Carbon Stars

The flip — why C/O is the decisive number

In almost every star you can see, oxygen is more abundant than carbon. In our Sun, the number ratio is C/O ≈ 0.55. That single fact governs what molecules can form in the cool photosphere. The carbon monoxide molecule, CO, has one of the strongest bonds in nature (dissociation energy 11.1 eV) and the temperatures at the surface of a red giant — below about 4,000 K — are low enough that nearly every available C-O pair survives as CO. Whichever element is the minority is consumed first; whichever is the majority has a leftover budget that does the rest of the molecular chemistry.

In a normal, oxygen-rich red giant, carbon is the minority. CO formation locks up all the carbon; oxygen has leftovers that go on to make TiO, VO, SiO and water vapour. The spectrum is dominated by TiO band heads, which is exactly what defines an M-type star like Betelgeuse. Flip the ratio so that carbon is the majority and you have a carbon star. CO formation now locks up all the oxygen; carbon has leftovers that build C₂, CN, CH, C₃, HCN and C₂H₂. The TiO bands vanish from the spectrum entirely — there is no free oxygen to form TiO — and are replaced by deep Swan bands of C₂ and red and violet bands of CN. The same molecule, CO, switches from being a carbon sink to an oxygen sink as you cross C/O = 1, and the rest of the molecular spectrum flips with it.

That one ratio is a sharp threshold. It is also the diagnostic line that the Harvard classification crossed when it labelled the C-type as a parallel sequence to the M-type rather than a temperature subclass within it, in the Morgan-Keenan extension of the 1940s that gave the modern C0–C9 (or C-N, C-R, C-J, C-H) labels.

How the AGB makes a carbon star — the third dredge-up

The only way to flip the surface C/O ratio of a star without external pollution is to bring up freshly-made carbon from below. That is what the third dredge-up does. On the AGB, the star burns hydrogen and helium in two thin shells around an inert carbon-oxygen core. The helium shell is thermally unstable: it ignites every 10⁴ – 10⁵ years in a runaway flash. After each helium-shell flash, the convective envelope reaches inward into the intershell region — the layer between the helium shell and the hydrogen shell — where helium burning has just produced fresh ¹²C, and where slow-neutron-capture reactions (the s-process) have built barium, lead and technetium.

The convection then carries this fresh ¹²C and s-process pollution up to the surface. Each dredge-up episode therefore enriches the photosphere by a small amount. If the star has enough envelope left and undergoes enough pulses, the cumulative enrichment eventually pushes C/O across 1. The star, having started as an M-type with TiO bands, passes through an S-type (where C/O ≈ 1 and ZrO bands appear because the s-process has produced free zirconium) and finally becomes a C-type N-star. The progression M → MS → S → SC → C traces the same envelope being enriched by dredge-up.

The minimum stellar mass for the third dredge-up to turn on at all depends sensitively on the convective-overshoot treatment in stellar evolution codes — values in the literature range from about 1.0 to 1.5 M☉. Above about 4 M☉ at near-solar metallicity, hot-bottom burning at the base of the convective envelope kicks in: the H-burning CN cycle running at the convective floor reconverts the dredged-up ¹²C to ¹⁴N before it reaches the surface, suppressing the flip. The carbon-star regime is therefore a mass window roughly 1.5 – 4 M☉ at solar metallicity, shifting to lower masses at low metallicity (where less carbon has to be added because there is less oxygen to overcome to begin with).

Three families of carbon stars

Not every carbon star is an AGB star caught in the act. The taxonomy has three distinct branches, set apart by where the carbon came from.

Class	Luminosity	T_eff	Origin	Diagnostic
N-type (classical)	AGB tip, ~10⁴ L☉	2,200 – 3,200 K	Intrinsic: third dredge-up	Tc in atmosphere, Mira-like variability
R-type (warm)	Red clump / lower RGB	3,500 – 5,000 K	He-core burning; not fully understood	Single, no Tc, normal CNO ratios
J-type	Like N but ¹³C-enriched	2,500 – 3,500 K	Disputed; many in binaries	Very low ¹²C/¹³C, often Li-rich
CH stars	Sub-giant / giant	4,000 – 5,000 K	Extrinsic: AGB mass transfer in binary	Strong CH band, metal-poor, binary
Dwarf carbon stars	Main sequence, M_V ~ 8 – 12	2,800 – 3,800 K	Mass transfer from former AGB companion	Halo kinematics, faint

The single most useful chemical clue is technetium. Tc has no stable isotopes; ⁹⁹Tc has a half-life of 2.1 × 10⁵ years and is produced by the s-process in the helium intershell. Its presence in a stellar photosphere is direct proof that the star itself was running the s-process within the last few hundred-thousand years — that is, it is an intrinsic AGB star. Tc absent? The carbon and s-process elements must have come from somewhere else (a now-dead AGB companion), and the star is extrinsic.

A short tour of famous carbon stars

• Y Canum Venaticorum (La Superba). A bright N-type AGB carbon star (visual magnitude 4.9–7.3, semi-regular variable), nicknamed by Father Angelo Secchi in the 1860s for its striking red colour. ¹²C/¹³C ≈ 3 places it close to a J-type. It hosts a detached dust shell at ~2.5×10⁵ AU, mapped in CO line emission, that records a mass-loss episode some 70,000 years ago.
• R Leporis (Hind's Crimson Star). Discovered by J. R. Hind in 1845 and described by him as "a drop of blood on a black field." A Mira-like long-period variable with a period near 430 days, visual magnitude swings 5.5–11.7, and a remarkably small atmospheric ¹²C/¹³C ratio (~ 30). Spectra show enormous C₂ and CN bands.
• V Hydrae. An extreme N-type Mira / semi-regular hybrid with a hot accretion-driven companion that is launching collimated bullets of plasma at hundreds of km/s — a probable preview of how some asymmetric planetary nebulae get their shapes. Mass-loss rate ~2×10⁻⁵ M☉/yr.
• TX Piscium. One of the brightest naked-eye carbon stars (magnitude 4.8 to 5.2), an irregular variable with intense C₂ and CN absorption. Frequently held up as a binocular target for amateurs because of its very red appearance.
• IRC +10216 (CW Leonis). The brightest infrared source outside the solar system at 5 µm. So heavily enshrouded in its own carbon dust that it is essentially invisible in the optical, while pouring out 10⁴ L☉ in the infrared. A natural laboratory for circumstellar molecular chemistry — more than 80 molecular species have been detected in its envelope.

Reading the spectrum

The carbon-star fingerprint in the visible is a sequence of band heads that step down toward the red. The two textbook Swan-band heads are:

C₂ Swan (0,0) band   5165 Å    green   (called the d³Πg − a³Πu transition)
C₂ Swan (0,1) band   4737 Å    blue
C₂ Swan (1,0) band   4715 Å    blue
CN violet (0,0)      3883 Å    violet
CN red system        7,900 Å   red
CH G-band            4300 Å    blue   (strong in CH stars)
C₃ band              ~4050 Å   violet

In the infrared, the diagnostic features change to molecular vibration-rotation bands: CO fundamental at 4.6 µm, HCN at 3.0 and 14 µm, C₂H₂ at 3.0 and 13.7 µm, and the silicon carbide (SiC) dust feature at 11.3 µm. The combination of CO 4.6 µm and SiC 11.3 µm uniquely identifies a dusty carbon-rich AGB envelope and is how surveys like 2MASS and AKARI pick carbon stars out of crowded fields in nearby galaxies.

The infrared two-colour diagram — typically J − K versus K, or equivalent — separates oxygen- and carbon-rich AGB stars cleanly because the dust composition shifts the slope of the spectral energy distribution. AGB carbon stars are pushed to extreme J − K (> 1.4) by their amorphous-carbon dust shells. This is the operational definition used in extragalactic surveys.

The dust factory — why galaxies need carbon stars

In the cool, slowly expanding envelope above an AGB carbon star, the gas drops below the condensation temperature for solid carbon (around 1,500 K) at a few stellar radii. Carbon atoms aggregate first into clusters, then into amorphous-carbon grains; silicon co-condenses as silicon carbide (SiC) where there is enough Si available. These grains absorb photons from the star efficiently, are accelerated outward by radiation pressure, and drag the gas with them through collisional coupling. The result is a cool, slow, dust-driven wind — characteristically 5–30 km/s, mass-loss rates from 10⁻⁷ to 10⁻⁴ M☉/yr.

The grains injected into the interstellar medium then become a major component of the galactic dust budget. Direct interstellar pickup of presolar carbonaceous grains in meteorites — graphite spherules, nanodiamonds, and especially SiC grains with isotopic compositions matching AGB nucleosynthesis — proves that material from individual carbon stars ended up in the solar nebula. Mainstream SiC grains in primitive meteorites have ²⁹Si/²⁸Si and ³⁰Si/²⁸Si signatures that single out AGB stars of 1.5 – 3 M☉ at slightly sub-solar metallicity as the dominant source.

Quantitatively, carbon-rich AGB winds contribute roughly half of the galactic carbon return rate at solar metallicity. The other major channels are massive-star winds and supernovae, but those produce a different mix of dust species (largely silicates). The carbon in your body — the carbon in every protein, in every nucleic acid, in every cell wall — almost certainly passed through one or more carbon-star envelopes before being incorporated into the molecular cloud that became the Sun.

Worked example: enrichment timescale

Take a 2 M☉ star reaching the AGB. Suppose its third dredge-up brings up an average of Δm_C ≈ 3 × 10⁻³ M☉ of fresh ¹²C per pulse, and that thermal pulses occur every t_TP ≈ 10⁵ years. To flip its photospheric C/O ratio from a starting value of C/O ≈ 0.55 (solar) to greater than 1, it has to add enough carbon to overcome the head-start oxygen has.

For a 0.5 M☉ envelope, solar abundances give about

m_O (envelope) ≈ X_O × M_env ≈ 0.01 × 0.5 M☉ = 5 × 10⁻³ M☉
m_C (envelope) ≈ X_C × M_env ≈ 0.003 × 0.5 M☉ ≈ 1.5 × 10⁻³ M☉

To reach C/O = 1 by mass-rate (close enough for an order of magnitude), the star must dredge up Δm_C ≈ m_O − m_C ≈ 3.5 × 10⁻³ M☉ of additional carbon — enough to roughly triple the starting envelope carbon mass.

N_pulses ≈ Δm_C(total) / Δm_C(per pulse) ≈ 3.5×10⁻³ / 3×10⁻³ ≈ 1.2 pulses
                                                  (very rough)
Time      ≈ N_pulses × t_TP ≈ 10⁵ years

So one to a few pulses can flip a low-metallicity envelope, while a solar-metallicity star may need ten or more — explaining why low-metallicity galaxies (such as the LMC and SMC) host proportionally more carbon stars per unit AGB population than the metal-rich solar neighbourhood.

Carbon stars as a distance indicator

Near the tip of the AGB, the bolometric luminosity of an intrinsic carbon star is nearly metallicity-independent, settling around M_bol ≈ −4.7. In the J band, the so-called J-band Carbon Star Method gives a mean absolute magnitude M_J ≈ −3.65 ± 0.1 mag for carbon stars in nearby Local Group dwarf galaxies. Combined with the ease of identifying carbon stars from infrared two-colour diagrams (J − K > 1.4), this makes them a robust distance indicator out to several Mpc, complementary to the tip of the red giant branch.

The advantage is that, unlike Cepheids, carbon stars are common at the old- and intermediate-age populations that dominate dwarf galaxies, and unlike RR Lyrae they are bright in the infrared, where dust extinction is minimal. They have been used to anchor distances to IC 1613, NGC 6822, the Magellanic Clouds, M31 dwarf satellites, and the more nearby galaxies of the Sculptor and M81 groups.

Variants and related populations

• S-type stars. Intermediate stage with C/O ≈ 1. ZrO bands appear because the s-process makes free zirconium, but oxygen is still slightly in excess so TiO has not yet vanished. The MS → S → SC sequence tracks rising C/O.
• Barium stars. Giants showing enhanced barium and other s-process elements but with C/O still < 1. The companion-mass-transfer cousins of CH stars: enriched by an AGB partner that has since become a white dwarf, but not enriched enough to flip C/O.
• Lithium-rich carbon stars. A subset of J-type stars with surface Li far above the photospheric solar value. Probably the signature of hot-bottom burning ("Cameron-Fowler mechanism") in the most massive AGB carbon stars, or perhaps engulfment of a planet.
• R Coronae Borealis stars. Hydrogen-deficient, helium- and carbon-rich supergiants. Not AGB carbon stars — they are the late evolutionary product of a merger between two white dwarfs (CO + He) — but spectroscopically carbon-rich enough to fit some of the same colour selections.
• Hot DQ white dwarfs. The dead progeny of massive AGB stars whose carbon-rich photospheres survived the planetary-nebula phase. C₂ Swan bands appear in the spectrum of a stellar remnant that no longer has fusion running anywhere.

Where carbon stars show up in modern astronomy

• Local Group dwarf galaxy populations. Carbon stars are bright, easy to count, and provide both age tracers (their 1.5–4 M☉ progenitors lived 0.1–2 Gyr) and a robust distance indicator. Survey papers routinely report the C/M ratio (number of carbon stars to oxygen-rich M-type AGB) as a coarse metallicity proxy.
• Galactic chemical evolution models. The carbon yield from intermediate-mass AGB stars sets the carbon-to-iron ratio history of a galaxy. The shape of the [C/Fe] vs [Fe/H] track in the Milky Way bulge and disc constrains the AGB IMF-integrated yield.
• Presolar grains in meteorites. SiC grains in Murchison and other primitive meteorites preserve the isotopic signature of individual carbon-star envelopes. They are a direct, in-hand sample of AGB nucleosynthesis dust, with no model extrapolation between source and detector.
• Massive planetary nebula progenitors. The shape of a planetary nebula reflects the geometry of the AGB superwind that produced it. The dustiest, most asymmetric, "carbon-fountain" carbon stars like V Hya appear to be transitioning into bipolar PNe driven by jets from a companion.
• Dust in primordial galaxies. JWST has begun resolving rest-frame near-IR emission from intermediate-age populations in z > 6 galaxies. The carbon yield from AGB carbon stars sets the floor of dust formation timescales — at z > 8 there has not been time for them to evolve, so any dust observed is supernova-driven.

Common pitfalls

• Confusing N-type AGB carbon stars with R-type warm carbon stars. Both show Swan bands, but they are at very different evolutionary stages — and the R-type origin is still debated, possibly involving He-flash mergers. Always check the luminosity and the presence of Tc.
• Mistaking heavy circumstellar reddening for low T_eff. A dusty carbon star can be much redder in colour than its photospheric T_eff would predict, because amorphous-carbon grains in the shell preferentially extinguish blue light. Spectral fits without a dust component will pull T_eff systematically too cool.
• Assuming all giants with strong CN are carbon stars. CN can be enhanced in oxygen-rich giants too; the diagnostic for a carbon star is the C₂ Swan band, not just CN. C/O must be measured (or inferred from molecular ratios) to be sure.
• Treating the C/O ratio as gradual. The molecular spectrum flips sharply at C/O = 1 because CO formation is a step function in the limit of strong CO bond. A 5% change in C/O around unity can be the difference between an S-type and an N-type spectrum.
• Ignoring metallicity dependence of the carbon-star phase. At low metallicity, less carbon needs to be dredged up to reach C/O > 1, so a much wider mass range of AGB stars becomes carbon stars. The Magellanic Clouds therefore have a higher C/M ratio than the solar neighbourhood, and the very-low-metallicity halo CH stars dominate the carbon-star population at [Fe/H] < −2.