Stellar Evolution
Thermohaline Mixing: Salt Fingers in the Stellar Interior
Roughly one-hundred-thousandth of a red giant's helium fuel is enough to turn its interior into a slow-motion version of the ocean's "salt fountains." When a low-mass star climbs the red giant branch, burning the primordial isotope helium-3 just above its hydrogen shell lowers the mean molecular weight there by a whisper — a few parts in ten thousand. That tiny inversion is enough to trigger a double-diffusive instability that grows long, thin downward-plunging plumes called salt fingers, physically identical to those oceanographers see below warm salty currents.
Thermohaline mixing is the name for this process: a "double-diffusive" instability in which heat and composition diffuse at wildly different rates. In stars it slowly rewrites the surface chemistry of ageing Sun-like stars, and it is the leading explanation for why old red giants show far more ¹³C, more nitrogen, and far less lithium and ³He than "classical" models — with no convection — predict.
- RegimeLow-mass giants (0.8–2.5 M☉) past the RGB bump
- Driven byμ-inversion from ³He(³He,2p)⁴He above the H-shell
- Key numberSurface ¹²C/¹³C falls from ~20 toward ~4–10
- MechanismHeat diffuses ~10⁶× faster than composition → salt fingers
- First describedOceans: Stern 1960; stars: Ulrich 1972, Charbonnel & Zahn 2007
- Observed withHigh-res spectroscopy (VLT/UVES, APOGEE, Kepler asteroseismology)
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
What it is and why it matters
Thermohaline mixing is a slow, non-convective transport process that operates in the radiative layer just above the hydrogen-burning shell of an evolved low-mass star. It is the stellar cousin of a phenomenon oceanographers have studied since the 1950s: when warm salty water sits above cooler fresh water, tiny "salt fingers" grow and carry salt downward even though the overall density profile looks stable. In a red giant the roles of temperature and salt are played by heat and chemical composition.
Why care? "Classical" stellar models — which only mix material inside convection zones — predict that after the first dredge-up a low-mass giant's surface abundances should freeze. But observations of field and cluster giants show the opposite: as stars climb past the red giant branch (RGB) bump, their surface ¹²C/¹³C ratio keeps falling, nitrogen rises, and lithium and ³He are destroyed. Thermohaline mixing supplies the missing transport, connecting a subtle nuclear side-effect to the observable chemistry of billions of ageing stars, and to the Galaxy's ³He budget.
The mechanism, step by step
The trigger is a nuclear accident of bookkeeping. On the main sequence, the outer wing of the hydrogen-burning region accumulates the primordial isotope ³He. As the star ascends the RGB after the bump, that ³He burns via the reaction ³He(³He, 2p)⁴He. This converts three particles (two ³He nuclei) into effectively fewer heavy particles plus protons, and locally lowers the mean molecular weight μ. The result is a thin layer where μ increases outward — a molecular-weight inversion sitting on top of the shell.
By itself that inversion is stable, because the temperature gradient holds it up (the Ledoux criterion is satisfied). The double-diffusive twist is that heat leaks out of a displaced blob far faster than its excess composition can. Nudge a heavier fluid element downward: it quickly reaches thermal equilibrium with its surroundings but keeps its high μ, so it stays denser and keeps sinking. It stretches into a long, thin finger — a salt finger — until turbulence shreds it. Millions of these fingers together slowly mix processed material outward to the envelope.
Characteristic numbers, scales, and the criterion
The controlling parameter is the density ratio R₀ = (∇ − ∇_ad)/(∇_μ), comparing the (stabilising) thermal stratification to the (destabilising) μ-inversion, together with the diffusivity ratio τ = κ_μ/κ_T. Salt fingers grow when 1 < R₀ < 1/τ. In a stellar radiative zone τ is minuscule — roughly 10⁻⁶ to 10⁻⁷, since radiative heat transport dwarfs the atomic/ionic diffusion of composition — so the finger regime is enormous.
The mixing is usually written as a diffusion coefficient D_th ∝ (κ_T/τ)·(∇_μ)/(∇ − ∇_ad), scaled by an efficiency tied to the finger aspect ratio α (length/width). Ulrich (1972) argued for long fingers (α ≈ 5–6, giving a large coefficient C_t ≈ 1000), the prescription Charbonnel & Zahn (2007) adopted to fit red giants; Kippenhahn, Ruschenplatt & Thomas (1980), by contrast, used a "blob" geometry (aspect ratio ~1) that gives a coefficient about 50× smaller (C_t ≈ 12). The μ-inversion is tiny — Δμ/μ ∼ 10⁻⁴ — yet, acting over ~10⁷–10⁸ yr of RGB ascent, it drives the surface ¹²C/¹³C from the post-dredge-up value near 20 down toward the CN-cycle equilibrium of ~4, with ~10 typical for low-mass Population I giants.
How it is observed and detected
Thermohaline mixing leaves no direct image — it is inferred from surface chemistry measured with high-resolution spectroscopy. The classic diagnostics are the carbon isotope ratio ¹²C/¹³C (from CN and CH molecular bands, e.g. ~2.3 μm CO or the 4200–4300 Å CH G-band), the [C/N] ratio, and the lithium 6708 Å line. Instruments include VLT/UVES and FLAMES-GIRAFFE for cluster giants, HARPS/HERMES for field stars, and enormous statistical samples from APOGEE (H-band, near-infrared) and LAMOST.
Globular-cluster giants (e.g. M3, M13, NGC 6791) are ideal test beds: same age and composition, so abundance trends versus luminosity isolate the mixing onset at the RGB bump. Kepler and K2 asteroseismology now pin down which giants are on the RGB versus the red clump, sharpening the ¹²C/¹³C-versus-evolution picture. A second, independent constraint is the Galactic ³He/H measured in H II regions with the Green Bank Telescope (the 8.665 GHz ³He⁺ hyperfine line): the low observed abundance requires exactly the extra ³He destruction thermohaline mixing provides.
Where it operates, and what it is not
Thermohaline mixing switches on in stars of about 0.8–2.5 M☉ — roughly Sun-like and lighter — but only after the RGB bump, once the advancing H-shell erases the composition discontinuity left by the first dredge-up and the ³He-burning μ-inversion can build. Its efficiency rises with decreasing stellar mass and metallicity, which is why metal-poor halo giants show the strongest ¹³C and lithium anomalies. It continues, more weakly, onto the asymptotic giant branch. Higher-mass stars develop convective cores and different shell structure and do not follow the same route.
It should not be confused with convective overshoot (ballistic penetration past a convective boundary), rotational mixing (meridional circulation and shear), or gravity-wave transport. All are candidate "extra mixing" mechanisms and may act together; thermohaline is unique in being a genuinely double-diffusive, μ-driven instability. It is also distinct from ordinary Schwarzschild/Ledoux convection, which needs an actual density inversion.
Open questions and significance
The physics is real, but the efficiency is hotly debated. One-dimensional prescriptions with long fingers (α ≈ 5–6) reproduce the abundance data, yet high-resolution 3D hydrodynamic simulations of finger convection (Denissenkov, Traxler & Garaud, and others) find much weaker mixing — fingers that are shorter and fatter — by one to two orders of magnitude. Reconciling the two is a leading problem: perhaps rotation, magnetic fields ("magneto-thermohaline"), or internal gravity waves boost the transport, as a 2025 study of wave-enhanced, rotation-boosted mixing suggests.
There are also hints that thermohaline mixing alone is not the full story near the RGB bump, with evidence for a separate, earlier mixing event. Resolving this matters beyond stellar physics: red giants set the ¹³C, ¹⁴N and ³He that they return to the interstellar medium, feeding Galactic chemical evolution and constraining Big Bang nucleosynthesis via the primordial ³He yield. Getting the mixing right also sharpens [C/N]-based ages for the millions of giants surveyed by APOGEE and Gaia.
| Property | Ocean (warm, salty over cool, fresh) | Red giant interior (above H-shell) | |
|---|---|---|---|
| Fast diffuser | Heat (κ_T) | Heat / radiation (κ_T) | |
| Slow diffuser | Salt (κ_S) | Chemical composition, μ (κ_μ) | |
| Destabilising gradient | Salinity increasing upward | Mean molecular weight μ increasing outward (inversion) | |
| Diffusivity ratio τ = κ_μ/κ_T | ~10⁻² (salt/heat in water) | ~10⁻⁶ to 10⁻⁷ (radiative interior) | |
| Finger scale | Few cm, minutes to hours | ~1 cm–100 m predicted; huge span vs pressure scale height | Uncertain by orders of magnitude |
| Net effect | Vertical salt/heat flux, staircases | Slow mixing of ¹³C, N, ³He, Li to the surface |
Frequently asked questions
What actually creates the salt fingers inside a red giant?
The burning of primordial helium-3 through the reaction ³He(³He,2p)⁴He in a thin layer just above the hydrogen-burning shell. This reaction lowers the local mean molecular weight, producing a region where μ increases outward — an inversion. Because heat diffuses far faster than composition, displaced heavy fluid elements sink as long thin fingers, mixing processed material toward the surface.
Why is it called 'thermohaline' if there is no salt in a star?
The name is borrowed from oceanography, where the same double-diffusive instability arises from competing temperature (thermo) and salinity (haline) gradients. Melvin Stern described the oceanic salt fingers in 1960. In a star, chemical composition (mean molecular weight) plays the role that salt plays in the ocean, and radiative heat plays the role of the ocean's warmth, so the physics is identical even though there is no NaCl.
How does thermohaline mixing change what we see at the surface?
It slowly transports CN-cycle products from near the shell up to the envelope. Observationally the surface ¹²C/¹³C ratio drops (from ~20 after first dredge-up toward ~4–10), nitrogen rises while carbon falls, and lithium and ³He are destroyed. These trends appear only in giants brighter than the RGB bump, which is why they cannot be explained by first dredge-up alone.
When during a star's life does it turn on?
Only after the star passes the red giant branch bump. At that point the hydrogen-burning shell has advanced into chemically homogeneous material, erasing the composition step left by the first dredge-up, so the ³He-burning molecular-weight inversion can finally establish itself and drive fingers. It is most efficient in the lowest-mass, most metal-poor giants, and continues weakly onto the asymptotic giant branch.
Who first proposed it for stars?
Robert Ulrich (1972) and Kippenhahn, Ruschenplatt & Thomas (1980) first worked out salt-finger mixing in stellar radiative zones. Corinne Charbonnel and Jean-Paul Zahn (2007) revived and popularised it, showing that a double-diffusive (rather than Rayleigh-Taylor) interpretation of the ³He-driven μ-inversion simultaneously reproduces the observed carbon, nitrogen, lithium and ³He behaviour of low-mass giants.
Is thermohaline mixing settled physics?
The mechanism is well established, but its efficiency is not. One-dimensional models with long, thin fingers fit the data, yet 3D simulations predict much weaker mixing from shorter, fatter fingers. Rotation, magnetic fields, and internal gravity waves may enhance it. There is also evidence that a separate mixing event operates near the RGB bump, so thermohaline mixing may not be the complete explanation of extra mixing.