Deep divefrom Episode 5Cosmology4 October 202625 min

The Helium Receipt From the First Three Minutes

A deep dive into one paper: what was known before, what the researchers did, what the numbers say and what they don't.

The paper: Loverde, Saravanan & Weiner — Big Bang For Your Helium Buck (arXiv)

Chapters

Transcript

The script the two AI voices read from. The times are worked out from the script, not heard, so a line can start a few seconds early or late.

Read the whole episode

Cold open

Host 1You heard the trailer — this is the whole story. Welcome to the Hastrup Review of Science — where artificial intelligence reads real science, so you don't have to pretend you understood the paper. One paper today: what a single, very careful measurement of helium tells us about the first three minutes of the universe, about cosmology's current arguments, and about a neutron that cannot agree with itself on how long it lives.

Host 2The paper is called Big Bang For Your Helium Buck, by Marilena Loverde and Murali Saravanan at the University of Washington and Zachary Weiner at the Perimeter Institute in Canada. It is a preprint on the arXiv dated September fourteenth, twenty twenty-six, and it has not been peer reviewed yet.

Host 1And it is not a new telescope result.

Host 2Correct. The measurement comes from a separate team using the Large Binocular Telescope. These three authors are theorists who take that number and work out what it changes when you combine it with maps of the oldest light in the sky.

Why this field exists

Host 1Start at the beginning, literally. Where does helium come from?

Host 2As established background: in the first few minutes after the Big Bang, the universe was hot and dense enough to fuse protons and neutrons into light nuclei, a process called big bang nucleosynthesis, or B B N. Almost every neutron that survived to that point ended up inside a helium-four nucleus. The result is that roughly a quarter of the ordinary matter in the universe, by mass, came out as helium, and almost all the rest as hydrogen.

Host 1And this was predicted before it was measured?

Host 2Also background: the idea goes back to work by Alpher, Gamow and others in the nineteen forties, and it became one of the pillars of the hot Big Bang model. The other pillar is the cosmic microwave background, the C M B, light released about three hundred eighty thousand years later when the universe cooled enough for electrons and nuclei to combine into neutral atoms.

Host 1Two snapshots of the early universe, then.

Host 2Two snapshots taken at very different times. The paper notes that nucleosynthesis happens at temperatures roughly five orders of magnitude hotter than when the microwave background was released. So if both agree, that is a strong consistency check on the whole picture in between.

How helium gets into the CMB

Host 1Why would the microwave background care about helium at all?

Host 2Because of electrons. Helium grabs its electrons earlier than hydrogen does, so the amount of helium sets how many free electrons are still floating around when hydrogen recombines. Free electrons scatter light, so that number shapes how the microwave background pattern was imprinted.

Host 1Where does it show up in the data?

Host 2Mostly in what cosmologists call the damping tail, the finest-scale ripples in the map. Light diffuses through the hot plasma and smears out the smallest fluctuations, and how far it diffuses depends on how many electrons it bounces off. The paper points out that helium is the only product of nucleosynthesis you need to model the microwave background at all.

Host 1So when cosmologists analyze those maps, what do they assume about helium?

Host 2By default they assume standard nucleosynthesis and calculate the helium fraction from theory. That is called a B B N consistent analysis. It is precise, but it quietly imports an assumption: that nothing unusual happened in the first few minutes.

What was known before

Host 1And if you refuse to make that assumption?

Host 2Then you let helium float as a free parameter, which the paper calls B B N agnostic. The microwave background can still pin it down a bit through the damping tail, but other parameters get much fuzzier. The paper says the precision on several key quantities degrades by factors of order one, meaning up to roughly double the error bars.

Host 1Give me one concrete example.

Host 2The effective number of neutrino species, written N eff, which counts all the radiation in the early universe in units of one neutrino. The Standard Model predicts three point zero four four. With current microwave background plus galaxy distance data, the uncertainty on it goes from zero point one one to zero point two three when you stop assuming standard nucleosynthesis.

Host 1So the cautious analysis is half as sharp.

Host 2Roughly, yes. The way out is to measure helium directly in the sky instead of calculating it. That is hard, because stars have been making extra helium ever since, so you need gas that has seen almost no stellar processing.

Host 1Which is what the Large Binocular Telescope team did.

Host 2They took high-quality spectra of glowing ionized gas in extremely metal-poor galaxies, places that have barely formed stars. Their primordial helium fraction is zero point two four five eight, plus or minus zero point zero zero one three, a precision of half a percent. The paper says that is two point three times more precise than the current recommended value from the Particle Data Group, partly thanks to a more uniform sample and better modeling of how the gas emits and transfers light.

The open question

Host 1So what was the question these theorists set out to answer?

Host 2Two questions, really. First: is a direct helium measurement now good enough that you can drop the nucleosynthesis assumption without paying for it in precision. Second: once you have that number, what does it rule in or out, from extra particles to the cosmological tensions to the neutron lifetime puzzle.

Host 1Why does the first one matter, if standard nucleosynthesis works?

Host 2Because new physics, even physics that never talks to ordinary matter, can change how much helium was made. If your flagship cosmology numbers secretly depend on the first three minutes being boring, you want to know that. And as we will get to, the neutron lifetime input itself is under dispute.

Method: the ingredients

Host 1Walk me through what they actually did. What data went in?

Host 2Microwave background temperature and polarization maps from three experiments: the Planck satellite's twenty eighteen release, the Atacama Cosmology Telescope's sixth data release, and the South Pole Telescope's latest release. They also used a combined lensing measurement from all three, and at times galaxy distance data from the Dark Energy Spectroscopic Instrument, DESI, from its second data release.

Host 1Galaxy distance data meaning?

Host 2Baryon acoustic oscillations, a standard ruler imprinted in how galaxies cluster. It measures the expansion history at much later times. They assume a flat universe and fix the total neutrino mass at sixty milli-electron-volts throughout, for simplicity.

Host 1And the helium measurement goes in how?

Host 2As a simple bell curve centred on zero point two four five eight with a width of zero point zero zero one three. Then they run every analysis four ways: helium free, helium from theory, helium free but constrained by the telescope, and both theory and telescope together.

Method: details that matter

Host 1Any technical traps along the way?

Host 2A nice one about bookkeeping. Astronomers quote helium as a fraction by number of nucleons, while microwave background codes use a fraction by mass, and the two differ because a helium nucleus weighs slightly less than four hydrogen nuclei. The paper notes the difference is the same size as the telescope's error bar, so mixing them up could bias results by about one standard deviation.

Host 1A rounding error that is as big as the measurement.

Host 2Exactly. They also built a compact formula for the theoretical helium yield, derived from tabulated results of a nucleosynthesis code called PRIMAT that ships with the standard cosmology software CAMB. They say it is accurate to five hundredths of a percent, an order of magnitude better than the measurement needs.

Host 1What does that formula say?

Host 2That helium barely depends on the density of ordinary matter, scaling with it to the power of about zero point zero four, but depends more strongly on the radiation, scaling with N eff to about the power zero point one seven. The physics is that more radiation means faster expansion, so the reactions that convert neutrons and protons freeze out earlier, fewer neutrons decay before fusion starts, and you end up with more helium.

Host 1So helium is basically a radiation meter.

Host 2A radiation meter, and also a neutron clock, which we will come back to. For the future, they also ran forecasts for an idealized microwave experiment limited only by cosmic variance, covering forty percent of the sky down to very fine angular scales.

Results: robustness

Host 1Results. Did the helium measurement rescue the cautious analysis?

Host 2Yes, completely, for current data. With helium left free, the measurement of the ordinary matter density is about one point four times less precise. Add the telescope's helium value and the result is identical to the standard analysis: one hundred times the baryon density comes out at two point two four seven nine, plus or minus about zero point zero zero nine five, in both cases.

Host 1And the neutrino number?

Host 2Same story. Helium free, with the microwave background and DESI, N eff is three point three zero with an uncertainty of about zero point two two. With the telescope's helium it becomes three point zero two, plus or minus zero point one one, exactly matching the standard nucleosynthesis analysis.

Host 1And with both theory and measurement together?

Host 2Two point nine eight two, plus or minus zero point zero seven two, which the paper calls a two and a half percent measurement. That is the tightest, but it is no longer agnostic, because it assumes standard nucleosynthesis again.

Host 1Is that consistent with the Standard Model's three point zero four four?

Host 2Within about one standard deviation, but the paper is careful here. Much of that agreement comes from the DESI data pulling the value up, while recent microwave data on their own, driven by the Atacama telescope's damping tail, prefer a value below three point zero four four at about two sigma. So the central value depends on which datasets you combine.

Results: inflation

Host 1What about inflation, the very first moment?

Host 2Inflation predicts the pattern of initial ripples, summarized by a number called the spectral index, n s. Leaving helium free nearly doubles its error bar, and the evidence that it differs from exactly one drops from about nine sigma to five or six. Add the telescope's helium and n s comes back to zero point nine six nine three, plus or minus zero point zero zero three four, identical to the standard analysis.

Host 1Does that settle anything about specific inflation models?

Host 2It closes one escape hatch. Current data disfavor two popular models, Starobinsky and Higgs inflation, at two sigma or more, driven by the Atacama telescope preferring a higher n s. The helium result shows that tension cannot be blamed on assumptions about nucleosynthesis, though the paper notes other explanations for the damping-tail preference remain open.

Results: the future

Host 1Will this measurement still be good enough for the next generation of telescopes?

Host 2Mostly. For the neutrino number and for the inflation parameters, the forecasts say the current precision is already nearly enough. For the ordinary matter density in the plain standard model, it is short by about a factor of two, and the abstract says no more than a factor of two improvement is needed in any case they considered.

Host 1So a modest upgrade, not a new instrument.

Host 2In terms of required precision, yes. Only if helium were measured to better than one part in ten thousand would it fully break the link between the neutrino number and the matter density, which is far beyond current capability.

A word from our sponsor

Host 1Is your universe cooling down too slowly? Then you are in luck.

Host 1The Hastrup Review of Science is brought to you by ice cube trays. Fill with water, place in the freezer, and wait for the phase transition.

Host 2I use mine every evening. Every compartment freezes out at the same time, which is more than you can say for the weak interactions.

Host 1They come in silicone and in plastic, with or without a lid, and in shapes from cubes to spheres. Widely available and very reasonably priced.

Host 2And they are completely free-streaming. Once the ice is out, nothing holds it back.

Host 1Ice cube trays. Our referral link is in the show notes.

Results: extra radiation

Host 1Back to the paper. What does helium say about undiscovered particles?

Host 2Many theories predict extra light particles left over from the early universe, often called dark radiation, which would push N eff above three point zero four four. For this part the authors leave out the DESI data because of its tension with the microwave background, and compare two combinations: Planck with the South Pole Telescope, and those two plus the Atacama telescope.

Host 1And they distinguish two kinds of extra radiation.

Host 2Free-streaming particles that fly around without colliding, like neutrinos, and fluid-like ones that interact strongly with each other. They leave different fingerprints on the microwave background, in the phase and height of the acoustic peaks. But helium only cares about the total amount, since it responds to the expansion rate.

Host 1What limits come out?

Host 2Using theory plus the telescope's helium, the extra radiation is capped at roughly zero point one two to zero point one three of a neutrino species at ninety-five percent confidence with Planck and the South Pole Telescope. Adding the Atacama data tightens that to roughly zero point zero seven five to zero point zero eight three. With both calibrations, the limits for the two kinds become nearly identical, which tells you the total amount of radiation dominates over the subtle fingerprints.

Host 1And from helium alone?

Host 2Helium with standard nucleosynthesis, plus a deuterium measurement to pin down the matter density, gives N_eff = 2.941 ± 0.092.

Results: did radiation change in between?

Host 1Could the amount of radiation have changed between the first three minutes and the release of the microwave background?

Host 2That is the next test. Helium measures the radiation at nucleosynthesis, the microwave background measures it at recombination, so you can fit two separate numbers. If some particle decayed or dumped energy in between, they would differ.

Host 1And do they?

Host 2The helium measurement pins the early value at two point nine six four, plus or minus about zero point one, consistent with the Standard Model to within one sigma. With Planck and the South Pole Telescope, the later value agrees too. With the Atacama data, the later value sits nearly two sigma low, which would point to new physics acting mostly around recombination rather than at nucleosynthesis.

Host 1Is that a hint of something?

Host 2The authors are cautious. Getting less radiation than the Standard Model requires changing neutrinos or heating photons, both strongly constrained by other experiments. And adding DESI data significantly opposes that low preference, so it rests on one dataset and should persist before anyone builds models on it.

Results: the cosmological tensions

Host 1Now the arguments. Remind me what the tensions are.

Host 2As background: the Hubble tension is the disagreement between the expansion rate inferred from the microwave background and the faster rate measured locally with supernovae and the distance ladder. The milder one is between the microwave background and DESI's galaxy distances, which prefer a somewhat lower matter fraction.

Host 1And the proposed fixes change the early universe.

Host 2Several do. The paper tests two families: extra radiation, and varying fundamental constants, specifically the electron mass or the fine-structure constant being slightly different before recombination. The key point is that in the simplest versions, those changes would also be present during nucleosynthesis, so helium can check them independently.

Host 1Start with extra radiation.

Host 2With Planck, the South Pole Telescope and DESI, the fit prefers some fluid-like extra radiation at about two sigma, which eases the DESI tension. Even then, the Hubble constant stays in about three and a half sigma tension with local measurements. Add the helium measurement and that extra radiation is squeezed out, so neither tension is meaningfully relieved.

Host 1And the Atacama data?

Host 2Independently rules out extra radiation as a fix, because it already prefers less radiation than the Standard Model.

Results: varying constants

Host 1What about changing the constants of nature?

Host 2If the electron were slightly heavier, or electromagnetism slightly stronger, early on, atoms would form earlier, and that shifts the inferred matter fraction and Hubble constant. With the microwave background and DESI, this can essentially eliminate the DESI tension, though not the Hubble tension.

Host 1What does helium say?

Host 2Helium responds weakly to the electron mass, scaling to about the power zero point four, so the telescope's constraint, a shift of minus one point four plus or minus one point three percent, has only a small effect. But it responds strongly to the fine-structure constant, with a power of about two and a half. There the helium constraint, minus zero point two three plus or minus zero point two one percent, is as precise as the microwave background's own, and it reduces the combined preference for a changed constant from two point nine sigma to one point eight.

Host 1So one candidate weakened, one barely touched.

Host 2Right, and that is why the abstract says helium precludes most but not all of the models they consider. The electron mass version survives this test.

Results: a note on deuterium

Host 1There is another light element. What about deuterium?

Host 2The authors argue helium carries most of the information. The telescope's helium value is twice as precise as current deuterium measurements, and deuterium predictions carry larger uncertainties from nuclear reaction rates, where different treatments currently disagree.

Host 1Did they check that claim?

Host 2They added deuterium to their electron mass fit and found it moved the result by only a third of a standard deviation and shrank the error by one and a half percent. They also push back on a recent claim of a generic tension between deuterium and the microwave background in Hubble-tension models, arguing it only holds for a narrow class like early dark energy, where the extra energy appears after nucleosynthesis.

Results: the neutron lifetime

Host 1Now the part from the trailer. What is the neutron lifetime anomaly?

Host 2A free neutron decays into a proton, an electron and an antineutrino in about fifteen minutes. Bottle experiments trap ultracold neutrons and count survivors, and the Particle Data Group average of eight of them is eight hundred seventy-eight point three seconds, plus or minus zero point four. Beam experiments that count the protons produced from a passing neutron beam get eight hundred eighty-eight point four five, plus or minus one point six five, which the paper says is six sigma discrepant.

Host 1Ten seconds apart, and both very precise.

Host 2Yes. If it is not an unknown error in one method, it might point to new physics, for example neutrons sometimes decaying into something invisible. A newer beam experiment that counts electrons instead gets eight hundred seventy-seven point two, with an uncertainty of about four seconds, which agrees with the bottles.

Host 1And helium can vote.

Host 2Because helium is essentially all the neutrons that survived until fusion began, the yield depends on the neutron lifetime, scaling with it to the power zero point seven three. Run nucleosynthesis backwards from the measured helium, and you get a lifetime of eight hundred seventy-three point zero seconds, plus or minus six point three.

Host 1Which side does that land on?

Host 2Within one sigma of the bottle experiments and the electron-counting beam, and two point four sigma below the proton-counting beam experiments. Because helium barely depends on the matter density, this answer does not care much whether you use the microwave background or deuterium to set that. The one assumption is that no new physics altered nucleosynthesis.

Why it matters

Host 1Step back. What is the principle here?

Host 2That a direct measurement can replace a theoretical assumption without losing precision. The headline numbers from the microwave background, the matter density, the shape of the initial ripples, the neutrino count, came out the same whether you trusted standard nucleosynthesis or measured helium directly. That is evidence those numbers are robust to whatever happened in the first three minutes.

Host 1What changes, and for whom?

Host 2For cosmologists, analyses that stay agnostic about the early universe no longer have to pay a precision penalty. For model builders, a set of proposed fixes to the cosmological tensions now has to explain why they would not have changed the helium yield, especially those that vary the fine-structure constant or add radiation. For neutron physicists, there is an independent, if blunt, data point from the sky.

Host 1And what does it not mean?

Host 2It does not resolve the Hubble tension or the DESI tension; it narrows the space of explanations. It does not prove the beam experiments are wrong, because two point four sigma is suggestive, not decisive, and the error bar is about fifteen times wider than the bottle average. And it does not mean new particles are ruled out, only that their total contribution is capped at roughly a tenth of a neutrino species.

Caveats

Host 1Honest caveats.

Host 2First, nearly everything rests on one helium measurement from one team. As background, extracting primordial helium from galaxy spectra involves modeling temperature, density and stellar absorption, and different groups' values have historically disagreed by more than their quoted errors, so a systematic shift in this one number would move many conclusions.

Host 1Second?

Host 2Dataset dependence. The authors repeatedly show that whether the neutrino count lands on, below or above the Standard Model depends on whether you include the Atacama telescope and DESI, which pull in opposite directions. Several results are therefore shown both ways, and readers should not pick the convenient one.

Host 1Third?

Host 2Model scope. They test minimal versions where new physics acts equally at nucleosynthesis and recombination; models where it switches on only later, like early dark energy, escape the helium test. They also fixed the neutrino mass, and they note that letting it vary changes how the varying-constant models behave.

Host 1And the neutron result?

Host 2It assumes standard nucleosynthesis, which is exactly the kind of assumption new neutron physics might break. So it is a consistency check, not an independent arbiter in the strongest sense. And this is a preprint, so the analysis has not yet been through peer review.

What to watch

Host 1What should we watch for next?

Host 2Independent helium measurements from other telescopes and teams that either match this half-percent value or do not. And whether someone pushes the precision by the factor of two the forecasts ask for.

Host 1On the cosmology side?

Host 2Whether the Atacama telescope's preference for less damping and the DESI tension survive new data, including from the Simons Observatory and the South Pole Observatory that the paper mentions. And in the lab, more neutron lifetime measurements with new methods, which could settle the beam and bottle split without needing the Big Bang at all.

Sign-off

Host 1That is the deep dive: Big Bang For Your Helium Buck, by Loverde, Saravanan and Weiner, a preprint on the arXiv. The link is in the show notes.

Host 2Every Friday this same feed carries the roundup episode: five new papers as trailers, and you pick which become deep dives. This has been the Hastrup Review of Science.