Stellar Structure and Nucleosynthesis
Hydrostatic stellar models, pp/CNO energy production with its solar-neutrino and helioseismic evidence, and stellar nucleosynthesis against the measured element abundances. A star is a self-gravitating fusion reactor in quasi-static equilibrium, and this chapter builds the standard model of one: the structure equations assembled by Eddington [Eddington:1926], the nuclear energy sources identified by Bethe and von Weizsäcker [Bethe:1939] [vonWeizsaecker:1938], and the synthesis of the elements in stellar interiors codified by Burbidge, Burbidge, Fowler and Hoyle [Burbidge:1957].
What earns the model its place in an evidence-based treatise is that the Sun's interior has been measured twice over, by two channels that did not have to agree: the neutrinos of the pp chain and CNO cycle, from the chlorine deficit [Davis:1968] to its resolution by flavour transformation [Ahmad:2002], and the acoustic oscillation spectrum of helioseismology [Leighton:1962]. The chapter feeds its endpoints to Compact Stars and Relativistic Astrophysics and its nuclear physics comes from Nuclear Forces and Nuclear Structure; the modern monograph is [Kippenhahn:2012].
The equations of stellar structure
Hydrostatic equilibrium and timescales
A star is a ball of gas whose mechanical state, to an accuracy that the end of this subsection will quantify, is rest: pressure balances self-gravity at every radius. The equations governing that balance are the fluid statics of Fluid Dynamics applied to a body whose gravitational field is its own.
A spherically symmetric fluid at rest under its own gravity satisfies
where \(m(r)\) is the mass interior to radius \(r\). Rests on Proposition 31.16.
Derives Proposition 52.1. Equation (52.1) is the definition of \(m(r)\) differentiated. For Equation (52.2), hydrostatic balance \(\nabla P=\rho\vect{g}\) (Equation (31.13)) needs the local gravitational acceleration, and for a spherically symmetric distribution that is \(g=-Gm(r)/r^{2}\) directed inward: the shells outside \(r\) exert no net force and the mass inside acts as if concentrated at the centre — Newton's shell theorem, which is Gauss's flux theorem applied to the Poisson equation \(\nabla^{2}\Phi=4\pi G\rho\) over a sphere of radius \(r\).
∎Two more equations make the system closed once the material properties are known. In a steady state, the luminosity \(L_{r}\) flowing outward through radius \(r\) grows by exactly the nuclear power released in each shell,
with \(\varepsilon\) the energy generation rate per unit mass (Section 52.2); and the temperature gradient is fixed by how that luminosity is carried — by radiation or by convection — which is the subject of Section 52.1.3. Together with an equation of state, an opacity \(\kappa(\rho,T)\) and a rate \(\varepsilon(\rho,T)\), Equations (52.1), (52.2) and (52.3) and the transport equation are the four structure equations assembled by Eddington [Eddington:1926].
Before anything is solved, Equation (52.2) already yields the theorem that governs the energetics of every star.
For any configuration obeying Equation (52.2) with \(P(R)=0\) at the surface,
Rests on Equation (52.2) and Proposition 52.1.
Derives Theorem 52.2. Multiply Equation (52.2) by \(4\pi r^{3}\) and integrate from centre to surface. The left side integrates by parts,
the boundary term vanishing at both ends. The right side is
since \(\dd m=4\pi r^{2}\rho\,\dd r\); and \(\int P\,\dd V=\int(P/\rho)\,\dd m\).
∎For a monatomic ideal gas the internal energy density is \(\tfrac{3}{2}P\), so Equation (52.4) reads \(2U_{\mathrm{int}}=-E_{\mathrm{grav}}\), and the total energy is
A star that loses energy therefore grows hotter: as \(E\) decreases, \(U_{\mathrm{int}}\) increases, half of the released gravitational energy being radiated and half stored as heat. This negative heat capacity is what makes a contracting protostar heat up until fusion ignites, and — combined with the extreme temperature sensitivity of the nuclear rates derived in Section 52.2.1 — it is what makes the burning stable: a transient over-production of energy expands the star, which cools it and throttles the reactions back.
The Sun radiates \(L_{\odot}=3.828\times 10^{26}\,\mathrm{W}\), and the radiometric ages of the oldest meteorites and terrestrial rocks show that it has been doing so, at a comparable rate, for about \(4.5\,\mathrm{Gyr}\). Neither chemical energy nor gravitational contraction can supply this for anything like that long. The conflict between the geological record and the physics of the day is what forced the conclusion that stars run on a subatomic energy source [Eddington:1920] [Eddington:1926]. Rests on Theorem 52.2.
Derivation. Derives Phenomenon 52.4. Three timescales settle the matter, and none of them requires solving a structure equation.
The Kelvin–Helmholtz time is how long a star could shine on the gravitational energy released in contracting to its present radius. For mass \(M\) and radius \(R\) that energy is of order \(GM^{2}/R\) (Equation (52.4)), so
about \(3\times 10^{7}\,\mathrm{yr}\). This is the number that set nineteenth-century physics against geology and against Darwin: it falls short of the age of the Earth by more than two orders of magnitude.
The nuclear time supposes instead that hydrogen fuses to helium, releasing a fraction \(0.007\) of the rest energy, and that the fuel available is the inner tenth of the mass, where the temperature is high enough for it to burn:
about \(10^{10}\,\mathrm{yr}\). This comfortably exceeds the observed \(4.5\,\mathrm{Gyr}\) and is of the right order to explain why stars of a solar mass are still on the main sequence in the oldest clusters — so the nuclear hypothesis is not merely sufficient, it is quantitatively right.
The dynamical time, \(t_{\mathrm{dyn}}\sim\left(G\bar{\rho}\right)^{-1/2}\), is under an hour for the Sun. Its separation from the other two by ten and by fourteen orders of magnitude is what licenses treating the star as exactly hydrostatic while its composition slowly changes, which is the assumption every model in this chapter rests on.
∎Polytropes and the Lane–Emden equation
The four structure equations couple mechanics to thermodynamics through the run of temperature. There is one family of models in which the mechanics closes on its own: assume a power-law relation between pressure and density. The classical treatment is [Chandrasekhar:1939].
A polytrope of index \(n\) is a self-gravitating sphere obeying Equations (52.1) and (52.2) with the equation of state
\(K\) and \(n\) constants.
Real stars realize particular indices exactly: a fully convective star is isentropic, and for a monatomic ideal gas constant entropy means \(P\propto\rho^{5/3}\), that is \(n=3/2\); a non-relativistic degenerate electron gas has the same exponent, and its relativistic limit \(P\propto\rho^{4/3}\) is \(n=3\) (Quantum Statistics, taken up for white dwarfs in Compact Stars and Relativistic Astrophysics); and a star in which the ratio of gas to total pressure is uniform is an \(n=3\) polytrope, as shown at the end of this subsection.
Write \(\rho=\rho_{\mathrm{c}}\theta^{n}\) with \(\rho_{\mathrm{c}}\) the central density, and \(r=\alpha\xi\) with
Then Equations (52.1) and (52.2) with Equation (52.7) are equivalent to
The surface is the first zero \(\xi_{1}\) of \(\theta\), so \(R=\alpha\xi_{1}\). Rests on Equations (52.1), (52.2) and (52.7).
Derives Theorem 52.6. Divide Equation (52.2) by \(\rho\), multiply by \(r^{2}\), differentiate, and use Equation (52.1):
For Equation (52.7),
using \(\rho^{1/n}=\rho_{\mathrm{c}}^{1/n}\theta\). Substituting into Equation (52.10) and setting \(r=\alpha\xi\),
which is Equation (52.9) exactly when \(\alpha\) is chosen as Equation (52.8). The boundary conditions state that the density is \(\rho_{\mathrm{c}}\) at the centre and that the pressure gradient vanishes there, as Equation (52.2) requires since \(m\sim r^{3}\).
∎Equation (52.9) has closed-form solutions for exactly three indices:
with \(\xi_{1}=\sqrt{6}\), \(\pi\) and \(\infty\) respectively; the \(n=5\) sphere has infinite radius but finite mass. Rests on Theorem 52.6 and Equation (52.9).
For \(n=1\) substitute \(u=\xi\theta\): Equation (52.9) becomes \(u''=-u\), so \(u=\sin\xi\) selects the solution regular at the origin with \(\theta(0)=1\). The \(n=0\) and \(n=5\) cases are verified by direct differentiation in the same way; for \(n=5\), \(\theta^{5}\xi^{2}\to0\) fast enough that the mass integral below converges even though \(\theta\) has no zero.
∎For every other index the equation is integrated numerically, which is elementary: it is a regular initial-value problem away from \(\xi=0\), started with the series \(\theta=1-\xi^{2}/6+n\xi^{4}/120-\cdots\). The two indices this chapter and Compact Stars and Relativistic Astrophysics need are (computed here, by that integration)
The mass follows from Equation (52.9) itself, with no further integration:
since the integrand is \(-\dd\left(\xi^{2}\theta'\right)/\dd\xi\). The ratio of central to mean density in Equation (52.11) is \(\xi_{1}/\left(3\abs{\theta'(\xi_{1})}\right)\), a pure number fixed by \(n\) alone.
Eliminating \(\rho_{\mathrm{c}}\) between Equation (52.12) (\(M\propto\rho_{\mathrm{c}}^{(3-n)/2n}\) at fixed \(K\)) and \(R=\alpha\xi_{1}\propto\rho_{\mathrm{c}}^{(1-n)/2n}\) gives the polytropic mass–radius relation
Two cases carry physics. For \(n=3/2\), \(M\propto R^{-3}\): a non-relativistic degenerate star shrinks as mass is added, the inverted mass–radius relation of white dwarfs (Compact Stars and Relativistic Astrophysics). For \(n=3\) the exponent diverges: the mass is independent of the radius, fixed by \(K\) alone — the degenerate ultra-relativistic case, where that unique mass is the Chandrasekhar limit (Equation (49.9)).
Eliminating \(K\) and \(\rho_{\mathrm{c}}\) instead in favour of \(M\) and \(R\) gives the central pressure,
with \(W_{3/2}=0.770\) and \(W_{3}=11.05\) (computed here from Equation (52.11)).
Eddington's model of a star [Eddington:1926] is the assumption that the ratio \(\beta\equiv P_{\mathrm{gas}}/P\) of gas pressure to total pressure is uniform, the remainder \(P_{\mathrm{rad}}=(1-\beta)P=aT^{4}/3\) being radiation pressure (Equation (68.2); \(a=4\sigma/c\)). Eliminating \(T\) between \(P_{\mathrm{rad}}=aT^{4}/3=(1-\beta)P\) and \(P_{\mathrm{gas}}=\rho k_{\mathrm{B}}T/(\mu m_{\mathrm{u}})=\beta P\) gives
a polytrope of index exactly \(3\), with \(K\) a known function of \(\beta\) and the mean molecular weight \(\mu\). Because the \(n=3\) mass Equation (52.12) depends on \(K\) alone, \(M\) determines \(\beta\): substituting gives Eddington's quartic equation, \(1-\beta\propto\beta^{4}\mu^{4}M^{2}\), which for the Sun (\(\mu=0.61\)) evaluates to \(1-\beta=4.1\times10^{-4}\) (computed here): radiation pressure is negligible in the Sun. At \(100\,M_{\odot}\) the same equation gives \(1-\beta\approx0.12\) — massive stars are increasingly radiation-supported, the structural face of the Eddington limit derived in Section 52.1.4.
The model's payoff is the solar centre with no nuclear physics and no transport theory at all. From Equations (52.11) and (52.14), computed here,
and, from the ideal-gas law with \(\mu=0.61\),
The detailed standard solar model, with its opacity tables, composition gradient and helioseismic calibration, gives \(T_{\mathrm{c}}=1.57\times 10^{7}\,\mathrm{K}\), \(\rho_{\mathrm{c}}\approx1.5\times 10^{5}\,\mathrm{kg}/\mathrm{m}^{3}\) and \(P_{\mathrm{c}}\approx2.3\times 10^{16}\,\mathrm{Pa}\) [Bahcall:2005] (quoted): a one-parameter-family polytrope lands within a factor of two of all three, and the temperature within twenty-five per cent. That \(T_{\mathrm{c}}\sim10^{7}\,\mathrm{K}\), secure already at this level, is what Section 52.2 needs.
Energy transport: radiation and convection
Deep inside a star, photons are absorbed and re-emitted after mean free paths of millimetres. Energy therefore leaks outward as a diffusive trickle, and the machinery is the transport theory of Kinetic Theory of Gases with the photon gas of Quantum Statistics as the carrier.
Where the photon mean free path \(\ell=1/(\kappa\rho)\) is small compared with the temperature scale height, the radiative energy flux is
equivalently, with \(L_{r}=4\pi r^{2}F\),
where \(\kappa\) is the Rosseland mean opacity (below). Rests on Equation (68.2).
Derives Proposition 52.9. Carriers of speed \(c\) and mean free path \(\ell\) random-walking down a gradient of their energy density \(u\) transport the net flux \(F=-\left(c\ell/3\right)\dd u/\dd r\) — the elementary mean-free-path transport argument of Kinetic Theory of Gases, the factor \(\tfrac13\) being the angular average. With matter and radiation locally equilibrated the photon energy density is that of a blackbody at the local temperature, \(u=aT^{4}\) (Equation (68.2)), and \(\ell=1/(\kappa\rho)\) defines the opacity \(\kappa\) as an absorption cross-section per unit mass. Substituting gives Equation (52.16).
Both hypotheses are extravagantly satisfied. In the solar interior \(\kappa\rho\sim10^{2}\,/\mathrm{m}\), so \(\ell\sim10^{-2}\,\mathrm{m}\) against a scale height of \(\sim10^{8}\,\mathrm{m}\): the anisotropy the flux represents, \(F/(cu)\), is of order \(10^{-11}\) (computed here with \(F\approx2.5\times 10^{8}\,\mathrm{W}/\mathrm{m}^{2}\) and \(T\approx5\times 10^{6}\,\mathrm{K}\) at mid-radius) — the radiation field is isotropic and Planckian to that accuracy, and the diffusion approximation is among the best in physics. Because real opacity depends on frequency, the \(\kappa\) appearing here must be the Rosseland mean: the transport-weighted harmonic mean, \(1/\kappa=\avg{\kappa_{\nu}^{-1}}\) weighted by \(\pp B_{\nu}/\pp T\), since the flux flows preferentially through the transparent frequency windows (the frequency-by-frequency statement is quoted; see [Kippenhahn:2012]).
∎Two opacity regimes recur. At high temperature the electrons are free and Thomson scattering dominates (Radiation and Scattering of Electromagnetic Waves): \(\kappa_{\mathrm{es}}=\sigma_{\mathrm{T}}(1+X)/2m_{\mathrm{u}} =0.034\,\mathrm{m}^{2}/\mathrm{kg}\) for hydrogen mass fraction \(X=0.7\) (computed here), independent of \(\rho\) and \(T\). At lower temperatures bound–free and free–free absorption dominate, with the law Kramers derived from his X-ray absorption theory, \(\kappa\propto\rho T^{-7/2}\) [Kramers:1923]. In cool stellar envelopes the opacity is instead governed by the H\(^{-}\) ion and rises steeply with temperature, which is what Section 52.3.3 will need.
Equation (52.16) also prices the journey: a photon random-walks \(N\sim(R/\ell)^{2}\) steps, taking \(t\sim R^{2}/(\ell c)\), which with the mean solar values above is of order \(10^{4}\,\mathrm{yr}\) (computed here), lengthening to order \(10^{5}\,\mathrm{yr}\) when the centrally concentrated run of \(\kappa\rho\) is used (quoted [Kippenhahn:2012]). Energy generated in the core today emerges from the photosphere in the far future; only the neutrinos of Section 52.5 leave in real time.
Radiative diffusion fails, and convection takes over, when the gradient it demands is too steep to be mechanically stable.
A chemically homogeneous stratification obeying Equation (52.2) is unstable to convection if and only if
where \(\nabla_{\mathrm{ad}}=1-1/\Gamma_{1}=2/5\) for a monatomic ideal gas [Schwarzschild:1906]. Radiative equilibrium is therefore stable exactly where
\(\nabla_{\mathrm{rad}}\) being the gradient Equation (52.17) written per unit of \(\ln P\) using Equation (52.2). Rests on Equations (52.2) and (52.17).
Derives Theorem 52.10. Displace a fluid parcel upward by \(\delta r\), slowly enough to stay in pressure balance with its surroundings (the sound-crossing time of a parcel is far shorter than any transport time) yet fast enough to exchange no heat (the radiative diffusion time across it is far longer); both inequalities hold by margins of many orders of magnitude, which is what makes the adiabatic parcel an accurate idealization. Moving adiabatically at equal pressure, its density becomes
with \(\Gamma_{1}=(\pp\ln P/\pp\ln\rho)_{s}\), while the ambient density at the new position is \(\rho+(\dd\rho/\dd r)\delta r\). The parcel keeps rising — instability — precisely when it arrives lighter than its surroundings:
For an ideal gas of uniform composition, \(\ln\rho=\ln P-\ln T+\text{const}\), and dividing by \(\dd\ln P/\dd r<0\) reverses the inequality, turning it into \(\nabla>1-1/\Gamma_{1}=\nabla_{\mathrm{ad}}\), which is Equation (52.18). In the stable case the restoring buoyancy makes the parcel oscillate at the Brunt–Väisälä frequency
positive exactly when Equation (52.18) fails; these are the buoyancy (g-)modes of a stellar interior, the acoustic counterpart being Section 52.6. In the unstable case the same expression is a growth rate, of order the dynamical time — so a superadiabatic stratification overturns essentially instantly on stellar timescales, and a convection zone sits pinned at \(\nabla\) only infinitesimally above \(\nabla_{\mathrm{ad}}\) except in its outermost, tenuous layers.
∎Equation (52.19) locates the convection zones. \(\nabla_{\mathrm{rad}}\) is large where the opacity is large or where \(L_{r}/m\) is large. Cool envelopes satisfy the first condition — in the Sun, hydrogen's partial ionization both raises \(\kappa\) enormously (H\(^{-}\)) and depresses \(\nabla_{\mathrm{ad}}\) below \(2/5\) (energy fed into ionization rather than temperature), so the outer envelope convects, from \(0.71\,R_{\odot}\) to the surface as helioseismology locates it (Section 52.6.2). Stars burning on the CNO cycle satisfy the second: Section 52.2.3 concentrates \(\varepsilon\) so sharply that \(L_{r}\) saturates near the centre, and such stars have convective cores instead. The two configurations exchange roles across the mass scale, and the difference propagates into everything from surface abundances to main-sequence lifetimes.
What convection actually carries, once running, is modelled — not derived. Böhm-Vitense's mixing-length theory [BoehmVitense:1958] posits parcels that rise a distance \(\ell_{\mathrm{m}}=\alpha_{\mathrm{MLT}}H_{P}\) (\(H_{P}\) the pressure scale height) before dissolving, and carries the flux implied by the buoyancy work over that distance. The single free parameter \(\alpha_{\mathrm{MLT}}\approx1.5\)–\(2\) is calibrated on the Sun, and this honest phenomenological confession sits inside every stellar evolution code. Deep convection is so efficient that the result barely matters there (\(\nabla\to\nabla_{\mathrm{ad}}\), an isentrope, whatever \(\alpha_{\mathrm{MLT}}\)); it matters within the superadiabatic surface layers, and through them the stellar radius. The turbulence being modelled belongs to Fluid Dynamics.
The mass–luminosity relation and the Eddington limit
With hydrostatics and radiative transport in hand, the structure equations yield their most consequential scaling law before any of them is solved.
Among main-sequence stars whose masses are measured directly, from the orbits of eclipsing and visual binaries, the luminosity rises as a steep power of the mass,
across the bulk of the main sequence [Eddington:1924b]. A star of ten solar masses is some thousands of times the more luminous. Set beside the fuel supply of Phenomenon 52.4, which grows only in proportion to the mass, this makes massive stars enormously short-lived, and it is what turns the main-sequence turnoff of a star cluster into a clock. Rests on Equation (52.2), Proposition 52.1, Equation (52.17) and Proposition 52.9.
Derivation. Derives Phenomenon 52.12. The exponent follows from three of the structure equations without any of them being solved, by comparing the scalings of their terms — a homology argument. Write \(M\) and \(R\) for the mass and radius and \(P\), \(\rho\sim M/R^{3}\), \(T\) for characteristic interior values.
Hydrostatic equilibrium (Equation (52.2)) scales as
For an ideal gas of mean molecular weight \(\mu\), that is \(P=\rho k_{\mathrm{B}}T/\left(\mu m_{\mathrm{u}}\right)\), and eliminating \(P\) and \(\rho\) leaves
Radiative diffusion carries the energy out (Equation (52.17)), which scales as
Substituting Equation (52.22), every factor of \(R\) cancels:
For an opacity independent of the local conditions — electron scattering, which dominates in hot interiors — this is \(L\propto M^{3}\), with no free parameter and no dependence on the radius at all. The observed exponent runs somewhat above three because at lower masses the opacity instead follows Kramers' law \(\kappa\propto\rho T^{-7/2}\) [Kramers:1923], which steepens the dependence.
The content of Equation (52.23) is worth stating plainly, because it is counter-intuitive: the luminosity of a main-sequence star is fixed by its mass and composition and does not refer to the energy source at all. A star shines as brightly as its own weight compels it to, and the nuclear reaction rate adjusts, by way of the central temperature, to supply exactly that.
∎The same radiation that carries the luminosity pushes on the matter carrying it, and there is a luminosity at which the push wins.
For fully ionized hydrogen, hydrostatic equilibrium is impossible above
(computed here) [Eddington:1926]. Rests on Equation (52.16).
Derives Proposition 52.13. A radiation flux \(F\) carries momentum flux \(F/c\), and an electron intercepts it with cross-section \(\sigma_{\mathrm{T}}\), feeling an outward force \(\sigma_{\mathrm{T}}F/c\); the protons, electrostatically tied to the electrons, supply almost all the weight, \(Gm_{\mathrm{p}} M/r^{2}\) per electron. With \(F=L/4\pi r^{2}\), both forces fall as \(1/r^{2}\), so their ratio is one number for the whole star:
At \(L>L_{\mathrm{Edd}}\) the effective gravity is outward at every radius and no hydrostatic atmosphere exists: the star sheds mass instead. Evaluating with the constants of Appendix B gives Equation (52.24).
∎Setting Equation (52.23) against Equation (52.24): \(L\) grows as \(M^{3}\) but \(L_{\mathrm{Edd}}\) only as \(M\), and the two meet near \(180\,M_{\odot}\) (computed here from the calibration \(L=L_{\odot}(M/M_{\odot})^{3}\)). Stars approaching that mass are radiation-pressure-dominated (Remark 52.8), loosely bound and violently unstable to mass loss, which is why the observed main sequence ends at masses of order \(10^{2}\,M_{\odot}\). The same limit, applied to accretion instead of fusion, caps the luminosity of matter falling onto compact objects and calibrates the brightest X-ray sources and quasars (Experiment: Black-Hole Observations).
Nuclear energy generation
Tunnelling and the Gamow peak
The temperature that hydrostatics dictates (Remark 52.8) is, by the standards of nuclear physics, absurdly cold. Two protons must approach to a few femtometres to fuse, against a Coulomb barrier of \(720\,\mathrm{keV}\) (Equation (107.99)), while \(k_{\mathrm{B}}T_{\mathrm{c}}=1.35\,\mathrm{keV}\): the barrier stands five hundred times above the mean thermal energy, and the Boltzmann factor for surmounting it classically, \(\ee^{-530}\sim10^{-230}\), forbids fusion in any classical gas of any astrophysical size. The resolution is quantum tunnelling through the barrier [Gamow:1928], applied to stellar interiors by Atkinson and Houtermans within a year [Atkinson:1929]. The machinery — the Gamow factor, the astrophysical \(S\) factor, and the thermal average over the Maxwell–Boltzmann tail — is built in Nuclear Forces and Nuclear Structure, whose central result, Theorem 107.85, this section uses as is: the reaction rate is dominated by a narrow window of collision energies (the Gamow peak) centred on \(E_{0}=\left(E_{G}(k_{\mathrm{B}}T)^{2}/4\right)^{1/3}\) (Equation (107.101)), of width \(\Delta E=4\sqrt{E_{0}k_{\mathrm{B}}T/3}\) (Equation (107.102)), where \(E_{G}\) is the Gamow energy of the colliding pair. For the Sun the worked numbers are Example 107.86: protons fuse at \(E_{0}\approx6\,\mathrm{keV}\), from far out on the thermal tail.
What this chapter must add is the temperature sensitivity, because everything structural — the segregation of burning stages, the pp/CNO divide, the existence of convective cores, the stability of the whole configuration — hangs on it.
With \(S(E)\) slowly varying, the thermally averaged rate per pair behaves as
so its local logarithmic slope is
Derives Proposition 52.14. The thermal average is \(\avg{\sigma v}\propto(k_{\mathrm{B}}T)^{-3/2}\int S(E)\, \ee^{-E/k_{\mathrm{B}}T-\sqrt{E_{G}/E}}\,\dd E\), and the saddle-point evaluation performed in the proof of Theorem 107.85 gives for the integral a Gaussian of height \(S(E_{0})\,\ee^{-3E_{0}/k_{\mathrm{B}}T}\) (the exponent at the peak is \(E_{0}/k_{\mathrm{B}}T+\sqrt{E_{G}/E_{0}} =3E_{0}/k_{\mathrm{B}}T\), using \(\sqrt{E_{G}/E_{0}}=2E_{0}/k_{\mathrm{B}}T\) from Equation (107.101)) and of width \(\Delta E\propto\sqrt{E_{0}k_{\mathrm{B}}T}\) (Equation (107.102)). Since \(E_{0}\propto T^{2/3}\), the width scales as \(T^{5/6}\), and the prefactors assemble to \(T^{-3/2}\times T^{5/6}=T^{-2/3}\), which is Equation (52.25). Differentiating: \(\tau=3E_{0}/k_{\mathrm{B}}T\propto T^{-1/3}\), so \(\dd\tau/\dd\ln T=-\tau/3\) and
| Reaction | $E_{G}$ (\(\mathrm{MeV}\)) | $E_{0}$ (\(\mathrm{keV}\)) | $\Delta E$ (\(\mathrm{keV}\)) | $\tau$ | $\nu$ |
|---|---|---|---|---|---|
| $p+p$ | 0.493 | 6.1 | 6.6 | 13.5 | 3.8 |
| $^{3}\mathrm{He}+{}^{4}\mathrm{He}$ | 26.9 | 23.1 | 12.9 | 51.2 | 16.4 |
| $p+{}^{7}\mathrm{Be}$ | 13.8 | 18.5 | 11.5 | 41.0 | 13.0 |
| $p+{}^{14}\mathrm{N}$ | 45.1 | 27.4 | 14.1 | 60.8 | 19.6 |
Table 52.1 turns Equation (52.26) into the architecture of stellar burning. The pp reaction runs as \(T^{4}\); the CNO bottleneck as \(T^{20}\); and higher-\(Z\) fuels, with their larger \(E_{G}\), switch on only at successively higher temperatures — which is why burning proceeds in discrete, well-separated stages (Section 52.2.4) rather than as a continuous smoulder. The steepness also closes the loop opened in Remark 52.3: a star that momentarily over-burns expands and cools, and a rate falling as the twentieth power of a falling temperature is a thermostat of extraordinary stiffness. A star is stable not despite the violence of its energy source but because of how sharply that source responds to the star's own mechanical state.
The pp chains
In the Sun, hydrogen burns predominantly by the proton–proton chains. The first step,
is a weak interaction — two protons form no bound state (Nuclear Forces and Nuclear Structure), so one must convert to a neutron during the fleeting collision — first computed by Bethe and Critchfield from Fermi's theory [Bethe:1938] (Weak Interactions). Its cross-section is some twenty orders of magnitude below direct measurement and is known only from theory; that a weak process gates the whole chain is why the Sun burns for \(10^{10}\) years instead of detonating, and the derivation of the total release \(Q=26.73\,\mathrm{MeV}\) per helium nucleus is Phenomenon 107.87 (Equation (107.104)). The deuteron is consumed within seconds, and the chain then completes along three branches.
| Branch | Reaction | $Q$ (\(\mathrm{MeV}\)) | $E_{\nu}$ (\(\mathrm{MeV}\)) |
|---|---|---|---|
| pp\,I | $p+p\to d+e^{+}+\nu_{e}$ | 1.442 | $\le0.420$, $\avg{E_{\nu}}=0.267$ |
| $d+p\to{}^{3}\mathrm{He}+\gamma$ | 5.493 | — | |
| $^{3}\mathrm{He}+{}^{3}\mathrm{He} \to{}^{4}\mathrm{He}+2p$ | 12.860 | — | |
| pp\,II | $^{3}\mathrm{He}+{}^{4}\mathrm{He} \to{}^{7}\mathrm{Be}+\gamma$ | 1.587 | — |
| $^{7}\mathrm{Be}+e^{-}\to{}^{7}\mathrm{Li}+\nu_{e}$ | 0.862 | 0.862 (90\%), 0.384 | |
| $^{7}\mathrm{Li}+p\to2\,^{4}\mathrm{He}$ | 17.346 | — | |
| pp\,III | $^{7}\mathrm{Be}+p\to{}^{8}\mathrm{B}+\gamma$ | 0.137 | — |
| $^{8}\mathrm{B}\to2\,^{4}\mathrm{He}+e^{+}+\nu_{e}$ | 18.07 | $\avg{E_{\nu}}\approx6.7$ |
The neutrinos are not detail: they are the audit trail. Each branch loses a different fraction of \(Q\) to neutrinos, which never thermalize (computed here from Table 52.2): pp\,I loses \(2\times0.267=0.53\,\mathrm{MeV}\), or \(2.0\,\mathrm{\%}\); pp\,II loses \(1.08\,\mathrm{MeV}\), \(4.0\,\mathrm{\%}\); pp\,III loses \(7.0\,\mathrm{MeV}\), fully \(26\,\mathrm{\%}\) of its \(Q\). Weighted by the branch shares, the Sun's neutrino loss is about \(2.3\,\mathrm{\%}\) (computed here): the photon luminosity corresponds to an effective \(26.1\,\mathrm{MeV}\) of heat per helium nucleus assembled. Two rare variants complete the bookkeeping: the pep reaction \(p+e^{-}+p\to d+\nu_{e}\), a monochromatic \(1.44\,\mathrm{MeV}\) line, and the hep reaction on \(^{3}\)He, the highest-energy solar neutrinos. Branch by branch, this entire table has now been read out of the Sun in real time by neutrino spectroscopy (Section 52.5.3; [Agostini:2018]), including the tiny pp\,III branch whose \(^{8}\)B neutrinos carried the solar-neutrino problem for thirty years (Section 52.5.1).
The CNO cycle
Where carbon, nitrogen and oxygen are present, hydrogen can also burn catalytically, in the cycle proposed independently by von Weizsäcker [vonWeizsaecker:1938] and worked out quantitatively by Bethe [Bethe:1939]:
| Reaction | $Q$ (\(\mathrm{MeV}\)) | $E_{\nu}$ (\(\mathrm{MeV}\)) |
|---|---|---|
| $^{12}\mathrm{C}+p\to{}^{13}\mathrm{N}+\gamma$ | 1.943 | — |
| $^{13}\mathrm{N}\to{}^{13}\mathrm{C}+e^{+}+\nu_{e}$ | 2.220 | $\avg{E_{\nu}}=0.707$ |
| $^{13}\mathrm{C}+p\to{}^{14}\mathrm{N}+\gamma$ | 7.551 | — |
| $^{14}\mathrm{N}+p\to{}^{15}\mathrm{O}+\gamma$ | 7.297 | — |
| $^{15}\mathrm{O}\to{}^{15}\mathrm{N}+e^{+}+\nu_{e}$ | 2.754 | $\avg{E_{\nu}}=0.997$ |
| $^{15}\mathrm{N}+p\to{}^{12}\mathrm{C}+{}^{4}\mathrm{He}$ | 4.965 | — |
The cycle's properties follow from Table 52.1. Its slowest link at stellar temperatures is \(^{14}\mathrm{N}(p,\gamma)^{15}\mathrm{O}\) — the largest Gamow energy among the captures, compounded by a small \(S\) factor — so in equilibrium the catalysts pile up as \(^{14}\)N: the CNO cycle is itself a nucleosynthesis process, and the cosmic abundance of nitrogen is largely its fossil [Burbidge:1957]. The same link sets the temperature exponent, \(\nu\approx20\) (Table 52.1), against \(\nu\approx4\) for pp. Two consequences:
First, the crossover. The pp rate, with its shallow slope, wins at low temperature; the CNO rate, at solar metallicity, overtakes it near \(T\approx1.7\times 10^{7}\,\mathrm{K}\) — just above the solar centre — corresponding to main-sequence masses above about \(1.3\,M_{\odot}\) (modern values, quoted [Kippenhahn:2012]). Bethe's own comparison, using the higher solar central temperature then accepted, placed the Sun itself on the CNO side [Bethe:1939]; improved solar models moved the divide just above it. The Sun sits below the divide: Borexino's direct detection of CNO neutrinos measures the cycle's contribution at about one per cent of the solar luminosity [Agostini:2020a], exactly the subordinate role the crossover predicts (Section 52.5.3).
Second, the core convection of Section 52.1.3. A \(T^{20}\) rate concentrates the burning into the innermost few per cent of the mass; \(L_{r}\) there saturates, \(\nabla_{\mathrm{rad}}\) of Equation (52.19) exceeds \(\nabla_{\mathrm{ad}}\), and every CNO-burning star carries a convective core — continuously refuelled and chemically homogenized, which lengthens its life and reshapes its track in Section 52.3.
Helium burning and beyond
Past helium the chart of nuclides has a double gap: no stable mass \(5\), no stable mass \(8\). Salpeter identified the two-step route through the unbound \(^{8}\)Be [Salpeter:1952], and Hoyle turned the observed abundance of carbon into the prediction of a nuclear excited state — the argument, stated from the nuclear-physics side in Remark 107.88, that this chapter records as a phenomenon because its input was astronomical.
Carbon is among the most abundant elements in the universe, yet the route to it is blocked twice: there is no stable nucleus of mass number \(5\) or \(8\), so helium burning must pass through the unstable \(^{8}\)Be, whose lifetime is of order \(10^{-16}\,\mathrm{s}\) [Salpeter:1952]. From the bare fact of carbon's abundance, Hoyle argued that the triple-alpha rate must be resonantly enhanced and therefore that \(^{12}\)C must possess a hitherto unknown excited state of spin-parity \(0^{+}\) lying just above the three-alpha threshold, near \(7.65\,\mathrm{MeV}\) of excitation. The state was then looked for, and found at that energy [Hoyle:1954]. It remains the clearest case in physics of a nuclear property predicted from an astronomical abundance. Rests on Definition 107.31 and Proposition 52.16.
Derivation. Derives Phenomenon 52.15. The threshold energy is fixed by the masses alone, and it is what makes the prediction quantitative. Three \(^{4}\)He nuclei have a combined atomic mass of \(3\times4.002602=12.007806\) in unified atomic mass units, while \(^{12}\)C has, by the definition of that unit, a mass of exactly \(12\). The excess is \(0.007806\) mass units, and with \(1\,\mathrm{u}=931.494\,\mathrm{MeV}/c^{2}\),
The three-alpha system therefore enters \(^{12}\)C at \(7.27\,\mathrm{MeV}\) of excitation, so the state Hoyle required at \(7.65\,\mathrm{MeV}\) sits \(0.38\,\mathrm{MeV}\) above the threshold.
That placement is the whole of the argument. Below the threshold the state could not act as a resonance for three free alpha particles at all. A few hundred \(\mathrm{keV}\) higher and the Boltzmann factor at helium-burning temperatures would have suppressed the rate, leaving a universe with very little carbon in it. The window is narrow, the abundance is large, and the state is where the abundance requires it to be — which is why the laboratory confirmation was taken as decisive rather than as a coincidence.
∎The resonance also fixes how helium burning responds to temperature, which is qualitatively unlike the nonresonant rates of Section 52.2.1.
With the capture fully resonant through the Hoyle state at \(E_{\mathrm{r}}=380\,\mathrm{keV}\) above the three-alpha threshold, the rate per unit volume behaves as
so its temperature exponent is \(\nu=E_{\mathrm{r}}/k_{\mathrm{B}}T-3\approx41\) at \(T=10^{8}\,\mathrm{K}\) (computed here). Rests on Equation (52.27).
Derives Proposition 52.16. For a resonance narrow on the thermal scale, the thermally averaged rate per pair follows from inserting a sharply peaked cross-section into the Maxwell–Boltzmann average (Kinetic Theory of Gases): the distribution contributes its value at the resonance, \(\propto(k_{\mathrm{B}}T)^{-3/2}\ee^{-E/k_{\mathrm{B}}T}\), and the energy integral over the resonance profile contributes a temperature-independent strength, so \(\avg{\sigma v}\propto T^{-3/2}\ee^{-E/k_{\mathrm{B}}T}\) with \(E\) the resonance energy in the collision frame.
The triple-alpha route applies this twice. First, \(\alpha+\alpha\rightleftharpoons{}^{8}\mathrm{Be}\): the \(^{8}\)Be ground state sits \(\Delta E_{1}=92\,\mathrm{keV}\) above two alphas, and because its formation and decay proceed through the same channel, the equilibrium population cancels the resonance strength entirely, leaving the Saha-like factor
Second, \(^{8}\mathrm{Be}+\alpha\to{}^{12}\mathrm{C}^{*}\) through the Hoyle state, which lies \(\Delta E_{2}=288\,\mathrm{keV}\) above the \(^{8}\mathrm{Be}+\alpha\) threshold: another factor \(T^{-3/2}\ee^{-\Delta E_{2}/k_{\mathrm{B}}T}\). The product is Equation (52.28) with \(E_{\mathrm{r}}=\Delta E_{1}+\Delta E_{2}=380\,\mathrm{keV}\) — necessarily the Hoyle state's height above three free alphas, Equation (52.27) subtracted from \(7.65\,\mathrm{MeV}\). Differentiating logarithmically: \(\nu=E_{\mathrm{r}}/k_{\mathrm{B}}T-3\), and at \(T=10^{8}\,\mathrm{K}\), where \(k_{\mathrm{B}}T=8.62\,\mathrm{keV}\), this is \(44-3=41\).
∎A rate scaling as \(T^{41}\) (and as the cube of the density) is effectively a switch: helium ignites abruptly near \(10^{8}\,\mathrm{K}\) and nowhere cooler. In stars below about \(2\,M_{\odot}\) the helium core reaches that temperature only after becoming electron-degenerate, and ignition under degeneracy is explosive — pressure barely responds to temperature, the thermostat of Remark 52.3 is disconnected, and the burning runs away as the helium flash until degeneracy is lifted (Quantum Statistics and Compact Stars and Relativistic Astrophysics).
Beyond the triple-alpha reaction, successive \(\alpha\) captures climb the ladder \(^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O}\) and onward. The carbon-to-oxygen ratio of the universe is set by the competition between the triple-alpha rate and the \(^{12}\mathrm{C}(\alpha,\gamma)\) rate, whose \(S\) factor at stellar energies remains among the least certain numbers in nuclear astrophysics — an honest caveat carried by every model of white dwarf composition. The later stages then follow in the sequence Table 52.4, each ignited by the next Coulomb barrier per Proposition 52.14, each faster than the last: partly because less energy remains per reaction (Phenomenon 107.32), and mainly because from carbon burning onward the core's thermal neutrino emission, not photon diffusion, carries away the power, decoupling the burning from the surface luminosity entirely.
| Stage | Principal products | $T$ (\(\mathrm{K}\)) | Duration |
|---|---|---|---|
| Hydrogen | $^{4}$He, $^{14}$N | $\sim4\times 10^{7}$ | $\sim$\,\(10^{7}\,\mathrm{yr}\) |
| Helium | $^{12}$C, $^{16}$O | $\sim2\times 10^{8}$ | $\sim$\,\(10^{6}\,\mathrm{yr}\) |
| Carbon | $^{20}$Ne, $^{23}$Na, $^{24}$Mg | $\sim8\times 10^{8}$ | $\sim$\,\(10^{3}\,\mathrm{yr}\) |
| Neon | $^{16}$O, $^{24}$Mg | $\sim1.5\times 10^{9}$ | $\sim$\,\(1\,\mathrm{yr}\) |
| Oxygen | $^{28}$Si, $^{32}$S | $\sim2\times 10^{9}$ | $\sim$\,\(0.3\,\mathrm{yr}\) |
| Silicon | iron-peak nuclei | $\sim3.5\times 10^{9}$ | $\sim$\,\(10^{-2}\,\mathrm{yr}\) |
Silicon burning is not a fusion of two silicon nuclei — that barrier is out of reach — but a photodisintegration rearrangement: at \(3.5\times 10^{9}\,\mathrm{K}\) the thermal photons strip nucleons and alphas off nuclei and the ensemble relaxes toward nuclear statistical equilibrium, which populates the binding-energy maximum (Proposition 107.41). There the energy release stops, by Phenomenon 107.32: an iron core can only absorb energy, by photodisintegration and electron capture, and its collapse — with everything that follows, neutron stars, supernovae and stellar black holes — is where this chapter hands the star to Compact Stars and Relativistic Astrophysics.
The Hertzsprung–Russell diagram and stellar evolution
The diagram
Everything observable about the surface of a distant star reduces, to first approximation, to two numbers: a luminosity and a surface temperature (measured as a colour or a spectral type). Plotting stars in that plane is due to Hertzsprung [Hertzsprung:1911] and, independently, Russell [Russell:1914], and the result is the single observable summary against which every stellar model is judged.
Stars do not fill the luminosity–temperature plane. About ninety per cent of them lie on one narrow band — the main sequence — running diagonally from hot and luminous to cool and faint; a separate population of giants sits cool yet luminous; white dwarfs sit hot yet faint, thousands of times below the main sequence [Hertzsprung:1911] [Russell:1914]. The bands are loci, not scatter: a star's position determines its other properties, and clusters of common age populate the diagram in sharply truncated patterns. Rests on Equation (68.2), Equation (52.23) and Phenomenon 52.12.
Derivation. Derives Phenomenon 52.17. The plane has a hidden third coordinate. A star radiates approximately as a blackbody, so (Equation (68.2))
and lines of constant radius are diagonals of slope four in \(\log L\) against \(\log T_{\mathrm{eff}}\). Every feature of the diagram is a statement about radii.
The main sequence is the locus of hydrogen burners, traced by mass. Homology fixes both of its coordinates as functions of \(M\) alone. The luminosity is Equation (52.23), \(L\propto M^{3}\). The radius follows from energy balance: the nuclear power \(L\propto M\bar{\rho}\,T^{\nu} \propto M^{2+\nu}R^{-3-\nu}\) must equal the transported \(L\propto M^{3}\), so
which for pp burning (\(\nu\approx4\), Table 52.1) is \(R\propto M^{0.43}\). Then Equation (52.29) gives \(T_{\mathrm{eff}}^{4}\propto M^{3}/R^{2}\propto M^{2.14}\), and eliminating \(M\),
(computed here): a single steep curve, parametrized by mass, with essentially no width — because homologous stars of one composition have no free parameter left. The main sequence is a mass sequence, not an age sequence.
The other bands are radius excursions. Giants are stars of ordinary mass at \(10\)–\(10^{2}\,R_{\odot}\) (Section 52.3.3); white dwarfs are stars of ordinary mass at \(\sim10^{-2}\,R_{\odot}\), the degenerate configurations of Compact Stars and Relativistic Astrophysics. Their separation from the main sequence by orders of magnitude in \(L\) at fixed \(T_{\mathrm{eff}}\) is just Equation (52.29) read along a vertical line. And a cluster — a population of one age and composition — fills the main sequence only up to the mass whose lifetime (Section 52.3.2) equals the cluster's age, which is the truncation observed.
∎Main-sequence lifetimes
The fuel scales as \(M\); the burn rate as \(M^{3}\) to \(M^{4}\) (Phenomenon 52.12). So the main-sequence lifetime,
falls as roughly \(M^{-2.5}\): \(10^{10}\,\mathrm{yr}\) for the Sun (Equation (52.6)), but only about \(3\times 10^{7}\,\mathrm{yr}\) at \(10\,M_{\odot}\) (computed here with \(\alpha=3.5\)). The most massive stars live for astronomically negligible times, which is why O stars are found only in regions of active star formation, and why the main-sequence turnoff of a cluster (Phenomenon 52.17) is a clock: reading the turnoff mass and applying Equation (52.32) dates the cluster. The oldest globular clusters date this way to about thirteen thousand million years (quoted; [Kippenhahn:2012]) — a lower bound on the age of the universe that cosmology must and does clear (Evidence-Based Cosmology).
What ends the main-sequence phase is not the exhaustion of hydrogen everywhere but the mechanics of the helium ash at the centre.
An isothermal, non-degenerate helium core of mass \(M_{\mathrm{c}}\) and molecular weight \(\mu_{\mathrm{c}}\), at temperature \(T_{\mathrm{c}}\) set by the surrounding envelope, can support that envelope only if its surface pressure does not exceed
\(c_{1}\) a constant of order unity; matching to the envelope demand caps the core mass fraction at
about ten per cent for a helium core beneath a hydrogen envelope (coefficient quoted from [Schoenberg:1942] [Kippenhahn:2012]). Rests on Theorem 52.2.
Derives Proposition 52.18. Repeat the integration by parts of Theorem 52.2 over the core alone: the boundary term no longer vanishes, leaving the virial theorem with a surface term,
the first equality for an isothermal ideal gas, \(c_{0}\) of order unity encoding the core's density profile. As a function of the core radius this is \(P_{\mathrm{s}}=A/R_{\mathrm{c}}^{3}-B/R_{\mathrm{c}}^{4}\) with \(A\propto M_{\mathrm{c}}T_{\mathrm{c}}/\mu_{\mathrm{c}}\) and \(B\propto GM_{\mathrm{c}}^{2}\): zero for a large diffuse core, zero again for a compressed one (gravity wins), with a maximum in between at \(R_{\mathrm{c}}^{*}=4B/3A\) of height \(P_{\max}=27A^{4}/256B^{3}\), which is Equation (52.33). The remarkable feature is the denominator: the more massive the core, the less pressure it can exert.
The envelope, meanwhile, demands of the core's surface a pressure of the star's own interior order, \(P_{\mathrm{env}}\sim GM^{2}/R^{4}\), at a temperature it also dictates, \(k_{\mathrm{B}}T_{\mathrm{c}}\sim G\mu_{\mathrm{env}}m_{\mathrm{u}}M/R\) (Equation (52.22)). Substituting both into \(P_{\max}\ge P_{\mathrm{env}}\), all factors of \(R\) cancel:
which is Equation (52.34) up to the coefficient, obtained by integrating the isothermal core structure exactly [Schoenberg:1942]. With \(\mu_{\mathrm{env}}\approx0.6\) and \(\mu_{\mathrm{c}}\approx1.3\), \(q_{\mathrm{SC}}\approx0.08\) (computed here from Equation (52.34)).
∎When shell burning pushes the helium core past \(q_{\mathrm{SC}}\), no isothermal equilibrium remains: the core contracts on its own Kelvin–Helmholtz timescale, and the star leaves the main sequence. (Below about \(2\,M_{\odot}\) the core evades the limit by going degenerate — degeneracy pressure needs no temperature — which is why low-mass stars linger as subgiants while more massive stars cross the gap quickly.) The envelope's response to core contraction is expansion — the “mirror” behaviour every detailed calculation shows [Kippenhahn:2012] — and the star moves right in the diagram, toward the giant region and the track derived next.
Before and after the main sequence
A star arriving at or leaving the main sequence spends its time fully or largely convective, and for such stars there is a wall in the HR diagram.
A fully convective star whose photospheric opacity follows the H\(^{-}\) law \(\kappa=\kappa_{0}\rho^{1/2}T^{9}\) (quoted fit, valid near \(4\times 10^{3}\,\mathrm{K}\) [Kippenhahn:2012]) satisfies
so its effective temperature is nearly independent of its luminosity: fully convective stars of given mass form an almost vertical line in the HR diagram, and the region cooler than that line is forbidden [Hayashi:1961]. Rests on Theorems 52.6 and 52.10.
Derives Proposition 52.19. Four ingredients, each one line.
(i) The interior is an \(n=3/2\) polytrope. Efficient convection enforces \(\nabla=\nabla_{\mathrm{ad}}\) (Theorem 52.10 and Remark 52.11): the star is isentropic, \(P=K\rho^{5/3}\), and the Lane–Emden mass relation Equation (52.12) with the scale Equation (52.8) fixes the adiabat's constant in terms of the bulk parameters: eliminating \(\rho_{\mathrm{c}}\) between \(M\propto K^{3/2}\rho_{\mathrm{c}}^{1/2}\) and \(R\propto K^{1/2}\rho_{\mathrm{c}}^{-1/6}\) leaves \(K\propto GM^{1/3}R\).
(ii) The adiabat in \((P,T)\). With the ideal-gas law \(\rho=\mu m_{\mathrm{u}}P/k_{\mathrm{B}}T\) substituted into \(P=K\rho^{5/3}\),
(iii) The photosphere. Hydrostatic equilibrium in the atmosphere, \(\dd P=(g/\kappa)\,\dd\tau\) per unit optical depth, integrates to \(P_{\mathrm{ph}}\,\kappa_{\mathrm{ph}}\simeq \tfrac{2}{3}\,g=\tfrac{2}{3}\,GM/R^{2}\) at the photosphere \(\tau=\tfrac23\). Writing the opacity law in \((P,T)\) via the ideal-gas law, \(\kappa\propto\rho^{1/2}T^{9}\propto P^{1/2}T^{17/2}\), this is \(P_{\mathrm{ph}}^{3/2}\,T_{\mathrm{eff}}^{17/2}\propto M/R^{2}\).
(iv) Match. The photosphere lies on the interior adiabat, so substituting (ii) into (iii):
Eliminating \(R\propto L^{1/2}T_{\mathrm{eff}}^{-2}\) (Equation (52.29)) gives \(T_{\mathrm{eff}}^{51}\propto M^{7}L^{1/2}\), which is Equation (52.35) (algebra verified numerically, computed here).
The exponent \(1/102\) is the result: as \(L\) changes a thousandfold, \(T_{\mathrm{eff}}\) moves by \(1000^{1/102}\approx7\) per cent (computed here). The forbidden region follows because a star pushed cooler than this track would have to be superadiabatic throughout its convective envelope, which convection erases on a dynamical timescale (Theorem 52.10). The extreme steepness of the H\(^{-}\) opacity in \(T\) is what pins the wall in place.
∎Proposition 52.19 organizes both ends of a star's life. A collapsing protostar becomes fully convective and descends the Hayashi line at nearly constant \(T_{\mathrm{eff}}\) (the Hayashi phase), then, as its interior heats and turns radiative, peels off toward the main sequence — reaching it on the Kelvin–Helmholtz timescale Equation (52.5), which is why \(3\times 10^{7}\,\mathrm{yr}\) is the right answer to a question Kelvin asked about the wrong energy source. After the main sequence, the sequence of Proposition 52.18 runs: the helium core contracts, hydrogen burns in a shell, the envelope expands and cools until it meets the Hayashi line, and the star, unable to cross it, is deflected upward along it — the red giant branch, at ever-growing radius and luminosity. Helium ignition follows (Proposition 52.16; as a flash below \(\sim2\,M_{\odot}\)), then core helium burning, and then the double-shell asymptotic giant branch, whose thermal pulses alternately ignite the two shells and drive the dredge-up episodes that carry freshly synthesized material to the surface — the fact that Phenomenon 52.21 turns into proof of nucleosynthesis in situ. Mass loss then strips the envelope (planetary nebula), leaving the carbon–oxygen white dwarf; stars above about \(8\,M_{\odot}\) instead burn through Table 52.4 to core collapse. Both endpoints, and the remnants they leave, are Compact Stars and Relativistic Astrophysics; the detailed evolutionary calculations behind this narrative are [Kippenhahn:2012].
Stellar nucleosynthesis
The abundance record
Nucleosynthesis theory has one empirical target: the measured abundances of the elements, compiled from the solar photosphere, primitive meteorites and stellar spectra into a single curve by Suess and Urey [Suess:1956], with the modern solar photospheric revision by Asplund and collaborators [Asplund:2009]. The light-element floor — hydrogen, helium, and the lithium plateau — is set earlier, by big-bang nucleosynthesis, and belongs to Evidence-Based Cosmology; everything above it is made in stars.
Element abundances measured in the solar photosphere, in primitive meteorites and in stellar spectra are not arbitrary: plotted against mass number they trace one reproducible curve with definite features [Suess:1956] [Asplund:2009]. Abundance falls by some ten orders of magnitude from hydrogen to the heaviest nuclei, but the fall is interrupted by a pronounced maximum at the iron group near mass number \(56\); nuclei that are whole multiples of the alpha particle stand systematically above their neighbours; and beyond iron there are two distinct pairs of peaks, offset from one another. This curve is the empirical target of every theory of the origin of the elements, and it is against it that the processes of [Burbidge:1957] were defined. Rests on Phenomenon 107.32 and Equation (107.66).
Derivation. Derives Phenomenon 52.20. The iron peak, the most conspicuous feature, follows from the binding energies alone. The binding energy per nucleon \(B/A\) rises steeply among the lightest nuclei, where surface effects and nucleon pairing dominate; reaches a broad maximum of about \(8.8\,\mathrm{MeV}\) in the iron-nickel region, near \(A=56\) to \(62\); and then declines slowly as the Coulomb repulsion of the growing proton number takes over (Nuclear Forces and Nuclear Structure). Fusing two nuclei releases energy exactly when it raises \(B/A\), so a self-gravitating fusion reactor liberates energy while it burns toward the iron group and would have to absorb energy to proceed past it.
Two consequences follow, and both are visible in the curve. Material processed through successive burning stages piles up where the energy release stops, which is the iron peak. And elements beyond iron cannot be made by fusion in a star at all: they require neutron capture, which faces no Coulomb barrier and so proceeds at any temperature, followed by beta decay [Burbidge:1957]. Whether capture is slow or fast compared with the intervening beta decays selects a different path across the chart of nuclides, and the two paths stall at different closed neutron shells — which is why the peaks beyond iron come in offset pairs rather than singly.
∎B$^{2}$FH and the eight processes
Burbidge, Burbidge, Fowler and Hoyle organized the synthesis of every nuclide into eight processes, each defined by the feature of Phenomenon 52.20 it exists to explain [Burbidge:1957]; Cameron's independent contemporaneous framework reached essentially the same taxonomy [Cameron:1957]. Table 52.5 lists them with their modern identifications.
| Process | Mechanism and site | Principal products |
|---|---|---|
| H burning | pp chains and CNO cycle (Sections 52.2.2 and 52.2.3) | $^{4}$He, $^{14}$N |
| He burning | triple-alpha and $\alpha$ captures (Section 52.2.4) | $^{12}$C, $^{16}$O |
| $\alpha$ | advanced burning stages (Table 52.4) | $^{20}$Ne to $^{40}$Ca |
| e | nuclear statistical equilibrium in silicon burning and supernovae | iron peak |
| s | slow neutron capture in AGB stars (Section 52.3.3) | heavy nuclides along the valley; peaks at $A\approx88$, $138$, $208$ |
| r | rapid neutron capture in explosive, neutron-rich events (Section 52.4.3) | neutron-rich heavies; peaks at $A\approx80$, $130$, $195$; actinides |
| p | photodisintegration of heavy seeds in supernovae | rare proton-rich isotopes |
| x | light-element synthesis, left open in 1957; now cosmic-ray spallation (Phenomenon 115.16) and big-bang nucleosynthesis (Evidence-Based Cosmology) | Li, Be, B |
The neutron-capture split deserves its derivation, because the abundance curve records it directly. A nucleus exposed to a neutron flux captures at a rate \(1/\tau_{n}\) set by the flux, and beta-decays toward stability at a rate \(1/\tau_{\beta}\). The two limits draw different paths across the chart of nuclides:
Slow (\(\tau_{n}\gg\tau_{\beta}\)): every unstable capture product decays before the next capture, so the path hugs the valley of beta stability. At the magic neutron numbers \(N=50\), \(82\), \(126\) (Theorem 107.52) the capture cross-sections drop sharply — a closed shell holds the next neutron only weakly — so material accumulates just at the magic nuclides of the valley: abundance peaks at \(A\approx88\), \(138\), \(208\). The site is the He-burning shells of AGB stars, where \((\alpha,n)\) reactions on \(^{13}\)C and \(^{22}\)Ne release the neutrons; Phenomenon 52.21 is the proof that this runs where the theory puts it.
Rapid (\(\tau_{n}\ll\tau_{\beta}\)): captures outrun decay and drive the material far to the neutron-rich side, until binding is so weak that photodisintegration balances capture. The path then climbs along waiting points at the same magic \(N\) — but on the neutron-rich side, where a given \(N\) is reached at lower \(A\). When the flux ceases, the nuclei decay back to the valley at roughly constant \(A\), leaving peaks shifted below the s-process peaks: \(A\approx80\), \(130\), \(195\). The twin offset peaks of Phenomenon 52.20 are therefore a direct spectrogram of the magic numbers, read twice at two neutron excesses — and the offset measures how far from stability the r-process ran.
The r-process demands an environment of extreme neutron density on a timescale of seconds. Burbidge, Burbidge, Fowler and Hoyle assigned it to supernovae; the one site now observed is the neutron-star merger of Section 52.4.3, with core-collapse supernovae remaining a possible contributor — an honest open fraction.
Direct evidence
The theory above is constrained by the abundance curve it reproduces, but two observations catch nucleosynthesis in the act.
The absorption spectra of certain cool giants of spectral class S contain the lines of technetium [Merrill:1952]. Technetium has no stable isotope, and the longest-lived of them has a half-life of about \(4\times10^{6}\) years — three orders of magnitude shorter than the lifetime of the star, and four shorter than the age of the Galaxy. The element cannot have been present when the star formed, and there is no mechanism by which it could have been acquired from outside. It is therefore being manufactured in the interior now and carried to the surface: direct observational proof that nucleosynthesis happens inside stars, published five years before the systematic theory [Burbidge:1957]. Rests on Phenomenon 107.2 and Proposition 52.19.
Derivation. Derives Phenomenon 52.21. Radioactive decay is exponential, \(N(t)=N_{0}2^{-t/t_{1/2}}\), so with \(t_{1/2}=4\times10^{6}\) years any technetium present when a star of age \(t=10^{9}\) years formed would by now have been reduced by the factor
This is not merely a small number: a solar mass holds about \(10^{57}\) nucleons, so the surviving population would be less than one atom in the entire star. The observed lines cannot be a relic of the star's formation under any assumption about its initial composition.
Since the atmosphere is far too cool for nuclear reactions to run there, the technetium must be produced in the deep interior and transported outward on a timescale short compared with \(t_{1/2}\). That requirement — a dredge-up operating within a few million years — is itself a constraint on the convective structure of evolved giants, and it is met by the thermal pulses of the asymptotic giant branch (Section 52.3.3). The observation therefore does double duty: it establishes in-situ synthesis and it dates the mixing.
∎The r-process has its own caught-in-the-act moment, sixty-five years later. The binary neutron-star merger GW170817 (Section 47.4) was followed by a kilonova whose optical and infrared light curve decayed on the timescales of freshly synthesized radioactive heavy nuclei — the thermal glow of an r-process event of about \(10^{-2}\) solar masses of ejecta, observed while it happened [Abbott:2017]. Technetium in a giant's atmosphere and radioactive heavy elements in merger debris bracket the two neutron-capture processes of Table 52.5: both have now been watched running. On the longest timescales the same processes write galactic history — each stellar generation enriches the gas that forms the next, which is why the ancient halo stars are metal-poor while the young disc is metal-rich, the abundance–age correlation already read as evolutionary in [Burbidge:1957].
Solar neutrinos
The chlorine experiment and the solar-neutrino problem
Everything in Sections 52.1 and 52.2 rests on physics happening at the centre of a star, under \(2\times 10^{16}\,\mathrm{Pa}\) of overburden that no photon crosses in less than millennia (Section 52.1.3). The neutrinos of Table 52.2 are the one probe that leaves the core at the speed of light, and Davis built the detector: six hundred tonnes of perchloroethylene deep underground, counting individual argon atoms produced by \(\nu_{e}+{}^{37}\mathrm{Cl}\to e^{-}+{}^{37}\mathrm{Ar}\) [Davis:1968]. The apparatus — installed \(1478\,\mathrm{m}\) below ground in the Homestake mine — with its procedure and data is registered with the other oscillation experiments in Section 114.2.2; here the Sun's side of the story is told. The chlorine threshold of \(0.814\,\mathrm{MeV}\) puts the dominant pp neutrinos out of reach: the measured rate is carried mainly by the rare \(^{8}\)B branch of Table 52.2, with a minority from \(^{7}\)Be — precisely the branches most sensitive to the model. Against Bahcall's contemporaneous standard-solar-model prediction [Bahcall:1968], Davis's first result was already low; twenty-five years of running settled the measured rate at \(2.56\pm0.23\) SNU, one SNU being \(10^{-36}\) captures per target atom per second, against a predicted \(8.1\pm1.2\) (Table 114.4; [Cleveland:1998] [Bahcall:2005]) — a ratio of \(0.32\pm0.05\), quoted.
A radiochemical detector deep underground, counting the argon atoms produced by neutrino capture on chlorine, measured a solar electron-neutrino rate about one third of the standard solar model's prediction [Davis:1968] [Bahcall:1968]. Every later experiment confirmed a deficit, and it stood for three decades: either the Sun, or the model of it, or the neutrino was misunderstood. The resolution came from a detector sensitive to both the electron-flavour flux and the flavour-blind total. The total agreed with the solar model, while the electron-flavour fraction was depleted by just the observed factor [Ahmad:2002]. The Sun was right and the stellar model was right; the neutrinos change flavour in flight. What had looked for thirty years like a failure of astrophysics became a precision confirmation of it — and, incidentally, the discovery that neutrinos have mass (Flavour Physics and Neutrinos). Rests on Equation (107.101), Equation (114.6), Proposition 52.14 and Phenomenon 52.25.
Derivation. Derives Phenomenon 52.22. Two quantitative points carried the thirty-year argument: why the deficit could be blamed on astrophysics for so long, and what measurement could not be so blamed.
The \(^{8}\)B flux is a thermometer of vicious sensitivity. The branch is fed through \(^{3}\mathrm{He}+{}^{4}\mathrm{He}\) (temperature exponent \(\nu=16.4\), Table 52.1), which builds the \(^{7}\)Be supply, and drained through \(p+{}^{7}\mathrm{Be}\) (\(\nu=13.0\)); the competing destruction of \(^{7}\)Be by electron capture is a weak process with almost no temperature dependence. At fixed density and composition the flux therefore responds with the sum of the two Gamow exponents, \(\Phi(^{8}\mathrm{B})\propto T_{\mathrm{c}}^{\,29}\) or so (computed here; detailed solar-model perturbation studies, which let the structure readjust self-consistently, soften the exponent somewhat without changing what follows). The lever is enormous: with that exponent, a central temperature just four per cent below the model's gives \(0.96^{29}\approx0.31\) (computed here) — the measured ratio. An error of one part in twenty-five in a fifteen-million-kelvin prediction was the entire “astrophysical solution”, and no chlorine measurement could exclude it. It was excluded from two independent directions: the gallium detectors reached the pp neutrinos, whose flux is pinned by the solar luminosity itself (Equation (114.6), the bound of Phenomenon 114.3), and still saw a deficit; and helioseismology (Section 52.6.2) measured the sound speed — \(c_{\mathrm{s}}^{2}\propto T/\mu\) — through the core to a few parts in a thousand, excluding a temperature error of that size by more than an order of magnitude.
The charged-current to neutral-current ratio is flavour-blind astrophysics-proof. On deuterium two channels are open (Section 114.4.4): the charged-current breakup \(\nu_{e}+d\to p+p+e^{-}\), which only \(\nu_{e}\) can drive at these energies, and the neutral-current breakup \(\nu+d\to p+n+\nu\), whose amplitude is identical for \(\nu_{e}\), \(\nu_{\mu}\) and \(\nu_{\tau}\) because the \(Z\) couples to the three lepton doublets with one strength (Weak Interactions). The neutral-current rate therefore measures the total active flux whatever the Sun does, the charged-current rate measures its electron-flavour part, and their ratio measures the surviving \(\nu_{e}\) fraction with every solar uncertainty cancelling. SNO measured the ratio near one third while the neutral-current total matched the solar-model prediction within its uncertainty (Phenomenon 114.5, Table 114.6; [Ahmad:2002]): the missing electron neutrinos were arriving with the wrong flavour, not missing. The oscillation mechanism, the matter effect that shapes the energy dependence, and the parameter determinations belong to Flavour Physics and Neutrinos and Experiment: Neutrino Oscillations.
∎Resolution: flavour transformation
What SNO delivered in 2002 was the comparison the derivation above calls for: in one target, during one exposure, the electron-flavour flux from the charged-current channel and the flavour-blind total from the neutral-current channel (Section 114.4.4). The electron fraction came out near one third; the total \(^{8}\)B flux agreed with the standard solar model within the model's own sixteen per cent uncertainty (Table 114.2) [Ahmad:2002]. Read as stellar physics, that is a precision confirmation delivered by the most temperature-sensitive observable the Sun emits: the \(T_{\mathrm{c}}^{\sim20}\) thermometer of Phenomenon 52.22 returned the model's central temperature to within a fraction of a per cent. The Sun was right; the neutrinos oscillate, with all that follows for particle physics developed in Flavour Physics and Neutrinos and measured across Experiment: Neutrino Oscillations.
Complete spectroscopy of solar fusion
The modern standard solar model — updated opacities, equation of state, and helioseismic calibration — is that of Bahcall, Serenelli and Basu [Bahcall:2005], whose predicted fluxes are Table 114.2. Borexino measured them: the pp, pep, \(^{7}\)Be and \(^{8}\)B branches separately and in real time [Agostini:2018], and in 2020 the CNO neutrinos [Agostini:2020a] (Section 114.6.1, Table 114.11).
Real-time spectroscopy in an ultra-pure liquid scintillator has measured the neutrino fluxes of the individual branches of the pp chain separately [Agostini:2018], and detected the neutrinos of the CNO cycle, which supplies about one per cent of the Sun's power [Agostini:2020a]. The nuclear energy generation rate reconstructed from those neutrinos agrees with the luminosity the Sun is observed to radiate in photons. The two are not measurements of the same thing at the same time: neutrinos leave the core immediately while photons diffuse out over some \(10^{5}\) years, so their agreement also says that the solar output has been steady across that interval. Rests on Equation (107.104) and Phenomenon 52.4.
Derivation. Derives Phenomenon 52.23. The total neutrino flux is fixed by the luminosity with essentially no model dependence, and that is the quantity these measurements check. The net effect of the pp chain, and equally of the CNO cycle, is
releasing \(Q=26.73\,\mathrm{MeV}\) per helium nucleus formed (Equation (107.104)) and producing exactly two neutrinos. Each such reaction therefore contributes \(Q\) to the luminosity and two neutrinos to the flux, so at the Earth's distance \(d\)
Numerically, with \(Q=26.73\,\mathrm{MeV}=4.283\times 10^{-12}\,\mathrm{J}\) and \(d=1.496\times 10^{11}\,\mathrm{m}\),
that is some \(6\times10^{10}\) per square centimetre per second — the statement made from the nuclear-physics side as Equation (107.105), and deployed as the luminosity constraint on every oscillation analysis in Equation (114.6).
Observe what went into this: the observed luminosity, the energy release of the net reaction, and the Earth–Sun distance. No opacity, no equation of state, no reaction rate, no central temperature. It is a slight overestimate, by a few per cent, because the neutrinos themselves carry away a small part of \(Q\) that never becomes light (the loss budget computed in Section 52.2.2). That the measured pp flux matches it is a check on the entire picture of hydrogen burning; the value of the branch-by-branch spectroscopy lies in what it adds beyond this bound — the distribution of the fusion rate among the branches, which does depend on the central temperature and the composition, and which is compared with the model term by term in Table 114.11.
∎Helioseismology
The five-minute oscillations
In 1962 Leighton, Noyes and Simon, Doppler-imaging the photosphere, found its surface heaving: patches thousands of kilometres across oscillating vertically with periods clustered around five minutes [Leighton:1962]. Ulrich proposed that these are not a surface phenomenon but standing acoustic waves — sound, trapped in a resonant cavity inside the Sun [Ulrich:1970] — and the proposal carried a sharp signature: trapped modes exist only on discrete ridges in the wavenumber–frequency plane. Deubner's observations resolved exactly those ridges [Deubner:1975]. The mechanics is the acoustics of Oscillations and Mechanical Waves and Fluid Dynamics on a stratified sphere, and the cavity can be exhibited with WKB tools (Ordinary Differential Equations and Sturm–Liouville Theory).
Linear adiabatic acoustic waves of angular frequency \(\omega\) and spherical-harmonic degree \(\ell\) in a stratified star propagate where
with \(c_{\mathrm{s}}\) the sound speed (Section 31.10.1), \(N\) the buoyancy frequency Equation (52.20), and
the acoustic cutoff of the near-isothermal atmosphere, \(H\) its pressure scale height. A mode is trapped between the inner turning radius \(r_{\mathrm{t}}\), where \(c_{\mathrm{s}}(r_{\mathrm{t}})/r_{\mathrm{t}} =\omega/\sqrt{\ell(\ell+1)}\), and the surface layer where \(\omega<\omega_{\mathrm{ac}}\); the trapped frequencies satisfy
\(n\) the radial order and \(\alpha\) a slowly varying surface phase. Rests on Equation (52.20), Theorem 9.78, Theorem 31.12 and Theorem 31.22.
Derives Proposition 52.24. The cutoff. In an isothermal atmosphere, \(\rho_{0}\propto\ee^{-z/H}\) with constant \(c_{\mathrm{s}}\) and \(H\). Linearize continuity, momentum and the adiabatic condition about this background for vertical motion, neglect buoyancy (\(\omega\gg N\)), and eliminate all variables in favour of \(\chi\equiv\delta p/\sqrt{\rho_{0}}\): the exponential factors cancel and the three equations collapse to
the \(\tfrac{1}{4H^{2}}\) arising as the square of the \(\rho_{0}^{-1/2}\) gradient. Below \(\omega_{\mathrm{ac}}\) the wave is evanescent: the thinning atmosphere reflects it back down. With photospheric values \(T_{\mathrm{eff}}=5772\,\mathrm{K}\), \(\mu\approx1.25\), \(g=274\,\mathrm{m}/\mathrm{s}^{2}\): \(c_{\mathrm{s}}\approx8.0\,\mathrm{km}/\mathrm{s}\), \(H=k_{\mathrm{B}}T/\mu m_{\mathrm{u}}g\approx140\,\mathrm{km}\), so \(\omega_{\mathrm{ac}}/2\pi\approx5\,\mathrm{mHz}\) — a period near three and a half minutes (computed here). Waves of longer period, the five-minute band included, cannot escape: the photosphere is a mirror.
The turning point. In the interior, WKB with \(\nabla_{\mathrm{h}}^{2}Y_{\ell m}=-\ell(\ell+1)Y_{\ell m}/r^{2}\) gives the local dispersion relation \(\omega^{2}=c_{\mathrm{s}}^{2}(k_{r}^{2}+k_{\mathrm{h}}^{2})\) with \(k_{\mathrm{h}}^{2}=\ell(\ell+1)/r^{2}\), i.e. Equation (52.37) (the cutoff term matters only near the surface). Descending, \(c_{\mathrm{s}}\) rises, refracting the ray horizontal: \(k_{r}\) vanishes where \(c_{\mathrm{s}}/r=\omega/\sqrt{\ell(\ell+1)}\) and the wave turns back up. High-\(\ell\) modes skim the envelope; low-\(\ell\) modes plunge nearly to the centre — which is what makes the spectrum an interior probe at every depth.
Quantization. A wave trapped between two turning surfaces interferes constructively when its radial phase integral is an integer-plus-phase multiple of \(\pi\) (Theorem 9.78), which is Equation (52.39): at each \(n\), a discrete curve \(\omega_{n}(\ell)\) — precisely the observed ridges of [Deubner:1975]. The observed power peaks near \(3.3\,\mathrm{mHz}\) because the modes are driven stochastically by the granulation of the convection zone (Remark 52.11), which shakes hardest at those frequencies (quoted [ChristensenDalsgaard:2002]).
∎The seismic Sun
Millions of mode frequencies \(\omega_{n\ell}\) have been measured, many to relative accuracies of \(10^{-5}\). Through Equation (52.39) each is an integral of \(1/c_{\mathrm{s}}\) along a different chord of the interior, and a sufficiently rich set of chords can be inverted.
The solar surface oscillates, with a velocity field whose power is concentrated near a period of five minutes [Leighton:1962]. The oscillations are not a surface phenomenon but the superposition of millions of trapped acoustic eigenmodes of the whole star [Ulrich:1970], as the discrete ridges observed in the wavenumber–frequency plane confirmed [Deubner:1975]. Inverting the measured mode frequencies yields the sound speed as a function of radius; it agrees with the standard solar model to a few parts in \(10^{3}\) throughout the radiative interior and locates the base of the convection zone [ChristensenDalsgaard:2002]. One discrepancy remains and is recorded here as unresolved: the inversions prefer the older, higher photospheric abundances to the downward-revised ones [Asplund:2009] [Bahcall:2005]. Rests on Equation (52.39) and Proposition 52.24.
Derivation. Derives Phenomenon 52.25. Away from the surface layers the cutoff term in Equation (52.37) is negligible, and Equation (52.39) becomes
The left side is data; the right side depends on \((n,\ell,\omega)\) only through the single combination \(w\). That millions of measured frequencies, spanning hundreds of \(\ell\) values, collapse onto one function \(F(w)\) is itself a sharp, parameter-free test of the acoustic-cavity picture, and the solar spectrum passes it (the collapse is known as the Duvall law; quoted [ChristensenDalsgaard:2002]).
The inversion is then classical. Let \(a(r)=c_{\mathrm{s}}/r\), monotonic in the region of interest, so that \(1/c_{\mathrm{s}}^{2}=1/(a^{2}r^{2})\) and the turning point is \(a(r_{\mathrm{t}})=w\). Changing variable from \(r\) to \(a\),
with \(a_{\mathrm{s}}\) the surface value. In the variables \(x=w^{-2}\), \(y=a^{-2}\) this is \(\tilde{F}(x)=\int_{x}^{y_{\mathrm{s}}}\sqrt{y-x}\,g(y)\,\dd y\) with \(g\ge0\), and differentiating once,
an integral equation of Abel type. Its classical solution is
verified by substituting back, exchanging the order of integration, and using \(\int_{y}^{x}\dd u/\sqrt{(x-u)(u-y)}=\pi\). Integrating \(G\) recovers \(r(a)\) by a quadrature, hence \(a(r)\), hence \(c_{\mathrm{s}}(r)\): the sound speed at every radius, from the mode frequencies alone, the surface phase \(\alpha(\omega)\) being fitted along the way — the honest caveat of the method, confined to the layers every mode shares.
What the inversion returns [ChristensenDalsgaard:2002] (quoted): a sound-speed profile agreeing with the standard solar model to a few parts in a thousand at every radius outside the innermost few per cent; the base of the convection zone — visible as the kink where the stratification leaves the adiabat of Theorem 52.10 — at \(r=0.713\,R_{\odot}\); the envelope helium abundance, read from the \(\Gamma_{1}\) depression in the helium ionization zone; and, from the rotational splitting of the modes, the internal rotation — differential through the convection zone, near-rigid beneath, with the shear concentrated in a thin tachocline at the interface, the presumptive seat of the solar dynamo (Plasmas and Magnetohydrodynamics). Since \(c_{\mathrm{s}}^{2}\propto T/\mu\), the profile pins the central temperature at the sub-per-cent level, which is what closed the astrophysical escape route in Phenomenon 52.22: the Sun's interior has been measured twice — acoustically and in neutrinos — and the two channels agree with each other and with the model. The one standing tension, between the seismic profile and the revised photospheric abundances [Asplund:2009] [Bahcall:2005], is recorded in the phenomenon above as unresolved.
∎