A star held up by a rule about counting

An ordinary star holds itself up by being hot. Fusion supplies energy, the energy maintains a temperature, the temperature maintains a pressure, and the pressure balances the weight of everything above. Stop the fusion and the whole arrangement fails: the star cools, the pressure drops, and it collapses.

A white dwarf has stopped. There is no fusion anywhere in it, and it has been cooling for as long as it has existed. It does not collapse.

What holds it up is not a thermodynamic pressure at all. It is a consequence of a counting rule — the Pauli exclusion principle, which forbids two electrons from occupying the same quantum state — and the pressure it produces does not care about temperature. A white dwarf at ten million kelvin and the same white dwarf cooled to a thousand are the same size.

Where the pressure comes from #

Compress a gas of electrons and the exclusion principle starts to bite.

Each electron occupies a cell in phase space of volume roughly $h^3$, and only two can share it — one for each spin. Squeeze the gas into a smaller volume and the available position-space shrinks, so to keep the phase-space cells distinct the momenta must spread out further. The electrons are forced to move faster, not because anything heated them but because there is nowhere slower left to be.

Fast-moving particles exert pressure. That is degeneracy pressure, and its two defining properties both follow from the derivation.

It does not depend on temperature. The momenta are set by the density and by the exclusion principle. Cooling the gas removes the thermal contribution and leaves the degenerate one untouched, which is why a white dwarf can cool indefinitely without shrinking.

It grows steeply with density. For a non-relativistic degenerate gas, $P \propto \rho^{5/3}$, which is stiffer than an ideal gas at fixed temperature.

Sirius B is the standard example and the numbers are worth having: about one solar mass inside a body the size of the Earth, which is a mean density around $10^{9}$ kg/m³. A teaspoon of it would weigh several tonnes. Under those conditions the electrons are moving at a substantial fraction of the speed of light with no temperature required.

The relation that runs backwards #

The mass–radius relation for a degenerate star has a sign that no ordinary object shares.

Balancing the degeneracy pressure against gravity for a non-relativistic gas gives

$$R \propto M^{-1/3}.$$

More mass, smaller star. Doubling the mass shrinks the radius by 21%, which is the reverse of the mass–radius behaviour of an ordinary star. Nothing in ordinary experience behaves like this — a bigger rock is a bigger rock — and the inversion is a direct consequence of the exponent in the equation of state. Gravity’s demand for pressure grows faster with mass than the degenerate gas’s supply does, so the star must contract to make up the difference.

That relation makes a white dwarf’s mass measurable from its radius, or the other way round, with no model of its interior beyond the equation of state. It also makes the diagnosis of degeneracy possible from a distance: any object whose radius falls as it gains mass is degenerate, whatever else is true of it. That inference — made from position on the diagram alone, before anyone knew what degeneracy was — is what forced the question. Eddington remarked of Sirius B that the message it was sending seemed absurd: a ton of material could be put in a matchbox. The absurdity was resolved by Fowler in 1926, who applied the newly formulated Fermi–Dirac statistics and showed that a degenerate electron gas provides exactly the pressure required.

The limit, and where it comes from #

Fowler’s solution has a flaw, and finding it made a nineteen-year-old’s reputation and cost him twenty years of argument.

As the mass rises the star shrinks, the density rises, and the electrons are forced to ever higher momenta. Eventually they approach the speed of light, and a relativistic gas is softer than a non-relativistic one: the pressure goes as $\rho^{4/3}$ rather than $\rho^{5/3}$.

That change of exponent is the whole story, because $4/3$ is exactly the exponent at which the pressure and the weight scale the same way with radius. Below it the star can always find an equilibrium; at it, the two balance at every radius and nothing decides the size; above the corresponding mass, gravity wins at every radius and there is no equilibrium at all.

The mass at which this happens is

$$M_{\text{Ch}} \approx \frac{5.836}{\mu_e^2}\,M_\odot,$$

which for a carbon–oxygen composition, $\mu_e = 2$, gives 1.46 solar masses. Above it, no degenerate configuration exists.

Subrahmanyan Chandrasekhar worked this out in 1930, aged nineteen, on the voyage from India to England. Eddington — by then the most influential astrophysicist alive, and the man who had posed the puzzle — rejected it publicly and repeatedly, arguing that some unknown physics must intervene to prevent so absurd a conclusion as unlimited collapse. The dispute damaged Chandrasekhar’s position for years and he moved to the United States. He received the Nobel Prize for the work in 1983, fifty-three years later.

What was actually measured #

The limit is a theoretical result and the observational support for it comes from three independent directions, which is worth setting out because a limit that nothing exceeds is a weak claim on its own.

The mass distribution. Several thousand white dwarfs have measured masses, from spectroscopic fits to their pressure-broadened hydrogen lines combined with the mass–radius relation. The distribution peaks sharply at 0.6 solar masses and has a hard upper edge: essentially nothing is found above about 1.35, and the handful of candidates above that are suspected mergers. The edge is where it is predicted to be.

Gravitational redshift. A photon climbing out of a white dwarf’s gravitational well loses energy, so its lines are shifted redward by an amount proportional to $M/R$. For Sirius B the shift is equivalent to 80 km/s — measured, and one of the earliest confirmations of general relativity. Combined with a radius from the star’s flux and distance, it gives a mass with no equation of state assumed at all: 1.02 solar masses, agreeing with the dynamical mass from its orbit around Sirius A.

Type Ia supernovae. A white dwarf accreting from a companion approaches the limit, and at about 1.38 solar masses carbon ignites in the degenerate interior. Because degenerate material does not expand when heated, there is no thermostat: the burning runs away, and the star is destroyed entirely in a few seconds.

That last case is where the limit becomes a measuring instrument. Every such explosion happens at nearly the same mass, so it releases nearly the same energy, so it reaches nearly the same peak luminosity — a standard candle, and the one that reaches furthest. The residual spread correlates with how fast the light curve fades, and correcting for it turns a rough candle into a good one.

The measurement that used it is the one that changed cosmology. Plotting the apparent brightness of distant type Ia supernovae against their redshift, two teams found in 1998 that the distant ones were fainter than a decelerating universe predicted — further away than expected — which means the expansion is accelerating — measured, as everything above the parallax ceiling is, through the distance modulus. That conclusion rests on the Chandrasekhar limit being the same everywhere and at every epoch, which is a claim about atomic physics being universal, and it is the assumption most often examined when the result is questioned.

What happens above the limit #

Exceeding the limit does not produce unlimited collapse, which is what Eddington could not accept and what turned out to be true in a modified form.

The collapse continues until a different degeneracy takes over. At densities around $10^{17}$ kg/m³ — a hundred million times a white dwarf’s — electrons and protons combine into neutrons, and the neutrons themselves become degenerate. A neutron star is held up by neutron degeneracy pressure, and it is about 12 kilometres across for a mass of 1.4 solar masses.

The same argument then repeats. Neutron degeneracy also has a maximum supportable mass — the Tolman–Oppenheimer–Volkoff limit — and above it nothing known stops the collapse. That limit is not calculable from first principles, because it depends on how nuclear matter behaves at densities beyond anything reproducible, and its value is one of the outstanding questions in physics. Observations bracket it: the heaviest well-measured neutron star is about 2.1 solar masses, and gravitational-wave detections of merging compact objects fill in the region above.

The general shape is a ladder of exclusions. Each one holds until the particles supplying it turn relativistic; then the equation of state softens to $\rho^{4/3}$, a maximum mass appears, and the next stage down takes over. Beyond the last rung there is nothing, and general relativity says the result is a black hole.

Cooling, which is the only thing left to do #

A white dwarf’s whole subsequent history is one process, and its simplicity makes it a clock.

There is no energy source. What the star radiates is the thermal energy stored in its ions — the electrons are degenerate and contribute almost nothing to the heat capacity, exactly as they contribute almost nothing to a metal’s. So the object is a hot lump of known mass and known radius, losing heat to space, and the rate follows from ordinary thermodynamics with no stellar structure in it beyond the equation of state.

That produces an age with a shorter chain of assumptions than any other stellar clock. The faintest white dwarfs in a population are the oldest, and the cut-off at the faint end of the white-dwarf sequence dates the population directly. Applied to the galactic disc it gives about 9 billion years; applied to globular clusters, 12 to 13 — agreeing with the turn-off ages obtained from completely different physics.

The cooling is not quite featureless, and the feature is the recent surprise. As the interior cools past a threshold, the carbon and oxygen ions crystallise — a genuine phase transition, releasing latent heat — and the star pauses in its cooling while it does. The pause makes a pile-up in the number of white dwarfs at a particular luminosity, and Gaia’s catalogue was large enough to show it as a distinct branch in the colour–magnitude diagram in 2019. A phase transition in the interior of a dead star was detected as a bump in the density of points on a scatter plot.

The generalisation: degeneracy is ordinary matter #

The mechanism sounds exotic and is not. Degeneracy pressure is what makes solid matter solid, and the astronomical cases differ only in degree.

The electrons in a metal are a degenerate gas. Their Fermi energy is a few electronvolts, corresponding to a temperature of tens of thousands of kelvin, so at room temperature they are almost entirely degenerate — the thermal energy is a fiftieth of the Fermi energy. That is why a metal’s electronic heat capacity is far smaller than a classical calculation predicts, a discrepancy that puzzled physics for decades and was resolved by exactly the statistics Fowler applied to Sirius B.

It is also, ultimately, why a table holds up a cup. The resistance of matter to compression is not electrostatic repulsion — atoms are neutral — but the cost of forcing electrons into states already occupied. A white dwarf is the same effect with gravity supplying the compression instead of a hand.

The scale of the difference is the interesting part. In a metal, a cubic centimetre contains $10^{23}$ electrons and the Fermi energy is a few eV. In a white dwarf, the same volume contains $10^{30}$ and the Fermi energy is around half an MeV — comparable to the electron’s own rest mass, which is precisely the condition for the relativistic softening. The Chandrasekhar limit exists because a white dwarf is dense enough to push a piece of solid-state physics into the relativistic regime.

Two ways to weigh one, and their disagreement #

The mass distribution quoted above rests on a spectroscopic method, and there is a second method, and for two decades they disagreed in a way worth recording.

The spectroscopic route fits the profiles of the hydrogen absorption lines. In an atmosphere at a surface gravity of $10^8$ centimetres per second squared — a hundred thousand times the Sun’s — the lines are enormously pressure-broadened, and their width is a direct measurement of that gravity. Combining a gravity with the mass–radius relation gives a mass.

The photometric route uses a parallax. A distance and an apparent brightness give a luminosity; a luminosity and a temperature give a radius through the Stefan–Boltzmann relation; and a radius gives a mass, again through the mass–radius relation. No spectrum is needed beyond a colour.

The two agreed for hot white dwarfs and diverged below about twelve thousand kelvin, where the spectroscopic masses came out systematically higher — by up to a fifth of a solar mass, which is enormous for a distribution whose whole width is a tenth.

The cause was traced to the model atmospheres rather than to the stars. Below that temperature the atmosphere becomes convective, and the one-dimensional mixing-length treatment used to compute the line profiles misrepresents the temperature structure. Three-dimensional convection calculations, when they became available, removed most of the discrepancy.

A systematic that appears below a threshold temperature and not above it is a statement about a physical transition, and in this case the transition was in the model rather than in the star.

And the same counting rule applied where the particles are neutrons rather than electrons, at a range of stiffnesses.

Where the model stops #

Zero temperature. The equation of state used here assumes complete degeneracy. Young, hot white dwarfs have a small thermal contribution and are slightly larger than the relation predicts.

No rotation, no magnetic field. A rapidly rotating white dwarf is supported partly by its own spin and can exceed the limit; some of the over-massive candidates are probably these, and some are merger products.

Uniform composition. $\mu_e$ enters squared, so a helium white dwarf ($\mu_e = 2$) and an iron one ($\mu_e = 2.15$) have limits differing by 14%.

Newtonian gravity. General relativity makes the star slightly harder to support, which lowers the limit a little and — more importantly — means that near the maximum the configuration becomes dynamically unstable somewhat before the Newtonian limit is reached.

The figure has a limitation worth being explicit about. It plots a sequence of equilibria, not a history: each point is a possible white dwarf, and a real star does not travel along the curve. A white dwarf accreting mass does move along it, which is the one case where the curve is also a path — and it is the case that ends in a supernova, at the right-hand edge where the curve meets the axis. A figure that is a locus everywhere except at the one place it matters is an awkward object, and the caption cannot fix it.

One last observation about what the limit is not. It is not a maximum mass for a star — main-sequence stars reach a hundred solar masses and more. It is a maximum for a configuration supported by degenerate electrons, and nothing above it is forbidden from existing; what is forbidden is existing in that state. The distinction matters because the limit is often quoted as though mass above 1.4 solar masses were impossible somewhere, when what it rules out is a particular way of holding that mass up.

The relation depends on the composition through one number, and it is worth drawing at the two extremes that number can take.

The ladder from here #

Later rungs on this anchor: the degenerate equation of state derived, in both regimes. The mass–radius relation from hydrostatic balance. The Chandrasekhar limit, computed. The relativistic softening and why $4/3$ is the critical exponent. White-dwarf cooling, and the age it gives. Crystallisation in the interior, and the split sequence Gaia found. Type Ia supernovae, their progenitors, and the light-curve corrections. Neutron stars and the TOV limit. The nuclear equation of state, and what gravitational-wave observations constrain about it. And the Fermi gas in a metal, where the same statistics were being worked out at the same time for entirely different reasons.

Chandrasekhar’s calculation takes about two pages. Eddington’s objection was that a star with no equilibrium available “should go on radiating and radiating and contracting and contracting until […] it gets down to a few kilometres radius, when gravity becomes strong enough to hold in the radiation, and the star can at last find peace” — which he offered as a reductio ad absurdum, and which is an accurate description of a black hole.