Appearance
12.2 — How Stars Die
The thermostat of Chapter 12.1 works only while there is hydrogen in the core. When it runs out, gravity has nothing left to push against, and what happens next is decided almost entirely by one number: the star's mass.
The ladder
| Initial mass | End state | Mechanism |
|---|---|---|
| <0.08M_\odot | Brown dwarf | Never ignites |
| 0.08–0.5M_\odot | Helium white dwarf | Never ignites helium |
| 0.5–8M_\odot | Carbon–oxygen white dwarf | Planetary nebula |
| 8–20M_\odot | Neutron star | Core collapse supernova |
| 20–40M_\odot | Black hole | Core collapse, possibly failed |
| >40M_\odot | Black hole | Direct collapse |
| 130–250M_\odot | Nothing | Pair instability |
The last row is the strangest and is covered below.
Red giants
The core runs out of hydrogen. Fusion stops there, pressure support fails, and the core contracts.
Contraction heats it — gravitational potential energy becoming thermal energy, the Kelvin–Helmholtz mechanism.
And that heats a shell around the core to fusion temperature. Hydrogen now burns in a thin shell surrounding an inert helium core.
The shell burns furiously, because it is hotter than the core ever was, and the star's luminosity rises by a factor of a thousand.
The envelope expands enormously. The reason is subtle and worth stating: the shell source is so concentrated that the envelope above it must expand to carry the flux, and the mirror principle — core contracts, envelope expands — is a robust result of stellar structure calculations.
The surface cools as it expands, so the star moves right and up on the HR diagram, becoming a red giant.
The Sun's future: in about 5 billion years it will expand to roughly 200 solar radii — engulfing Mercury and Venus and probably Earth — and shine at about 2000 times its present luminosity.
The helium flash
The inert helium core keeps contracting until it becomes degenerate (Chapter 7.7): the electrons are packed so tightly that exclusion pressure, not thermal pressure, is holding it up.
And degenerate matter has no thermostat.
Chapter 12.1's stability argument required pressure to depend on temperature. In a degenerate gas:
P \propto \rho^{5/3}, \qquad \text{independent of } T
So when helium ignites at 10^{8} K, the released energy raises the temperature and the core does not expand. Higher temperature means a faster reaction, which means more energy, which means higher temperature.
\boxed{\text{Runaway.}}
The helium flash releases about 10^{11} solar luminosities for a few seconds — comparable to an entire galaxy.
And nothing is visible from outside. All of it goes into lifting the degeneracy, and once the core becomes non-degenerate it expands, cools, and settles into stable helium burning. The surface does not notice for thousands of years.
Only stars below about 2M_\odot have this. Heavier ones ignite helium before the core degenerates, so it is a smooth transition.
The triple alpha process
Helium fusion has a problem. Beryllium-8, the obvious product of two alpha particles, is unbound — it falls apart in 10^{-16} seconds.
^4\text{He}+^4\text{He} \rightleftharpoons ^8\text{Be}
So the route to carbon requires a third alpha to arrive during that 10^{-16} s window:
^8\text{Be}+^4\text{He} \to ^{12}\text{C}+\gamma
Effectively a three-body reaction, which is why it needs densities of 10^{8} kg/m³ and temperatures of 10^{8} K.
And it should still be far too slow.
Fred Hoyle's prediction, 1953. Carbon exists in abundance, so the reaction must be much faster than the naive rate. Hoyle argued that carbon-12 must have an excited state at almost exactly the combined energy of $^8$Be plus $^4$He, providing a resonance that enormously enhances the rate.
He predicted the energy: about 7.65 MeV.
Willie Fowler's group at Caltech looked, reluctantly, and found it at 7.654 MeV.
This is the most famous anthropic prediction in physics — Hoyle reasoned from the existence of carbon (and of himself) to a specific nuclear energy level, and was right. It is also often overstated: the level's position is not as finely tuned as sometimes claimed, and shifting it by a few percent would change carbon abundance without eliminating it.
What is not overstated is that all the carbon in the universe passes through this one resonance.
Shell burning and the onion
A massive star repeats the cycle. Core exhausts, contracts, heats, ignites the next fuel, while the previous fuel burns in a shell.
The result is an onion structure, with successively heavier elements in successively deeper shells.
| Fuel | Temperature | Duration for 25M_\odot |
|---|---|---|
| Hydrogen | 4\times10^{7} K | 7 million years |
| Helium | 2\times10^{8} K | 700,000 years |
| Carbon | 8\times10^{8} K | 600 years |
| Neon | 1.6\times10^{9} K | 1 year |
| Oxygen | 1.8\times10^{9} K | 6 months |
| Silicon | 2.5\times10^{9} K | 1 day |
Look at the timescale collapse. Seven million years for hydrogen, one day for silicon.
Two reasons. Each stage releases less energy per unit mass (the binding curve flattens towards iron, Chapter 9.2), and above about 5\times10^{8} K neutrino losses dominate. Pair production of neutrinos in the hot core carries energy away directly, since neutrinos escape immediately, so the core must burn faster and faster just to hold itself up.
And silicon burning makes iron.
\boxed{\text{Iron is the end. There is no energy left to extract.}}
Chapter 9.2's binding energy curve peaks at iron-56 and nickel-62. Fusing iron absorbs energy.
Core collapse
The iron core grows to about 1.4M_\odot — the Chandrasekhar limit — and electron degeneracy can no longer hold it.
What happens next takes less than a second.
Photodisintegration. At 10^{10} K the gamma rays are energetic enough to break iron nuclei apart:
\gamma+^{56}\text{Fe} \to 13\,^4\text{He}+4n
This absorbs 124 MeV per nucleus — undoing in an instant the energy released over millions of years — and removes the radiation pressure supporting the core.
Neutronisation. Electrons are forced into protons:
p+e^- \to n+\nu_e
This removes the degenerate electrons that were providing the pressure, and it floods the core with neutrinos.
Collapse. The core falls inward at up to 0.25c, going from Earth-sized to 20 km across in about one second.
Bounce. At nuclear density, 2.3\times10^{17} kg/m³ (Chapter 9.1), the strong force becomes strongly repulsive and the collapse halts abruptly. The infalling material rebounds, launching a shock wave outwards.
And the shock stalls. It runs into the still-infalling outer core and loses energy to photodisintegration. Simulations for decades could not make the star explode.
The neutrinos are what revive it. About 10^{58} neutrinos are released, carrying 3\times10^{46} J — 99 % of the total energy of the event. Even though they interact barely at all, depositing 1 % of that behind the stalled shock is enough to restart it.
Modern three-dimensional simulations including convection, turbulence and the standing accretion shock instability now produce explosions reliably, and this was a genuinely open problem until the 2010s.
The energy budget
| Channel | Energy | Fraction |
|---|---|---|
| Neutrinos | 3\times10^{46} J | 99 % |
| Kinetic energy of ejecta | 10^{44} J | 1 % |
| Light | 10^{42} J | 0.01 % |
The visible supernova — briefly outshining an entire galaxy — is one part in ten thousand of the energy released.
And SN 1987A confirmed it (Chapter 8.5): 24 neutrinos detected three hours before the light arrived, with a total inferred energy matching the gravitational binding energy of a neutron star.
Nucleosynthesis
Where the elements come from, and the answer differs by element.
Big Bang (Chapter 12.6): hydrogen, helium, a little lithium.
Stellar fusion: everything up to iron.
Beyond iron requires energy input, so it happens only in extreme environments.
The s-process (slow neutron capture). In red giants, neutrons are captured one at a time, with beta decay between captures. Builds up to bismuth, and produces about half the elements heavier than iron — strontium, barium, lead.
The r-process (rapid neutron capture). A burst of neutrons so intense that a nucleus captures many before it can decay. Makes gold, platinum, uranium, thorium.
Where does the r-process happen? This was uncertain for fifty years, and GW170817 settled it (Chapter 6.10). The neutron star merger's optical counterpart faded exactly as predicted for radioactive decay of freshly made heavy elements, and about ten Earth masses of gold were made in that single event.
\boxed{\text{The gold in a wedding ring was made when two neutron stars collided.}}
And the carbon, oxygen, nitrogen and calcium in your body were made inside stars that died before the Sun formed.
Carl Sagan's phrase — "we are made of star stuff" — is literally correct and quantitatively checkable. By mass, a human is about 65 % oxygen, 18 % carbon and 10 % hydrogen. The hydrogen is primordial, from the Big Bang. Everything else was made in a star.
White dwarfs and the Chandrasekhar limit
A star below about 8M_\odot never reaches carbon ignition. It sheds its envelope as a planetary nebula — a name coined by Herschel because they looked like planetary discs in early telescopes, and nothing to do with planets — and leaves a hot carbon–oxygen core.
That core is a white dwarf, supported entirely by electron degeneracy pressure (Chapter 7.7).
Deriving the limit
Chapter 7.7 gave the Fermi energy for n electrons per unit volume:
E_F = \frac{\hbar^2}{2m_e}(3\pi^2n)^{2/3}
Non-relativistic degeneracy pressure:
P = \frac{2}{5}nE_F \propto n^{5/3} \propto \rho^{5/3}
Hydrostatic equilibrium (Chapter 12.1) requires the central pressure to scale as:
P_c \sim \frac{GM^2}{R^4}
And \rho \sim M/R^3, so:
\frac{GM^2}{R^4} \sim \left(\frac{M}{R^3}\right)^{5/3} = \frac{M^{5/3}}{R^5}
R \sim M^{-1/3}
\boxed{\text{More massive white dwarfs are } smaller.}
A counterintuitive result and it is correct, and it is what makes the limit exist.
Now push to high mass. As R shrinks, n rises, and the Fermi energy rises. When E_F approaches m_ec^2 = 0.511 MeV, the electrons become relativistic, and then:
E_F \approx \hbar c(3\pi^2n)^{1/3} \quad\Longrightarrow\quad P \propto n^{4/3} \propto \rho^{4/3}
Repeat the balance:
\frac{GM^2}{R^4} \sim \frac{M^{4/3}}{R^4}
The R cancels completely.
GM^2 \sim M^{4/3} \quad\Longrightarrow\quad M^{2/3} \sim \frac{1}{G} \quad\Longrightarrow\quad M = \text{constant}
There is exactly one mass at which relativistic degeneracy pressure balances gravity, and it depends on nothing but fundamental constants:
M_{\text{Ch}} \approx \frac{1}{\mu_e^2}\left(\frac{\hbar c}{G}\right)^{3/2}\frac{1}{m_p^2}
Evaluate it. With \mu_e = 2 for carbon and oxygen (two nucleons per electron):
\left(\frac{\hbar c}{G}\right)^{1/2} = \left(\frac{(1.055\times10^{-34})(3\times10^{8})}{6.674\times10^{-11}}\right)^{1/2} = (4.742\times10^{-16})^{1/2} = 2.178\times10^{-8}\ \text{kg}
That is the Planck mass. Cubing and dividing by m_p^2:
\frac{(2.178\times10^{-8})^3}{(1.673\times10^{-27})^2} = \frac{1.033\times10^{-23}}{2.799\times10^{-54}} = 3.69\times10^{30}\ \text{kg}
M_{\text{Ch}} \approx \frac{3.69\times10^{30}}{4}\times(\text{numerical factor} \approx 3.1) = 2.86\times10^{30}\ \text{kg} = 1.44\,M_\odot
\boxed{M_{\text{Ch}} = 1.44\,M_\odot}
Chandrasekhar derived this in 1930, aged 19, on the boat from Madras to Cambridge.
Eddington, then the most influential astrophysicist alive, publicly ridiculed it at a Royal Astronomical Society meeting in 1935, saying there ought to be a law of nature to prevent a star behaving in this absurd way. He was wrong, the humiliation drove Chandrasekhar to leave stellar structure for decades, and he received the Nobel Prize in 1983.
Note what the derivation contains: \hbar, c, G and m_p. A maximum stellar mass built from quantum mechanics, relativity, gravity and nuclear physics — one of the few places all four meet in a single number.
Type Ia supernovae
A white dwarf accreting from a companion approaches the Chandrasekhar limit.
At about 1.38M_\odot the core ignites carbon. And because the matter is degenerate, there is no thermostat (Chapter 12.1).
Runaway. The entire star fuses in about one second, releasing 10^{44} J and completely destroying itself. No remnant is left.
Because the trigger mass is always the same, the peak luminosity is nearly the same — about 10^{10}L_\odot.
\boxed{\text{A standard candle, and the third rung of the distance ladder (Chapter 11.9).}}
The Phillips relation reduces the residual scatter: brighter supernovae decline more slowly, and correcting for the decline rate gives distances good to about 6 %.
These are what found dark energy (Chapter 12.7).
And the progenitor question is not fully settled. The single-degenerate scenario has a white dwarf accreting from a normal star; the double-degenerate has two white dwarfs merging. Both probably occur, and which dominates matters for whether the standard-candle calibration drifts with cosmic time — which is a live concern for cosmology.
Neutron stars
Above the Chandrasekhar limit, electron degeneracy fails and the collapse continues until neutron degeneracy takes over at nuclear density.
The numbers are extreme:
| Property | Value |
|---|---|
| Mass | 1.4–2.2 M_\odot |
| Radius | 10–13 km |
| Density | 4\times10^{17} kg/m³ |
| Surface gravity | 2\times10^{12} m/s² |
| Escape velocity | 0.5c |
| Magnetic field | 10^{8}–10^{11} T |
| Surface temperature | 10^{6} K |
A teaspoon weighs a billion tonnes. The whole Sun's mass in a sphere the size of a city.
Surface gravity is 2\times10^{11}g. Dropping something from 1 metre would deliver it at 1400 km/s.
And the structure is layered: an iron crust, then a region where nuclei are squeezed into rods and sheets — nuclear pasta, genuinely the technical term — then a superfluid neutron interior, and a core whose composition is unknown and may be quark matter.
Pulsars
Jocelyn Bell Burnell found the first in 1967 as a regular pulse every 1.337 seconds in radio data. It was labelled LGM-1 — Little Green Men — until a second was found in a different part of the sky.
The mechanism: the neutron star's magnetic axis is tilted from its rotation axis, and beamed radiation sweeps past like a lighthouse.
Why they spin so fast. Angular momentum conservation (Chapter 1.8). A star of radius 7\times10^{8} m rotating once a month collapses to 10 km:
\frac{\omega_f}{\omega_i} = \left(\frac{R_i}{R_f}\right)^2 = \left(\frac{7\times10^{8}}{10^{4}}\right)^2 = (7\times10^{4})^2 = 4.9\times10^{9}
A one-month period becomes 2.6\times10^{6}/4.9\times10^{9} = 5\times10^{-4} s. Half a millisecond, which is why the fastest known pulsars spin at 716 times per second.
And the magnetic field intensifies the same way. Magnetic flux is conserved, so B \propto 1/R^2, giving the same factor of 5\times10^{9}.
Millisecond pulsars are the most precise clocks known, stable to one part in 10^{15} — comparable to atomic clocks, and Chapter 6.10 described how timing arrays of them detect gravitational waves.
The Hulse–Taylor binary pulsar provided the first proof that gravitational waves exist (Chapter 6.10).
The maximum neutron star mass is uncertain because the equation of state of matter above nuclear density is unknown. Observations constrain it to about 2.2M_\odot, and GW170817's merger product provided one of the tightest limits.
And Chapter 6.8 explained the vicious mechanism: in general relativity pressure gravitates, so making the matter stiffer eventually helps gravity rather than resisting it. There is no equation of state that can support a star above about 3M_\odot, whatever the nuclear physics.
Pair instability
The strangest entry in the mass table.
In a star of 130 to 250 solar masses, the core reaches 3\times10^{9} K while still burning oxygen. At that temperature the gamma rays are energetic enough to produce electron–positron pairs:
\gamma+\gamma \to e^++e^-
This converts radiation into mass, removing the pressure that was supporting the star.
The core collapses, which raises the temperature, which triggers explosive oxygen burning throughout the entire core at once.
The star is completely destroyed. No neutron star, no black hole, nothing at all — the entire mass is dispersed, including up to 50 solar masses of nickel-56, making these the brightest supernovae known.
\boxed{\text{Stars between 130 and 250 solar masses leave no remnant of any kind.}}
This produces the pair-instability mass gap, an expected absence of black holes between about 50 and 130 solar masses. GW190521, at 142 solar masses, landed squarely in it (Chapter 6.10) — which is why it was such a surprise, and why the favoured explanation is that it was itself the merger of two earlier merger products.
Where this shows up in your life
Every atom in you heavier than helium was made in a star, and the heaviest ones in a neutron star merger.
Iodine in your thyroid, iron in your blood, calcium in your bones, phosphorus in your DNA — all stellar nucleosynthesis.
Type Ia supernovae measure the universe and discovered dark energy.
Pulsars provide clocks, test general relativity, and were used to encode the Earth's position on the Pioneer plaque.
And Betelgeuse will explode, sometime in the next 100,000 years, from 550 light years away. It will be as bright as the full Moon for months, visible in daylight, and entirely harmless.
What the next chapter fixes
Above about 20 solar masses, neither electron nor neutron degeneracy can stop the collapse, and nothing known can. Chapter 12.3 follows what happens then: the Schwarzschild and Kerr solutions applied to real objects, the photon sphere and the shadow, what tidal forces do to an infalling observer, how accretion makes quasars the brightest steady objects in the universe, Hawking radiation derived at the level of its actual argument, the information paradox, and the two images that were finally taken.