s-Process Nucleosynthesis

Past the iron wall

Fusion can build the periodic table only as far as iron. Beyond ⁵⁶Fe, the binding energy per nucleon turns over — additional fusion costs energy rather than releasing it, and any furnace hot enough to fuse heavier nuclei would also dissociate them. Yet there are thousands of stable isotopes between iron and uranium, and they exist in the Sun, on Earth, and in your bloodstream. Two mechanisms account for nearly all of them, and neither is fusion. Both are neutron capture. Both were named in the seminal 1957 review by Burbidge, Burbidge, Fowler and Hoyle, often called B²FH. The fast one — call it the r-process — happens in cataclysmic events: core-collapse supernovae and neutron-star mergers, with neutron densities so high (~ 10²⁰ to 10²⁴ neutrons per cm³) that captures pile up faster than β decay can keep pace. The slow one — the s-process — is the subject of this page.

The slow walk: one neutron at a time

The s-process operates wherever a star can generate free neutrons at moderate density (~ 10⁷ to 10¹¹ per cm³) inside material containing pre-existing iron-group "seed" nuclei. The cycle is simple:

(Z, A) + n  →  (Z, A+1) + γ
(Z, A+1)    →  (Z+1, A+1) + e⁻ + ν̄_e    (if (Z, A+1) is β-unstable)

Every capture adds one mass unit (A → A+1) without changing the proton number Z. If the new isotope is stable, it sits and waits for the next neutron — typically 10² to 10⁵ years away at AGB intershell densities. If it is unstable, it β-decays in seconds, minutes, or years to (Z+1, A+1) before the next capture arrives. Average a few thousand of these steps together and you trace a path that runs adjacent to the valley of beta stability, branching upward in Z at every short-lived intermediate.

That single picture — capture, wait, decay, capture again — is the entire microphysics of the s-process. What changes from star to star is the neutron source, the seed abundance, the density and the duration. The path is determined by nuclear physics alone.

Where the neutrons come from

Free neutrons are not lying around inside stars. They have to be produced by a reaction that liberates them from nuclei. Two reactions account for essentially the entire stellar s-process budget:

Reaction	Site	Temperature	Branch	Mass range
¹³C(α,n)¹⁶O	AGB intershell, interpulse phase	~ 10⁸ K (~ 8 keV)	main s-process	A ≈ 90 – 208
²²Ne(α,n)²⁵Mg	Massive-star He core / C shell	~ 3 × 10⁸ K (~ 23 keV)	weak s-process	A ≈ 60 – 90
²²Ne(α,n)²⁵Mg	AGB thermal pulse	~ 3 × 10⁸ K	secondary boost	branch-point isotopes

The dominant source for heavy s-process products is ¹³C(α,n)¹⁶O, which has a low Coulomb barrier and operates efficiently at the relatively cool temperatures characteristic of AGB intershell layers. The ¹³C itself is made when protons from above mix downward into a carbon-rich pocket and react with ¹²C via ¹²C(p,γ)¹³N(β⁺)¹³C. This "¹³C pocket" is the engine of the main s-process. The Coulomb barrier for the α + ¹³C reaction is just low enough that it ignites at 10⁸ K, a temperature regularly reached in the long, quiet interpulse periods between AGB thermal pulses.

In massive stars (≳ 8 M☉), the picture is different. ¹³C is destroyed quickly by ¹³C(α,n) earlier in the evolution; the relevant neutron source there is ²²Ne(α,n)²⁵Mg, which only activates at the much higher temperatures (~ 3 × 10⁸ K) reached during core helium burning and especially shell carbon burning. The hotter source releases neutrons in shorter, more intense bursts and tops out around A ~ 90 before the seed material is exhausted. This is the "weak" s-process; it dominates the Sr/Y/Zr region and contributes much of the solar abundance of light heavy elements.

Inside an AGB star

Asymptotic giant branch stars — Sun-like stars in their late, swollen, doubly-shell-burning phase — host the main s-process. The interior structure during this stage is layered: an inert C/O core, a thin helium-burning shell, an intershell layer of He and ¹²C, a thin hydrogen-burning shell, and a vast convective envelope above. The action is in the intershell.

An AGB star alternates between two states. For most of its time (the interpulse phase, ~ 10⁴–10⁵ years), the H-burning shell ticks along quietly, depositing fresh helium into the intershell. Periodically (every 10⁴ to 10⁵ years), enough He accumulates to ignite explosively in a thermal pulse — a runaway He flash that briefly dumps L ~ 10⁸ L☉ into a thin shell. Convection from the pulse extends upward; in subsequent expansion, the base of the envelope reaches down into former intershell material (the "third dredge-up") and mixes processed elements to the surface.

The s-process happens between pulses, in the quiet ¹³C pocket. Neutrons are released at densities of ~ 10⁷ cm⁻³ and bathe the seed nuclei for thousands of years. After each pulse, dredge-up exposes the surface to fresh s-process material; over many cycles, the envelope accumulates substantial s-enrichment. Stellar winds — at AGB mass-loss rates of 10⁻⁵ to 10⁻⁴ M☉/yr — then carry that enriched gas into the interstellar medium. After tens to hundreds of thousand-year cycles, the envelope is gone, the star becomes a planetary nebula and a cooling white dwarf, and what remains of the s-process products has been donated to the next generation of stars.

Walking the chart of nuclides

If you draw a chart of nuclides (N on the horizontal axis, Z on the vertical), the valley of stability is a diagonal band running from the lower left to the upper right. The s-process path stays inside this band. Starting from ⁵⁶Fe, the trajectory goes ⁵⁶Fe → ⁵⁷Fe → ⁵⁸Fe → ⁵⁹Fe(β⁻, 44 d) → ⁵⁹Co → ⁶⁰Co(β⁻, 5.3 y) → ⁶⁰Ni, and so on. At every step there's a choice: capture one more neutron, or wait for the current isotope to β-decay. The β-decay half-life is what controls the choice. In the canonical s-process, β half-lives are seconds to years and capture times are millennia, so β decay always wins on the unstable side.

There are interesting exceptions — branch points where the half-life is comparable to the capture time. At ⁸⁵Kr (half-life 10.7 yr at 8 keV), ⁹⁵Zr (64 d), ¹³⁴Cs (2.06 y), ¹⁵¹Sm (90 y), ¹⁷⁹Hf and others, the flow splits: some nuclei β-decay, some capture another neutron first. The ratio of products at a branch point is a thermometer/densitometer for the s-process site — if the neutron density were higher, more flow would push through the capture channel. Isotopic ratios in pre-solar SiC grains preserved in meteorites carry exquisite records of these branch points and are among the strongest constraints on AGB model parameters.

Magic numbers and abundance peaks

The s-process abundance curve is not smooth. Solar abundance vs. mass number shows three sharp local peaks:

Peak	Mass number	Closed shell	Signature elements
Light s-peak	A ≈ 88	N = 50	Sr, Y, Zr
Heavy s-peak	A ≈ 138	N = 82	Ba, La, Ce, Nd
Lead peak	A = 208	N = 126	²⁰⁸Pb (also Z = 82 — doubly magic)

These mass numbers correspond exactly to the magic neutron numbers — the closed-shell configurations of the nuclear shell model, analogous to the closed electron shells of the noble gases. At a closed neutron shell, the (n,γ) cross-section drops by roughly an order of magnitude. Capture is suppressed; flow piles up. Material accumulates at the magic-number isotopes until enough flux integrates over enough time to push through, and the resulting solar abundance is correspondingly enhanced.

The lead peak is special. Once the flow reaches ²⁰⁸Pb (doubly magic, both N = 126 and Z = 82), it stalls. Any α decay above ²⁰⁸Pb returns the flow toward the same peak — so a leaky population of unstable transbismuth isotopes feeds back into ²⁰⁸Pb rather than continuing. As a result, the s-process produces almost no Th or U; those elements are essentially pure r-process. The Pb peak is itself a useful chronometer: lead-rich, metal-poor stars show that early-Universe AGB enrichment from low-metallicity progenitors preferentially terminates at the Pb peak (a "strong-component" s-process driven by very few seeds).

Merrill's technetium — proof in real time

The strongest single piece of observational evidence for s-process operation comes from a 1952 paper by Paul W. Merrill (ApJ 116, 21). Using high-dispersion spectroscopy at Mount Wilson, Merrill identified absorption lines of technetium in the spectra of S-type red giants. Technetium has no stable isotopes; its longest-lived isomer, Tc-99, has a half-life of 211 000 years. Cosmologically and even galactically, that is instantaneous. For Tc to appear in a stellar atmosphere, it must have been synthesised inside that star within the last few hundred thousand years and then dredged up to the surface — exactly the cadence predicted for AGB s-process plus third dredge-up.

No competing mechanism could be made to work. Slow neutron capture, predicted in detail by Cameron and by B²FH a few years later, was the unique explanation. Merrill's observation thus elevated the s-process from a theoretical proposal to a directly observed astrophysical process. Subsequent detections of Tc in many AGB stars — including, by Smith and Lambert (1985, 1986), correlations between Tc presence and overall s-process enrichment — have only deepened the case.

Decomposing the solar abundance curve

Imagine taking the solar abundance distribution and partitioning each isotope into s-process and r-process contributions. For some nuclides this is easy. An isotope shielded from the r-process by a stable neighbour on its β-decay chain — for instance ¹⁵⁰Sm shielded by ¹⁵⁰Nd, or ¹⁵²Gd shielded by ¹⁵²Sm — can only have come from the s-process: it is an s-only isotope. Conversely, isotopes that the s-process cannot reach because the path forks around them via a short-lived bypass — for instance ⁹⁶Zr blocked by ⁹⁵Zr's branching — are essentially r-only.

Element-by-element s-process fractions in the solar mix (approximate, from Sneden, Cowan and Gallino 2008 and Bisignano 2003 decompositions):

Element	Z	s-process fraction	r-process fraction	Notes
Sr	38	~ 85 %	~ 15 %	Light s-peak
Y	39	~ 92 %	~ 8 %	Light s-peak
Zr	40	~ 80 %	~ 20 %	Light s-peak
Ba	56	~ 85 %	~ 15 %	Heavy s-peak
La	57	~ 75 %	~ 25 %	Heavy s-peak
Ce	58	~ 80 %	~ 20 %	Heavy s-peak
Eu	63	~ 5 %	~ 95 %	Canonical r-process tracer
Pb	82	~ 90 %	~ 10 %	Lead peak
Th	90	~ 0 %	~ 100 %	Beyond Pb — pure r-process
U	92	~ 0 %	~ 100 %	Beyond Pb — pure r-process

Using stellar spectroscopy, this decomposition can be inverted: the ratio of an s-tracer (Ba) to an r-tracer (Eu) in any given star tells you whether its heavy-element enrichment came preferentially from AGB winds (high Ba/Eu) or from r-process events (low Ba/Eu). Stars with extreme Ba/Eu or Eu/Ba are spectroscopic fossils of single dominant enrichment events.

CEMP stars and AGB mass transfer

About 20 percent of metal-poor halo stars ([Fe/H] < −2) are carbon-enhanced. Many of these CEMP (Carbon-Enhanced Metal-Poor) stars also show extreme s-process enrichment — Ba abundances elevated by factors of 100 to 1000 relative to scaled solar. These are CEMP-s stars, and the leading interpretation is binary mass transfer: a more massive companion went through its AGB phase, dumped C-rich, s-process-rich material onto the surviving lower-mass star (now observed), then died as a white dwarf. CEMP-s stars are therefore archaeological records of early-Universe AGB nucleosynthesis at very low metallicity.

Their patterns reveal something striking. Low-metallicity AGB stars produce a much heavier-biased s-process distribution, dominated by lead — because at low Fe abundance there are few seed nuclei per neutron and the flow runs all the way to ²⁰⁸Pb before exhausting itself. The most extreme lead-rich CEMP-s stars have [Pb/Fe] > 3, an enhancement of 1000× over solar Pb/Fe, with [Ba/Fe] still strongly elevated. They confirm the "strong-component" s-process predicted by Gallino, Busso and collaborators in the 1990s — the same nuclear physics, run at low metallicity, redistributes the abundance pattern toward higher masses.

The cosmic budget

How much of your body and your planet came from the s-process? Earth's barium and strontium are mostly s-process. The lead in your batteries is ~ 90% s-process. The cerium that polishes your camera lens, the lanthanum in nickel-metal-hydride batteries, the zirconium in your dentist's drill — predominantly s-process. Cumulatively, of every kilogram of elements heavier than iron on Earth, roughly half was forged by slow neutron capture inside long-vanished red giants whose dusty winds eventually mixed into the solar protostellar nebula 4.6 billion years ago.

The other half came from the r-process — kilonovae and core-collapse supernovae. Together, these two mechanisms split the heavy-element periodic table almost cleanly down the middle. Without both, the chemistry that supports life would be impossible: gold for jewellery and electronics from r-process; barium for medical imaging from s-process; copper and zinc for biochemistry from a mixture of processes earlier in the chain. The s-process is half the heritage of the world above iron — and Merrill's technetium proves it's still running, right now, in the swollen envelopes of dying Sun-like stars across the galaxy.

Variants and components

• Main s-process. AGB intershell, ¹³C(α,n)¹⁶O at ~ 10⁸ K. Produces A ≈ 90 to 208, the bulk of Ba, La, Ce, Nd, Sm, Pb in the solar mix. Operates over ~ 10⁵ years per pulse cycle.
• Weak s-process. Massive-star He core and C shell, ²²Ne(α,n)²⁵Mg at ~ 3 × 10⁸ K. Produces A ≈ 60 to 90, dominates Sr, Y, Zr. Operates on ~ 10³–10⁴ yr timescales.
• Strong s-component. Inferred to explain ²⁰⁸Pb production in the Sun; requires extremely low metallicity AGB progenitors with very few seeds per neutron. Identified in lead-star CEMP-s spectra.
• i-process (intermediate). Neutron density 10¹⁴ to 10¹⁵ cm⁻³ — between s and r. Proposed for some early-Universe stars, possibly in late thermal pulses of low-metallicity AGB stars or H-ingestion events; explains some CEMP abundance patterns the pure s-process cannot reproduce.
• Branch-point thermometers. Specific isotopic ratios at branching points (e.g. ¹³⁴Cs, ¹⁵¹Sm) probe neutron density and temperature; their values in pre-solar SiC grains constrain the conditions inside individual long-dead AGB stars.

Where the s-process shows up observationally

• S-type red giants. Cool AGB stars with surface enrichment in Zr, Y, La, Ba and visible Tc. The original Merrill (1952) targets.
• Barium stars. G/K giants on the red-giant branch with strong Ba II lines. Long thought puzzling — too small to be running AGB themselves — and explained as mass-transfer binaries from now-WD companions that did go through AGB.
• Carbon stars (C/N/J/R types). AGB stars with surface C/O > 1 from dredge-up, often s-process enhanced.
• CEMP-s stars. Carbon-Enhanced Metal-Poor halo stars enriched by binary AGB mass transfer at very low metallicity. Sample most-extreme s-process patterns including lead stars with [Pb/Fe] > 3.
• Pre-solar SiC grains. Microscopic silicon-carbide condensates found in meteorites, carrying isotopic signatures of single AGB stars. Spectacularly precise constraints on branch-point physics.
• Planetary nebulae. The shed AGB envelope, observed before it dissolves into the ISM. Surface abundances reflect the integrated s-process enrichment of the progenitor.

Common pitfalls

• "Slow" is relative. The s in s-process means slow compared to β decay of the relevant intermediate isotopes — typically capture times of 10²–10⁵ yr versus β half-lives of seconds to years. On terrestrial timescales, both are absurdly slow.
• The s-process does not start from iron alone. Iron is the dominant seed because it has the highest abundance in the iron group, but every isotope from ²⁰Ne up to ⁵⁶Fe can absorb neutrons. In low-metallicity AGB stars with few seeds per neutron, the flow loads up onto the few available seeds and pushes them all the way to lead.
• Branch points are not just decoration. They are how observers infer the neutron density. Mistaking a branch ratio for a pure abundance ratio without temperature/density corrections leads to inconsistent AGB models.
• The ¹³C pocket is not a free parameter. Its mass and profile are set by mixing physics at the convective envelope boundary that is still poorly modelled. Reasonable AGB s-process predictions can vary by factors of several depending on assumed ¹³C pocket structure.
• Don't conflate s-process with all of stellar nucleosynthesis. The s-process is one chain, responsible for ~ half of the heavy elements. CNO, He, carbon and oxygen burning, p-process, r-process — each has its own site, conditions and products. They overlap in time and even in some stars (massive stars run weak s-process plus pre-supernova hydrostatic burning) but are physically distinct.