Radiative Temperature Gradient

Deep inside a star the temperature runs to millions of kelvin and the plasma glows with a fierce thermal radiation field. Where the gas is stable against overturning, that radiation is how the star ships its energy outward — not by streaming freely, but by the slow, groping random walk of photons that are absorbed and re-emitted a countless number of times before they ever reach the surface. A photon does not so much fly out of the Sun as diffuse out, and diffusion needs a slope. The radiative temperature gradient is exactly that slope: the value of dT/dr for which the diffusive leak of heat down the temperature gradient carries precisely the luminosity the star must transport at that radius. Set the diffusive flux equal to the required flux and the gradient is fixed by four local quantities — the Rosseland mean opacity κ, the density ρ, the temperature T, and the enclosed luminosity $L_r$ . This single relation is one of the four coupled equations of stellar structure, sitting alongside hydrostatic equilibrium, mass continuity, and energy generation. Recast in the dimensionless logarithmic form $\nabla_{\mathrm{rad}}=(\mathrm{d}\ln T/\mathrm{d}\ln P)_{\mathrm{rad}}$ , it becomes the yardstick against which the adiabatic gradient $\nabla_{\mathrm{ad}}$ is measured. That comparison — the Schwarzschild criterion — decides the deepest structural question a star faces at every radius: does radiation carry the heat, or must the gas boil? This article owns that gradient: where it comes from, what sets its size, and why its steepening is what lights the convective fire in stellar envelopes and cores.

Key takeaways

In a radiative zone, energy leaks outward by photon diffusion, and the temperature gradient self-adjusts so the diffusive flux carries exactly the local luminosity $L_r$ .
The gradient follows from the diffusion limit of radiative transfer: dT/dr = −3κρ $L_r$ / (16π a c T³ r²), steepened by opacity and flux, flattened by temperature.
The dimensionless logarithmic form $\nabla_{\mathrm{rad}}$ = 3κ $L_r$ P / (16π a c G m T⁴) is the quantity actually compared against the adiabatic gradient.
When $\nabla_{\mathrm{rad}}$ exceeds $\nabla_{\mathrm{ad}}$ ≈ 0.4, the layer is convectively unstable — the Schwarzschild criterion — and the true gradient is then pinned near $\nabla_{\mathrm{ad}}$ , not at $\nabla_{\mathrm{rad}}$ .
Opacity spikes at hydrogen and helium ionization zones drive $\nabla_{\mathrm{rad}}$ up, which is why the Sun turns convective in its outer 29% by radius (base at 0.71 R_⊙).
Key idea: the radiative gradient is the temperature slope photon diffusion must have to carry exactly the local luminosity $L_r$ ; it scales as κρ $L_r$ /(T³r²), so opacity and flux steepen it while temperature flattens it — and its excess over $\nabla_{\mathrm{ad}}$ is what triggers convection.

The core insight

The radiative gradient is the temperature slope that photon diffusion must have to carry exactly the local luminosity $L_r$ ; it scales as κρ $L_r$ /(T³r²), so opacity and flux steepen it and temperature flattens it, and the moment it would exceed the adiabatic gradient the layer convects instead.

Where this lives today

The radiative gradient is not a historical relic — it is evaluated billions of times in every run of a stellar-evolution code. Modern codes such as MESA carry equation (3) at each of thousands of mesh points, look up κ from OPAL or OP opacity tables as a function of the local composition, temperature, and density, compare $\nabla_{\mathrm{rad}}$ against $\nabla_{\mathrm{ad}}$ (or the Ledoux criterion where composition gradients matter), and switch between the radiative gradient and a mixing-length convective gradient accordingly. The stakes are real: the size of a convective core, set precisely by where $\nabla_{\mathrm{rad}}$ falls back below $\nabla_{\mathrm{ad}}$ , fixes how much fuel a massive star can burn and therefore its main-sequence lifetime, and the treatment of "convective overshoot" past that boundary is one of the largest remaining uncertainties in stellar-age determinations. Helioseismology has turned the base of the solar convection zone into a precision test: the sound-speed inversions place it at 0.713 ± 0.001 R_⊙, and matching that number is a stringent constraint on the interior opacity — the "solar abundance problem," in which revised (lower) metallicities spoil the once-perfect agreement, is fought out precisely through κ in the radiative gradient. Asteroseismology with Kepler and TESS now extends the same test to thousands of other stars, reading off the depth of their convective envelopes from the oscillation frequencies and feeding it straight back to the opacities and gradients that this article's equations encode.

The physical picture

Imagine standing inside the Sun at half a solar radius. In every direction the plasma is blindingly bright, and to an excellent approximation the radiation field is isotropic — the same brightness whichever way you look. But not quite the same: the side facing the hot core is a hair brighter than the side facing the cool surface, because the material a mean free path inward is slightly hotter and radiates a touch more. That tiny fore-aft imbalance, no more than a part in $10^{9}$ or so, is the entire net outward flux. Everything else cancels. And the size of the imbalance is set by how fast the temperature falls with radius: the steeper dT/dr, the larger the leftover flux. Photons take a step of length ℓ = 1/(κρ), get absorbed and re-emitted in a random new direction, and repeat — a random walk whose statistical drift down the temperature gradient is the luminosity.

Figure

Photons random-walk outward through the opaque interior, each free flight of length ℓ = 1/(κρ) ≈ 0.006 cm near the solar centre, absorbed and re-emitted in a random direction. A net outward flux F = L_r / (4π r²) survives only because the hot inner side (T + dT) is very slightly brighter than the cool outer side (T), an anisotropy proportional to the temperature gradient dT/dr.

Because diffusion is inefficient when the mean free path is short, opacity is the throttle on the whole process. The radiative "conductivity" scales as T³/(κρ): hot, transparent matter conducts heat radiatively with ease and needs only a gentle gradient, while cool, opaque matter is a poor radiative conductor and demands a steep one. Push the opacity up far enough — as happens where hydrogen and helium partially ionize in a stellar envelope, or in cool molecule-rich atmospheres — and the gradient the star would need to keep transporting $L_r$ by radiation alone becomes so steep that it exceeds the adiabatic value. At that point the layer can no longer stay put: it overturns, and convection takes over. The second figure is the quantity a stellar modeller actually computes at every mesh point — the run of $\nabla_{\mathrm{rad}}$ through the Sun and the radius where it crosses $\nabla_{\mathrm{ad}}$ .

Figure

The two gradients through the Sun. The adiabatic gradient ∇_ad ≈ 0.40 is nearly flat (green); the radiative gradient ∇_rad (blue) is small in the hot deep interior but climbs steeply outward as opacity rises and temperature falls. Where ∇_rad first exceeds ∇_ad, at r = 0.71 R_⊙, the Schwarzschild criterion is met and the base of the solar convection zone begins (amber). Interior of that radius is radiative, exterior is convective.

The core equations

Start from the flux, because the gradient is defined by what the flux must carry. In the optically thick interior the radiation field is nearly isotropic, and the transfer equation collapses to a diffusion law: the flux is minus a "conductivity" times the gradient of the radiation energy density $u=aT^4$ . This is Fick's law for a photon gas whose step length is one mean free path, ℓ = 1/(κρ).

F_{\mathrm{rad}} = -\frac{c}{3\kappa\rho}\,\frac{\mathrm{d}(aT^4)}{\mathrm{d}r} = -\frac{4ac\,T^3}{3\kappa\rho}\,\frac{\mathrm{d}T}{\mathrm{d}r} \qquad (1)

where: $F_{\mathrm{rad}}$ — radiative energy flux (erg cm⁻² s⁻¹); $\kappa$ — Rosseland mean opacity (cm² g⁻¹); $\rho$ — mass density; $a=4\sigma/c\approx7.566\times10^{-15}$ erg cm⁻³ K⁻⁴ — radiation constant; $c$ — speed of light; ℓ = 1/(κρ) — photon mean free path. Reading it: the flux flows down the temperature gradient (hence the minus sign), it grows with the radiative conductivity $4acT^3/(3\kappa\rho)$ , and it is choked by opacity — double κ and you halve the flux a given gradient can carry. Now demand that this diffusive flux deliver exactly the luminosity threading the sphere at radius $r$ , namely $F=L_r/(4\pi r^2)$ . Solving for the slope gives the dimensional radiative temperature gradient — the central result of this article.

\frac{\mathrm{d}T}{\mathrm{d}r} = -\frac{3}{16\pi a c}\,\frac{\kappa\rho}{T^3}\,\frac{L_r}{r^2} \qquad (2)

where: $L_r$ — luminosity (energy per second) flowing through radius $r$ ; $T$ — local temperature; $r$ — radial coordinate. Reading it: the required slope steepens with opacity κ, density ρ, and flux $L_r/r^2$ , and flattens hard with temperature as $T^{-3}$ . A hot, transparent, low-flux region needs only a whisper of a gradient; a cool, opaque, high-flux region needs a cliff. The dimensional form mixes $r$ , ρ, and $L_r$ in a way that is awkward to compare against thermodynamics. Divide equation (2) by the equation of hydrostatic equilibrium, dP/dr = −Gmρ/r², and the density and $r^2$ cancel cleanly, leaving a dimensionless logarithmic derivative — the form every stellar-structure code carries.

\nabla_{\mathrm{rad}} \equiv \left(\frac{\mathrm{d}\ln T}{\mathrm{d}\ln P}\right)_{\mathrm{rad}} = \frac{3}{16\pi a c G}\,\frac{\kappa\,L_r\,P}{m\,T^4} \qquad (3)

where: $\nabla_{\mathrm{rad}}$ — dimensionless radiative gradient; $P$ — total pressure (gas + radiation); $m$ — mass enclosed within radius $r$ (i.e. $M_r$ ); $G$ — gravitational constant. Reading it: $\nabla_{\mathrm{rad}}$ is the fractional change in $\ln T$ per fractional change in $\ln P$ that radiation alone would produce. It is not necessarily the actual gradient in the star — it is the gradient radiation would need to carry all of $L_r$ . That distinction is the whole game, because there is a competitor: convection. A displaced blob that rises and expands adiabatically follows $\nabla_{\mathrm{ad}}$ . If the ambient (radiative) gradient is shallower than adiabatic, the blob sinks back and the layer stays radiative; if it is steeper, the blob keeps rising — runaway. This is the Schwarzschild criterion.

\nabla_{\mathrm{rad}} > \nabla_{\mathrm{ad}} \quad\Longrightarrow\quad \text{convectively unstable}, \qquad \nabla_{\mathrm{ad}} = \frac{\gamma-1}{\gamma} = \frac{2}{5}\ \ (\text{monatomic ideal gas}) \qquad (4)

where: $\nabla_{\mathrm{ad}}$ — adiabatic gradient; γ — adiabatic index (5/3 for a monatomic ideal gas, giving $\nabla_{\mathrm{ad}}=0.4$ ). Reading it: compare the two numbers at each radius. Below the threshold the region is radiative and equation (2) governs its thermal structure; above it the region convects and the actual gradient is held down close to $\nabla_{\mathrm{ad}}$ , with only a tiny superadiabatic excess where convection is efficient. Finally, the same physics sets a ceiling. Rearranging (2) for the luminosity a radiative structure can push shows that radiation exerts an outward force κF/c per unit mass; when that force equals gravity, no larger luminosity can be carried in hydrostatic balance — the Eddington luminosity.

L_{\mathrm{Edd}} = \frac{4\pi c\,G\,m}{\kappa} \qquad (5)

where: $L_{\mathrm{Edd}}$ — Eddington luminosity; κ — opacity (electron scattering, κ ≈ 0.34 cm² g⁻¹, in the classic case). Reading it: as $L_r\to L_{\mathrm{Edd}}$ the radiative gradient and the radiation-pressure gradient blow up together and the layer is on the verge of being blown off. For the Sun with electron-scattering opacity $L_{\mathrm{Edd}}\approx3.8\times10^{38}$ erg s⁻¹, about $4\times10^{4}\,L_\odot$ — the Sun runs safely far below it.

Derivation AThe radiative gradient from photon diffusion

The goal is to obtain the dimensional gradient dT/dr = −(3/16πac)(κρ/T³)( $L_r$ /r²) from first principles, treating the interior radiation as a diffusing photon gas. Deep inside a star the optical depth across a scale height is enormous, τ ≫ 1, so the specific intensity is nearly isotropic and can be expanded to first order in a small directed part. We integrate that anisotropy to get the flux, cast the result as Fick's law for the energy density $aT^4$ , and equate it to the flux that carries the enclosed luminosity through a spherical shell — isolating the slope at the end.

Step 1. Write the intensity as an isotropic part plus a small dipole, I(θ) ≈ I₀ + I₁ cos θ, where θ is measured from the outward radial direction and I₁ ≪ I₀. Why: at τ ≫ 1 the field is almost perfectly isotropic; only the cos θ piece survives the angular integral that defines the net flux, so keeping it to first order is both necessary and sufficient.

Step 2. Integrate the radial component of intensity over solid angle: the isotropic part cancels by symmetry and the net flux is F = ∫ I cos θ dΩ = (4π/3) I₁. Why: this identifies the physical carrier of the flux as the tiny fore-aft brightness difference I₁, not the huge isotropic background — the whole luminosity lives in the anisotropy.

Step 3. The transfer equation in the diffusion limit relates that anisotropy to the gradient of the energy density, giving Fick's law F = −(c/3κρ) du/dr with u = aT⁴. Why: solving the transfer equation to first order shows the flux is a conductivity c ℓ /3 (with ℓ = 1/κρ) times the driving gradient — the factor 1/3 is the same one that gives radiation pressure P_rad = u/3.

Step 4. Apply the chain rule to u = aT⁴, since d(aT⁴)/dr = 4aT³ dT/dr, to write F = −(4acT³/3κρ) dT/dr. Why: this converts a gradient in energy density into the temperature gradient we actually want to solve for, and exposes the strong T³ radiative-conductivity dependence.

Step 5. Impose energy conservation through a spherical shell: the flux crossing radius r must sum to the enclosed luminosity, F = $L_r$ /(4π r²). Why: this is the boundary condition that turns an identity about fluxes into a statement about the star — it forces the local slope to be whatever it takes to carry $L_r$ .

Step 6. Set the two expressions for F equal and solve, giving dT/dr = −3κρ $L_r$ /(16π a c T³ r²). Why: this is the required gradient; its dimensions are K cm⁻¹ and it matches Carroll and Ostlie equation (10.68), the standard form used throughout stellar-structure theory. *The dimensions check out: [κρL_r/(acT³r²)] = K cm⁻¹, confirming the gradient has the right units. The factor of 1/3 traces directly to the isotropic-averaged radiation pressure, the same factor that gives P_rad = u/3.*

Go deeper · Derivation B · 2 minLogarithmic form and the Schwarzschild criterion

Derivation BLogarithmic form and the Schwarzschild criterion

The aim is to recast the dimensional gradient into the dimensionless $\nabla_{\mathrm{rad}}$ and read off the condition for convective instability. Working in the $\ln T$ versus $\ln P$ plane removes the explicit radius and density and puts the radiative slope on the same footing as the adiabatic slope, so the two can be compared by a single inequality. We divide by hydrostatic equilibrium to eliminate ρ and r², multiply by P/T to nondimensionalize, then argue physically why the larger of the two competing gradients does not simply win but instead switches on convection.

Step 1. Divide equation (2) by hydrostatic equilibrium dP/dr = −Gmρ/r². The density ρ and the geometric factor r² cancel between numerator and denominator, leaving dT/dP = 3κ $L_r$ /(16π a c G m T³). Why: this is the crucial simplification — the messy structural quantities ρ and r drop out, and the slope in the natural thermodynamic variables depends only on κ, $L_r$ , m, and T.

Step 2. Multiply by P/T to form the logarithmic derivative, $\nabla_{\mathrm{rad}}$ = (P/T) dT/dP = 3κ $L_r$ P/(16π a c G m T⁴). Why: d ln T/d ln P is dimensionless and directly comparable to $\nabla_{\mathrm{ad}}$ , which is what makes this the working form in every stellar-structure integration.

Step 3. Recognize that the actual gradient obeys $\nabla_{\mathrm{actual}}$ = min( $\nabla_{\mathrm{rad}}$ , ≈ $\nabla_{\mathrm{ad}}$ ): if radiation can carry $L_r$ with $\nabla_{\mathrm{rad}} < \nabla_{\mathrm{ad}}$ the zone stays radiative, otherwise convection intervenes and pins the gradient just above $\nabla_{\mathrm{ad}}$ . Why: $\nabla_{\mathrm{rad}}$ is a demand, not a description — when the demand exceeds what a stable stratification permits, a second transport channel opens rather than the demand being met.

Step 4. State the instability condition, $\nabla_{\mathrm{rad}} > \nabla_{\mathrm{ad}}$ : a blob displaced upward rises adiabatically along $\nabla_{\mathrm{ad}}$ and, in a superadiabatic surrounding, stays hotter and less dense than its new surroundings, so buoyancy keeps pushing it — runaway convection. Why: this is the Schwarzschild criterion, the single inequality that partitions every star into radiative and convective zones and locates the boundaries of convective cores and envelopes. *The P/T factor is the standard nondimensionalization that converts dT/dP into the dimensionless logarithmic derivative d ln T/d ln P.*

Go deeper · Derivation C · 1 minThe Eddington luminosity as the radiative ceiling

Derivation CThe Eddington luminosity as the radiative ceiling

The purpose is to show that the very equation governing radiative transport also caps how much luminosity a star can carry, by deriving the Eddington luminosity $L_{\mathrm{Edd}}$ = 4π c G m/κ directly from the flux. Radiation carries not only energy but momentum, so a diffusing flux exerts an outward force on the matter it pushes through; when that radiative force approaches the local gravity, hydrostatic equilibrium can no longer hold and the layer is on the brink of being ejected. We compute the radiative acceleration from the flux, balance it against gravity, and solve for the critical luminosity.

Step 1. The radiation-pressure gradient that drives the flux is dP_rad/dr = −(κρ/c) F, since each gram of matter absorbs momentum at a rate κF/c from the radiation streaming past it. Why: this is the momentum form of the same diffusion law used for energy — opacity couples the matter to the photon momentum just as it couples it to photon energy.

Step 2. Substitute F = $L_r$ /(4π r²) to get the outward radiative acceleration per unit mass, g_rad = κ $L_r$ /(4π r² c). Why: dividing the force by the mass isolates the acceleration that competes directly with gravity, independent of how much matter is present.

Step 3. Set the radiative acceleration equal to the gravitational acceleration Gm/r², the point at which net support vanishes: κ $L_r$ /(4π r² c) = Gm/r². Why: this is the marginal condition — below it gravity wins and the star is bound, above it radiation would blow the layer outward, so equality defines the maximum stable luminosity.

Step 4. The r² cancels; solve for the critical luminosity, $L_{\mathrm{Edd}}$ = 4π c G m/κ. Why: this Eddington luminosity is the ceiling implicit in the radiative-transport equation — a star trying to carry more than $L_{\mathrm{Edd}}$ radiatively cannot remain in hydrostatic equilibrium, which caps the luminosity and hence the mass of stable stars. *For the Sun with electron-scattering opacity κ = 0.34 cm² g⁻¹, this gives $L_{\mathrm{Edd}}$ ≈ 3.8×10³⁸ erg s⁻¹ ≈ 4×10⁴ $L_\odot$ .*

Limiting cases

How the required radiative gradient responds when one local condition is pushed to an extreme, holding the others fixed:

Case	Result	Meaning
High opacity	$\nabla_{\mathrm{rad}} \propto \kappa$	Opaque matter — ionization zones, cool molecular gas — forces a steep gradient; this is the usual trigger for envelope convection.
High flux / luminosity	$\nabla_{\mathrm{rad}} \propto L_r$	Centrally concentrated energy generation (CNO cycle) makes $L_r/m$ huge in the core, steepening $\nabla_{\mathrm{rad}}$ and producing convective cores in massive stars.
High temperature	$\nabla_{\mathrm{rad}} \propto T^{-4}$ (fixed P)	Hot deep interiors are superb radiative conductors; the gradient stays shallow and the region remains radiative.
Optically thick (diffusion) limit	$\tau \gg 1,\quad F = -\dfrac{c}{3\kappa\rho}\dfrac{\mathrm{d}(aT^4)}{\mathrm{d}r}$	The whole formalism assumes many mean free paths per scale height; it degrades near the photosphere where τ ~ 1.
Approaching Eddington	$L_r \to L_{\mathrm{Edd}} = 4\pi cGm/\kappa$	The radiative and radiation-pressure gradients diverge together; the layer loses hydrostatic support and is on the verge of being driven off.

Common trap∇_rad is the actual temperature gradient of the star

It is not — and this is the single most common misreading. $\nabla_{\mathrm{rad}}$ is the gradient that would be required to carry the entire local luminosity $L_r$ by radiation alone. In a radiative zone that demand is met and $\nabla_{\mathrm{rad}}$ is indeed the true slope. But wherever $\nabla_{\mathrm{rad}}>\nabla_{\mathrm{ad}}$ , the stratification is convectively unstable and the gas overturns: convection then carries most of the flux, and the actual gradient drops to just a hair above $\nabla_{\mathrm{ad}}$ (in the deep, efficient-convection interior) or somewhere between $\nabla_{\mathrm{ad}}$ and $\nabla_{\mathrm{rad}}$ (in the inefficient near-surface layers, where mixing-length theory is needed). Quoting a value of $\nabla_{\mathrm{rad}}$ = 3 in a convective envelope does not mean the temperature is plunging at three times the adiabatic rate — it means the region convects, and $\nabla_{\mathrm{rad}}$ is merely the size of the radiative demand that convection is relieving.

Worked examples

1 · How long does a photon take to leak out of the Sun? (confidence builder)

Estimate the time for radiant energy to random-walk from the solar centre to the surface, given a representative interior mean free path. Take the solar radius R_⊙ = 7.0×10¹⁰ cm and c = 3.0×10¹⁰ cm s⁻¹.

Step 1. Fix the step length. Near the centre κ ≈ 1.2 cm² g⁻¹ and ρ ≈ 150 g cm⁻³, so ℓ = 1/(κρ) = 1/(1.2 × 150) ≈ 0.006 cm — a photon flies barely a tenth of a millimetre before being absorbed. Density falls steeply outward, so a representative mean free path over the whole star is larger; take ℓ ≈ 1 cm as an order-of-magnitude average.

Step 2. A random walk of N steps of length ℓ covers a net displacement of only √N ℓ, so escaping a distance R requires N = (R/ℓ)² steps. The total path length is L_path = N ℓ = R²/ℓ, and the time is that path divided by the speed of light: t = R²/(ℓ c).

Step 3. Insert the numbers: t = (7.0×10¹⁰)² / (1 × 3.0×10¹⁰) = 4.9×10²¹ / 3.0×10¹⁰ ≈ 1.6×10¹¹ s ≈ 5000 yr. A more carefully weighted interior model raises this to of order 10⁴–10⁵ yr, but the estimate makes the point: the energy now leaving the Sun was generated in the core tens of thousands of years ago. The diffusion is slow precisely because the mean free path is minute — the same opacity that appears in $\nabla_{\mathrm{rad}}$ .

2 · The base of the solar convection zone (fully worked)

At the base of the Sun's convective envelope, standard solar models give T ≈ 2.2×10⁶ K, P ≈ 6.0×10¹³ dyn cm⁻², Rosseland opacity κ ≈ 20 cm² g⁻¹, enclosed mass m ≈ 0.98 M_⊙ = 1.95×10³³ g, and essentially the full luminosity $L_r$ = L_⊙ = 3.83×10³³ erg s⁻¹. Compute $\nabla_{\mathrm{rad}}$ and compare it to $\nabla_{\mathrm{ad}}$ .

Step 1. Assemble the constants: a = 7.566×10⁻¹⁵ erg cm⁻³ K⁻⁴, c = 3.0×10¹⁰ cm s⁻¹, G = 6.674×10⁻⁸ cgs, and 16π = 50.27. The working formula is equation (3), $\nabla_{\mathrm{rad}}$ = 3κ $L_r$ P / (16π a c G m T⁴).

Step 2. Numerator: 3 κ $L_r$ P = 3 × 20 × (3.83×10³³) × (6.0×10¹³) = 1.38×10⁴⁹.

Step 3. Denominator, built up factor by factor: 16π a = 50.27 × 7.566×10⁻¹⁵ = 3.80×10⁻¹³; times c gives 1.14×10⁻²; times G gives 7.62×10⁻¹⁰; times m gives 1.49×10²⁴. Then T⁴ = (2.2×10⁶)⁴ = 2.34×10²⁵, so the full denominator is 1.49×10²⁴ × 2.34×10²⁵ = 3.48×10⁴⁹.

Step 4. Divide: $\nabla_{\mathrm{rad}}$ = 1.38×10⁴⁹ / 3.48×10⁴⁹ = 0.40. Compare with the monatomic ideal-gas value $\nabla_{\mathrm{ad}}$ = (γ−1)/γ = 2/5 = 0.40. The two gradients are equal — which is exactly the definition of the base of the convection zone: $\nabla_{\mathrm{rad}} = \nabla_{\mathrm{ad}}$ marks the marginal radius, here 0.71 R_⊙.

Step 5. Read the trend. Just outside this radius, temperature keeps falling and the H and He ionization opacity keeps climbing, so $\nabla_{\mathrm{rad}} \propto \kappa/T^4$ shoots above 0.40 and the gas convects. Just inside, the hotter, more transparent plasma has $\nabla_{\mathrm{rad}} < 0.40$ and stays radiative. The single crossing of two numbers fixes one of the most important structural boundaries in the Sun, in agreement with helioseismology to about one percent in radius.

Physics of the portrait

The radiative gradient and the criterion that decides when it must yield to convection both trace to one man, whose name is attached to the inequality every stellar model still evaluates.

Origins · 1 minKarl Schwarzschild and how this was discovered

Karl Schwarzschild

1873–1916 · radiative equilibrium and the criterion for convection

Karl Schwarzschild is remembered by most for the exact black-hole solution he found in the trenches of the First World War, but to stellar physics his decisive contribution came a decade earlier. In a 1906 study of the Sun's atmosphere he formulated the idea of radiative equilibrium — the notion that a stable stellar layer transports its heat by radiation, with the temperature distribution set by the balance of absorbed and re-emitted photons rather than by convection. In doing so he wrote down the condition that distinguishes the two regimes: a layer stays radiative only if the temperature gradient it would need is shallower than the adiabatic gradient a rising parcel would follow, and convects otherwise. That comparison is the Schwarzschild criterion, $\nabla_{\mathrm{rad}}>\nabla_{\mathrm{ad}}$ , and it remains the first test applied at every mesh point of a modern stellar-evolution code. Schwarzschild built on the recognition — pressed by R. A. Sampson and soon systematized by Eddington — that radiation, not convection, dominates the deep interior, and his framework was completed when Svein Rosseland defined in 1924 the correctly weighted mean opacity that makes the diffusion equation exact. Schwarzschild died in 1916 of an illness contracted at the front, at forty-two; his son Martin became one of the twentieth century's great stellar-structure theorists, carrying the family name into the numerical models that his father's criterion still governs.

History

1894

— R. A. Sampson argues that radiation, not convection, carries the flux through a stellar interior

1906

— Karl Schwarzschild formulates radiative equilibrium and the criterion for convective instability in the Sun's atmosphere

1924

— Svein Rosseland defines the harmonic mean opacity that makes the radiative diffusion equation exact

1926

— Arthur Eddington publishes The Internal Constitution of the Stars, cementing radiative transport as the basis of stellar structure

1958

— Erika Böhm-Vitense develops mixing-length theory, supplying the actual gradient once convection sets in

Going further

Texts. Carroll and Ostlie, An Introduction to Modern Astrophysics, 2nd ed., §10.4 (Energy Transport and Thermodynamics) · Kippenhahn, Weigert and Weiss, Stellar Structure and Evolution, 2nd ed., Ch. 5 and §6.1 · Hansen, Kawaler and Trimble, Stellar Interiors, 2nd ed., Ch. 4 · Prialnik, An Introduction to the Theory of Stellar Structure and Evolution, Ch. 5. Key numerical anchors for this article: the radiation constant a = 4σ/c ≈ 7.566×10⁻¹⁵ erg cm⁻³ K⁻⁴; the dimensional gradient dT/dr follows from equating F = −(4acT³/3κρ) dT/dr to $L_r$ /(4πr²); the logarithmic form $\nabla_{\mathrm{rad}}$ = 3κ $L_r$ P/(16πacGmT⁴) follows by dividing by hydrostatic equilibrium; $\nabla_{\mathrm{ad}}$ = (γ−1)/γ = 2/5 for a monatomic ideal gas; the worked base-of-convection-zone value gives $\nabla_{\mathrm{rad}}$ ≈ 0.40 at T = 2.2×10⁶ K, P = 6.0×10¹³ dyn cm⁻², κ = 20 cm² g⁻¹, m = 0.98 M_⊙; and $L_{\mathrm{Edd}}$ = 4πcGm/κ ≈ 3.8×10³⁸ erg s⁻¹ for the Sun.

Prerequisites. Opacity and Radiative Transfer in Stars · Hydrostatic Equilibrium in Stars · Radiation Pressure in Stellar Interiors · Equation of State for Stellar Matter. Leads to. Adiabatic Temperature Gradient · Schwarzschild and Ledoux Criteria for Convection · Convective Energy Transport · Equations of Stellar Structure.

Ask the tutor: Ask me to walk through why the Sun turns convective at exactly 0.71 R_⊙, or to compute $\nabla_{\mathrm{rad}}$ for a massive-star core and show how a convective core forms.

Is ∇_rad the temperature gradient the star actually has?

Only in radiative zones. $\nabla_{\mathrm{rad}}$ is the gradient that radiation would need to carry the whole local luminosity by diffusion. Where that value stays below the adiabatic gradient, radiation obliges and $\nabla_{\mathrm{rad}}$ is the real slope. But where $\nabla_{\mathrm{rad}}$ would exceed $\nabla_{\mathrm{ad}}$ , the layer is convectively unstable and overturns; convection then carries most of the flux and the actual gradient falls back to just above $\nabla_{\mathrm{ad}}$ . So a large computed $\nabla_{\mathrm{rad}}$ in an envelope is a signal that the region convects, not a statement that the temperature is really plunging that fast.

Why does the temperature appear as T³ (or T⁴) in these equations?

Because radiation is a T⁴ phenomenon. The energy density of a thermal photon gas is u = aT⁴, so its gradient carries a factor 4aT³ — that is the T³ in the flux and in dT/dr. When you nondimensionalize to $\nabla_{\mathrm{rad}}$ and pull a factor of T out through the pressure and the logarithmic derivative, it becomes T⁴ in the denominator. Physically the strong temperature dependence means the radiative conductivity T³/(κρ) rises steeply with depth: the hot interior conducts heat radiatively with great ease and needs only a gentle slope, while the cool outer layers are radiatively "insulating" and demand a steep one — which is why convection appears in envelopes, not cores of low-mass stars.

What exactly is the difference between the radiative and adiabatic gradients?

They answer two different questions. $\nabla_{\mathrm{rad}}$ = d ln T/d ln P is the ambient slope that radiative diffusion imposes on the star to move $L_r$ outward — it depends on opacity, luminosity, mass, pressure, and temperature. $\nabla_{\mathrm{ad}}$ = (γ−1)/γ is the slope a blob of gas would follow if you displaced it and let it expand without exchanging heat — it is a pure thermodynamic property of the gas (0.4 for a monatomic ideal gas). Comparing them is a stability test: if the surroundings fall in temperature faster than a rising blob does (∇_rad > ∇_ad), the blob stays buoyant and convection runs; if slower, the blob sinks back and the layer is stable.

Why does high opacity push a region toward convection?

Opacity is the resistance in the radiative "circuit." Since $\nabla_{\mathrm{rad}} \propto \kappa$ , doubling the opacity doubles the temperature gradient radiation needs to force the same luminosity through. Opacity peaks sharply where hydrogen and helium are partially ionized, because bound-free and bound-bound absorption become enormously effective there. In a stellar envelope those ionization zones sit at just the right temperature range, so κ spikes, $\nabla_{\mathrm{rad}}$ leaps above $\nabla_{\mathrm{ad}}$ , and the gas has no choice but to convect. This is why essentially all cool stars, including the Sun, have convective outer envelopes even though their deep interiors are radiative.

Where does the diffusion approximation break down?

Right where you can see it — at the photosphere. The whole derivation assumes the optical depth across a scale height is huge (τ ≫ 1), so the radiation field is nearly isotropic and the photon mean free path is minuscule compared to the distance over which conditions change. That holds superbly through the deep interior. But as radiation approaches the surface the mean free path grows until, at τ ~ 1, a photon can escape to space in a single step; the field becomes strongly anisotropic (bright inward, dark outward) and the simple Fickian flux is no longer valid. There one must solve the full radiative-transfer equation with a proper stellar-atmosphere model, which is a separate branch of the subject.

How does this gradient connect to the Eddington luminosity?

Through the momentum of the diffusing photons. The same flux that carries energy also carries momentum, exerting an outward force κF/c per unit mass on the plasma it pushes through. Writing the radiative gradient in terms of $L_r$ and κ and asking when that radiative force matches gravity gives $L_{\mathrm{Edd}}$ = 4πcGm/κ — the maximum luminosity a star can transport radiatively while staying in hydrostatic equilibrium. So the radiative-transport equation quietly sets a ceiling: as $L_r$ approaches $L_{\mathrm{Edd}}$ the gradient and radiation-pressure gradient diverge, and the layer is on the verge of being driven off. This ceiling is what ultimately limits the luminosity, and hence the mass, of stable stars.