Stellar nuclear fusion

Stellar nuclear fusion converts light nuclei into more tightly bound nuclei and supplies most main-sequence stellar luminosity.

Why can a star shine for so long, and why do hotter stellar cores favor different nuclear reactions? A useful theory must supply both an adequate energy reservoir and the right stellar energy-generation rate. Nuclear binding energy supplies the reservoir; the thermonuclear reaction rate, stellar composition and stellar structure determine how quickly it is spent. The proton–proton chain and CNO cycle are two routes for hydrogen burning with very different temperature sensitivities.

In his 1967 lecture, Hans Albrecht Bethe contrasted the short Kelvin-Helmholtz cooling time with long ages inferred from terrestrial and stellar evidence.^[ref]

For an original solar-scale estimate, choose mass $M=1.99\times 10^{30}\mathrm{kg}$, radius $R=6.96\times 10^8\mathrm{m}$ and luminosity $L=3.83\times 10^{26}\mathrm{W}$. The characteristic Kelvin-Helmholtz cooling time $G{M}^2/(RL)$ is about $31\mathrm{Myr}$; the precise gravitational reservoir has a structure-dependent coefficient. If a fraction $f=0.10$ of the mass can burn its initial hydrogen mass fraction $X=0.70$, with a chosen rest-energy conversion fraction $\eta =0.007$, the fuel estimate isThis assumes fixed luminosity and neglects neutrino losses. It demonstrates the reservoir advantage, rather than predicting the Sun's age or its full stellar evolution. Having enough fuel still does not identify a reaction: the fuel must release energy at a rate compatible with hydrostatic equilibrium and radiative transfer.

Hans Albrecht Bethe and Charles Louis Critchfield calculated deuterium formation from two protons in 1938, combining quantum tunnelling with a weak conversion of a proton into a neutron. Their calculated nuclear reaction cross-section was a theoretical input to hydrogen burning.^[ref] In 1939, Hans Albrecht Bethe analysed the catalytic carbon–nitrogen route and compared required stellar core temperatures with stellar structure estimates in Table VIII. His reported solar agreement used historical abundance and structure assumptions; it should not be read as a modern demonstration that the CNO cycle dominates the Sun.^[ref]

Hans Albrecht Bethe's 1967 lecture credits Carl Friedrich von Weizsäcker's independent 1938 carbon–nitrogen cycle discovery, distinguishing it from calculation of rates and temperature dependence.^[ref] The 1967 Nobel Prize in Physics recognized Hans Albrecht Bethe's contributions to nuclear reaction theory, especially stellar energy production.^[ref]

Laboratory nuclear reaction cross-sections were inputs, not measurements of an artificial star. The 1939 paper converted thick-target yields to nuclear reaction cross-sections using a range–energy relation and tabulated capture data.^[ref] A concrete apparatus example is the Cockcroft-Walton lithium disintegration experiment, with an accelerated proton beam, lithium target, zinc sulfide screen and beam-blocking control. Its nuclear reaction differs from both stellar burning routes. The 1967 lecture stresses extrapolation below laboratory access and tests for nearby nuclear levels.^[ref]

William Alfred Fowler's 1983 lecture connects laboratory nuclear reaction cross-sections to stellar nucleosynthesis and credits Fred Hoyle's stellar synthesis framework.^[ref] This complements Bethe stellar energy-production theory: a fuel reservoir alone cannot predict surviving stellar chemical abundances.

Transient beryllium-8 must capture an alpha particle before dissociation; subsequent electromagnetic decay of the Hoyle state can stabilize carbon-12 despite dominant breakup.

In 1953, David Noel Ferguson Dunbar and collaborators used accelerated deuterons on nitrogen-14, populating carbon-12 through ${}^{14}\mathrm{N}(d,\alpha ){}^{12}\mathrm{C}$. Magnetic analysis resolved outgoing alpha particle groups. At $620\mathrm{keV}$ incident energy and laboratory angle $90$ degrees, they reported excitation $7.68\pm 0.03\mathrm{MeV}$, with yield $6\%$ of the $4.43\mathrm{MeV}$ group's yield. The stronger group supplied an internal comparison. The paper's opening explicitly relates the search to Fred Hoyle's abundance argument. This is evidence for a nuclear level from a different entrance reaction than stellar fusion.^[ref]

The inference uses energy and momentum conservation: at fixed incident energy and observation angle, more energy retained as carbon-12 excitation leaves less outgoing kinetic energy. A resolved particle group therefore identifies a nuclear level. This laboratory nuclear line supports an input required by a stellar synthesis model. It does not reproduce core helium burning, measure a complete stellar carbon yield, or by itself determine the stabilizing decay probability.

In a dilute thermal helium gas with rapid formation–decay balance of beryllium-8, write its small equilibrium abundance schematically as $n_8\propto n_{\alpha }^2{T}^{-3/2}\exp [-{\Delta }_8/(k_{B}T)]$. A narrow isolated Hoyle state gives a second-step coefficient proportional to $T^{-3/2}\exp [-{\Delta }_C/(k_{B}T)]$. Multiplication yieldswith the stabilizing decay strength absorbed into the coefficient. Here $k_B$ is the Boltzmann constant. The two positive offsets count energy relative to their respective entrance thresholds, not relative to the carbon-12 ground state. A resonance enhances capture at the relevant energy; the Boltzmann factor still selects how many encounters reach it. This is a sequential narrow-level approximation with fixed nuclear widths, excluding plasma screening and changes of composition or mass density.

William Alfred Fowler lists the Hoyle state $7.654\mathrm{MeV}$ above the carbon-12 ground state, $0.2875\mathrm{MeV}$ above beryllium-8 plus an alpha particle, and $0.3794\mathrm{MeV}$ above three alpha particles.^[ref] Subtraction gives ${\Delta }_8=0.0919\mathrm{MeV}$ for the diagram.

Normalize the calculated rate at $T_0=10^8\mathrm{K}$ and $E_r=0.3794\mathrm{MeV}$. Holding alpha particle number density and the decay strength fixed gives the original sensitivity calculationAt $T_0$, a deliberately chosen shift $\delta E=+100\mathrm{keV}$ multiplies the rate by $9.12\times 10^{-6}$; $-100\mathrm{keV}$ gives $1.10\times 10^5$. These offsets are hypothetical perturbations, not alternative measured nuclear levels. The reference curve has local temperature exponent of a thermonuclear reaction $-3+E_r/(k_B{T}_0)=41.03$: raising temperature by $10\%$ gives a factor $41.12$ in this fixed-environment calculation. A real star changes its stellar structure and composition as it burns, so these factors do not directly predict final carbon abundance or a constraint on life.

Temperature-sensitive nuclear fusion can heat material faster than cooling and expansion can regulate it, accelerating the burning. In a classical nova, hydrogen accumulated on a white dwarf ignites; initial electron degeneracy pressure helps suppress the usual expansion response. The eruption can eject the envelope without destroying the underlying white dwarf.

A thermal distribution of relative velocities weights a nuclear cross-section to determine the reaction rate per volume. Identical reactants require the denominator two to avoid counting each pair twice. Specific deposited power is $r_{12}{Q}_{\mathrm{deposited}}/\rho$. Charged-particle quantum tunnelling through a Coulomb barrier produces a Gamow peak, while resonances and plasma screening can alter the rate.

If a charged-particle rate coefficient has $\lambda \propto {\eta }^2{e}^{-\eta }$ with $\eta \propto T^{-1/3}$, its logarithmic temperature derivative is $(\eta -2)/3$. This is a local slope evaluated at fixed composition. It increases as temperature falls, so extrapolating a single power law across a large temperature range can be inaccurate.

The astrophysical S factor separates the dominant Coulomb barrier suppression from a charged-particle nuclear reaction cross-section $\sigma (E)$:Here $E$ is relative kinetic energy, $\mu$ is reduced mass, $\alpha$ is the fine-structure constant, and $Z_1,Z_2$ are atomic numbers. The astrophysical S factor has units of energy times area. Factoring out a barrier does not remove all nuclear physics: nuclear levels and reaction mechanisms can still make $S(E)$ vary strongly.

For a dilute nonrelativistic Maxwell-Boltzmann velocity distribution of relative motion, multiplying the relative-energy density by speed and nuclear reaction cross-section cancels the explicit factors of $E$, givingThus the Gamow peak describes a rate-integral kernel, not a histogram of reacting stars or even the unweighted particle population. Plasma screening and departures from the assumed Maxwell-Boltzmann velocity distribution require additional physics.

For a nonresonant charged-particle thermonuclear reaction rate, the thermal high-energy tail favors low $E$ while the Coulomb barrier penetration factor favors high $E$. Their product peaks in an intermediate energy window. Resonant reactions such as the Triple-alpha process require an additional resonance treatment.

For a slowly varying nuclear factor and penetration factor $\exp [-2\sqrt{E_B/E}]$, the thermal integral is ${\int }_0^{\infty }{e}^{-\Phi (E)}dE$, where $\Phi (E)=E/(k_{B}T)+2\sqrt{E_B/E}$. The Gamow peak occurs at $E_0=[E_B(k_{B}T{)}^2{]}^{1/3}$, with $\Phi (E_0)=3E_0/(k_{B}T)$ and ${\Phi }^{\prime \prime }(E_0)=3/(2k_{B}T{E}_0)$. The Gaussian integral gives $\sqrt{4\pi k_{B}T{E}_0/3}{e}^{-3E_0/(k_{B}T)}$. Combined with the Maxwellian prefactor $(k_{B}T{)}^{-3/2}$ this produces the displayed thermonuclear reaction rate at fixed reactant density. The approximation assumes $E_B\gg k_{B}T$ and excludes strong resonances or rapidly varying nuclear factors.

The Gaussian approximation to a nonresonant thermonuclear rate already gives the saddle and curvature. In its notation $E_G=4E_B$, so the same result readsThe full width ${\Delta }_{1/e}$ joins the two points at $e^{-1}$ of the peak of the approximating Gaussian function; it is neither its standard deviation ${\sigma }_G$ nor its full width at half maximum $2\sqrt{2\log 2}{\sigma }_G$. These are local Gaussian approximation widths, not exact equal-height endpoints of the asymmetric kernel.

Why keep the astrophysical S factor inside the integral? If $S(E)$ is positive and smooth, stationarity of $\log S(E)-\Phi (E)$ gives a leading peak displacement $\delta E\simeq {\sigma }_G^2(d\log S/dE{)}_{E_0}$. The slow-variation test is therefore $|d\log S/dE|{\sigma }_G\ll 1$, together with small curvature corrections. More generally any energy-dependent prefactor shifts the saddle; a temperature-dependent prefactor also changes the local temperature exponent of a thermonuclear reaction. A sharp nuclear feature can invalidate this expansion altogether.

Choose $T=15\mathrm{MK}$ and $E_G=493\mathrm{keV}$ for an illustrative proton pair. Then $k_{B}T=1.293\mathrm{keV}$, $E_0=5.905\mathrm{keV}$ and ${\Delta }_{1/e}=6.380\mathrm{keV}$. The relative Maxwell-Boltzmann velocity distribution has energy-density maximum $k_{B}T/2=0.646\mathrm{keV}$, far below the reaction kernel's maximum. Now choose the deliberately artificial enhancementNumerical integration raises the rate by about $54.2\%$ at the same temperature, even though the enhancement is away from $E_0$. This is a sensitivity experiment, not a claimed proton–proton nuclear level or a physical nuclear line-shape model. It illustrates why extrapolating $S(E_0)$ without checking nearby structure can fail.

At very high temperatures, photodisintegration and proton, neutron and alpha captures rearrange silicon-group material through networks approaching quasi-statistical equilibrium. The result is an iron-group mixture determined by electron fraction and weak-reaction timescales. Direct silicon-plus-silicon fusion is not the usual mechanism. The approach to maximal nuclear binding energy per nucleon ends sustained hydrostatic fusion power.

Oxygen fusion in sufficiently hot evolved stellar cores has several reaction channels, including the displayed silicon-plus-alpha channel and channels producing phosphorus and protons. Its ashes feed silicon burning, while stellar neutrino energy loss strongly affects the evolution timescale.

High-temperature photodisintegration of neon-20 releases an alpha particle and oxygen-16. Capture of that alpha particle by another neon-20 nucleus produces magnesium-24. The breakup consumes energy, but the combined reactions can release net energy. Other captures modify the ashes in a real network.

A helium-burning shell converts helium into carbon and oxygen outside a helium-exhausted stellar core. Thin-shell burning on the Asymptotic giant branch can develop a thermal pulse of an asymptotic-giant-branch star.

The helium mass fraction is the fraction of a material's mass carried by helium nuclei. Hydrogen burning raises it, whereas core helium burning lowers it while producing carbon and oxygen.

The stellar energy-generation rate $ϵ$ is power produced per unit mass. Spherical energy conservation is $d{L}_r/dr=4\pi r^2\rho ϵ$.

The hydrogen mass fraction is the fraction $X$ of a stellar material's mass carried by hydrogen nuclei.

Hydrogen burning converts hydrogen into helium through the proton–proton chain or CNO cycle.

A hydrogen-burning shell converts hydrogen into helium in a layer surrounding a hydrogen-exhausted stellar core. It moves outward in enclosed mass as it supplies fresh helium to that core; its movement in radius need not have the same sign.

The triple-alpha process fuses three helium-4 nuclei into carbon-12 through an unstable beryllium-8 intermediate. Its steep temperature dependence can make helium-shell burning thermally unstable. The resonant triple-alpha temperature sensitivity isolates the role of the Hoyle state, while laboratory evidence for the Hoyle state explains how a particle spectrum tests the required nuclear level.

This excited carbon-12 resonance lies close to the three-alpha threshold and enables the Triple-alpha process to make carbon efficiently. A transient beryllium-8 nucleus captures an alpha particle into the resonance; electromagnetic decay then stabilizes carbon-12. A small radiative branching probability still matters because the resonance enhances formation.

The CNO mass fraction is the combined mass fraction of carbon, nitrogen, and oxygen nuclei available to catalyse the CNO cycle.

The proton–proton chain converts hydrogen to helium and dominates hydrogen burning in stars near and below the Sun's mass. Near $2\times 10^7\mathrm{K}$, its specific energy generation is roughly proportional to $\rho T^4$.

The proton–proton chain first produces deuterium by a weak reaction and then helium-3 by proton capture. pp I ends with ${}^3\mathrm{He}+{}^3\mathrm{He}\to {}^4\mathrm{He}+2p$. pp II passes through beryllium-7 electron capture to lithium-7 and then proton capture. pp III passes through boron-8 beta-plus decay and beryllium-8 breakup. Each branch produces one net helium-4 nucleus from four protons, but the neutrino spectrum and deposited heat differ. In pp II the net weak reactions involve one emitted positron and one captured electron.

Present solar hydrogen is most depleted in the central burning region and rises toward the envelope. helium-3 is efficiently destroyed in the hot centre, increases outward as its equilibrium abundance rises, and develops an off-centre maximum where equilibration becomes slow. Still farther out its production becomes weak, leaving a smaller envelope abundance. Convective mixing tends to flatten the outer profiles. This qualitative pattern is not itself a quantitative solar model.

An effective Proton–proton chain network can track protons, deuterium, helium-3, helium-4 and lithium-7 after eliminating a rapidly adjusting beryllium intermediate. Identical-reactant event rates include a factor one half. Stoichiometric changes must count both helium-4 products of lithium-7 proton capture. The weighted baryon mass density $n_1+2n_2+3n_3+4n_4+7n_7$ is conserved by the nuclear reactions at fixed volume.

With $A={\lambda }_{33}$, $B={\lambda }_{34}{n}_4$ and $C={\lambda }_{11}{n}_1^2/2$, helium-3 evolves as ${\dot{n}}_3=C-A{n}_3^2-B{n}_3$ after deuterium elimination. The positive root is its unique attracting equilibrium for a fixed background. In the pp-I limit it reduces to $n_1\sqrt{{\lambda }_{11}/(2{\lambda }_{33})}$, making equilibrium abundance strongly temperature dependent.

A small abundance displacement from the helium-3 equilibrium abundance obeys $\dot{x}=-x/\tau$ to linear order. Rapid relaxation relative to background evolution allows the abundance to track its equilibrium. If the equilibrium drifts, the displacement equation also contains $-{\dot{n}}_{3e}$. In the pp-I limit the rate is $n_1\sqrt{2{\lambda }_{11}{\lambda }_{33}}$, so both production and destruction temperature dependences matter.

Equilibrium fails where the helium-3 relaxation time becomes comparable to stellar age. In a fixed-density pp-I scaling with reaction exponents four and sixteen, the equilibrium abundance scales as $T^{-6}$ and the relaxation time as $T^{-10}$. Central normalization then places the transition near six million kelvin; using the full supplied exponential factors gives a somewhat higher estimate. A precise radius or temperature requires a stellar mass density/composition profile and the second branch contribution.

In the pp-I limit, helium-3 relaxation time is ${\tau }^{-1}=n_1\sqrt{2{\lambda }_{11}{\lambda }_{33}}$. At fixed proton number density, each thermonuclear reaction rate coefficient obeys $\lambda (T)/\lambda (T_0)=(T/T_0{)}^{-2/3}\exp [-\eta (T_0)((T_0/T{)}^{1/3}-1)]$. Taking their geometric mean gives the displayed relaxation ratio with $\bar{\eta }=[{\eta }_{11}(T_0)+{\eta }_{33}(T_0)]/2$. A local power-law approximation instead gives $\tau \propto T^{-(\alpha +\beta )/2}$. These are estimates rather than a complete radial model: the proton number density, composition and pp-II branch change along a real stellar composition profile.

deuterium proton capture is much faster than the weak reaction that produces deuterium. Its abundance therefore adjusts so that ${\lambda }_{21}{n}_2{n}_1\simeq {\lambda }_{11}{n}_1^2/2$. Substituting this balance eliminates the fast intermediate from the slower proton and helium-3 evolution equations. This is a quasi-steady approximation, not an assumption that hydrogen itself is in nuclear equilibrium.

The CNO cycle catalyses hydrogen fusion through carbon, nitrogen, and oxygen nuclei. Near $2\times 10^7\mathrm{K}$, its specific energy generation is approximately proportional to $\rho T^{17}$ and is therefore strongly concentrated near a star's centre.

If proton–proton chain heating scales as $X_H^2\rho T^{{\nu }_{pp}}$ and CNO cycle heating as $X_H{Z}_{\mathrm{CNO}}\rho T^{{\nu }_{\mathrm{CNO}}}$, their ratio at fixed hydrogen abundance scales as $Z_{\mathrm{CNO}}{T}^{{\nu }_{\mathrm{CNO}}-{\nu }_{pp}}$. Equating the rates proves the displayed crossover-temperature relation. It applies to a local power-law approximation, with unchanged coefficient normalizations apart from the catalyst abundance.

A hotter stellar core can favor a higher-barrier route despite fewer catalyst nuclei. To isolate this competition, prescribe a local hydrogen burning toy model at fixed mass density $\rho$, with reference $T_0=15\mathrm{MK}$ andChoose $E_{G,pp}=493\mathrm{keV}$ and $E_{G,\mathrm{CNO}}=45000\mathrm{keV}$, rounded barrier parameters for a proton pair and proton capture on nitrogen-14. These prescribe the temperature dependence, while the following freely chosen heating normalization incorporates effective nuclear factors and deposited energy:Here $X$ is hydrogen mass fraction, $X_{\mathrm{CNO}}$ is CNO mass fraction, and ${ϵ}_{\ast }$ is an arbitrary positive unit of stellar energy-generation rate. For $X=0.70$ and $X_{\mathrm{CNO}}=0.01$, the reference ratio is $0.01$ by construction. It is not a measured solar ratio. The CNO cycle bottleneck is represented by one effective rate, without evolving the existing proton-proton chain branches or catalyst network.

At $T_0$, the logarithmic temperature slopes are $({\tau }_{pp,0}-2)/3=3.90$ and $({\tau }_{\mathrm{CNO},0}-2)/3=19.90$. The larger Coulomb barrier gives a steeper fractional response. Because these slopes change with temperature, this example retains the exponential rather than extending two fixed power laws. Writing $R_0=0.01(X_{\mathrm{CNO}}/0.01)(0.70/X)$ and $D={\tau }_{\mathrm{CNO},0}-{\tau }_{pp,0}$ givesprovided the bracket is positive. With the chosen composition, $T_{\ast }=20.30\mathrm{MK}$. Reducing $X_{\mathrm{CNO}}$ to $0.001$ at unchanged $X$ shifts it to $23.91\mathrm{MK}$. For $X_{\mathrm{CNO}}=0$, this catalytic route supplies no heat in this model. These are local toy crossovers: a whole star's luminosity requires integrating its stellar energy-generation rate through its stellar composition profile and stellar structure. Nuclear features, plasma screening and changes of the dominant bottleneck limit the model's extrapolation.

A branch of the CNO cycle proceeds through proton captures on nitrogen-15 and oxygen-16, beta-plus decay of fluorine-17, and ${}^{17}\mathrm{O}(p,\alpha ){}^{14}\mathrm{N}$. Returning through oxygen-15 and nitrogen-15 closes the cycle. It also catalyzes the conversion of four protons into helium-4, with branch-dependent neutrino loss.

The principal CNO cycle follows ${}^{12}\mathrm{C}(p,\gamma ){}^{13}\mathrm{N}$, beta-plus decay to carbon-13, proton capture to nitrogen-14, proton capture to oxygen-15, beta-plus decay to nitrogen-15, and ${}^{15}\mathrm{N}(p,\alpha ){}^{12}\mathrm{C}$. The carbon seed is regenerated. The ordinary bottleneck is ${}^{14}\mathrm{N}(p,\gamma ){}^{15}\mathrm{O}$, so nitrogen-14 accumulates in cycle equilibrium.

The stellar thermostat is the negative feedback between temperature-sensitive nuclear burning and hydrostatic expansion: excess burning expands and cools a star, while deficient burning permits contraction and heating.

Core helium burning begins when a hydrogen-exhausted stellar core becomes hot enough for the Triple-alpha process. Subsequent alpha capture on carbon produces oxygen, leaving a carbon-oxygen core after central helium exhaustion.

A helium flash is a thermally unstable ignition of helium in an electron-degenerate stellar core. A nondegenerate intermediate-mass core can expand and regulate ignition, so a usual isolated star of five solar masses begins core helium burning without this flash.

Carbon burning fuses carbon nuclei at temperatures of order $10^9\mathrm{K}$. In intermediate-mass stars it can ignite off-centre under partial electron degeneracy and propagate inward as a burning flame.

The slow neutron-capture process builds nuclei heavier than iron when neutron captures occur more slowly than typical beta decays. Thermal pulses in asymptotic-giant-branch stars produce characteristic enhancements of elements such as strontium, zirconium, and barium.