The Cosmic Neutrino Background: A 1.95 K Bath from the First Second

About one second after the Big Bang — at a temperature of approximately 10 billion Kelvin and an age the universe could measure with a slow stopwatch — the rate of weak interactions among the dense plasma of particles fell below the rate of cosmic expansion. Neutrinos, which had been in thermal equilibrium with everything else, suddenly found themselves no longer interacting with anything at a meaningful rate. They decoupled. They have been free-streaming through the expanding universe ever since.

The relic neutrinos from that decoupling are still here, 13.8 billion years later. They fill space at a number density of approximately 336 per cubic centimetre — about a hundred million times less than the cosmic microwave background photons but identical to them in spirit: a fossil of the first moments of the universe, persisting because the universe became too dilute to absorb them.

This is the cosmic neutrino background, often abbreviated CνB or CNB. It is the second-oldest direct relic of the early universe (younger only by 380,000 years than the CMB photons, but older by ~13.7 billion years than every star). It has never been directly detected. Every piece of evidence we have for it comes indirectly — from how it shapes other observables. And the next generation of experiments may, finally, see it.

Decoupling and free-streaming

In the first second after the Big Bang, the universe was hot enough that neutrinos were in thermal equilibrium with the surrounding plasma of electrons, positrons, and photons through reactions like: $\nu +e^{-}\leftrightarrow \nu +e^{-}\text{(elastic scattering)}$ $e^{+}+e^{-}\leftrightarrow \nu +\bar{\nu }\text{(pair conversion)}$ The rate of these reactions per neutrino, $\Gamma$, depends on the cross-section (which scales as $G_F^2{T}^2$ at temperatures $T\ll m_W$) and the target density (which scales as $T^3$). Therefore $\Gamma \sim G_F^2{T}^5$.

The expansion rate of the universe, set by general relativity for a radiation-dominated cosmology, scales as $H\sim T^2/M_{Pl}$ where $M_{Pl}$ is the Planck mass.

Neutrinos remain in equilibrium as long as $\Gamma \gtrsim H$. The two rates cross at: $T_{\text{dec}}\sim {\left(\frac{M_{Pl}}{G_F^2}\right)}^{1/3}\approx 1\text{ MeV}\approx 10^{10}\text{ K}$ At and below this temperature, the universe expands faster than neutrinos can interact. They decouple from the rest of the plasma and propagate freely thereafter, redshifting with the cosmic expansion.

For comparison, photons remain coupled to matter through Thomson scattering off free electrons until the universe is about 380,000 years old (recombination). Neutrinos therefore decouple about 14 orders of magnitude earlier than photons. This makes the CνB the oldest relic radiation field accessible (in principle) to direct observation.

The temperature relation

A subtle quantum-mechanical effect distinguishes the CνB from the CMB. Shortly after neutrino decoupling but before $T\sim 100$ keV, the universe cooled enough that electron-positron pairs annihilated: $e^{+}+e^{-}\to \gamma +\gamma$ This annihilation released energy into the photon plasma, raising its temperature relative to the already-decoupled neutrino plasma. By entropy conservation in the photon-electron-positron sector, the photon temperature is now larger than the neutrino temperature by a fixed factor: $\frac{T_{\nu }}{T_{\gamma }}={\left(\frac{4}{11}\right)}^{1/3}\approx 0.7138$ With the measured CMB temperature $T_{\gamma }=2.7255$ K, the predicted CνB temperature is $T_{\nu }=1.9454\text{ K}$ This number is the canonical value of the CνB temperature today. It is a prediction of the standard cosmological model, not a measurement; no apparatus has directly determined T_ν.

The corresponding number density for three neutrino species (each with two helicity states for relativistic neutrinos, including antiparticles) is: $n_{\nu }=\frac{6}{11}{n}_{\gamma }=336{\text{ cm}}^{-3}$ Half are neutrinos, half are antineutrinos, distributed equally across the three flavors.

The mean kinetic energy per neutrino, assuming a Fermi-Dirac thermal distribution at $T=1.95$ K, is approximately $⟨E⟩=\frac{7{\pi }^4}{180\zeta (3)}{k}_B{T}_{\nu }\approx 5\times 10^{-4}\text{ eV}=0.5\text{ meV}$ This is one to two orders of magnitude below the lightest neutrino mass eigenstate. Since at least two of the three mass eigenstates are heavier than the kinetic energy, the CνB neutrinos are non-relativistic today — the only known cosmologically significant relic radiation field that is not relativistic.

Indirect evidence: BBN

The Big Bang Nucleosynthesis epoch ($\sim 100$ seconds, $T\sim 100$ keV) produced the primordial abundances of helium, deuterium, and lithium. The yields depend on the neutron-to-proton ratio at the start of nucleosynthesis, which in turn depends on the expansion rate at $T\sim 1$ MeV. The expansion rate at that epoch is set by the relativistic energy density, which has contributions from photons, electron-positron pairs, and neutrinos.

The BBN-relevant parameter is “the effective number of relativistic species at decoupling”, $N_{\text{eff}}$. For three Standard Model neutrinos with the precise standard decoupling, the predicted value is $N_{\text{eff}}=3.044$ (the small departure from 3.000 reflects subtle corrections from incomplete neutrino-electron decoupling and finite-temperature QED effects).

Comparing measured primordial abundances of ${}^4$He, deuterium, and the few remaining usable observations of ${}^7$Li with theoretical predictions yields $N_{\text{eff}}=2.91\pm 0.17$ from BBN alone, consistent with the standard prediction. This was one of the earliest pieces of evidence that the number of light neutrino species was three.

Indirect evidence: the CMB

The cosmic microwave background, observed by COBE, WMAP, and Planck, encodes information about the radiation density at recombination through the position and amplitude of the acoustic peaks in the temperature and polarization power spectra. Adding extra relativistic species — beyond the three Standard Model neutrinos — would shift the peak positions and modify the damping tail. The Planck 2018 data analysis yields: $N_{\text{eff}}=2.99\pm 0.17$ in stunning agreement with the standard-cosmology prediction. Future observations from Simons Observatory and CMB-S4 should reduce the uncertainty by roughly a factor of three, sharply constraining any beyond-Standard-Model relativistic species (sterile neutrinos, axions, dark photons).

Indirect evidence: structure formation

Massive neutrinos free-stream through forming density perturbations and suppress the small-scale matter power spectrum. Galaxy redshift surveys (BOSS, eBOSS, DESI) measure this suppression and translate it to a constraint on the sum of neutrino masses, currently $\sum m_{\nu }\lesssim 0.12$ eV. The same observations are sensitive to the number of neutrino species through the integrated Sachs-Wolfe effect and the matter power spectrum normalization. Combined CMB + galaxy clustering gives consistent results: three (or essentially three) Standard Model neutrino species, with the rest of cosmological evolution behaving exactly as the relic-neutrino picture predicts.

The direct-detection challenge

Direct detection of CνB neutrinos is astonishingly difficult. The kinetic energies are sub-meV. The interaction cross-sections — already tiny at MeV energies — scale roughly as $E^2$ for elastic scattering, putting them ~10⁸ times smaller than for solar neutrinos. The differential rate per kg of target is unmeasurable by any conventional technique.

The one possible loophole is Stodolsky’s mechanism: capture of CνB neutrinos via induced beta decay on a target nucleus that is already energetically poised to undergo beta decay. The classic candidate is tritium: ${\nu }_e+{}^3\text{H}\to {}^3\text{He}+e^{-}$ This is the inverse reaction of normal tritium beta decay. The reaction has no kinematic threshold — even an arbitrarily low-energy neutrino can induce it, because the daughter ³He is more stable than the parent ³H by 18.6 keV of kinetic energy, which is supplied entirely by the rest-mass difference rather than the neutrino energy.

The signature is an electron with energy above the standard tritium beta-decay endpoint by roughly 2 m_ν — typically a few hundred meV. Distinguishing this electron from the much larger background of standard beta-decay endpoint electrons requires energy resolution at the 0.1-eV level, far beyond current technology but conceivably reachable with cyclotron radiation emission spectroscopy.

The PTOLEMY project (Princeton Tritium Observatory for Light, Early-universe, Massive-neutrino Yield) is developing this technique. Operating with 100 g of atomic tritium adsorbed on graphene substrates, PTOLEMY aims for the energy resolution required to resolve the CνB capture line from the beta endpoint background. The current demonstrator is a long way from a science-grade experiment, but proof-of-principle work is ongoing.

A successful PTOLEMY-class detection would be one of the great experimental physics achievements of the century.

Why it matters

Direct CνB detection would do three things, none of them small.

Confirm a 60-year-old cosmological prediction directly. Indirect evidence is overwhelming, but direct is direct. The CνB has been a theoretical certainty since the late 1960s; making it an observational fact would close a long-standing loop.

Probe absolute neutrino mass independently. The CνB capture rate depends sensitively on the local neutrino density, which can be enhanced by gravitational clustering for massive neutrinos. A measured rate could constrain the absolute mass scale through this clustering enhancement.

Test the standard cosmological model at the second after the Big Bang. All current observations probe the universe at recombination (380,000 years) or later. The CνB carries information from one second — three orders of magnitude earlier than the CMB. Any deviation between expected and observed CνB properties would point to new physics in the very early universe.

The relic neutrino bath is, in a precise sense, the oldest message we can hope to read. Whether we ever read it directly is one of the open questions of experimental physics.