose outer envelopes are convective rather than radiative — cooler stars like the Sun — the flux does not track gravity so steeply; Lucy (1967) derived β ≈ 0.08 from convective atmosphere models. Real stars fall between these limits depending on where the radiative/convective boundary sits in the envelope.

Worked example: how hot is each pole of Regulus?

Let us put numbers on it. Treat the surface as a Roche model — a point-mass potential plus the centrifugal term — and use the von Zeipel law with β = 0.25. The polar-to-equatorial temperature ratio is then just the ratio of effective gravities raised to the β power:

T_pole / T_eq = (g_pole / g_eq)^β = (g_pole / g_eq)^0.25

For Regulus, interferometric models (McAlister et al. 2005; Che et al. 2011) give an equatorial radius about 32% larger than the polar radius and an equatorial rotation velocity around 86% of the critical breakup value. With those parameters the Roche model yields an equatorial effective gravity several times smaller than the polar value. Plugging in a representative ratio g_pole / g_eq ≈ 5 and a polar temperature of about 15,000 K:

T_eq = T_pole × (g_eq / g_pole)^0.25
     = 15,000 K × (1/5)^0.25
     = 15,000 K × 0.669
     ≈ 10,000 K

So the poles run roughly 15,000 K while the equator sags to about 10,000 K — a polar–equatorial difference of several thousand kelvin, in line with the published interferometric values (Regulus's modeled polar T_eff is near 15,400 K and equatorial near 10,300 K). The same machinery applied to Achernar, which rotates closer to critical, gives an even steeper gradient and the largest oblateness of any well-studied star. Because flux scales as the fourth power of temperature, a 50% temperature contrast pole-to-equator is roughly a factor of five contrast in surface brightness — the poles genuinely outshine the equator.

Regimes: radiative, convective and near-critical rotation

The single exponent β is a convenient shorthand, but the physics splits into distinct regimes:

• Hot radiative rotators (β ≈ 0.25). O, B, A and early-F stars have radiative envelopes and follow von Zeipel's flux–gravity proportionality closely. These are where gravity darkening is strongest and most cleanly imaged: Regulus, Vega, Altair, Achernar, Alderamin.
• Cool convective rotators (β ≈ 0.08). Stars with deep convective envelopes redistribute heat more efficiently across the surface, weakening the gravity–temperature link. Lucy's value β ≈ 0.08 applies here. The Sun, rotating once a month, shows no measurable gravity darkening at all.
• Near-critical rotation (power law breaks down). As a star approaches breakup, the equatorial g_eff plunges toward zero and the simple T_eff ∝ g_eff^0.25 form overstates the contrast and misbehaves at the equator. The modern Espinosa Lara & Rieutord (2011) model derives the flux from the divergence-free energy transport in the Roche potential rather than assuming a fixed exponent, giving an effective local β that varies from ~0.25 at the pole down toward smaller values at the equator. This is the standard for the fastest rotators today.

Quantitative analysis: from the Roche potential to the limb

The standard model treats the star's gravity as that of a central point mass M (a good approximation because the mass is centrally concentrated) plus rotation, giving the Roche potential at colatitude θ and radius r:

Φ(r, θ) = − G M / r − ½ Ω² r² sin²θ.

The stellar surface is an equipotential Φ = const. Setting the polar and equatorial points on the same equipotential gives the classic Roche result that the equatorial radius can exceed the polar radius by at most 50%:

R_eq,crit = 1.5 × R_pole   (at critical rotation, ω = 1).

The effective gravity is the gradient of the potential, g_eff = |∇Φ|, which at the equator becomes g_eq = G M / R_eq² − Ω² R_eq. Define the dimensionless rotation rate ω = Ω / Ω_crit (so Regulus has ω ≈ 0.86). The local effective temperature follows from von Zeipel:

T_eff(θ) = T_pole × [ g_eff(θ) / g_pole ]^0.25.

To predict what an observer actually measures, you integrate the local blackbody (or model-atmosphere) intensity over the visible, foreshortened, limb-darkened surface at the relevant inclination. A pole-on observer sees mostly hot polar flux and infers a high temperature and small radius; an equator-on observer sees the cool bulge across the limb and infers a lower temperature and larger radius. The discrepancy is exactly why a single number for "the temperature of Vega" is ambiguous without specifying the geometry — and why Vega's role as a calibration standard had to be revisited once it was recognized as a pole-on rapid rotator.

Observational status: imaging the gradient

Gravity darkening was a theoretical prediction for eighty years before instruments could resolve it on a single star. The breakthrough came with optical and infrared long-baseline interferometry, which combines light from telescopes separated by tens to hundreds of meters to reach milliarcsecond angular resolution — enough to resolve the disks of the nearest bright stars. Key results:

• Altair (van Belle et al. 2001; Monnier et al. 2007, CHARA/MIRC): the first rotationally flattened star to be imaged, showing the oblate shape and a measurable hot pole.
• Regulus (McAlister et al. 2005, CHARA): equatorial radius ~32% larger than polar, ω ≈ 0.86, polar–equatorial T difference of several thousand kelvin; later refined by Che et al. (2011).
• Vega (Aufdenberg et al. 2006; Peterson et al. 2006): resolved as a pole-on rapid rotator near 88% of breakup, resolving its calibration puzzle.
• Achernar (Domiciano de Souza et al. 2003, VLTI): the most oblate star known, axis ratio ≈ 1.5, near-critical rotation.

Independent confirmation comes from exoplanet transits across oblate, gravity-darkened hosts. When a planet crosses a star whose poles are brighter than its equator, the transit light curve is distorted asymmetrically; KOI-13 b and Kepler-13 Ab (Barnes 2009; Barnes et al. 2011; Szabó et al. 2011) show this signature, and modeling it yields both the gravity-darkening exponent and the planet's true spin–orbit obliquity. Asteroseismology of fast rotators adds further constraints on the internal structure that sets β.

Fast rotators compared

Star	Spectral type	ω = Ω/Ω_crit	R_eq / R_pole	T_pole − T_eq (approx)
Regulus (α Leo)	B8 IVn	~ 0.86	~ 1.32	~ 5,000 K
Vega (α Lyr)	A0 V	~ 0.88	~ 1.13	~ 2,000 K
Altair (α Aql)	A7 V	~ 0.74	~ 1.20	~ 1,500 K
Achernar (α Eri)	B6 Vpe	~ 0.96	~ 1.5	several × 10³ K
Alderamin (α Cep)	A8 V	~ 0.83	~ 1.26	~ 1,500 K
Rasalhague (α Oph)	A5 III	~ 0.89	~ 1.20	~ 1,800 K
The Sun	G2 V	~ 0.004	~ 1.00002	negligible

The trend is clean: the closer ω is to 1, the more oblate the star and the steeper its pole-to-equator temperature gradient. The Sun, anchoring the bottom row, rotates so slowly that its oblateness is a part in fifty thousand and its gravity darkening is utterly unmeasurable — which is exactly why it took rapidly rotating early-type stars to reveal the effect.

Where gravity darkening matters

• Fundamental stellar parameters. The "effective temperature" of a fast rotator is a flux-weighted average over a non-uniform surface, so the inferred radius, luminosity, gravity, age and surface abundances all depend on inclination unless the gravity-darkened model is fit explicitly.
• Photometric calibration. Vega's status as a near-zero-magnitude standard had to be re-evaluated once it was recognized as a pole-on rapid rotator with a non-uniform disk; gravity darkening shifts the spectral energy distribution.
• Exoplanet obliquities. Gravity-darkened transit modeling of oblate hosts (KOI-13, Kepler-13A) measures the sky-projected and true spin–orbit angle photometrically, independent of the Rossiter–McLaughlin spectroscopic method.
• Be stars and decretion disks. Near-critical rotators with weak equatorial gravity more easily launch material into circumstellar disks, linking gravity darkening to the Be-star phenomenon.
• Stellar evolution. Reduced equatorial flux and rotational mixing alter mass loss, chemical transport and lifetimes for the most massive rotating stars, feeding into models of how they end their lives.

Common pitfalls and misconceptions

• "The equator is darker because it's farther from the core." Distance is only part of it. The dominant new effect is the centrifugal reduction of effective gravity, which is what von Zeipel's flux law actually responds to — not geometric distance alone.
• "Gravity darkening makes the whole star fainter." No — it redistributes brightness. The poles brighten as the equator dims. The integrated luminosity changes only modestly; the surface brightness map is what becomes strongly non-uniform.
• Using β = 0.25 for a cool, convective star. The radiative value applies to hot early-type envelopes. For convective stars β ≈ 0.08; using 0.25 overstates the gradient by a factor of three.
• Pushing the von Zeipel power law to critical rotation. Near breakup the equatorial gravity vanishes and the simple power law misbehaves; use the Espinosa Lara & Rieutord (2011) flux model instead.
• Reading one temperature off a spectrum. A fast rotator has no single surface temperature; quoting one without the inclination and the gravity-darkening model is meaningless for a star like Vega or Achernar.
• Confusing gravity darkening with limb darkening. Limb darkening is the dimming toward the edge of any stellar disk due to viewing the cooler upper atmosphere at a glancing angle; gravity darkening is a real, latitude-dependent change in surface temperature from rotation. Both shape an observed disk, but they are distinct effects.