Stellar Thermohaline Mixing

Stellar thermohaline mixing (often termed the thermohaline instability or astrophysical salt-finger convection) is a secular double-diffusive fluid instability that operates inside thermally stable stellar radiative zones whenever an inverted chemical composition gradient places plasma of higher mean molecular weight μ above plasma of lower mean molecular weight (∇_μ ≡ dlnμ / dlnP < 0). First adapted from physical oceanography to stellar interiors by Roger K. Ulrich (1972) and Rudolf Kippenhahn, Gregor Ruschenplatt, and Hans-Christoph Thomas (1980), the instability arises because radiative photon diffusion equalizes temperature perturbations across sinking heavy plumes much faster than microscopic ion diffusion can equalize their chemical composition. Within astronomy, stellar thermohaline mixing explains the abrupt surface depletion of lithium (⁷Li), carbon (¹²C/¹³C), and helium-3 (³He) in low-mass red giants above the red-giant-branch luminosity bump, regulates surface abundance patterns in carbon-enhanced metal-poor (CEMP) binary stars, and redistributes planetary debris accreted onto white dwarf atmospheres.

Thermodynamic Stability Window

In a hydrostatic stellar radiative zone (dP / dr = −ρ g), a layer is dynamically stable against rapid overturning convection under the Ledoux criterion when the square of the total Brunt–Väisälä buoyancy frequency N² = N_T² + N_μ² is positive, which corresponds to:

∇_rad < ∇_ad + (φ / δ) ∇_μ

where ∇_rad is the radiative logarithmic temperature gradient (dlnT / dlnP), ∇_ad = (∂lnT / ∂lnP)_S,μ is the adiabatic gradient, δ = −(∂lnρ / ∂lnT)_P,μ, and φ = (∂lnρ / ∂lnμ)_P,T.

Suppose a nuclear reaction, off-center shell flash, or external accretion event deposits material with higher mean molecular weight atop lighter interior plasma, creating a negative composition gradient (∇_μ < 0, since pressure P increases downward while μ increases upward). Even if the layer remains dynamically stable under the Ledoux criterion because the strongly sub-adiabatic thermal stratification (∇_ad − ∇_rad > 0) outweighs the inverted composition gradient:

0 < (φ / δ) |∇_μ| < ∇_ad − ∇_rad

the layer is still secularly unstable to thermohaline mixing.

Role of Rapid Radiative Heat Exchange

In oceanography, warm salty water resting atop cold fresh water is dynamically stable if the thermal expansion of the warm upper layer makes it slightly less dense than the cold lower layer. However, because heat diffuses through water ~100 times faster than dissolved salt ions, a small downward perturbation of warm salty water loses its heat to the cold surroundings while retaining its excess salinity, becoming denser than the ambient water and sinking continuously as narrow vertical "salt fingers."

In a stellar radiative zone, the disparity between thermal diffusivity K_T = 4 a c T³ / (3 κ ρ² c_P) and microscopic ion diffusivity D_ion is far more pronounced than in terrestrial oceans, with an inverse Lewis number τ ≡ D_ion / K_T ~ 10⁻¹⁰ to 10⁻⁶:

1. When a fluid parcel of higher molecular weight μ + Δμ at the top of the layer is displaced downward by Δr < 0 into lower-μ surroundings, adiabatic compression initially heats the parcel above the ambient sub-adiabatic temperature (ΔT > 0), providing upward thermal buoyancy that opposes its downward motion.
2. Radiative photon diffusion rapidly bleeds this excess thermal energy ΔT into the cooler ambient plasma over the thermal relaxation time τ_th ~ l² / K_T of a finger of horizontal width l, whereas the parcel retains its higher molecular weight Δμ.
3. Once thermal equilibrium with the surroundings is reached (ΔT → 0), the parcel's density excess Δρ / ρ ≈ φ (Δμ / μ) > 0 is unopposed, causing the heavy finger to sink steadily while adjacent lower-μ plumes rise.

The Ulrich–Kippenhahn–Zahn Prescription

In one-dimensional stellar evolution codes, thermohaline transport is modeled as a radial diffusion process with effective diffusivity D_th proportional to the thermal diffusivity K_T and the magnitude of the negative μ-gradient:

D_th = C_t K_T [ (φ / δ) |∇_μ| / (∇_ad − ∇) ] = C_t K_T R₀⁻¹

where R₀ ≡ (∇_ad − ∇) / [ (φ / δ) |∇_μ| ] > 1 is the density ratio (which must satisfy 1 < R₀ < τ⁻¹ for the instability to operate) and C_t = (8/3) π² α_th² is a dimensionless geometry constant governed by the aspect ratio α_th = L_f / d_f (length-to-diameter ratio) of the convective fingers:

• Kippenhahn et al. (1980) "Blob" Geometry (α_th ≈ 1): Treating the sinking elements as equant spherical blobs diluted by shear yields a small coefficient C_t ≈ 12, producing relatively slow mixing.
• Ulrich (1972) and Charbonnel & Zahn (2007) Slender "Finger" Geometry (α_th ≈ 5–6): Treating the elements as elongated vertical salt fingers with aspect ratio α_th ≈ 5 yields C_t ≈ 658 to 1,000, increasing the mixing speed by Nearly two orders of magnitude.

Three-Dimensional Magnetohydrodynamic Calibrations

Direct 3D numerical simulations by Pascale Garaud, Justin Brown, and Stephan Stellmach (2013) demonstrated that at the tiny Prandtl numbers (Pr = ν / K_T ~ 10⁻⁶) of stellar interiors, secondary parasitic shear instabilities saturate non-magnetic thermohaline fingers at lower aspect ratios (C_t ≈ 10–50) unless large-scale toroidal or poloidal magnetic fields (B ~ 10²–10⁴ G) stabilize the finger walls against Kelvin–Helmholtz disruption, restoring transport rates comparable to C_t ~ 100–1,000.

The ³He(³He, 2p)⁴He Reaction in Red Giants Above the Luminosity Bump

A central triumph of stellar thermohaline theory was proposed by Eggleton, Dearborn, and Lattanzio (2006) and quantified by Corinne Charbonnel and Jean-Paul Zahn (2007) to solve a long-standing abundance anomaly in low-mass red-giant-branch (RGB) stars (M ≈ 0.8–2.0 M_☉). During the main sequence, incomplete proton–proton chain burning builds up a rich pocket of helium-3 (³He) at intermediate radii, which is later homogenized throughout the convective envelope during the first dredge-up.

As the red giant ascends the RGB, its thin hydrogen-burning shell advances outward and crosses the composition discontinuity left behind by the deepest penetration of the first dredge-up (producing the RGB luminosity bump on the Hertzsprung–Russell diagram). In the outer wing of the hydrogen shell, ³He fuses via:

³He + ³He → ⁴He + 2 ¹H

Because this reaction converts two nuclei of mass 3 (6 nucleons shared across 2 ions, mean mass per particle = 3 amu excluding electrons) into three nuclei (one ⁴He plus two free protons ¹H, 6 nucleons shared across 3 ions, mean mass per ion = 2 amu), it locally decreases the mean molecular weight μ by Δμ / μ ≈ −10⁻⁴ just above the main hydrogen-burning shell. Consequently, heavier envelope gas sits atop this lighter ³He-burning layer (∇_μ < 0), triggering thermohaline salt fingers that bridge the radiative zone between the burning shell and the convective envelope. This conveyor belt:

1. Destroys >90% of the excess ³He produced by low-mass stars, reconciling stellar nucleosynthesis yields with the primordial Big Bang ³He abundance measured in Galactic H II regions;
2. Lowers the surface carbon isotope ratio ¹²C/¹³C from ~25 after first dredge-up down to the observed RGB tip values of 6–12 in globular clusters and open clusters; and
3. Depletes surface lithium (⁷Li) and boosts surface nitrogen (¹⁴N) in over 95% of evolved red giants brighter than the luminosity bump.

Why does the ³He(³He, 2p)⁴He reaction lower the mean molecular weight whereas normal fusion raises it?

Most fusion reactions fuse lighter nuclei into fewer heavier nuclei (e.g., 4 ¹H → ⁴He reduces 4 ions and 4 electrons to 1 ion and 2 electrons, raising μ from 0.50 to 1.33). By contrast, ³He + ³He → ⁴He + 2 p takes 2 helium-3 nuclei (total 6 particles including 4 electrons, μ = 6/6 = 1.00) and splits them into 3 ions (one ⁴He + two protons, total 7 particles including 4 electrons, μ = 6/7 ≈ 0.857), increasing the number of free particles per unit mass and lowering μ.

Why does thermohaline mixing wait until the red-giant luminosity bump to begin?

Before the luminosity bump, the retreating first dredge-up envelope leaves behind a steep positive composition gradient (∇_μ > 0) that is three orders of magnitude larger than the tiny Δμ / μ ≈ −10⁻⁴ inversion produced by ³He burning. Only after the advancing hydrogen-burning shell erases this composition discontinuity at the luminosity bump can the ³He inversion establish ∇_μ < 0 in a chemically homogeneous radiative zone.

1. Ulrich, R. K. (1972). "Thermohaline Convection in Stellar Interiors." The Astrophysical Journal, 172, 165–177. https://doi.org/10.1086/151336
2. Kippenhahn, R., Ruschenplatt, G., & Thomas, H.-C. (1980). "The time scale of thermohaline mixing in stars." Astronomy and Astrophysics, 91, 175–180. https://ui.adsabs.harvard.edu/abs/1980A%26A....91..175K
3. Charbonnel, C., & Zahn, J.-P. (2007). "Thermohaline mixing: a physical mechanism governing the photospheric composition of low-mass giants." Astronomy & Astrophysics, 467(1), L15–L18. https://doi.org/10.1051/0004-6361:20077274
4. Wachlin, F. C., Vauclair, G., Vauclair, S., & Althaus, L. G. (2017). "Importance of thermohaline mixing for the chemical composition of stars and white dwarfs." Astronomy & Astrophysics, 601, A13. https://doi.org/10.1051/0004-6361/201629790