The first three minutes

Here is a fact that should be harder to believe than it is: roughly a quarter of every ordinary atom in your body, in the Sun, in the most distant galaxy we can see, is helium that was not made in any star. It was forged in the first few minutes after the Big Bang, when the whole universe was hot and dense enough to run as a single fusion reactor, and then it was switched off forever once everything cooled. That helium is a fossil. It predates the first star by hundreds of millions of years, it cannot be unmade, and we can calculate how much of it there should be from nuclear physics you could fit on a napkin.

The astonishing part is not the story. The astonishing part is the number. Theory says the primordial helium should be 24.7% of ordinary matter by mass. Go and measure it in the most pristine, least-processed gas we can find, dwarf galaxies so metal-poor that almost nothing has been cooked in stars since the beginning, and you get 24.6%. That agreement, to within about a percent, is not a vague "the Big Bang basically works". It is a quantitative prediction about the composition of the universe, made from the temperature it had when it was three minutes old, confirmed by light that left a galaxy before the Earth existed.

Let me try to convince you the number is forced, not fitted.

A reactor with a cooling problem

To make helium you need to fuse protons and neutrons. The early universe had both, in vast quantities, slammed together at billions of kelvin. So why didn't all of it become helium? Why is the universe still three-quarters hydrogen?

The answer is a race. The universe is fusing nuclei and cooling at the same time, and cooling wins. Everything about primordial nucleosynthesis is a competition between reaction rates, which want to build heavier nuclei, and the expansion of space, which is dropping the temperature and density out from under those reactions before they can finish the job.

So the whole problem reduces to a stopwatch. How many neutrons are around when fusion finally fires, and what fraction of them get locked into helium before the universe cools too far? Get those two numbers and you have the helium abundance. Everything else is detail.

Freeze-out: the universe forgets how to make neutrons

Rewind to about one second after the Big Bang, temperature near 10 billion kelvin, or about 1 MeV in the units particle physicists prefer. The cosmos is a soup of photons, electrons, positrons, neutrinos, and a thin sprinkling of protons and neutrons. At these energies the weak force keeps protons and neutrons interconverting freely: a proton plus an electron becomes a neutron plus a neutrino, and the reverse, back and forth, fast.

While those reactions are fast, the ratio of neutrons to protons is set by thermodynamics. A neutron is slightly heavier than a proton, by $\Delta m = 1.293$ MeV, and nature charges you that energy difference to make the heavier one. So the populations follow a Boltzmann factor: the colder it gets, the rarer neutrons become.

$$\frac{n}{p} = e^{-\Delta m / T}, \qquad \Delta m = 1.293\ \text{MeV}$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

Now bring in the expansion. The weak reactions that maintain this balance have a rate $\Gamma_{\text{wk}}$ that falls steeply as the universe cools, much faster than the expansion rate $H$ does. At some point the reactions can no longer keep up with the expanding, thinning plasma. The conversions effectively stop. Physicists call this freeze-out, and it happens when the two rates cross.

$$\Gamma_{\text{wk}} < H \quad \Rightarrow \quad \frac{n}{p} \approx e^{-\Delta m / T_f}, \quad T_f \approx 0.8\ \text{MeV}$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

Freeze-out lands at roughly $T_f \approx 0.8$ MeV. Plug that into the Boltzmann factor and you get a neutron-to-proton ratio of about $1/6$. The universe has, in effect, forgotten how to convert protons into neutrons. Whatever neutrons exist at this instant are nearly all the neutrons that will ever be available to build nuclei.

There's one more twist before fusion starts. A free neutron is unstable; it decays into a proton with a half-life of about 10 minutes. Between freeze-out and the start of nucleosynthesis a few minutes elapse, and in that gap some neutrons quietly decay away. The ratio slides from about $1/6$ down to roughly $1/7$. That decay is a slow leak, and it's racing the same stopwatch: the sooner fusion starts, the fewer neutrons are lost. As it happens, fusion starts just in time to save most of them.

The deuterium bottleneck

You'd think that with neutrons and protons both present at MeV temperatures, helium would assemble immediately. It can't, and the reason is a single fragile nucleus standing in the doorway.

The first step toward helium is making deuterium: one proton plus one neutron, the simplest compound nucleus. But deuterium is weakly bound, held together by only 2.22 MeV. In a bath of energetic photons, any deuteron that forms is instantly blown apart again by a gamma ray before it can react further. This is the deuterium bottleneck. Until the photon bath cools enough that deuterons survive, no helium can be built, because there's no stepping stone to stand on.

When the gate finally opens, the whole episode runs at $T_{\text{BBN}} \sim 10^{9}$ K (about 0.1 MeV) and roughly three minutes in. Everything then happens fast. Deuterium piles up, then rapidly fuses up the chain to helium-3, tritium, and finally the very tightly bound helium-4. Helium-4 is a dead end: it's so stable, and the gaps in the chart of nuclei at mass 5 and mass 8 are so unforgiving, that almost everything funnels into helium-4 and stops there. Below 0.1 MeV the reactions themselves fall behind the expansion and freeze out, just as the weak interactions did earlier, and the reactor goes dark.

That mass gap is why the Big Bang made no carbon, no oxygen, no iron. There is no stable nucleus at mass 5 or mass 8 to bridge across, and by the time you'd need three-body collisions to leap the gap the universe is too thin and too cold for them to happen. The early universe builds hydrogen, helium, a trace of lithium, and then quits. Every heavier element in existence, every atom of the calcium in your bones and the iron in your blood, was made later, inside stars. Cosmology hands you the first two rungs of the periodic table; stars build the rest.

A quarter, from one ratio

Now the payoff. Once deuterium survives, nearly every available neutron ends up bound into a helium-4 nucleus, because helium-4 is the most tightly bound light nucleus and the chain runs straight to it. So the helium abundance is essentially a counting exercise. Take the neutrons; pair each with a proton; the result is helium.

A helium-4 nucleus holds 2 neutrons and 2 protons. If the neutron-to-proton ratio is $n/p$, then for every pair of neutrons you consume two protons to match, and the mass fraction of helium works out to a clean formula.

$$Y_p \approx \frac{2(n/p)}{1 + (n/p)}$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

Put $n/p \approx 1/7$ into equation (3): the numerator is $2/7$, the denominator is $8/7$, and the ratio is $2/8 = 0.25$. A quarter. The helium fraction of the universe is, to a first approximation, just the geometry of pairing up a population of neutrons that's one-seventh the size of the proton population. No fine-tuning, no free parameters in this estimate. The neutron-to-proton ratio came from freeze-out physics; the helium fraction is what arithmetic does to it.

The universe is a quarter helium because, when the weak force let go, there was one neutron for every seven protons. Pair them up, and a quarter of the mass ends up in helium. That is the whole calculation, and it predates the first star.

This is the result that made Gamow, Alpher, and Herman's hot Big Bang model credible in the late 1940s, long before the cosmic microwave background was found. Their 1948 work predicted that a hot, dense early universe would naturally bake in roughly 25% helium, matching the abundance astronomers actually saw, and it predicted leftover radiation at a few kelvin as the afterglow. The radiation temperature they guessed, around 5 K, turned out to be 2.725 K when Penzias and Wilson stumbled onto it in 1965. Not exact, but the right idea and the right scale from a back-of-envelope calculation two decades early.

The one dial that sets everything

The 25% estimate is robust because helium barely cares about the details. But the other light elements, deuterium, helium-3, lithium-7, are exquisitely sensitive to a single number: the baryon-to-photon ratio, the count of ordinary-matter particles per photon in the universe.

$$\eta \equiv \frac{n_b}{n_\gamma} \approx 6 \times 10^{-10}$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

That tiny number, about six baryons for every ten billion photons, is the master dial. Turn $\eta$ up and there's more matter packed into each volume, so reactions run faster and more completely: more deuterium gets burned up into helium, leaving less leftover deuterium behind. Turn $\eta$ down and reactions stall earlier, leaving more unburned deuterium stranded. The effect is dramatic. Drop $\eta$ by a factor of 10 and the leftover deuterium abundance jumps by a factor of roughly 50. Deuterium is the most sensitive gauge we have; cosmologists call it a baryometer.

So the framework is over-determined, which is the strongest position a theory can be in. One parameter, $\eta$, predicts the abundances of four different light nuclei across nine orders of magnitude. And that same parameter is pinned by a completely separate observation. There is no freedom left to fudge.

Holding the fossil up to the light

Step back and look at what we can now do. Take the standard model of nucleosynthesis, feed it the baryon density that the CMB independently measures, and let it predict the primordial helium fraction. The modern calculation gives a sharp number.

$$Y_p = 0.24709 \pm 0.00017 \;\; (\text{theory}) \quad \text{vs.} \quad Y_p = 0.2458 \pm 0.0013 \;\; (\text{obs.})$$

Copy LaTeXwith $$ delimitersCopy mathexpression only

To measure that observed value, astronomers hunt for the most chemically pristine gas in the universe: metal-poor dwarf galaxies where almost no stellar processing has happened since the beginning, so the helium you see is nearly all primordial. Take the helium fraction in galaxies of varying metal content, and extrapolate back to zero metals, to the gas as the Big Bang left it. That extrapolation lands at $Y_p = 0.2458$, against a theoretical $0.24709$. The two agree to within a percent, comfortably inside the error bars.

Helium by mass

≈ 25%

of all ordinary matter, forged before any star

Theory vs observation

0.247 vs 0.246

primordial helium mass fraction, agreeing to ~1%

When it happened

~3 minutes

after the Big Bang, at about a billion kelvin

Take a moment with what this is. A measurement of the light from galaxies billions of light-years away matches a calculation based on the temperature the universe had when it was three minutes old, using nuclear cross-sections measured in terrestrial labs and a baryon density read off the afterglow of the Big Bang. Three independent windows onto the same instant, and they line up. The agreement between the nucleosynthesis baryon density and the CMB baryon density holds the standard cosmological model together to within about two standard deviations.

What the fossil tells us

There's a quiet philosophical payload here worth sitting with. We cannot rewind the universe and watch the first three minutes. There is no recording, no laboratory that reaches a billion kelvin and cosmic densities at once. And yet we can say, with sub-percent confidence, what was happening then, because the era left a fossil that survives intact: the helium, the deuterium, the trace of lithium. They are unchanged relics of a time we can never directly observe, and they let us reconstruct it the way a palaeontologist reconstructs an animal from bones.

The reconstruction even flags its own gaps honestly. The predicted lithium-7 abundance sits a factor of three above what we measure in old stars, the so-called lithium problem, and it has stubbornly refused to resolve for decades. That isn't a failure of the framework; it's the framework being precise enough that a factor-of-three discrepancy in one element is a genuine, sharp puzzle rather than noise. A theory that matches three elements to the percent and misses the fourth by a clean factor is far more interesting, and far more scientific, than one that vaguely fits everything.

So the next time you read that the universe is about three-quarters hydrogen and a quarter helium, remember that the second number is not a measured accident. It is a prediction, made from the neutron-to-proton ratio at one second, the deuterium binding energy, and the rate at which space was expanding when the cosmos was a single fusion reactor. The reactor ran for a few minutes, switched off, and left its ash everywhere. We are still reading it. Some food for thought, the next time you breathe in a little helium fossil from the first three minutes.

Reading further

• Alpher, Bethe & Gamow, The Origin of Chemical Elements (Phys. Rev. 73, 1948). The two-page paper that started it all: a hot early universe that bakes in roughly 25% helium and leaves a few-kelvin afterglow, decades before the CMB was found.
• Weinberg, The First Three Minutes (Basic Books, 1977). The definitive popular account of nucleosynthesis, and the source of the framing this post borrows; still the clearest narrative of the freeze-out and the deuterium bottleneck.
• Pitrou, Coc, Uzan & Vangioni, Precision big bang nucleosynthesis with improved Helium-4 predictions (arXiv:1801.08023, 2018). The modern calculation behind $Y_p = 0.24709 \pm 0.00017$, with the nuclear-reaction-rate detail that makes the sub-percent comparison possible.
• Cyburt et al., Big Bang Nucleosynthesis: 2015 (Nucl. Phys. A 944). A thorough review of the theory and the data, including the baryon-to-photon ratio, the CMB cross-check, and the lingering lithium problem.