Nuclear Fusion in Stars — The Burning Stages

A star is a self-regulating thermonuclear reactor held in check by gravity. The core has to produce exactly enough energy, every second, to support the weight of everything above it. Too little — and the core contracts, heating up until fusion speeds back up. Too much — and the core expands, cooling until fusion slows down. This negative feedback is called hydrostatic equilibrium, and every burning stage described below is a different solution to the same equation: what nuclear reactions can a gravitationally confined plasma run, fast enough, at the temperature its own weight forces on it?

The answer changes as the star evolves because fuel runs out and the core must heat up to reach the next reaction. Each successive fuel requires a higher Coulomb barrier to cross — the electrostatic repulsion between two positive nuclei that have to touch before the strong force can bind them. Heavier nuclei carry more protons, so the barrier grows and each stage demands a hotter, denser core. That single fact is what drives the whole sequence of burning stages and sets a clock on the star's life.

Everything that follows — the pp chain in the Sun, the CNO cycle in hot massive stars, the helium flash, the onion-shell structure of a supergiant, the iron core that triggers a supernova — is just a star looking for the next way to stay in balance.

Fusion releases energy because a bound nucleus weighs slightly less than the sum of its separated nucleons. The difference — the mass defect — is converted to radiation via E = Δm c². Plot the binding energy per nucleon against mass number A and you get one of the most important graphs in astrophysics.

The curve rises steeply from hydrogen, peaks around Fe-56 / Ni-62 at about 8.8 MeV/nucleon, and falls away for heavier nuclei. Two consequences follow, and they dictate everything a star does:

Two positively charged nuclei repel each other. To fuse they have to approach within roughly 1–2 fm — the range of the strong force. The electrostatic potential barrier at that distance is enormous: for two protons it is about V_C ≈ 1 MeV. The core of the Sun is only at T ≈ 15 × 10⁶ K, which corresponds to an average thermal energy of just kT ≈ 1.3 keV — nearly a thousand times too little to climb the barrier classically.

Stars work because nuclei don't need to climb the barrier. They tunnel through it. The probability that a particle with energy E tunnels through a Coulomb barrier is captured by the Gamow factor:

The parameter η (the Sommerfeld parameter) grows with the product of the nuclear charges Z₁Z₂ and falls with relative velocity v. Two consequences matter for stars:

The raw nuclear physics is packaged into the astrophysical S-factor S(E), a smooth function (for non-resonant reactions) that absorbs the trivial Coulomb and geometric dependencies, leaving a quantity that experimentalists can extrapolate down to stellar energies from laboratory measurements. Reaction rates in stellar models are computed from S(E), temperature, and density — not from classical collision estimates.

The Gamow peak sits where the rising tunneling probability meets the falling Maxwell-Boltzmann tail. Its centroid E₀ is typically 5–30 kT, far above the thermal average — so it is the rare, fast particles that do the fusing.

Because the Gamow peak sits so far up the thermal tail, reaction rates depend violently on temperature. It is customary to write ε ∝ ρ T^ν and read off the exponent ν:

The extreme T-dependence of the CNO cycle is the reason a star just 1.3× the Sun's mass can be ten times more luminous: a small bump in core temperature multiplies the power output enormously.

In stars below about 1.3 M☉, core temperatures are too low for the CNO cycle to matter, and the proton-proton (pp) chain dominates. Net reaction:

Of that 26.73 MeV, the neutrinos escape the star almost immediately and carry off 0.6–2% of the energy depending on which branch fires. The rest heats the plasma. The first step is the rate-limiting bottleneck of the entire chain:

pp-I — ~86% of the time in the Sun

Step (1) is astonishingly slow. It requires not just Coulomb tunneling but a simultaneous weak interaction that converts one proton into a neutron — and weak processes are rare. The mean lifetime of a proton against pp fusion in the Sun's core is about 10¹⁰ years. This bottleneck is why the Sun will last ~10 Gyr rather than seconds. Step (2) is fast: deuterium burns so quickly it is essentially invisible. Step (3) is the other slow one (two rare ³He nuclei must find each other).

pp-II — ~14% of the time

pp-III — ~0.1% of the time

The three branches converge on the same net result but produce neutrinos with different energy spectra. Solar neutrino observatories (Super-Kamiokande, SNO, Borexino, JUNO) measure each branch independently — this is how we know the Sun's core temperature to ~1% and how the solar neutrino deficit was resolved into the discovery of neutrino oscillations.

The Sun converts about 600 million tonnes of hydrogen per second into helium via these chains, with ~4.3 million tonnes of that mass going directly into radiation as E = mc². Over a 10-billion-year main-sequence life that consumes roughly 10% of the Sun's total hydrogen — only the convective or near-convective core participates.

In stars hotter than about 18 MK at the core — typically more massive than ~1.3 M☉ — a different route dominates. Pre-existing carbon, nitrogen, and oxygen act as catalysts, cycling through a series of proton captures and β-decays that consume four protons and leave behind one ⁴He, regenerating the catalyst. The standard CNO-I cycle:

The net reaction is identical to the pp chain — four protons into one helium — but notice two extra features. First, the ¹⁴N(p,γ)¹⁵O step is the slowest, so nitrogen piles up at the expense of the other catalysts. Stars running the CNO cycle eventually develop a core depleted in ¹²C and enriched in ¹⁴N — an abundance shift later dredged to the surface by convection in red giants, and directly observable in spectra.

Second, the CNO rate scales roughly as T¹⁷ near solar temperatures. A factor-of-two jump in core temperature multiplies the rate by ~130,000. The transition from pp dominance to CNO dominance happens over a narrow mass range near ~1.3 M☉. This is the physical origin of the kink in the main sequence between F- and A-type stars, and why the convective core that CNO burning produces changes the whole internal structure.

CNO-II, NeNa, and MgAl

At hotter temperatures (> 50 MK) the cycle has secondary branches. CNO-II branches at ¹⁵N: instead of emitting an α, it captures a proton to produce ¹⁶O, which then proceeds through ¹⁷F and ¹⁷O before rejoining. At even higher T (classical novae, hot AGB stars) the NeNa cycle and MgAl cycle run in parallel, producing telltale isotopes like ²⁶Al (a gamma-ray tracer of recent massive-star activity) and ²²Na. These cycles don't contribute much energy but they rearrange element abundances in ways we can measure.

When core hydrogen is exhausted, a helium core of degenerate or near-degenerate helium sits inside a hydrogen-burning shell. The core contracts, heats, and eventually ignites a remarkable reaction. The obvious pathway — just fuse two alphas — does not work, because ⁸Be is unbound. It exists for only ~10⁻¹⁶ s before flying apart. Yet at ~100 MK, enough ⁸Be forms at any instant in thermal equilibrium for a third α to occasionally arrive in time:

Net: 3 ⁴He → ¹²C + γ with Q = 7.275 MeV. Only 0.7% of the mass is converted — about one tenth as efficient per gram as hydrogen burning. The rate scales roughly as ε₃α ∝ ρ² T⁴⁰ — the cube is because three particles must meet, and the T⁴⁰ is the reason helium ignition is essentially explosive in degenerate cores (the helium flash in low-mass red giants, where for a few minutes the core luminosity exceeds 10¹¹ L☉ — all absorbed internally, never reaching the surface).

Triple-alpha burning does not happen in isolation. As soon as ¹²C accumulates, it captures another alpha to form ¹⁶O — another reaction that shapes the universe:

The ratio of ¹²C to ¹⁶O produced during helium burning depends delicately on the ¹²C(α,γ)¹⁶O rate, which is probably the single most important nuclear astrophysics reaction still not measured to the precision we would like. It dictates the final C/O ratio in white dwarfs, the composition of massive-star cores going into supernovae, and the abundance of oxygen (the third most common element in the universe) versus carbon.

At helium-burning temperatures, further α-captures (onto ²⁰Ne, ²⁴Mg, ²⁸Si, …) are much slower and don't contribute significantly. To go further, the star has to heat up again.

Only stars born above ~8 M☉ ever reach these stages. Lower-mass stars end with a degenerate C/O core that never ignites — they become white dwarfs. The massive-star sequence is a series of shorter, hotter, stranger burning stages separated by brief contractions, stacking the star's interior into concentric shells of different fuel:

Carbon burning (~600 MK)

Carbon burning lasts a few hundred years in a typical massive star — but almost all of that energy is carried off by neutrinos, not photons. At these temperatures thermal pair production (γγ → e⁺e⁻) followed by e⁺e⁻ annihilation into a neutrino pair becomes the dominant core cooling channel. The star's surface luminosity barely twitches even as its core is hemorrhaging energy. From here on every later burning stage loses energy overwhelmingly to neutrinos, which is why the timescales get so short.

Neon burning (~1.2 GK) — photodisintegration

At a gigakelvin the thermal photon field is so energetic that high-energy γ-rays can break nuclei apart. Neon burning begins with photodisintegration of ²⁰Ne rather than a direct fusion:

Net: two neons make one oxygen and one magnesium. The α released by photodisintegration is captured by another ²⁰Ne, liberating energy. Neon burning lasts perhaps a year in a massive star.

Oxygen burning (~2 GK)

A few months of oxygen burning produces a core of ²⁸Si and a stew of intermediate-mass nuclei. The core has now run through four successive fuels (H, He, C, O) and is building a silicon-group envelope — but neutrino losses are now so high that the core contracts visibly on human timescales.

At temperatures above ~3 GK, the ²⁸Si + ²⁸Si Coulomb barrier is still too high for a direct fusion. Instead, silicon burning proceeds by a quasi-equilibrium: photons chip α-particles and nucleons off existing nuclei, and those fragments are reabsorbed by heavier targets. Dozens of reactions run in parallel, flowing net material up the binding-energy curve toward iron:

The network climbs in steps of A = 4 (adding an alpha at a time) from ²⁸Si through ³²S, ³⁶Ar, ⁴⁰Ca, ⁴⁴Ti, ⁴⁸Cr, ⁵²Fe, and ends at ⁵⁶Ni — which is the end of the line because further α-captures on ⁵⁶Ni are slow and unproductive. After the supernova explosion, ⁵⁶Ni β-decays (τ = 6 d) to ⁵⁶Co, which β-decays (τ = 77 d) to stable ⁵⁶Fe. That radioactive decay chain is what powers the light curve of a Type Ia supernova for months after the explosion.

As burning deepens, the population of all these isotopes converges to the distribution that maximizes entropy at the given temperature and density — a state called nuclear statistical equilibrium (NSE). In NSE, abundances are determined entirely by binding energy, and the iron-peak nuclei (⁵⁴Fe, ⁵⁶Fe, ⁵⁶Ni, ⁵⁸Ni) dominate because they sit at the top of the curve.

Why it stops at iron

Once the core is iron-group, the binding-energy curve turns over. Fusion no longer releases energy — it absorbs it. At the same time the temperature is high enough that photodisintegration starts to destroy iron nuclei:

Every endothermic reaction steals energy from the pressure supporting the core. Meanwhile, densities are so high (~10⁹ g/cm³) and the Fermi energy of core electrons so large that electron capture onto protons becomes favorable:

This removes electrons (which were providing degeneracy pressure) and converts protons to neutrons. Both effects accelerate the collapse. The inner iron core, now below the Chandrasekhar limit it was relying on, free-falls inward at a significant fraction of the speed of light. A bounce at nuclear density launches the neutrinos and shock wave that become a core-collapse supernova. The star that took ten million years to get here is gone in a second.

Fusion cannot build elements heavier than iron. They come from a different mechanism altogether: neutron capture. Neutrons have no charge, no Coulomb barrier, and they can drift up to a nucleus and stick. The captured nucleus then decays by β⁻ emission to the next element on the periodic table:

Whether this path produces mainly samarium, europium, gold, platinum, or uranium depends on how rapidly neutrons arrive relative to how long the intermediate isotopes survive against β-decay. That distinction produces two very different nucleosynthesis regimes:

A smaller p-process (also called γ-process) accounts for the rare proton-rich isotopes the s and r processes can't reach. It runs in the shocked outer layers of core-collapse supernovae.

You can point a telescope at a representative of every burning stage described above in one evening.

Pair a bright example with the physics above and the night sky stops being a gallery of pretty points. Every photon that reaches your eyepiece was paid for by a nuclear reaction with a measurable cross-section, happening right now, in a core whose temperature and composition you can look up on each star's detail page.