Stellar Metallicity: Reading the Chemical Fingerprint of Stars

When a chemist hears "metal", they think iron, copper, gold — solid, conducting, sitting on the left of the periodic table. When an astronomer says "metal", they mean anything heavier than helium. Carbon is a metal. Oxygen is a metal. Neon, by all rights a noble gas, is also a metal in this dialect. The terminology is jarring at first, but there's a reason for it: in stars, what matters isn't the chemistry of an element but whether it was around at the dawn of time or whether it had to be cooked up later.

Hydrogen and helium were forged in the first three minutes after the Big Bang, in a runaway fusion called Big Bang nucleosynthesis. Together they account for about 98% of the atomic mass in the universe — and almost all the rest is helium-4. A trace of lithium snuck through too. Then nucleosynthesis ran out of steam, the universe cooled below the threshold for fusion, and the cosmic recipe locked in: roughly 75% hydrogen, 24% helium, and 1% everything else. That "everything else" had to wait for stars.

The metallicity of a star is the fraction of its mass made of those everything-else elements. Astronomers call it Z, and for the Sun, Z = 0.014 — the Sun is 1.4% metals by mass. Hydrogen accounts for about 73.8% of the Sun's mass (called X), helium about 24.8% (Y), and that 1.4% rounding error is everything from lithium to uranium.

Z is fine for averages, but for individual stars astronomers prefer a more sensitive instrument: the iron abundance, expressed in the ratio [Fe/H]. The brackets and the slash are a logarithmic convention that lets you compare two ratios with one number:

Iron is a stand-in for "all metals" because its spectrum is dense with absorption lines that are easy to measure precisely. In practice [Fe/H] tracks the overall metal content well enough that astronomers use the words "metal-rich" and "high [Fe/H]" almost interchangeably.

The remarkable thing about that last number is that nothing in the laws of physics requires it to exist at all. Stars with [Fe/H] below −5 are statistically extreme. Each one is a survivor — a small star that formed nearly 13 billion years ago in a galaxy that had barely begun to enrich itself, and which has been quietly burning ever since. Finding them is one of the great treasure hunts of modern astronomy, and we'll come back to that hunt in the last section.

For most of human history, the obvious working hypothesis was that the Sun was made of stuff like the Earth — rock, iron, maybe a fiery surface above some molten interior. The 19th-century chemists who first split sunlight through a prism saw thousands of dark absorption lines, and they cleverly identified them by matching laboratory spectra: hydrogen, sodium, calcium, iron, magnesium. The Sun had iron in it, just like the Earth did. Therefore, the reasoning went, the Sun was something like the Earth.

That seemed reasonable until 1925, when a 25-year-old graduate student at Harvard named Cecilia Payne-Gaposchkin submitted her doctoral thesis. She had taken Meghnad Saha's then-new ionization theory — which related the strength of a spectral line to the temperature, pressure, and abundance of the element producing it — and applied it carefully to the spectra of dozens of stars. She came out the other side with a stunning conclusion: the Sun and the other stars are not Earth-like at all. They are made overwhelmingly of hydrogen and helium. The metals — including the iron we see so clearly in solar spectra — are present in trace amounts, less than a few percent.

Her advisor, Henry Norris Russell, urged her to soften the claim. "It is clearly impossible," he wrote her, that the abundance of hydrogen could be a million times that of the metals. She added a hedge to her thesis at his suggestion. Four years later, Russell did the calculation himself and confirmed she was right. Hydrogen really is a million times more abundant than iron in the Sun, by atom count. He published a generous paper crediting her, but generations of textbooks gave him the credit anyway.

It was the moment astronomy discovered what stars are made of. In one stroke, the universe shifted from being a thinly scattered version of the Earth to being a vast hydrogen-helium ocean with traces of everything else floating in it. The chemistry of the cosmos had a new floor plan.

Her result also implied something the field hadn't yet grasped: if stars are mostly hydrogen, and if they shine for billions of years, then that hydrogen has to be the fuel. The path was now clear from "what are stars made of" to "how do they shine" — fusion of hydrogen into helium, with traces of heavier elements being cooked along the way. Every later question about metallicity flows from this 1925 foundation.

The next twist in the story came from one of those happy accidents that science occasionally hands out. In 1943, World War II had blacked out the city of Los Angeles for fear of Japanese air raids. The dark skies above Mount Wilson — sitting on a ridge at 1,742 m above the LA basin — became the darkest they had been in decades. A German-born astronomer named Walter Baade, classified as an enemy alien and barred from war work, had Mount Wilson largely to himself. He pointed the 100-inch Hooker telescope at the Andromeda Galaxy and exposed photographic plates for hours.

He resolved something nobody had managed before: he could see individual stars in M31's central bulge, not just in its outer spiral arms. And the stars in those two regions were not the same. The bulge stars were mostly red giants — old, cool, evolved stars. The spiral-arm stars were hotter, bluer, and dotted with bright young stars and HII regions where new stars were still forming. The two regions weren't just spatially distinct; they were chemically and dynamically distinct, with different ages and different histories.

Baade called them Population I (the disk: young, metal-rich, dusty, full of new stars) and Population II (the halo and bulge: ancient, metal-poor, gas-poor, no recent star formation). The Sun is a Population I star — a relatively young, metal-rich resident of the galactic disk. Globular clusters are Population II — most of their stars are ten billion years old and were already on the main sequence when the Sun was condensing out of its molecular cloud.

The two-population scheme was an immediate hit. It explained why the disk was thin (gas can collapse into a disk; old, dynamically heated stars cannot) and why the halo was puffy (no gas left to dissipate orbits). It explained why disk stars contain heavy elements (they formed from gas already enriched by earlier supernovae) and why halo stars contain almost none (they formed before there had been many supernovae). It explained why the Cepheid period-luminosity relation, calibrated by Henrietta Leavitt on Pop I Cepheids in the Magellanic Clouds, gave the wrong distance to M31 when applied to M31's Pop II RR Lyrae stars: the two populations have different physics.

For Nightbase observers, this is something you can see for yourself. Aim binoculars or any small telescope at Andromeda on a dark night. The bright central core glows yellow-white — that's the Pop II bulge, dominated by old red giants. As you sweep outward to the disk, the colour shifts subtly cooler and bluer, and at the spiral-arm radii you'll find the patchy texture of dust lanes and HII regions where Pop I stars are still being born. With averted vision and a 10-inch scope, M31's two-toned face becomes one of the most direct demonstrations of stellar populations you can witness from your back garden.

Knowing that stars sort into populations is one thing. Knowing why — what chemistry actually links a star's age to its iron content — is another, and that connection wasn't pinned down until 1957.

In that year, four authors — Margaret Burbidge, Geoffrey Burbidge, William Fowler, and Fred Hoyle — published a 104-page paper in Reviews of Modern Physics titled "Synthesis of the Elements in Stars". Within the field it is universally referred to by the authors' initials, B²FH. It is one of the most influential papers in 20th-century astrophysics. In it, the four authors laid out the entire menu of nuclear processes by which a star can build heavy elements out of light ones, and in so doing explained the relative abundances of every isotope in the periodic table.

The argument went something like this. A star spends its main-sequence life fusing hydrogen into helium in its core (the p-p chain and the CNO cycle). When the hydrogen in the core runs out, the core contracts, heats, and starts fusing helium into carbon and oxygen via the triple-alpha process. Massive stars then move on to fusing carbon into neon, magnesium, silicon; silicon into iron; and at iron the chain stops, because iron-56 is the most tightly bound nucleus and any further fusion costs energy rather than releasing it. The collapsing iron core triggers a supernova, and in the supernova's explosive nucleosynthesis the elements heavier than iron are forged through the rapid capture of free neutrons (the r-process). Slower neutron captures in the cores of giant stars produce a different signature (the s-process), responsible for elements like barium and lead.

B²FH put it all together. Every element heavier than lithium is the fossil of a particular nuclear pathway in a particular kind of star. Carbon and oxygen come from helium burning in red giants. Iron-peak elements come from massive-star core fusion and from Type Ia supernovae. Gold and uranium come from neutron-star mergers and explosive r-process events. The periodic table you remember from chemistry class is a map of stellar deaths.

Once you grasp B²FH, the link between metallicity and age becomes obvious. Each generation of stars dies and seeds the interstellar medium with metals it has cooked. The next generation forms from gas slightly richer in those metals. Over time, the average metallicity of the gas in a galaxy creeps up. So a star's [Fe/H] is, very roughly, a clock — high [Fe/H] means the star formed late in cosmic history, low [Fe/H] means it formed early. The relation is messier than a clock at the level of any individual star (some regions of the galaxy enrich faster than others; some stars migrate; mergers stir the pot), but on the population level the trend is unmistakable.

This is also why the absence of certain populations is informative. Pop III stars — the ones that supposedly formed from primordial gas with [Fe/H] = −∞ — should mark the very first generation. They are predicted by theory but have never been directly observed. The leading explanation is that they were exclusively very massive (because primordial gas, lacking metal-line cooling, couldn't fragment into low-mass clumps), and they all died as supernovae or collapsed to black holes within a few million years of forming. None survived to become red dwarfs we could see today. The James Webb Space Telescope is searching for their light at redshift z > 10, and may have caught the chemical fingerprint of the very first stellar deaths in a few of its targets.

How do astronomers actually measure [Fe/H] for a star they've never visited? They use the same trick the 19th-century chemists used: a prism. Spread a star's light across a long strip of wavelengths and an unbroken rainbow appears, except for the dark gaps where atoms in the star's atmosphere have absorbed photons of specific energies. Each element imprints a unique pattern of absorption lines on the spectrum, like a chemical barcode.

The deeper a metal line, the more atoms of that element are sitting in the star's atmosphere absorbing photons. By measuring how much light is missing — the line's equivalent width, the width of an imaginary rectangular notch with the same total absorption — astronomers infer how many atoms are doing the absorbing, and from there the ratio to hydrogen. Modern echelle spectrographs split a single star's light into hundreds of orders, exposing thousands of individual lines per spectrum. Tools like MOOG, Turbospectrum, and iSpec compare the observed line strengths to synthetic model atmospheres and back out [Fe/H], the temperature, the surface gravity, and a dozen individual element abundances. Modern survey programs — APOGEE, GALAH, the Gaia-ESO survey — have measured [Fe/H] for millions of stars across the Milky Way.

The most useful diagnostic features in a stellar spectrum, especially for cool yellow stars like the Sun, are these:

• Ca II H and K at 3934 and 3968 Å — two black trenches that are the deepest features in the violet end of any solar-type spectrum. So strong they remain visible in even the most metal-poor halo stars, and the ratio of their depth to the surrounding continuum is a very sensitive metallicity indicator.
• The G-band near 4300 Å — a forest of CH molecular lines that fades dramatically as carbon abundance drops. The original "G" line, named by Joseph Fraunhofer when he first catalogued the solar spectrum.
• The Mg b triplet at 5167–5184 Å — a beautiful trio of magnesium lines that astronomers use to break degeneracies between metallicity and surface gravity in unresolved galaxies.
• The Na D doublet at 5890–5896 Å — sodium, easy to see in any small spectroscope.
• A blizzard of Fe I lines scattered through the green and yellow — the workhorse iron lines from which [Fe/H] is actually computed.

Try it — the metallicity slider. Drag the [Fe/H] slider below to see how a synthetic spectrum changes as a star's metal content shifts from −3 dex (a halo subdwarf) to +0.5 dex (a metal-rich disk giant). Watch the calcium and iron lines deepen as the slider moves right, while hydrogen lines barely flinch. The synthetic colour swatch on the right shows the resulting B−V drift — the star's slight reddening as more lines suck blue photons out of the spectrum.

This brings up a counter-intuitive consequence of how spectra work: because absorption lines are concentrated in the blue and ultraviolet end of the spectrum, a star with more metals is redder than a star at the same temperature with fewer metals. The effect is called line blanketing — the carpet of metal lines acts as an opacity blanket that suppresses blue continuum flux. A halo subdwarf at 6000 K is noticeably bluer than a disk star at 6000 K, even though they have identical effective temperatures. If you're trying to estimate a star's temperature from its B−V colour alone, you need to correct for metallicity — and conversely, metallicity itself becomes measurable from a star's deviation from the colour you'd expect at its spectral type.

For more on how spectra are taken apart line by line, see stellar absorption spectra. For the photometric side — how a star's colour reveals temperature — see the B−V colour index.

There is no better laboratory for stellar populations than a globular cluster. Unlike most groups of stars, a globular's members all formed from the same molecular cloud at roughly the same time, sharing a common metallicity and common age. Once you know the cluster's distance, you can plot a colour–magnitude diagram (CMD) — basically a Hertzsprung-Russell diagram of one cluster — and read its age and metallicity directly off the shape of the diagram.

The Milky Way has about 150 known globulars, and they are distributed in a roughly spherical halo around the galactic centre. Almost all of them are 10 to 13 billion years old, and almost all of them are metal-poor. The most metal-poor globulars (M15, M92, NGC 5466) sit near [Fe/H] = −2.3 — about 0.5% of the Sun's iron content. The most metal-rich (47 Tucanae, NGC 6388, NGC 6624) reach [Fe/H] ≈ −0.7 — still well below solar.

You can see the metallicity differences with your own eyes if you compare clusters at the eyepiece. M13, the Great Hercules Cluster, has [Fe/H] ≈ −1.5 and a slightly warmer cast to its giant branch. 47 Tucanae (visible only from the southern hemisphere) is much more metal-rich at [Fe/H] ≈ −0.7, and its red giants are noticeably more orange in long-exposure colour images. M15 in Pegasus and M92 in Hercules are both metal-poor at [Fe/H] ≈ −2.3, with bluer horizontal-branch stars dominating their visual appearance.

Why does metallicity affect a cluster's appearance? Two big reasons. First, line blanketing again — metal-rich stars at the same temperature are slightly redder, so a metal-rich cluster's red-giant branch sits at lower temperatures and is visibly redder. Second, the horizontal branch — the strip of helium-burning stars that lies above the main sequence — is bluer in metal-poor clusters. Why? In metal-poor stars the envelope is more transparent, allowing the star to puff up to lower densities and shed more of its envelope on the red-giant branch, which leaves a smaller, hotter remnant on the horizontal branch. The "second-parameter problem" of why two clusters with the same [Fe/H] sometimes have different horizontal branches is one of the open puzzles of stellar evolution; the leading suspects are age, helium content, and the cluster's mass.

The most spectacular globular for our purposes is Omega Centauri, the king of the southern sky. Visible to the naked eye as a fuzzy 4th-magnitude star in Centaurus, it's the largest and most luminous globular in the Milky Way — about 3.5 million solar masses. And here's the wonderful complication: ω Cen's stars don't share a single metallicity. They span [Fe/H] from −2.0 all the way to −0.6, divided into at least three or four distinct populations. The leading hypothesis is that ω Cen isn't a globular cluster at all but the stripped core of an ancient dwarf galaxy the Milky Way ate billions of years ago, with each population representing a distinct burst of star formation in the dwarf's history before its outer layers were tidally torn away.

If you're far enough south to point a scope at it, ω Cen is one of the most rewarding objects in the sky — and now you have a reason beyond its sheer visual splendour to spend an evening on it.

Astronomers organise everything in this article into three populations, each named for its place in the chronology of cosmic chemistry:

• Population III — the hypothetical first stars. Born from pristine Big Bang gas, [Fe/H] = −∞ in principle. None survive today (they were all very massive and died young). Their existence is inferred from the very low but nonzero metallicity floor we see in even the oldest known halo stars, which must have formed from gas already polluted by some generation before them. JWST is hunting their ghost light at redshift z > 10.
• Population II — the second generation. Born from gas barely enriched by Pop III deaths. [Fe/H] roughly −1 to −4. Globular cluster stars, halo subdwarfs, and the oldest stars in the galactic bulge. Today they live on highly inclined orbits that take them far above and below the plane of the Milky Way. Most are old red dwarfs and red giants; the metal-poor main-sequence subdwarfs are slightly bluer than their disk counterparts at the same temperature.
• Population I — the modern crowd. Born from gas heavily enriched by countless previous supernovae. [Fe/H] roughly −0.5 to +0.5. Disk stars, open clusters, OB associations, and the Sun. They orbit in the disk on nearly circular paths, sometimes with traces of the molecular cloud they came from — gas, dust, even traces of birth-companion stars in young moving groups.

Walter Baade's original two-population scheme has, over the decades, been split into finer subdivisions: thin-disk Pop I, thick-disk intermediate, halo Pop II, bulge stars (which span a wide metallicity range and don't fit neatly into either Pop I or Pop II). The galactic archaeology programs of the 21st century — Gaia, APOGEE, GALAH, the upcoming 4MOST and PLATO surveys — are turning this into a continuous map of the galaxy in chemical-dynamical space. You can now point at almost any star in the local volume and read off not just its position but its birth radius in the disk, its likely age, and the kind of nucleosynthesis events (Type Ia supernovae vs. core-collapse vs. neutron-star mergers) that contributed to its composition.

For a real-time feed of which populations are well-placed tonight, see the Tonight view, or browse the full sky on the interactive star map. The catalog detail page for any object lists its constellation, magnitude, and — for cluster-bearing objects — typical [Fe/H], so you can plan a night around chemically-themed targets.

A final thought to leave with. Every clear night, the photons reaching your eye from M31 left that galaxy 2.5 million years ago, when Homo habilis was just figuring out how to chip flint. The photons reaching you from M13 left it 22,000 years ago, somewhere in the Last Glacial Maximum. The photons from a halo subdwarf in the local solar neighbourhood, with [Fe/H] = −2 and an age of 12.5 billion years, left that star a few decades ago — but the iron those photons reveal in its atmosphere is older than every other element in your body. Looking at a star is, in a real sense, looking back through the entire chemical history of the universe. And [Fe/H] is the meter stick that lets you read the date.