Horizontal Branch

After the flash, settling down

Picture a star of one solar mass that has just finished a billion-year climb up the red giant branch. Its hydrogen-burning shell has gradually deposited helium ash onto a degenerate, electron-supported core. When the core finally reaches T ≈ 10⁸ K, helium ignites — not gently but in a runaway flash that releases roughly 10⁴¹ J in minutes. Almost none of that energy escapes the star. Instead it goes into lifting the core out of electron degeneracy: the core expands, the density drops, and what was a relativistically stiff Fermi gas becomes an ordinary thermal plasma again.

What survives is a quiet star in a new configuration. The core, now non-degenerate, burns helium via the triple-alpha reaction at a steady rate. A thin shell of hydrogen continues to fuse just outside the core. The envelope, which the helium flash mildly contracted, has settled into a stable structure with surface temperature somewhere between 5000 K and 30000 K and luminosity locked near L ≈ 40-50 L_⊙. This is the horizontal branch: a star roughly 50 times brighter than the Sun, lasting about 100 million years, burning helium without drama. The dramatic finale of the red giant branch is over.

Why the band is horizontal

The defining curiosity of the HB is that every star sits at essentially the same luminosity even though their surface temperatures span a factor of six. To see why, count what is locked and what is free.

Locked: the helium core mass. The helium flash always triggers at the same physical condition — degenerate matter heated to T ≈ 10⁸ K — so the core that emerges from the flash has nearly the same mass regardless of the star's main-sequence mass. Detailed models give M_core ≈ 0.47 M_⊙, varying by less than 0.01 M_⊙ across the entire 0.5-2.2 M_⊙ range of HB progenitors.

Because the luminosity of a core-He-burning star scales steeply with core mass (roughly L ∝ M_core¹⁰ in this regime, much steeper than the main-sequence relation), a fixed core mass implies a fixed luminosity. Hence the horizontal stripe.

Free: the envelope mass. A star that climbed the RGB more vigorously, or lived in a denser cluster where binary mass transfer was efficient, or had higher initial helium content, will arrive on the HB with a thinner hydrogen envelope. Envelope thickness controls radius (thick → bloated → cool, thin → compact → hot), and therefore Teff varies along the stripe while L stays nearly constant. The horizontal branch is a one-parameter family in core-mass × fixed-luminosity, decorated along the abscissa by the leftover envelope.

Worked example: where does a 0.65 M_⊙ HB star sit?

Take a globular cluster star whose initial main-sequence mass was 0.85 M_⊙. Standard RGB models say it loses about 0.20 M_⊙ via the Reimers wind before the helium flash. After the flash, the leftover mass is

M_total = M_initial − M_lost = 0.85 − 0.20 = 0.65 M_⊙
M_core ≈ 0.47 M_⊙ (set by flash)
M_envelope = M_total − M_core = 0.18 M_⊙

An envelope of 0.18 M_⊙ places the star near the middle of the HB. Models predict L ≈ 45 L_⊙ and Teff ≈ 6500 K — squarely inside the instability strip. The star will pulsate as an RRab variable with a period near 0.55 days, and its mean visual magnitude M_V ≈ +0.50 marks it as a standard candle. If the same progenitor had lost more mass — say 0.40 M_⊙ instead of 0.20 — the envelope would be only 0.02 M_⊙, the star would be a hot 25000 K extreme blue-HB star, and the pulsation would disappear (the instability strip cuts off near 7500 K).

The instability strip and RR Lyrae

Between roughly 6000 K and 7500 K the HB crosses the classical instability strip — a narrow temperature range where stars become pulsationally unstable through the kappa mechanism. Helium ionises and recombines once per cycle in a layer just below the surface; the resulting opacity variation extracts energy from the radiation field and pumps the fundamental radial mode. Stars in this strip — RR Lyrae variables — pulsate with periods of 0.2 to 1.0 days and amplitudes up to 1.5 magnitudes in V.

The astrophysical importance is enormous. Because every RR Lyrae sits at essentially the same intrinsic luminosity (M_V ≈ +0.5, fine-tuned by a small period-luminosity-metallicity correction), an apparent magnitude measurement gives a distance modulus directly. RR Lyrae are visible at 1 Mpc — out to the Magellanic Clouds, the dwarf spheroidals, and across the Local Group. They are the rung between trigonometric parallax and Cepheids on the distance ladder.

Walter Baade exploited this in 1944 to identify two stellar populations and reclassify the distance scale: the RR Lyrae in M31 were systematically fainter than expected, revealing that M31 was twice as far as Hubble had estimated. The Hubble constant dropped by a factor of two overnight.

The second-parameter problem

If metallicity alone determined HB morphology, every globular cluster of similar [Fe/H] should have a similar distribution of HB stars along the stripe. They do not. The classical example is the pair of clusters M3 (with predominantly red HB) and M13 (with predominantly blue HB) — both have nearly identical metallicities, ages, and chemical compositions, yet M13's HB is much hotter overall. Something else — the "second parameter" — controls HB morphology.

The leading candidates:

• Age. A cluster older by 1-2 Gyr produces stars with lower turnoff mass, more RGB mass loss, thinner envelopes, and bluer HB. Age is now thought to be the dominant second parameter for most pairs.
• Helium abundance Y. Many globular clusters host multiple populations with He-enhanced sub-populations (Y ≈ 0.32 vs primordial 0.245). He-enhanced stars produce bluer HBs at the same metallicity.
• Rotation. Faster RGB rotation enhances mass loss; only a few clusters show clear signatures.
• Cluster density. Crowded clusters host more binary interactions that strip envelopes catastrophically; blue HB excess in dense clusters supports this.

The current consensus is that the second parameter is not a single variable but a cluster-by-cluster combination of age and helium variations, with binary stripping contributing in the densest systems. The original M3-M13 puzzle is now mostly explained by an age difference of about 1.5 Gyr.

Comparison: where the HB sits in stellar evolution

Phase	Energy source	L (L_⊙)	Teff (K)	Duration	HR location
Main sequence (1 M_⊙)	Core H burning	1	5778	10 Gyr	MS band
Subgiant	H-shell ignition	3-30	5000-5500	10-50 Myr	Crosses HR diagram rightward
Red giant branch (RGB tip)	H shell + degenerate He core	2000-3000	3500-4000	500 Myr	Upper right, climbing
Helium flash	Runaway 3α in degenerate core	(internal, brief)	—	Minutes	(not directly visible)
Horizontal branch	Quiet 3α core + H shell	40-50	5000-30000	100 Myr	Horizontal stripe at M_V ≈ +0.5
Asymptotic giant branch	Double shell (H + He)	1000-10000	3000-3500	5-100 Myr	Upper right, paralleling RGB

How the horizontal branch ends

Core helium runs out after about 10⁸ years. At that point a layer of carbon and oxygen ash has built up in the center, with a thin helium-burning shell surrounding it. The carbon-oxygen core is again inert and starts to contract; the surrounding helium shell ignites; the envelope expands. The star leaves the HB rightward and upward on the HR diagram, climbing the asymptotic giant branch — a second, brighter giant phase that ends in thermal pulses, intense mass loss, a planetary nebula, and ultimately a white dwarf.

For globular cluster stars the AGB phase lasts only a few tens of millions of years. The complete sequence — main sequence to white dwarf — for a 0.8 M_⊙ Population II star is roughly 12 Gyr main sequence, 500 Myr RGB, 100 Myr HB, 30 Myr AGB, then white dwarf cooling forever after. The horizontal branch is the last bright, stable phase before the long decline.

Common pitfalls

• "Horizontal branch" is not a literal horizontal line. In V vs (B−V) color-magnitude diagrams it is approximately horizontal because M_V is nearly constant; in true (log L, log Teff) HR diagrams it actually rises slightly toward higher temperatures because hotter HB stars are a bit more luminous (small bolometric correction effect). The visual horizontality is partly a coincidence of how astronomical magnitudes are defined.
• Higher-mass stars do not produce a horizontal branch. Stars above 2.2 M_⊙ ignite helium quietly in non-degenerate cores. They go through a core-He-burning phase — sometimes called the "clump" or, for massive stars, the "blue loop" — but at much higher luminosity (≈ 100-1000 L_⊙) and not as a horizontal stripe.
• RR Lyrae are not on the HB by coincidence. The instability strip is a temperature region (set by hydrogen and helium ionisation in the envelope), not a luminosity region. Where the strip intersects the HB is where RR Lyrae appear; where it intersects the supergiant region is where Cepheids appear.
• The 0.47 M_⊙ core is not the Chandrasekhar limit. Chandrasekhar is 1.4 M_⊙ for relativistic electrons. The 0.47 M_⊙ value is set by the temperature at which 3α ignites in degenerate helium — a much lower density than Chandrasekhar conditions.
• Globular clusters have many HB stars at once. A "snapshot" cluster has stars at every evolutionary phase. The HB appears clearly populated because stars spend ~10⁸ yr there, long enough that several percent of cluster members are caught in this phase.

Observational handles on the HB

• Cluster color-magnitude diagrams. HST and Gaia have measured HBs in over 150 Galactic globular clusters with sub-percent photometry. The morphological diversity is one of the cleanest probes of multiple stellar populations.
• RR Lyrae light curves. Ground-based time-domain surveys (OGLE, ASAS-SN, ZTF) have catalogued tens of thousands of RR Lyrae out to the Magellanic Clouds. Periods and amplitudes give distance and metallicity simultaneously.
• Asteroseismology of red HB stars. Kepler observations of HB stars in the field (so-called red clump stars in Pop I) measure oscillation power spectra that reveal the core mass and helium content directly — a powerful test of stellar models.
• Spectroscopic Y measurements. Helium absorption lines in hot HB stars yield direct measurements of envelope helium content, testing the He-enhanced multiple-population scenario.
• Cluster age-metallicity-HB correlations. The morphology of a cluster HB is a sensitive function of age, allowing differential ages between clusters to be measured to ≈ 0.5 Gyr from HB statistics alone.