Asymptotic Giant Branch and Thermal Pulses

When a star of between roughly one and eight solar masses runs out of helium in its core, it is not yet finished. It builds an inert carbon–oxygen core, no larger than the Earth but supported by the same electron degeneracy pressure that holds up a white dwarf, and wraps it in two concentric nuclear furnaces — a helium-burning shell just above the core and a hydrogen-burning shell farther out. This double-shell structure defines the Asymptotic Giant Branch, so named because in the Hertzsprung–Russell diagram the track climbs steeply and asymptotically approaches the earlier Red Giant Branch. The star swells to hundreds of solar radii, cools to about $3000\,\mathrm{K}$ , and shines at $10^3$ – $10^4\,L_\odot$ . The two shells cannot both burn steadily. A geometrically thin burning shell is thermally unstable: it cannot expand enough to cool itself when it heats up, so the helium shell ignites in violent, recurrent flashes — the thermal pulses — separated by long quiescent stretches during which the hydrogen shell alone supplies the light. Each flash drives a convective tongue through the helium-rich intershell, cooks fresh $^{12}\mathrm{C}$ , and liberates free neutrons that build half of the elements heavier than iron by slow neutron capture. Afterward the swollen envelope reaches down and dredges that new material to the surface, turning ordinary stars into carbon stars. The AGB does not end by igniting the next fuel. A dusty, radiation-driven superwind strips the envelope faster than the degenerate core can grow, truncating it well below the Chandrasekhar mass; the naked hot core then lights up a planetary nebula and cools forever as a white dwarf. This article assembles that story from its three governing relations: the thin-shell instability that fires the pulses, the core-mass–luminosity relation that paces the climb, and the mass loss that writes the ending.

Key takeaways

The Asymptotic Giant Branch is the second red-giant ascent, powered by a hydrogen shell and a helium shell burning around an inert, electron-degenerate carbon–oxygen core.
A geometrically thin burning shell is thermally unstable, so the helium shell ignites in recurrent flashes — thermal pulses — with the threshold $D/r \lesssim 1/4$ and interpulse spacing of $10^4$ – $10^5\,\mathrm{yr}$ .
The surface luminosity is tied almost entirely to the degenerate core mass through the Paczyński core-mass–luminosity relation, $L \approx 5.9\times10^4\,L_\odot\,(M_c/M_\odot - 0.52)$ .
Each pulse drives a convective pocket that runs the slow neutron-capture (s-)process using neutrons from $^{13}\mathrm{C}(\alpha,n)^{16}\mathrm{O}$ and $^{22}\mathrm{Ne}(\alpha,n)^{25}\mathrm{Mg}$ , then dredges $^{12}\mathrm{C}$ and heavy elements to the surface.
A dust-driven superwind reaching $\dot M \sim 10^{-4}\,M_\odot\,\mathrm{yr^{-1}}$ strips the envelope before the core nears the Chandrasekhar mass, so single AGB stars die as CO white dwarfs, not supernovae.
Key idea: A geometrically thin nuclear-burning shell cannot expand enough to lower its own temperature, so the AGB helium shell ignites in recurrent runaway flashes — thermal pulses — that mix, dredge up, and s-process new elements while the core-mass–luminosity relation governs the star's slow, terminal climb.

The core insight

A geometrically thin nuclear-burning shell is thermally unstable: it cannot expand enough to lower its temperature when energy generation increases, so the helium shell in the double-shell AGB structure ignites in recurrent runaway flashes (thermal pulses) that mix, dredge up, and s-process new elements while a core-mass–luminosity relation governs the star's slow climb.

Where this lives today

AGB stars are the workshops where much of the periodic table's cosmic chemistry is finished, and their fingerprints are now read at extraordinary precision. Presolar silicon-carbide grains recovered from primitive meteorites like Murchison carry isotopic ratios of carbon, nitrogen, and heavy s-process elements that match, grain by grain, the intershell composition predicted for carbon-rich AGB stars — literal dust from dead stars, dated to before the Solar System formed. ALMA and JWST now resolve the dusty, spiralling winds around nearby AGB stars directly, testing the pulsation-plus-dust superwind picture that equation (6) sketches, while Gaia has delivered enormous samples of long-period variables — Miras and semiregulars — whose infrared period–luminosity relations serve as standard candles across the Local Group. On the theory side the central open problem is the origin and size of the $^{13}\mathrm{C}$ pocket that supplies the s-process neutrons: it hinges on how convection mixes at the boundary of the pulse-driven convective zone, a fluid-dynamics question that models cannot yet compute from first principles and that also feeds the newly recognized "i-process." Every stellar-population synthesis model — used to weigh galaxies and read their star-formation histories — carries the TP-AGB as one of its largest remaining uncertainties.

The physical picture

Picture two nuclear furnaces stacked in a thin onion shell around a dead, degenerate ember. The outer hydrogen furnace steadily rains helium ash down onto the inner helium furnace. When enough ash has piled up and grown hot enough, the helium furnace ignites — but because it lives in a shell too thin to puff outward and cool itself, the ignition runs away for a few years before the whole intershell finally expands and quenches it. That heartbeat repeats every $10^4$ – $10^5$ years: one beat is a thermal pulse. The first figure is the anatomy the whole article rests on, with the real temperatures and sizes of each layer attached.

Figure

Cutaway of an AGB star (schematic, radii logarithmic): an Earth-sized degenerate CO core of about 0.6 M_⊙ sits beneath a thin He-burning shell that reaches 1–3×10⁸ K, a He/¹²C intershell that hosts the s-process, a hydrogen shell near 7×10⁷ K, and a vast convective envelope extending to about 300 R_⊙ with T_eff ≈ 3000 K and L ≈ 5×10³ L_⊙

The thin-shell instability is the mirror image of ordinary stellar stability. Normally a layer that burns faster expands, cools, and throttles back — the negative feedback that makes the Sun a steady thermostat. In a thin shell the geometry breaks that feedback: the overlying weight fixes the pressure regardless of how the shell expands, so heating does not relieve the burning, and temperature and reaction rate spiral upward until the whole intershell finally expands enough to break the geometry. The consequence is the second, punchline figure: a saw-tooth history in which the helium shell is nearly dormant for tens of thousands of years and then erupts to $10^7$ – $10^8\,L_\odot$ for a few years, extinguishing the hydrogen shell and opening the door for dredge-up.

Figure

The thermal-pulse cycle in luminosity: the H shell burns quiescently near 10⁴ L_⊙ for an interpulse of 10⁴–10⁵ yr, then the He shell flashes to 10⁷–10⁸ L_⊙ for a few years, the expansion extinguishes the H shell, and the convective envelope reaches in for third dredge-up before the cycle repeats

The dredge-up that follows each flash is a conveyor belt: the deep convective envelope dips into freshly cooked material and hauls carbon and heavy s-process elements to the surface, transforming the star pulse by pulse. The ending is delivered by mass loss — pulsations levitate cool gas to where soot-like dust condenses, starlight pushes on the dust, and a dusty gale peels the envelope away, leaving the hot core to ionize a planetary nebula.

The core equations

The single most useful fact about an AGB star is that its luminosity barely knows about its envelope — it is set by the mass of the degenerate core underneath. Paczyński's core-mass–luminosity relation captures this near-linear dependence over the whole AGB range.

L = 5.9\times 10^4\,L_\odot\left(\frac{M_c}{M_\odot} - 0.52\right) \qquad (1)

where: $L$ — surface (quiescent H-shell) luminosity of the AGB star; $M_c$ — mass of the degenerate CO core; $L_\odot$ — solar luminosity. Reading: a core of $0.6\,M_\odot$ gives $L\approx4700\,L_\odot$ and a core of $0.9\,M_\odot$ gives $L\approx2.2\times10^4\,L_\odot$ , so as the core slowly grows the star brightens on a fixed track — the envelope mass hardly enters. The reason the helium shell flashes rather than burns steadily begins with the ferocious temperature sensitivity of triple-alpha burning near $10^8\,\mathrm{K}$ — the amplifier that turns any small heating into a runaway.

\epsilon_{3\alpha} \propto \rho^2\,T^{40} \quad (T \sim 10^8\,\mathrm{K}) \qquad (2)

where: $\epsilon_{3\alpha}$ — specific energy-generation rate of triple-alpha burning ( $\mathrm{erg\,g^{-1}\,s^{-1}}$ ); $\rho$ — density in the helium shell; $T$ — temperature, with local exponent $\nu = \partial\ln\epsilon/\partial\ln T \approx 40$ near $10^8\,\mathrm{K}$ . Reading: a mere $2\%$ rise in temperature nearly doubles the burning rate, so any layer that heats when it expands is doomed to run away. That amplifier is only dangerous because the geometry of a thin shell removes the usual safety valve. The instability criterion is a statement about the shell's fractional thickness.

\text{thermally unstable} \;\Longleftrightarrow\; \frac{D}{r} \lesssim \frac{1}{4} \qquad (3)

where: $D$ — geometric thickness of the burning shell; $r$ — its radius; the helium shell on the TP-AGB has $D/r \ll 1$ . Reading: a thin shell cannot expand enough to cool, so heating raises rather than lowers its temperature — the derivation below turns this into the exact factor $(r/D - 4)$ . Inside the flash-driven convective pocket, free neutrons appear even though the gas is helium-rich, not neutron-rich, because alpha captures on light seed nuclei release them. The dominant source in low-mass AGB stars runs in a thin $^{13}\mathrm{C}$ pocket between pulses.

{}^{13}\mathrm{C} + {}^4\mathrm{He} \rightarrow {}^{16}\mathrm{O} + n \qquad (4)

where: $n$ — the free neutron released for slow neutron capture; the companion source $^{22}\mathrm{Ne}(\alpha,n)^{25}\mathrm{Mg}$ switches on at the higher temperatures of massive-AGB flashes. Reading: these are $(\alpha,n)$ reactions, so the same helium that fuels the flash also seeds the nucleosynthesis of heavy elements. Whether those neutrons build the s-process pattern or the r-process pattern is decided by a race between capture and decay. The "slow" in s-process means capture is slow compared with beta decay.

\tau_n \gg \tau_\beta \;\Rightarrow\; \text{s-process path hugs the valley of stability} \qquad (5)

where: $\tau_n$ — neutron-capture timescale for a given isotope (set by the neutron density $N_n \sim 10^{7}$ – $10^{10}\,\mathrm{cm^{-3}}$ ); $\tau_\beta$ — beta-decay timescale of the unstable capture product. Reading: because unstable nuclei decay before capturing again, the path hugs stable isotopes, and cross-section bottlenecks at the neutron-magic numbers build the Sr–Y–Zr, Ba–La–Ce, and Pb abundance peaks. Finally, the AGB ends by mass loss: the pre-superwind rate follows a Reimers-type scaling, and the terminal dusty wind is bounded by the momentum the starlight can supply.

\dot M_{\rm R} = -\eta\,\frac{L R}{M}\,(4\times10^{-13}\,M_\odot\,\mathrm{yr^{-1}}),\qquad \dot M v_\infty \lesssim \frac{L}{c} \qquad (6)

where: $\dot M_{\rm R}$ — Reimers mass-loss rate with $L,R,M$ in solar units; $\eta \sim 0.4$ – $1$ — a dimensionless efficiency; $v_\infty$ — terminal wind speed; $c$ — speed of light; the right inequality is the single-scattering photon-momentum bound. Reading: on the AGB, $L$ , $R$ climb and $M$ falls, so $\dot M$ accelerates into a superwind that removes the envelope in $\sim10^4$ – $10^5\,\mathrm{yr}$ .

Derivation AWhy a thin burning shell is thermally unstable

The goal is to show, from hydrostatic balance and an ideal-gas equation of state alone, that a geometrically thin nuclear-burning shell responds to a small expansion by heating up rather than cooling down, and to derive the exact threshold $D/r \lesssim 1/4$ that decides when this happens — the root cause of every thermal pulse on the AGB. The whole argument turns on one geometric fact: the shell's inner boundary rests on a nearly incompressible degenerate core, so when the shell expands it thickens outward while its base stays put.

Step 1. Fix the pressure from the overburden. Hydrostatic balance makes the pressure at the burning layer equal to the weight per unit area above it, $P \approx G M_r\,\Delta m_{\rm above}/(4\pi r^4)$ , where $r$ is the fixed radius of the shell's base on the core. A small outward displacement $\Delta r$ of the overlying column therefore gives $\Delta P/P = -4\,\Delta r/r$ . Why: the overlying weight, not the shell's own state, sets the pressure — the safety valve a thick layer has and a thin shell loses, because here $r$ barely moves.

Step 2. Impose mass conservation. The shell mass $\Delta m = 4\pi r^2 \rho\, D$ is fixed, and since the base radius $r$ is pinned to the incompressible core, only the thickness and density respond: $\Delta\rho/\rho = -\Delta D/D$ . The expansion goes entirely into thickening, $\Delta D = \Delta r$ , so $\Delta\rho/\rho = -\Delta r/D$ . Why: pinning the inner boundary makes $D$ , not $r$ , the small quantity — density drops by the large factor $r/D$ times the fractional displacement.

Step 3. Use the ideal-gas equation of state $P \propto \rho T$ , so $\Delta P/P = \Delta\rho/\rho + \Delta T/T$ , giving $\Delta T/T = \Delta P/P - \Delta\rho/\rho$ . Why: the temperature is whatever the equation of state demands once the externally fixed pressure and the geometrically fixed density are both set.

Step 4. Substitute Steps 1 and 2: $\dfrac{\Delta T}{T} = -\dfrac{4\,\Delta r}{r} + \dfrac{\Delta r}{D} = \left(\dfrac{r}{D} - 4\right)\dfrac{\Delta r}{r}$ . When $D/r$ is small the coefficient $r/D - 4$ is large and positive, so an expansion ( $\Delta r > 0$ ) raises the temperature. Why: this single line is the whole instability — the geometry has flipped the sign of the temperature response relative to a normal, thick, self-regulating layer.

Step 5. Read off the threshold. The response is destabilizing when $r/D - 4 > 0$ , that is $D/r \lesssim 1/4$ . Fed into the triple-alpha amplifier $\epsilon_{3\alpha}\propto T^{40}$ , a positive $\Delta T$ boosts the burning, which drives more expansion and more heating — a thermonuclear runaway that halts only once the whole intershell has swollen enough to push $D/r$ back above the threshold. Why: the number $1/4$ is what separates the steadily burning shells of earlier evolution from the flashing helium shell of the TP-AGB. *Note: the clean substitution gives $\Delta T/T = (r/D - 4)\,\Delta r/r$ and threshold $D/r\lesssim1/4$ , the standard Schwarzschild–Härm result; some short treatments incorrectly quote the factor $(r/D-2)$ from the same steps.*

Go deeper · Derivation B · 1 minThe core-mass–luminosity relation

Derivation BThe core-mass–luminosity relation

The aim is to explain why an AGB star's luminosity depends almost entirely on the mass of its hidden degenerate core and hardly at all on the huge envelope wrapped around it, arriving at the Paczyński relation $L \approx 5.9\times10^4\,L_\odot\,(M_c/M_\odot - 0.52)$ . The logic is a chain of dependencies: the degenerate core's structure is fixed by its mass alone, that structure fixes the temperature of the shell burning on its surface, and the extreme temperature sensitivity of shell burning fixes the luminosity.

Step 1. Start from the mass–radius relation of a non-relativistic electron-degenerate core, $R_c \propto M_c^{-1/3}$ . A more massive core is smaller, so its surface gravity and gravitational potential are determined by $M_c$ and nothing else. Why: degeneracy pressure does not care about temperature, so the core's mechanical structure decouples completely from the burning shells and the envelope above it.

Step 2. Set the shell temperature by hydrostatic support at the core surface, $k_B T_{\rm sh} \sim G M_c \mu m_H/R_c \propto M_c/M_c^{-1/3} = M_c^{4/3}$ . The thin hydrogen shell sits in the core's steep potential well, so its temperature climbs steeply with core mass. Why: the shell has no freedom to choose its own temperature — it must be hot enough to hold up the overlying envelope at the core's gravity, which $M_c$ alone dictates.

Step 3. Let the burning rate saturate. CNO hydrogen burning has $\epsilon \propto T^{\sim16}$ , so the shell adjusts its density and thickness until it produces exactly the luminosity consistent with the core-fixed temperature $T_{\rm sh}(M_c)$ ; the envelope rearranges around this but cannot change it. Why: with such a steep rate law the shell luminosity is effectively pinned by temperature, so a relation $L \approx L(M_c)$ is forced.

Step 4. Linearize against detailed models. Full stellar-evolution calculations give a nearly straight $L$ – $M_c$ line over $0.5 \lesssim M_c/M_\odot \lesssim 1.4$ , reproduced by the boxed formula; the intercept at $M_c = 0.52\,M_\odot$ marks the minimum core mass that can sustain the shell. Why: the microphysics predicts linearity, and calibrating the slope and offset to models turns the scaling argument into the quantitative tool used to read a star's core mass off its luminosity.

L \approx 5.9\times10^4\,L_\odot\left(\frac{M_c}{M_\odot}-0.52\right)

As a check: $M_c = 0.6\,M_\odot \Rightarrow L \approx 4.7\times10^3\,L_\odot$ .

Go deeper · Derivation C · 2 minHelium accumulation and the pulse cycle

Derivation CHelium accumulation and the pulse cycle

The purpose here is to estimate, from the energy budget alone, how much helium the hydrogen shell must pile up before the next flash, how long that takes, and how brief and bright the resulting eruption is — turning the qualitative saw-tooth of the second figure into numbers. We equate the energy radiated during the quiescent interpulse to the hydrogen burned that supplies it, then release the accumulated helium in a flash and read off its peak luminosity.

Step 1. During the interpulse the surface luminosity $L$ is supplied entirely by the hydrogen shell, which burns hydrogen at yield $Q_H \approx 6.4\times10^{18}\,\mathrm{erg\,g^{-1}}$ and converts it almost mass-for-mass into helium ash added to the intershell. The helium accumulated over an interpulse of duration $t_{\rm ip}$ is $\Delta M_{\rm He} \approx L\,t_{\rm ip}/Q_H$ . Why: every joule the star radiates between flashes is paid for by hydrogen fusion, so the radiated energy directly measures the helium delivered to the dormant shell below.

Step 2. Put in numbers for a typical TP-AGB star: $L = 10^4\,L_\odot = 3.8\times10^{37}\,\mathrm{erg\,s^{-1}}$ and $t_{\rm ip} = 10^4\,\mathrm{yr} = 3.16\times10^{11}\,\mathrm{s}$ give $\Delta M_{\rm He} \approx (3.8\times10^{37})(3.16\times10^{11})/(6.4\times10^{18}) \approx 1.9\times10^{30}\,\mathrm{g} \approx 10^{-3}\,M_\odot$ . Why: this fraction-of-a-percent of a solar mass is the fuel charge for one flash, and it is why the core grows only slowly — by roughly $10^{-3}\,M_\odot$ per pulse.

Step 3. Ignite it. When the base of the helium layer reaches $\sim10^8\,\mathrm{K}$ , triple-alpha lights in the thermally unstable thin shell, releasing $Q_{\rm He} \approx 5.9\times10^{17}\,\mathrm{erg\,g^{-1}}$ . Even if only a few percent of the accumulated helium burns in the flash proper, $E_{\rm flash} \sim (0.05)\,\Delta M_{\rm He}\,Q_{\rm He} \sim 10^{47}\,\mathrm{erg}$ is liberated in a convective pocket. Why: helium burning yields about ten times less energy per gram than hydrogen burning, yet concentrating it into a thin shell over a few years is what makes the local luminosity enormous.

Step 4. Convert to a peak luminosity. Releasing $\sim10^{47}\,\mathrm{erg}$ over $t_{\rm flash}\sim1\,\mathrm{yr} = 3.2\times10^{7}\,\mathrm{s}$ gives $L_{\rm He}^{\rm peak}\sim3\times10^{39}\,\mathrm{erg\,s^{-1}}\approx10^{6}$ – $10^{8}\,L_\odot$ locally. This drives the convective tongue that enables s-processing, then the whole star expands, extinguishes the H shell, and permits third dredge-up before the cycle repeats. Why: the flash luminosity never emerges from the surface — it is spent expanding the star — which is exactly why the observable surface light barely flickers while the interior briefly rivals a galaxy's worth of stars.

\Delta M_{\rm He} \approx \frac{L\,t_{\rm ip}}{Q_H}\approx 10^{-3}\,M_\odot,\qquad L_{\rm He}^{\rm peak}\sim 10^{6}\text{–}10^{8}\,L_\odot

These estimates — $\Delta M_{\rm He}\approx10^{-3}\,M_\odot$ per interpulse and $L_{\rm He}^{\rm peak}\sim10^{7}$ – $10^{8}\,L_\odot$ — are consistent with detailed stellar-evolution models.

Limiting cases

The AGB and its instabilities are cleanest when read through their limiting regimes, from the geometry that fires the pulses to the mass that decides the ending.

Case	Result	Meaning
Thin-shell instability	$D/r \lesssim 1/4$	a burning shell thinner than a quarter of its radius heats on expansion; the He shell has $D/r \ll 1$ and flashes
Low-mass AGB $(M \lesssim 1.5\,M_\odot)$	few pulses, $\mathrm{C/O}\to1$	small envelope, modest dredge-up; these can become carbon stars and leave $\sim0.55\,M_\odot$ CO white dwarfs
Intermediate-mass AGB $(4 \lesssim M/M_\odot \lesssim 8)$	$T_{\rm base}\gtrsim5\times10^7\,\mathrm{K}$	hot-bottom burning runs CNO in the envelope base, turning $^{12}\mathrm{C}$ into $^{14}\mathrm{N}$ , suppressing carbon stars and making $^7\mathrm{Li}$
Superwind / termination	$\dot M \to 10^{-4}\,M_\odot\,\mathrm{yr^{-1}}$	dust-driven wind strips the envelope faster than the core grows, ending the AGB before $M_c$ nears $M_{\rm Ch}$
Very low neutron density	$N_n \sim 10^{7}\,\mathrm{cm^{-3}}$	the $^{13}\mathrm{C}$ pocket runs the slow process, building s-process peaks near stability rather than the r-process

Common trapThe word asymptotic, and the two different helium flashes

Two traps recur. First, the "asymptotic" in Asymptotic Giant Branch is a statement about the track in the HR diagram — it climbs and approaches the earlier Red Giant Branch from the blue side — not about any mathematical limit of the star's evolution or a limiting mass. Second, and more damaging: thermal pulses are not the core helium flash. The core helium flash is a single event at the tip of the Red Giant Branch that ignites helium in a degenerate core and lifts that degeneracy once and for all. Thermal pulses are recurrent helium-shell flashes on the TP-AGB, in a shell that is non-degenerate but geometrically thin; they repeat every $10^4$ – $10^5\,\mathrm{yr}$ and are driven by the thin-shell geometry of equation (3), not by degeneracy. Same fuel, entirely different instability.

Worked examples

1 · Luminosity and interpulse helium of an AGB star (fully worked)

An AGB star has a degenerate CO core of $M_c = 0.6\,M_\odot$ . Find its luminosity from the core-mass–luminosity relation, then estimate how much helium its hydrogen shell delivers to the intershell during a $t_{\rm ip} = 10^4\,\mathrm{yr}$ interpulse.

Step 1. Apply equation (1): $L = 5.9\times10^4\,L_\odot\,(0.60 - 0.52) = \mathbf{4.7\times10^{3}\,L_\odot}$ . A modest core already outshines the Sun several thousand-fold, and a $0.9\,M_\odot$ core would reach $2.2\times10^4\,L_\odot$ — the star brightens as the core grows.

Step 2. Convert to cgs: $L = 4.7\times10^3 \times 3.83\times10^{33} = 1.8\times10^{37}\,\mathrm{erg\,s^{-1}}$ , with $t_{\rm ip} = 10^4\,\mathrm{yr} = 3.16\times10^{11}\,\mathrm{s}$ .

Step 3. Every joule radiated is paid for by hydrogen burning at $Q_H = 6.4\times10^{18}\,\mathrm{erg\,g^{-1}}$ , and the burned hydrogen becomes helium ash: $\Delta M_{\rm He} \approx L\,t_{\rm ip}/Q_H = (1.8\times10^{37})(3.16\times10^{11})/(6.4\times10^{18}) = 8.9\times10^{29}\,\mathrm{g}$ .

Step 4. In solar masses: $\Delta M_{\rm He} = 8.9\times10^{29}/1.99\times10^{33} \approx \mathbf{4.5\times10^{-4}\,M_\odot}$ . Each interpulse deposits a few $\times10^{-4}\,M_\odot$ of helium onto the dormant shell, and the core grows by a comparable amount per cycle — which is why it creeps toward, but is truncated well short of, the Chandrasekhar mass.

2 · Peak luminosity of a single thermal pulse (fully worked)

Suppose $5\%$ of the $\Delta M_{\rm He} \approx 5\times10^{-4}\,M_\odot$ accumulated above burns during a flash lasting $t_{\rm flash} \approx 1\,\mathrm{yr}$ . Estimate the peak helium-burning luminosity and compare it to the star's surface luminosity.

Step 1. Mass actually burned in the flash: $\Delta M_{\rm burn} = 0.05 \times 5\times10^{-4}\,M_\odot = 2.5\times10^{-5}\,M_\odot = 2.5\times10^{-5}\times1.99\times10^{33}\,\mathrm{g} = 5.0\times10^{28}\,\mathrm{g}$ .

Step 2. Energy released at $Q_{\rm He} = 5.9\times10^{17}\,\mathrm{erg\,g^{-1}}$ : $E_{\rm flash} = (5.0\times10^{28})(5.9\times10^{17}) = \mathbf{2.9\times10^{46}\,\mathrm{erg}}$ .

Step 3. Divide by the flash duration $t_{\rm flash} = 1\,\mathrm{yr} = 3.16\times10^{7}\,\mathrm{s}$ : $L_{\rm He}^{\rm peak} = 2.9\times10^{46}/3.16\times10^{7} = 9.2\times10^{38}\,\mathrm{erg\,s^{-1}} = \mathbf{2.4\times10^{5}\,L_\odot}$ , rising to $10^{7}$ – $10^{8}\,L_\odot$ if a larger fraction burns in a shorter spike.

Step 4. Compare with the surface: the star still radiates only $L \approx 5\times10^{3}\,L_\odot$ from its photosphere. The flash luminosity is far larger but stays buried — it is spent expanding the intershell and envelope, so an observer sees the surface light barely dip rather than flare. The interior briefly outshines the surface by orders of magnitude, yet the star looks almost unchanged.

3 · Why mass loss wins the race (fully worked)

An AGB star has an envelope of $M_{\rm env} = 0.5\,M_\odot$ and a superwind $\dot M = 10^{-4}\,M_\odot\,\mathrm{yr^{-1}}$ . Its core grows at $\mathrm{d}M_c/\mathrm{d}t \approx 10^{-7}\,M_\odot\,\mathrm{yr^{-1}}$ (one interpulse charge per $10^4\,\mathrm{yr}$ ). Show that the envelope is stripped long before the core could reach $M_{\rm Ch} = 1.4\,M_\odot$ .

Step 1. Envelope-removal time: $t_{\rm strip} = M_{\rm env}/\dot M = 0.5/10^{-4} = \mathbf{5\times10^{3}\,\mathrm{yr}}$ . The superwind clears the entire envelope in only a few thousand years.

Step 2. Core growth in that time: $\Delta M_c = (\mathrm{d}M_c/\mathrm{d}t)\,t_{\rm strip} = 10^{-7}\times5\times10^{3} = \mathbf{5\times10^{-4}\,M_\odot}$ . The core barely moves — less than a thousandth of a solar mass.

Step 3. Time the core would need to grow from $0.6\,M_\odot$ to $M_{\rm Ch} = 1.4\,M_\odot$ at that rate: $t_{\rm Ch} = (1.4 - 0.6)/10^{-7} = 8\times10^{6}\,\mathrm{yr}$ , about $1600$ times longer than $t_{\rm strip}$ .

Step 4. Mass loss removes the fuel supply roughly three orders of magnitude faster than the core can grow, so a single AGB star never reaches the Chandrasekhar mass by this route. It ends as a $\sim0.6$ – $0.7\,M_\odot$ CO white dwarf, not a supernova — the superwind, not nuclear physics, writes the ending.

Physics of the portrait

The thermal pulse is a thin-shell instability, and the person who first showed that a nuclear-burning shell could be thermally unstable — and who, more than anyone, turned stellar evolution into a computational science — was Martin Schwarzschild.

Origins · 1 minMartin Schwarzschild and how this was discovered

Martin Schwarzschild

1912–1997 · stellar evolution and the thin-shell instability

Martin Schwarzschild was the son of Karl Schwarzschild, the astronomer who first solved Einstein's field equations, and he inherited both his father's gift and his exile: born in Potsdam in 1912, he fled Nazi Germany in 1936, spent a year in Norway, and settled in the United States, eventually at Princeton for the rest of his career. More than any other single figure he made stellar structure a quantitative, computational discipline. His 1958 monograph Structure and Evolution of the Stars used the new digital computers to trace how stars change over time, and the "Schwarzschild criterion" for the onset of convection still bears his name. In 1965, with Richard Härm, he analyzed the helium-burning shell of a late-stage low-mass star and discovered that such a geometrically thin shell is thermally unstable — it cannot expand enough to regulate itself, so it must burn in recurrent flashes. That Schwarzschild–Härm instability is exactly the engine of the thermal pulses that define the TP-AGB, drive dredge-up, and power the s-process. He also pioneered high-altitude astronomy with the Stratoscope balloon telescopes and, late in life, turned to the dynamics of elliptical galaxies. Building on Eddington's conviction that stars are governed by physics we can write down, Schwarzschild supplied both the numerical machinery and the key instability analysis that turned the abstract picture of a dying star into the pulse-by-pulse theory used today.

History

1955

— Sanders and others recognize double-shell burning in post-helium-core stars

1965

— Schwarzschild and Härm identify the thermal instability of a thin helium-burning shell

1966

— Weigert computes the first stellar models showing recurrent helium-shell flashes

1970

— Paczyński derives the core-mass–luminosity relation for AGB stars

1975

— Iben maps thermal pulses, third dredge-up, and s-processing along the AGB

1983

— Iben and Renzini review AGB evolution and beyond

Going further

Texts. Carroll and Ostlie, An Introduction to Modern Astrophysics, 2nd ed., ch. 13 (post-main-sequence evolution) · Kippenhahn, Weigert and Weiss, Stellar Structure and Evolution, 2nd ed., chs. 33–34 (shell burning and thermal pulses) · Herwig, F., "Evolution of Asymptotic Giant Branch Stars," ARA&A 43 (2005) 435 · Iben and Renzini, "Asymptotic Giant Branch Evolution and Beyond," ARA&A 21 (1983) 271 · Busso, Gallino and Wasserburg, "Nucleosynthesis in AGB Stars," ARA&A 37 (1999) 239. The core-mass–luminosity relation has coefficient $5.9\times10^4\,L_\odot$ and offset $0.52\,M_\odot$ . The thin-shell threshold $D/r\lesssim1/4$ is derived from hydrostatic balance and the ideal-gas law, recovering the standard Schwarzschild–Härm (1965) result. Key numerical values used throughout: temperature exponent $\nu\approx40$ for triple-alpha near $10^8\,\mathrm{K}$ ; burning yields $Q_H\approx6.4\times10^{18}\,\mathrm{erg\,g^{-1}}$ and $Q_{\rm He}\approx5.9\times10^{17}\,\mathrm{erg\,g^{-1}}$ ; s-process neutron densities $N_n\sim10^{7}$ – $10^{10}\,\mathrm{cm^{-3}}$ ; and mass-loss rates spanning $10^{-7}$ – $10^{-4}\,M_\odot\,\mathrm{yr^{-1}}$ . These give: $L(M_c=0.6\,M_\odot)\approx4.7\times10^3\,L_\odot$ ; interpulse helium $\Delta M_{\rm He}\sim10^{-3}\,M_\odot$ ; flash peak $L_{\rm He}\sim10^{6}$ – $10^{8}\,L_\odot$ ; and an envelope-stripping-to-core-growth timescale ratio of $\sim10^{3}$ .

Prerequisites. Helium Burning and the Triple-Alpha Process · Helium Flash and the Horizontal Branch · Post-Main-Sequence Evolution and the Red Giant Branch · The CNO Cycle · White Dwarfs and Electron Degeneracy Pressure. Leads to. White Dwarfs and Electron Degeneracy Pressure · The Chandrasekhar Mass · Advanced Burning Stages in Massive Stars.

Ask the tutor: Ask me to walk through the thin-shell instability algebra step by step, or to work out how many thermal pulses a given star survives before its superwind strips the envelope.

What does "asymptotic" actually mean in Asymptotic Giant Branch?

It refers to the shape of the evolutionary track in the Hertzsprung–Russell diagram, not to anything about the star's ultimate fate. As a star begins its second ascent up the cool, luminous side of the diagram, its track climbs and appears to approach — asymptotically, from the blue — the earlier Red Giant Branch it traced during its first ascent. Early observers named the branch for that geometric near-convergence of the two tracks at high luminosity. It has nothing to do with a limiting mass, a limiting luminosity, or a mathematical asymptote in time. The name is a good reminder that much of stellar-evolution vocabulary was fixed by the appearance of tracks and cluster diagrams long before the underlying physics was understood.

How is a thermal pulse different from the core helium flash?

They share a fuel — helium — but almost nothing else. The core helium flash happens once, at the tip of the Red Giant Branch, when helium ignites in a degenerate core whose pressure does not respond to temperature; the runaway there is driven by degeneracy and ends by lifting it, after which the star settles onto the horizontal branch and burns helium quietly. Thermal pulses come much later, on the AGB, in the helium shell. That shell is not degenerate; its instability is purely geometric, the thin-shell instability of $D/r\lesssim1/4$ . Because the geometry is restored after each flash, the pulses recur, every $10^4$ – $10^5$ years, dozens of times in a star's life, driving dredge-up and s-processing. One is a single degeneracy-driven event; the other is a recurring geometry-driven cycle.

Where do the neutrons for the s-process come from if there is no free neutron source?

From alpha-capture reactions on light nuclei, which quietly release neutrons as a by-product. The dominant channel in low-mass AGB stars is $^{13}\mathrm{C}(\alpha,n)^{16}\mathrm{O}$ , operating in a thin " $^{13}\mathrm{C}$ pocket" that forms when a little hydrogen is mixed into the carbon-rich intershell after a dredge-up, seeding $^{13}\mathrm{C}$ that then burns with helium during the following interpulse at relatively low temperature. In more massive AGB stars a second source, $^{22}\mathrm{Ne}(\alpha,n)^{25}\mathrm{Mg}$ , switches on at the higher temperatures reached inside a flash. Both are $(\alpha,n)$ reactions, so the same helium that fuels the pulses also supplies the neutrons, and the neutron densities stay low ( $10^{7}$ – $10^{10}\,\mathrm{cm^{-3}}$ ) — which is exactly why the process is slow.

Why does "slow" neutron capture build a completely different set of elements than "rapid" capture?

Because the competition is between capturing another neutron and beta-decaying. In the s-process the neutron density is so low that whenever a nucleus captures a neutron and becomes unstable, it has plenty of time to beta-decay back toward stability before the next capture. The reaction path therefore creeps along the valley of stable isotopes, and it piles up at nuclei with neutron-magic numbers ( $N=50, 82, 126$ ) whose tiny capture cross-sections act as bottlenecks — producing the characteristic abundance peaks at strontium–yttrium–zirconium, barium–lanthanum–cerium, and lead. In the r-process the neutron flux is so intense that nuclei capture many neutrons before they can decay, driving far off stability into very neutron-rich territory and building a different pattern. The AGB s-process and the explosive r-process together make roughly half the elements heavier than iron each.

Why is mass loss, and not another nuclear ignition, what ends the AGB?

Because the star runs out of envelope before its core can grow enough to ignite anything new. The luminosity climbs with core mass, the radius swells, and pulsations lift cool gas to where dust condenses; radiation pressure on that dust drives an increasingly fierce wind that becomes a "superwind" at $\dot M \sim 10^{-4}\,M_\odot\,\mathrm{yr^{-1}}$ . As the worked example shows, that wind strips the whole envelope in a few thousand years, while the degenerate core grows only $\sim10^{-3}\,M_\odot$ per pulse and would need millions of years to reach the carbon-ignition mass. The envelope simply disappears first. Once the envelope drops below a small fraction of a solar mass, the star leaves the AGB, crosses to high temperature at nearly constant luminosity, ionizes the ejected gas as a planetary nebula, and settles onto the white-dwarf cooling track.

What is hot-bottom burning, and which stars show it?

In the more massive AGB stars — above roughly $4\,M_\odot$ — the base of the deep convective envelope is hot enough, $\gtrsim5\times10^7\,\mathrm{K}$ , to run nuclear reactions itself. This "hot-bottom burning" activates the CNO cycle right at the envelope's foot, so any $^{12}\mathrm{C}$ dredged up from a thermal pulse is promptly converted to $^{14}\mathrm{N}$ before it can accumulate. These stars are therefore prevented from becoming carbon stars (their surface stays oxygen-rich), and they can synthesize $^7\mathrm{Li}$ through the Cameron–Fowler beryllium-transport mechanism, sometimes becoming bright lithium-rich giants. Hot-bottom burning thus splits the AGB into two chemically distinct families: low-mass stars that turn carbon-rich, and intermediate-mass stars that stay nitrogen-rich and lithium-bright.

Why does the surface barely brighten during a flash that reaches 10⁸ solar luminosities?

Because that colossal luminosity is generated deep inside and never escapes as light — it is consumed as mechanical work. When the helium shell flashes, the sudden energy release expands the overlying intershell and envelope, pushing mass outward against gravity rather than streaming out as radiation. The photosphere, hundreds of solar radii away, responds only feebly; in fact the surface luminosity typically dips slightly after a pulse (the small feature the second figure marks), because the hydrogen shell that actually powers the surface is temporarily extinguished by the expansion. The flash announces itself not in brightness but in chemistry — the convective pocket it drives, and the dredge-up that follows, quietly rewrite the star's surface composition over the ensuing millennia.