Negative Heat Capacity of Stars

Hand a laboratory gas a joule of heat and it warms; take a joule away and it cools. A star does the opposite. Remove energy from a self-gravitating ball of ideal gas — let it shine into empty space — and it gets hotter. This is not a subtle correction or an exotic regime; it is the ordinary behaviour of every non-degenerate star, following in three lines from the virial theorem applied to hydrostatic equilibrium. The theorem locks the internal kinetic energy $K$ to the gravitational potential energy $U$ through $2K + U = 0$ , so the total energy is $E = K + U = -K$ . Because the thermal energy of an ideal gas is proportional to temperature, $E \propto -T$ : driving $E$ down drives $T$ up. The gravitational field behaves as a reservoir that pays out twice whatever the star radiates — half escapes as light, half stays as heat. Read as a heat capacity, $C \equiv \mathrm{d}E/\mathrm{d}T < 0$ , and this single sign organizes enormous tracts of stellar physics. It makes a fuel-less protostar contract, heat, and ignite rather than fade quietly; it turns the main sequence into a self-correcting thermostat that keeps hydrogen burning steady for gigayears; and when it fails — in an electron-degenerate core where pressure no longer tracks temperature — fusion detonates, as in the helium flash and the Type Ia supernova. The same negative heat capacity drives the gravothermal catastrophe in globular clusters and dark-matter halos. It is the signature of long-range attractive gravity, where the thermodynamics of a bound star parts ways with everything on a laboratory bench.

Key takeaways

The virial theorem for a steady, self-gravitating ideal-gas star gives $2K + U = 0$ , so $E = K + U = -K = \tfrac{1}{2}U < 0$ : a bound star's total energy equals minus its thermal energy.
Because $E = -K$ and $K \propto T$ , losing energy ( $E$ becomes more negative) forces $T$ to rise — the star has an effective negative heat capacity, $C = \mathrm{d}E/\mathrm{d}T < 0$ .
When the star radiates, only half the liberated gravitational energy leaves as light; the other half heats the interior, so radiating-at-the-surface makes the core hotter.
This drives pre-main-sequence Kelvin–Helmholtz contraction toward ignition ( $t_{\mathrm{KH},\odot}\approx3\times10^{7}$ yr) and makes the main sequence a self-regulating thermostat.
The effect requires a non-degenerate equation of state $P\propto\rho T$ ; when electron degeneracy makes $P$ nearly independent of $T$ , the virial link breaks and fusion can run away.
For a polytrope, $E = -\dfrac{3\gamma-4}{3(\gamma-1)}\,|U|$ , which is $-\tfrac12|U|$ at $\gamma=5/3$ and vanishes at $\gamma=4/3$ — the dynamical-instability threshold.
Key idea: For a virialized ideal-gas star $E = -K \propto -T$ ; radiating energy lowers $E$ and therefore raises $T$ — a system with negative heat capacity, contracting and heating as it loses energy.

The core insight

For a virialized ideal-gas star $E = -K \propto -T$ ; radiating energy lowers $E$ and therefore raises $T$ . The star behaves as a system with negative heat capacity, contracting and heating as it loses energy.

Where this lives today

Negative heat capacity is a working tool across contemporary astrophysics. Stellar-evolution codes such as MESA build it into every pre-main-sequence track, following a protostar down its Hayashi and Henyey paths as contraction heats the core to ignition on the Kelvin–Helmholtz timescale — the physics that sets where young stars land on the Hertzsprung–Russell diagram. In the explosive channel it underlies the cosmological distance ladder: a mass-accreting or merging carbon–oxygen white dwarf has a degenerate, heat-capacity-positive core, so carbon ignition runs away into a Type Ia supernova, and the near-uniformity of those detonations made them the standardizable candles behind the 1998 discovery of cosmic acceleration. In stellar dynamics, N-body simulations of globular clusters watch the gravothermal catastrophe unfold — a core losing energy, contracting, heating, and collapsing until binary stars halt it — and the same instability is now invoked for self-interacting dark-matter halos.

The physical picture

Picture the star as a loaded spring. Gravity is the compression that wants to collapse the ball; the thermal pressure of the hot gas is the restoring push holding it open. When the star radiates it drains a little of that thermal push, so the spring compresses and the star contracts. Here is the twist ordinary springs lack: compressing a self-gravitating ball releases gravitational energy — exactly twice what the star just lost. Half of that windfall is paid straight back out as the next instant's radiation; the other half has nowhere to go but into the random motion of the gas. The gas speeds up, the temperature climbs, and the drained pressure is restored at a higher temperature than before. The star cannot cool itself by shining. Shining is how it heats up. The sign is the whole story, and the cleanest way to feel it is to track $E$ . A bound star has $E < 0$ ; "losing energy" means $E$ slides further below zero, so $|E|$ grows. Since $E = -K$ , a growing $|E|$ is a growing $K$ , hence a growing temperature. The minus sign in $E = -K$ is not a bookkeeping detail — it is the phenomenon. The first figure follows one joule through the split so the "twice out, half back" accounting becomes a single picture.

Figure

Energy bookkeeping of a contracting star: gravity releases twice what is radiated, so half the released energy leaves as light L and half stays as heat, and while the total energy E falls the temperature T rises — the virial law 2K + U = 0 forcing C = dE dT below zero

This is why stars are self-regulating furnaces rather than bombs. Nudge the core with extra fusion and you add energy: $E$ rises toward zero, so $T$ falls, so the temperature-sensitive reactions throttle back down. It is also why a fuel-less star cannot park itself and cool to space: every joule it radiates only reheats it, and the only exit is for electron degeneracy to sever the virial link. The second figure makes the heating quantitative: for a contracting solar-mass protostar the virial relation $\langle T\rangle \propto 1/R$ carries the core up across the hydrogen-ignition threshold.

Figure

Virial mean temperature of a contracting one-solar-mass protostar: with ⟨T⟩ = (α 3)(μ m_H k_B)(GM R) the core temperature rises as 1 over R while the star shrinks, crossing the ~10⁷ K hydrogen-ignition line near R ≈ 0.7 R_sun — contraction alone drives the star to fusion

The core equations

The starting point is the virial theorem for a body in hydrostatic equilibrium under its own gravity. For a steady, self-gravitating ideal gas it ties the total thermal energy to the total gravitational energy with a fixed coefficient, and everything else in this article follows from it.

2K + U = 0 \qquad (1)

where: $K$ — total internal (thermal) kinetic energy, $K>0$ ; $U$ — total gravitational potential energy, $U<0$ for a bound star. Reading: the thermal energy is exactly half the binding magnitude, $K = \tfrac12|U|$ — a self-gravitating gas cannot be arbitrarily hot or cold, its temperature is pinned by its gravity. Adding the two pieces gives the total energy, and this is where the surprising sign is born: it comes out equal to minus the thermal energy, and to half the (negative) gravitational energy.

E = K + U = -K = \tfrac{1}{2}U \qquad (2)

where: $E$ — total energy, thermal plus gravitational; $E<0$ confirms the star is bound. Reading: a bound star's total energy is the negative of its heat content — pump heat in and $E$ rises toward zero (less bound); drain heat out and $E$ sinks (more bound). Differentiating $E=-K$ with respect to temperature defines the effective heat capacity, and the minus sign carries straight through.

C \equiv \frac{\mathrm{d}E}{\mathrm{d}T} = -\frac{\mathrm{d}K}{\mathrm{d}T} < 0 \qquad (3)

where: $C$ — heat capacity of the whole virialized star; $T$ — characteristic internal temperature with $K\propto T$ . Reading: an ordinary gas has $\mathrm{d}K/\mathrm{d}T>0$ , so $C$ is negative — the defining statement of the phenomenon. Setting the rate of energy loss equal to the luminosity gives the energy budget of a contracting star and exposes the "half and half" split.

L = -\dot{E} = +\dot{K} = -\tfrac{1}{2}\dot{U} \qquad (4)

where: $L$ — surface luminosity; $\dot{E},\dot{K},\dot{U}$ — time derivatives of the total, thermal, and gravitational energies. Reading: the gravitational energy released per unit time is $-\dot U = 2L$ ; exactly half ( $L$ ) is radiated and half ( $\dot K = L$ ) heats the gas. To make the temperature explicit, write $K$ for an ideal gas and $U$ for a self-gravitating ball and solve for the mean temperature — the relation plotted in the second figure.

\langle T\rangle = \frac{\alpha}{3}\,\frac{\mu m_H}{k_B}\,\frac{G M}{R} \qquad (5)

where: $M$ — stellar mass; $R$ — radius; $\mu$ — mean molecular weight ( $\approx0.6$ for ionized solar material); $m_H$ — hydrogen mass; $k_B$ — Boltzmann constant; $\alpha$ — structure constant of order unity ( $3/5$ for a uniform sphere, $\approx3/2$ for an $n=3$ polytrope). Reading: $\langle T\rangle\propto M/R$ , so contracting the star ( $R\downarrow$ ) raises its temperature — negative heat capacity as a formula. Finally, replacing the monatomic ideal gas by a general polytrope of index $\gamma$ shows how the binding energy — and the very existence of a bound state — depends on the equation of state.

E = -\frac{3\gamma - 4}{3(\gamma-1)}\,|U| \qquad (6)

where: $\gamma$ — adiabatic index. Reading: at $\gamma=5/3$ this is $-\tfrac12|U|$ (bound, $C<0$ ); as $\gamma\to4/3$ (radiation- or relativity-dominated) $E\to0$ and the star becomes marginally bound and dynamically unstable.

Derivation AThe virial theorem from hydrostatic equilibrium

The goal is to earn equation (1) rather than assert it — to show that $2K+U=0$ is a direct integral of the hydrostatic-equilibrium condition holding inside any static star, with no thermodynamics smuggled in. We start from force balance, weight it by a volume factor chosen so the gravitational side assembles into the potential energy, integrate across the star, and use the ideal-gas relation $P=\tfrac23u$ to turn the pressure integral into the thermal energy $K$ . The payoff is a fixed link between heat and gravity for every star in equilibrium.

Step 1. Write hydrostatic equilibrium, the balance between the outward pressure gradient and the inward pull of the mass $m(r)$ interior to radius $r$ : $\mathrm{d}P/\mathrm{d}r = -Gm(r)\rho/r^2$ . Why: this is the only physical input — a star that is neither expanding nor collapsing must have its pressure gradient exactly cancel gravity at every radius.

Step 2. Multiply both sides by $4\pi r^3$ and integrate from centre to surface: $\int_0^R 4\pi r^3\,(\mathrm{d}P/\mathrm{d}r)\,\mathrm{d}r = -\int_0^R \dfrac{Gm(r)}{r}\,4\pi r^2\rho\,\mathrm{d}r$ . Why: the factor $4\pi r^3$ is chosen so the right-hand side, using $\mathrm{d}m = 4\pi r^2\rho\,\mathrm{d}r$ , becomes $-\int_0^M (Gm/r)\,\mathrm{d}m = U$ , the gravitational potential energy.

Step 3. Integrate the left-hand side by parts. The boundary term $[4\pi r^3 P]_0^R$ vanishes since $P(R)=0$ and $r^3\to0$ at the centre, leaving $-3\int_0^R P\,4\pi r^2\,\mathrm{d}r = -3\int P\,\mathrm{d}V$ . Why: integration by parts trades the pressure gradient for the pressure itself, and the vanishing surface term — the star being self-contained — is what keeps the result clean.

Step 4. Equate the two sides: $-3\int P\,\mathrm{d}V = U$ . For a non-relativistic ideal gas the pressure is two-thirds of the internal energy density, $P=\tfrac23 u$ , so $3\int P\,\mathrm{d}V = 2\int u\,\mathrm{d}V = 2K$ , giving $2K + U = 0$ . Why: the relation $P=\tfrac23u$ converts the mechanical pressure integral into thermal energy — the single place the ideal-gas assumption enters, and the place degeneracy later breaks the chain.

Go deeper · Derivation B · 1 minFrom the virial theorem to a negative heat capacity

Derivation BFrom the virial theorem to a negative heat capacity

The goal is to show, with the sign made unmistakable, that a virialized ideal-gas star has $\mathrm{d}E/\mathrm{d}T<0$ , and to see how it generalizes as the equation of state changes. We combine the virial theorem with the ideal-gas expression for $K$ to get $E$ as an explicit function of temperature, differentiate, compare the magnitude with the ordinary gas heat capacity, and extend to a general index to expose the $\gamma=4/3$ boundary where the bound state ceases to exist.

Step 1. Use $E=-K$ with the ideal-gas thermal energy $K = \tfrac32 (M/\mu m_H)k_B\langle T\rangle$ to write $E = -\tfrac32 (M/\mu m_H)k_B\langle T\rangle$ . Why: this expresses the total energy directly in terms of temperature, so the heat capacity is now just a derivative — gravity has already done its work through the minus sign.

Step 2. Differentiate at fixed mass and composition: $C = \mathrm{d}E/\mathrm{d}\langle T\rangle = -\tfrac32 M k_B/(\mu m_H)$ , manifestly negative. Why: nothing on the right depends on temperature, so the sign is exact for the whole virialized star, not a limit or approximation.

Step 3. Compare magnitudes. The ordinary heat capacity of the same gas is $C_V = +\tfrac32 M k_B/(\mu m_H)$ , so the star's effective capacity is $C = -\tfrac12 C_V$ : half the size, opposite sign. Why: the factor $\tfrac12$ is the "half and half" split quantified — of every unit you remove, gravity releases two, so the net thermal change is half a unit and reversed.

Step 4. Repeat for a polytrope with $P=(\gamma-1)u$ . The virial theorem becomes $3(\gamma-1)K + U = 0$ , so $K = |U|/[3(\gamma-1)]$ and $E = K+U = -\dfrac{3\gamma-4}{3(\gamma-1)}|U|$ . Why: this recovers $E=-\tfrac12|U|$ at $\gamma=5/3$ and shows $E\to0$ as $\gamma\to4/3$ — the binding energy, and with it the stabilizing negative heat capacity, evaporates where radiation pressure or relativistic degeneracy softens the gas toward $\gamma=4/3$ .

Go deeper · Derivation C · 2 minThe thermonuclear thermostat and its failure in degenerate matter

Derivation CThe thermonuclear thermostat and its failure in degenerate matter

The goal is to turn negative heat capacity into a stability criterion for nuclear burning and identify the assumption whose failure converts stability into catastrophe. We perturb a burning core with a small excess of energy generation and track the temperature response in two cases: an ideal-gas core, where the virial link holds, and a degenerate core, where pressure no longer depends on temperature. The temperature-sensitive rate $\varepsilon\propto T^\nu$ ( $\nu\sim4$ for the pp chain, $\sim16$ for CNO) then either damps the perturbation or amplifies it into a runaway.

Step 1. In an ideal-gas core, add a burst of nuclear energy so $\delta E>0$ . Negative heat capacity gives $\delta T = -\delta E/|C| < 0$ : the core expands and cools. Why: counterintuitive but decisive — extra heat makes a virialized core colder, because the added energy loosens the binding and the gas expands and drops in temperature.

Step 2. Feed the cooler temperature back into the rate. Since $\varepsilon\propto T^\nu$ with $\nu$ large, a small $\delta T<0$ drops $\varepsilon$ substantially, pushing burning back toward equilibrium. Why: the loop $\delta E>0 \Rightarrow \delta T<0 \Rightarrow \delta\varepsilon<0$ closes with the opposite sign to the disturbance — negative feedback, and the reason main-sequence stars burn placidly for billions of years even with a ferociously $T$ -sensitive CNO cycle.

Step 3. Now make the core electron-degenerate, so $P\approx P(\rho)$ , nearly independent of $T$ . Adding nuclear energy no longer expands the core, because the supporting pressure ignores temperature; the virial link $2K=-U$ is broken and the effective heat capacity turns positive. Why: degeneracy severs the one relation, $P\propto\rho T$ , on which the whole argument rested — the equation-of-state assumption flagged in the first derivation.

Step 4. Track the perturbation in the degenerate case: $\delta E>0 \Rightarrow \delta T>0 \Rightarrow \delta\varepsilon>0 \Rightarrow \delta T>0 \Rightarrow \cdots$ . With no expansion to cool the gas, heat raises the temperature, which raises the rate, which adds more heat — a thermonuclear runaway. Why: the loop that stabilized the ideal-gas core now closes with the same sign as the disturbance — the physics of the helium flash and of the Type Ia carbon detonation; it halts only when the temperature climbs high enough to lift the degeneracy and restore $P\propto\rho T$ .

Limiting cases

The sign and magnitude of the heat capacity are set by the equation of state through the adiabatic index $\gamma$ . The four regimes below span the canonical stable star to the marginally bound object on the edge of collapse.

Case	Result	Meaning
Ideal monatomic gas, $\gamma=5/3$	$E=\tfrac12 U=-K<0$ , so $C=\mathrm{d}E/\mathrm{d}T<0$	The canonical star: bound, stable, negative heat capacity. Describes most non-degenerate interiors dominated by gas pressure.
Radiation-influenced, $\gamma\to4/3$	$E\to0$ , binding vanishes	As radiation pressure grows the gas softens toward $\gamma=4/3$ ; the total energy approaches zero and the star becomes only marginally bound and dynamically fragile. Very massive stars sit near this edge.
Degenerate electron gas	$P\not\propto\rho T$ , so $C>0$ effectively	Degeneracy pressure barely depends on temperature, decoupling energy loss from $T$ . The virial link breaks; the object cools at nearly fixed structure — the basis of white-dwarf cooling and of the helium-flash and Type Ia runaways.
Ultra-relativistic gas, $\gamma=4/3$ exactly	$2K+U=0$ with $K=3\!\int\! P\,\mathrm{d}V$ gives $E=0$	No net binding energy is stored; the configuration teeters on the edge of collapse — the physics behind the Chandrasekhar limit and pair-instability supernovae.

Common trapThe star has negative heat capacity, so surely a scoop of its gas does too.

It does not. Negative heat capacity is a property of the whole self-gravitating, virialized star, never of a local parcel of gas. Isolate a small blob and it has an ordinary positive heat capacity $C_V>0$ : heat it and it warms, exactly as in a laboratory. The sign flips only when you include the gravitational potential energy of the entire configuration and enforce hydrostatic equilibrium ( $2K=-U$ ), so the star must restructure itself in response to any energy change — and that restructuring is where gravity injects its "second joule." Two cautions ride along: the result assumes a non-degenerate ideal gas ( $P\propto\rho T$ ), which degeneracy breaks; and "the surface radiates and cools" is fully consistent with "the interior heats" — they happen together, which is the whole point.

Worked examples

1 · The Kelvin–Helmholtz timescale of the Sun (fully worked)

Before fusion was understood, the Sun's only conceivable power source was gravitational contraction. Estimate how long it could shine on that alone, using the virial energy budget, and say what the answer implied historically.

Step 1. With no nuclear source the luminosity is paid for by the total energy reservoir, and equation (4) says half the released gravitational energy is radiated: $L=-\dot E = -\tfrac12\dot U$ . The reservoir is $|E|=\tfrac12|U|=\tfrac12\alpha\,GM^2/R$ , so $t_{\mathrm{KH}} = |E|/L \approx \tfrac{\alpha}{2}\,GM^2/(RL)$ . With $\alpha$ of order unity, the order-of-magnitude estimate is $t_{\mathrm{KH}}\sim GM^2/(RL)$ .

Step 2. Insert solar values $M_\odot=1.99\times10^{30}$ kg, $R_\odot=6.96\times10^{8}$ m, $L_\odot=3.83\times10^{26}$ W, and $G=6.674\times10^{-11}$ N m² kg⁻². The numerator is $GM_\odot^2 = (6.674\times10^{-11})(1.99\times10^{30})^2 = 2.64\times10^{50}$ J·m. The denominator is $R_\odot L_\odot = (6.96\times10^{8})(3.83\times10^{26}) = 2.67\times10^{35}$ W·m.

Step 3. Divide: $t_{\mathrm{KH}} = 2.64\times10^{50}/2.67\times10^{35} = 9.9\times10^{14}$ s. With $1\ \mathrm{yr}=3.16\times10^{7}$ s this is $t_{\mathrm{KH},\odot}\approx3.1\times10^{7}$ years — only about thirty million years, far shorter than the geological age of the Earth already known in the nineteenth century. That mismatch ruled out gravitational contraction as the Sun's present power source and forced the search that ended at nuclear fusion. Contraction can power the Sun for only ~3×10⁷ yr, so today's Sun must be nuclear-powered.

2 · Virial temperature and the road to ignition (fully worked)

A collapsing cloud of one solar mass has no nuclear source yet. Show that as it contracts its mean temperature rises, and find the radius at which it reaches the ~10⁷ K needed to ignite hydrogen. Use $\mu=0.6$ and $\alpha\approx3/2$ ( $n=3$ polytrope).

Step 1. From equation (5), group the radius-independent factor $A \equiv \tfrac{\alpha}{3}\,\tfrac{\mu m_H}{k_B}\,GM$ . With $\alpha/3=0.5$ , $\mu m_H/k_B = (0.6)(1.673\times10^{-27})/(1.381\times10^{-23}) = 7.27\times10^{-5}$ K·s²·m⁻², and $GM=1.33\times10^{20}$ m³·s⁻², this gives $A = 4.83\times10^{15}$ K·m, so $\langle T\rangle = A/R$ .

Step 2. Evaluate at $R=R_\odot=6.96\times10^{8}$ m: $\langle T\rangle = A/R = 4.83\times10^{15}/6.96\times10^{8} = \mathbf{6.9\times10^{6}}$ K. About seven million kelvin is the right order for the Sun (its central value is higher, $\sim1.5\times10^{7}$ K, because the core is hotter than the mean).

Step 3. Solve for the radius at ignition, $\langle T\rangle = 1.0\times10^{7}$ K: $R = A/\langle T\rangle = 4.83\times10^{15}/10^{7} = 4.8\times10^{8}$ m $\approx 0.7\,R_\odot$ — the crossing marked in the second figure. By the virial law contraction does not merely accompany heating; it is the heating, and it drives the star inexorably to the fusion threshold. ⟨T⟩ ∝ 1/R, giving ~7×10⁶ K at R⊙ and ignition (~10⁷ K) near 0.7 R⊙.

3 · Why a fusion overshoot does not detonate the Sun

The core over-produces energy by a small fraction, momentarily raising the total energy of the virialized core by $\delta E>0$ . Show the temperature response and estimate the size of the star's effective heat capacity.

Step 1. By equation (3) the response is $\delta T = \delta E/C$ with $C<0$ , so $\delta E>0$ gives $\delta T<0$ : the core cools. The cooler gas burns more slowly ( $\varepsilon\propto T^\nu$ ), damping the overshoot — the thermostat in action.

Step 2. Put a number on $C$ : $C = -\tfrac32 M k_B/(\mu m_H) = -1.5\times(1.99\times10^{30})(1.381\times10^{-23})/[(0.6)(1.673\times10^{-27})] \approx$ $-4.1\times10^{34}$ J/K. The negative sign is the whole content: dumping energy into the star lowers its temperature, which keeps hydrogen burning steady for billions of years instead of running away. The core self-corrects because $C<0$ ; numerically $C\approx-4\times10^{34}$ J/K for the Sun.

Physics of the portrait

Negative heat capacity took two centuries and three disciplines to assemble: the thermodynamics of the virial theorem, the astrophysics of stellar energy sources, and the statistical mechanics of self-gravitating systems. Its portrait is not a face but the idea itself — the virial law $E=-K$ and the strange sign it forces on gravity's thermostat.

Origins · 1 minnegative heat capacity and how this was discovered

negative heat capacity

the virial law E = −K, so radiating energy raises the temperature · gravitational thermodynamics of stars and star clusters

Negative heat capacity was discovered not as a theorem but as a puzzle about the Sun. In the 1850s Hermann von Helmholtz and, independently, William Thomson (Lord Kelvin) proposed that the Sun shines by slowly contracting, converting gravitational energy into heat and light — and Kelvin used it to estimate a solar and terrestrial age of only tens of millions of years, on a famous collision course with Darwin's geology. The engine underneath was clarified by Rudolf Clausius, whose virial theorem of 1870 gave the relation $2K+U=0$ that pins a bound star's thermal energy to its gravity. Arthur Eddington, in his 1926 The Internal Constitution of the Stars, wove virial reasoning into the first coherent theory of stellar structure, making explicit that a star must heat as it contracts. The most startling face emerged in stellar dynamics: Vladimir Antonov showed in 1962 that a self-gravitating gas has no maximum-entropy state, and Donald Lynden-Bell and Roger Wood, in 1968, named the "gravothermal catastrophe," proving a cluster core has genuinely negative specific heat and must collapse as it loses energy. The same minus sign that dooms Kelvin's short timescale, stabilizes the main sequence, and detonates degenerate cores also collapses globular-cluster cores — one sign unifying stars and the systems of stars.

History

1854

— Helmholtz proposes that the Sun shines by gravitational contraction

1862

— Kelvin estimates the Sun's age at tens of millions of years from contraction

1870

— Clausius formulates the virial theorem, fixing 2K + U = 0

1926

— Eddington's Internal Constitution of the Stars builds stellar stability on virial reasoning

1968

— Lynden-Bell and Wood name the gravothermal catastrophe and its negative specific heat

Going further

Texts. Carroll & Ostlie, An Introduction to Modern Astrophysics, 2nd ed., §2.4 (virial theorem) and Ch. 10–13 (stellar structure and evolution) · Kippenhahn, Weigert & Weiss, Stellar Structure and Evolution, 2nd ed., Ch. 3 (virial theorem) and Ch. 33–34 (helium flash) · Prialnik, An Introduction to the Theory of Stellar Structure and Evolution, Ch. 2 · Binney & Tremaine, Galactic Dynamics, 2nd ed., §4.8 (virial theorem, gravothermal catastrophe) · Mo, van den Bosch & White, Galaxy Formation and Evolution, Ch. 5 (virial relations of self-gravitating systems) · Dodelson, Modern Cosmology (virial reasoning in structure formation).

Provenance. A verifier should: (1) rederive $2K+U=0$ by integrating hydrostatic equilibrium $\mathrm{d}P/\mathrm{d}r=-Gm\rho/r^2$ against $4\pi r^3$ (Carroll & Ostlie §2.4); (2) confirm $E=-K=\tfrac12U$ and $C=\mathrm{d}E/\mathrm{d}T=-\tfrac32 Mk_B/(\mu m_H)<0$ , equal to $-\tfrac12 C_V$ ; (3) check the polytrope result $E=-\dfrac{3\gamma-4}{3(\gamma-1)}|U|$ and its $\gamma=4/3$ limit (Kippenhahn/Weigert/Weiss Ch. 3); (4) recompute $t_{\mathrm{KH},\odot}=GM_\odot^2/(R_\odot L_\odot)\approx3\times10^{7}$ yr and the virial temperature $\langle T\rangle=A/R$ with $A\approx4.8\times10^{15}$ K·m using $M_\odot=1.99\times10^{30}$ kg, $R_\odot=6.96\times10^{8}$ m, $L_\odot=3.83\times10^{26}$ W, $\mu=0.6$ , $\alpha=3/2$ . Gravothermal-catastrophe statements follow Binney & Tremaine §4.8. Only standard solar constants and textbook structure constants are used.

Prerequisites. Hydrostatic Equilibrium in Stars · Equation of State for Stellar Matter · Equations of Stellar Structure. Leads to. Energy Generation Equation · Radiation Pressure in Stellar Interiors · Adiabatic Temperature Gradient.

Ask the tutor: Ask me why removing a single joule from the Sun makes it hotter, or to trace the same virial minus sign from a protostar's ignition to a globular cluster's core collapse.

How can a star "cool by radiating" and "heat up" at the same time?

Both are true at once. "Cool by radiating" is about the surface: the star sends photons into space and loses total energy, $\dot E<0$ . "Heat up" is about the interior temperature: because $E=-K$ , a falling $E$ means a rising $K$ and thus a rising $T$ . The reconciler is the gravitational reservoir — each unit radiated is matched by two units released as the star contracts, one replacing what was radiated and the second going into thermal motion. So the star loses total energy while its gas gets hotter: "energy" (total, including gravity) and "temperature" (thermal energy alone) move in opposite directions.

Why exactly one half — why is the split of released gravitational energy 50/50?

It is the number 2 in the virial theorem, seen from the energy side. Because $E=\tfrac12 U$ , we have $\dot E=\tfrac12\dot U$ , so the star radiates at $L=-\dot E=-\tfrac12\dot U$ while gravity releases $-\dot U$ in total — a radiated fraction of exactly $\tfrac12$ . The remaining half, $\dot K=-\dot E=L$ , heats the gas. Physically the factor two is fixed by the $1/r$ form of gravity: for a monatomic ideal gas the thermal energy is always half the binding-energy magnitude, so any change in binding splits equally between restoring the thermal energy and escaping as light.

Does negative heat capacity violate the second law of thermodynamics?

No. The usual theorem that $C>0$ assumes a system whose energy is the sum of its parts — short-range interactions. A self-gravitating star breaks that assumption: gravity is long-range, so you cannot cut the star into independent subsystems each carrying their share of energy. Antonov and later Lynden-Bell made this precise — a self-gravitating gas has no global maximum-entropy state, and the negative specific heat is the fingerprint of that fact. Entropy still increases overall; only the naive "heat capacity must be positive" theorem, which assumes short-range forces, fails to apply.

Why does this make the main sequence stable but degenerate cores explosive?

Both behaviours come from asking what happens after a small excess of fusion. In an ideal-gas core the extra energy raises $E$ , and negative heat capacity turns that into a drop in temperature; the fiercely $T$ -sensitive reactions slow down, and the disturbance dies away — a thermostat. In an electron-degenerate core the pressure barely depends on temperature, so the extra energy cannot expand and cool the gas; instead the temperature rises, which speeds up the reactions, which raises the temperature further — a runaway. The difference is entirely in the equation of state: $P\propto\rho T$ (ideal gas) keeps the virial thermostat working, while $P\approx P(\rho)$ (degenerate) disconnects it. The runaway stops only when heating finally lifts the degeneracy.

Is the star's heat capacity really the same size as the gas's, just negative?

Not the same size — it is exactly half, and negative. The ordinary heat capacity of the gas alone is $C_V=+\tfrac32 Mk_B/(\mu m_H)$ , but once gravity and hydrostatic equilibrium are enforced the star's effective capacity is $C=-\tfrac12 C_V$ . The factor $\tfrac12$ is again the "half and half" split: of every unit you remove, gravity supplies two, so the net thermal change is half a unit and goes the wrong way. The magnitude, not just the sign, carries the imprint of gravity paying out double.

Does negative heat capacity apply beyond individual stars?

Yes — it is a property of any system bound by long-range $1/r$ attraction, not of stars specifically. A globular cluster, treated as a gas of stars, obeys the same $2K+U=0$ with $K$ now the kinetic energy of stellar motions and its "temperature" the velocity dispersion. Such a cluster has negative heat capacity, so a core losing energy (via two-body relaxation) contracts and grows hotter — its stars move faster — which drives still more energy outward. This gravothermal catastrophe collapses the core until binary stars halt it, and the same reasoning is now applied to self-interacting dark-matter halos.