Hold out your hand. The calcium stiffening the bones inside it and the iron ferrying oxygen through the blood beneath the skin were not made on Earth, and they were not made in the Big Bang. They were forged in the crushing cores of stars that lived and died before the Sun was even born, blown out across the galaxy when those stars exploded, and eventually swept up into the cloud that became our solar system. Carl Sagan's line is literally, atomically true: we are made of star-stuff.

The Big Bang was a poor blacksmith. In its first few minutes it fused hydrogen into a large dose of {}^{4}\text{He} (about a quarter of the mass) and a whisker of deuterium, {}^{3}\text{He} and {}^{7}\text{Li} — and then the universe thinned and cooled too fast to make anything more. There is no stable nucleus of mass 5 or 8 to act as a stepping stone, so the chain stalled at helium. Essentially every atom heavier than lithium — the carbon in your cells, the oxygen you breathe, the silicon in glass, the gold in a ring — was built afterwards, one fusion reaction at a time, in stars. This page is about how that furnace works: the proton–proton chain, the CNO cycle, and the long climb up the periodic table that ends, abruptly and fatally, at iron.

The one curve that tells the whole story

Every question about which reactions release energy and which cost it is answered by a single graph: the binding energy per nucleon. Pull a nucleus apart into its separate protons and neutrons and you must supply energy; that energy is the nucleus's total binding energy B. Divide by the number of nucleons A and you get B/A, a measure of how tightly — how economically — each nucleon is held. A larger B/A means a lower-energy, more stable, more deeply bound nucleus.

Read the shape. The curve rises steeply from hydrogen, jumps at the unusually stable {}^{4}\text{He}, and keeps climbing through carbon, oxygen and silicon to a broad maximum at the iron group — {}^{56}\text{Fe} sits near 8.8\ \text{MeV} per nucleon (its neighbour {}^{62}\text{Ni} is a hair higher, but iron-56 is the famous landmark). After the peak it declines gently all the way to uranium.

Now the punchline. When two light nuclei fuse, the product sits higher on the curve than the reactants — its nucleons are more tightly bound — and the surplus binding energy is released, as the kinetic energy and radiation that make a star shine. This works all the way up the left-hand slope: fusion of nuclei lighter than iron releases energy. But once you reach the peak, there is nowhere higher to go. Fusing iron into something heavier would move downhill on the curve, meaning it would absorb energy rather than release it. Iron is the ash at the bottom of the fire: the point where a star's fusion furnace can extract no more.

Step one: the proton–proton chain

In a star like the Sun, the core is a plasma at roughly 1.5\times 10^{7}\ \text{K} — hot enough that bare protons occasionally collide hard enough to fuse, despite their mutual electric repulsion (they cheat past the Coulomb barrier by quantum tunnelling). The dominant route in low-mass stars is the proton–proton (pp) chain, which welds four protons into one helium-4 nucleus in three moves:

1. Two protons fuse and one immediately turns into a neutron (via the weak force), making deuterium: {}^{1}\text{H} + {}^{1}\text{H} \rightarrow {}^{2}\text{H} + e^{+} + \nu_{e}. This step is agonisingly slow — it needs the weak interaction — and it is the bottleneck that sets how long the Sun lives.
2. The deuterium grabs another proton to make helium-3: {}^{2}\text{H} + {}^{1}\text{H} \rightarrow {}^{3}\text{He} + \gamma.
3. Two helium-3 nuclei meet and fuse, spitting two protons back out: {}^{3}\text{He} + {}^{3}\text{He} \rightarrow {}^{4}\text{He} + 2\,{}^{1}\text{H}.

Run steps 1 and 2 twice to feed one step 3, add up everything, and the two positrons annihilate with electrons — the net effect is beautifully simple:

Step one, again: the CNO cycle

Turn up the heat and Nature finds a faster way to do exactly the same job. In stars more massive and hotter than the Sun, the four-protons-into-helium conversion runs mostly through the CNO cycle, in which pre-existing carbon, nitrogen and oxygen act as catalysts. A carbon-12 nucleus captures protons one at a time, marching through {}^{13}\text{N}, {}^{13}\text{C}, {}^{14}\text{N}, {}^{15}\text{O}, {}^{15}\text{N} (with a couple of \beta^{+} decays along the way), and the last step ejects a helium-4 and hands the carbon-12 back, unchanged and ready to go again:

The carbon is a catalyst, not a fuel — it is regenerated every loop, so a tiny seed of C, N and O can process an enormous amount of hydrogen. The net reaction and the net energy release are identical to the pp chain (4\,{}^{1}\text{H} \rightarrow {}^{4}\text{He}, \approx 26.7\ \text{MeV}). What differs wildly is the temperature sensitivity. Because the CNO reactions must overcome the much larger Coulomb barriers of carbon and nitrogen nuclei, the rate is ferociously steep:

against the pp chain's gentle T^{4}. That single contrast explains a great deal of stellar astrophysics. Below about 1.3\,M_{\odot} the core is cooler and the pp chain wins; above it the CNO cycle takes over and, because its output is so violently concentrated at the very centre, it drives convection in the core (the star has to stir vigorously to carry the heat away). It also means massive stars burn their fuel prodigiously fast and live short, brilliant lives — a 20\,M_{\odot} star may last only a few million years, while the Sun's pp chain will keep it steady for ten billion.

Building the heavier elements: helium ash and the onion

Hydrogen fusion — pp or CNO — only fills the core with helium. To go further the star must get hotter still, and that happens naturally: once the core hydrogen is spent, the core contracts under gravity, heats up, and when it reaches about 10^{8}\ \text{K} a new reaction ignites, the triple-alpha process:

This is a small miracle. Beryllium-8 is unstable and falls apart in about 10^{-16}\ \text{s}, so ordinarily nothing can catch it in time. The reaction works only because there is a fortunate excited state of carbon-12 — the Hoyle resonance, predicted by Fred Hoyle precisely because carbon exists — at just the right energy to make the second capture likely. Without it, the universe would have almost no carbon, and no carbon-based chemistry (or biologists to notice).

From carbon onward, a massive star climbs the curve by successive alpha captures and advanced burning stages, each igniting as the previous fuel runs out and the core contracts and heats:

• Helium burning → carbon and oxygen;
• Carbon burning → neon, sodium, magnesium;
• Neon and oxygen burning → magnesium, silicon, sulphur;
• Silicon burning → the iron group (nickel, cobalt, iron).

Each stage runs hotter, faster and shorter than the last. In the final days the star develops an onion-shell structure: hydrogen fusing in a thin outer skin, then shells of helium, carbon, neon, oxygen and silicon burning at ever-greater temperatures, all wrapped around a growing core of inert iron ash. And there the fire meets its wall.

The iron catastrophe and the elements beyond

Silicon burning dumps iron-56 into the core, and iron, sitting at the very peak of the B/A curve, is a fuel that will not burn: fusing it absorbs energy instead of releasing it. The core's fire goes out. With no fusion to hold it up against gravity, once the iron core exceeds about 1.4\,M_{\odot} (the Chandrasekhar mass) it collapses catastrophically — in under a second, an Earth-sized core implodes to a city-sized ball of neutrons — and rebounds as a core-collapse supernova, blasting the star's layered elements out into space to seed the next generation of stars and planets.

So where do gold, platinum, uranium and the rest of the heavier-than-iron elements come from, if fusion can't make them? By a completely different mechanism: neutron capture. A neutron has no charge, so it feels no Coulomb barrier and can simply be absorbed by an existing nucleus, which then \beta^{-}-decays to bump up its atomic number. There are two flavours:

• The s-process ("slow"): neutrons are captured slowly compared with the decay time, so the nucleus decays between captures and creeps up the valley of stability. It runs in the gently pulsing envelopes of AGB stars and builds roughly half the elements heavier than iron (barium, lead, and much more).
• The r-process ("rapid"): a torrential neutron flux dumps captures onto a nucleus faster than it can decay, driving it far off stability before it cascades back — the only way to make the heaviest elements like gold, the lanthanides, uranium and thorium. It needs the most extreme environments known: core-collapse supernovae and, spectacularly, neutron-star mergers (one of which, GW170817, was caught in 2017 forging a planet's worth of gold and lit up telescopes across the spectrum).

Put it all together and the periodic table becomes a map of cosmic history: hydrogen and most helium from the Big Bang; carbon through iron fused in stellar cores over the light-element slope of the binding-energy curve; and everything beyond iron built grain by grain in dying stars and colliding neutron stars by neutron capture. Star-stuff, all of it.

Worked examples

Example 1 — the Sun eats itself, four million tonnes a second. The Sun radiates power (luminosity) L_{\odot} = 3.8\times 10^{26}\ \text{W}. All of that energy comes from converting mass to energy via E = mc^{2}. What rate of mass is being turned into pure energy? Rearranging for the mass-loss rate:

The Sun converts about 4 billion kilograms — four million tonnes — of its own mass into sunlight every second. (Note this is the mass that vanishes as energy; the Sun actually processes some 6\times 10^{11}\ \text{kg} of hydrogen per second, of which that 0.7\% is the part that becomes energy.)

Example 2 — where the 26.7 MeV comes from. Weigh the ingredients of one pp-chain helium nucleus. Four protons have total mass 4 \times 1.007276\ \text{u} = 4.029104\ \text{u}. The helium-4 nucleus they make weighs only 4.002602\ \text{u} (this is the atomic mass minus its two electrons, close enough for our purposes). The missing mass — the mass defect — is

Convert it to energy using the handy factor 1\ \text{u} = 931.5\ \text{MeV}/c^{2}:

Add back the roughly 2\ \text{MeV} released when the two positrons annihilate with electrons and you recover the full \approx 26.7\ \text{MeV}. And the fraction of the original mass converted? \Delta m / m \approx 0.0265 / 4.029 \approx 0.0066 \approx 0.7\% — exactly the figure quoted for hydrogen burning. A star is a device for cashing in that 0.7\%, atom by atom, for billions of years.