Cosmological Neutrino Mass: Reading the Sum from the Sky

The simplest cosmological measurement of neutrino mass uses no detector, no underground laboratory, and no nuclear physics. It uses the cosmic microwave background, galaxy redshift surveys, and the imprint that neutrinos with mass leave on the universe’s large-scale structure.

This indirect approach has, over the past decade, become the most stringent constraint on the absolute neutrino mass scale. Current bounds — Σ m_ν < 0.12 eV from Planck CMB + galaxy clustering data — are tighter than the best laboratory measurement (KATRIN’s 0.45 eV) by a factor of about 4. They are also model-dependent: cosmological assumptions enter, and modifications to standard ΛCDM (extra relativistic species, modified gravity, time-varying dark energy) can shift the bound by factors of 2 or more.

The future is more dramatic. Within the next decade, surveys like DESI, Euclid, and LSST should push the bound to below 0.04 eV — at which point the cosmological measurement will either detect a positive signal or rule out the inverted neutrino mass ordering. Either outcome will be foundational.

This post explains how the cosmological neutrino mass bound works, what it currently constrains, and where it is heading.

Why cosmology cares about neutrino mass

In standard cosmology, three species of neutrinos contribute to the universe’s energy budget. They froze out at temperatures of about 1 MeV — one second after the Big Bang — and have been free-streaming through the expanding universe ever since.

For a long time, neutrinos remained relativistic. Their kinetic energies were thermal, scaling with the universe’s temperature, while their rest masses (likely below 0.1 eV) were not yet relevant. As long as kinetic energy dominates rest mass, neutrinos behave like radiation.

At some point in cosmological history, the neutrino’s thermal energy fell below its rest mass, and it became non-relativistic. The transition occurred at a temperature roughly equal to $m_{\nu }/3$. For $m_{\nu }=0.1$ eV, this corresponds to redshift $z\approx 200$ — well after recombination ($z\approx 1100$) but well before galaxies formed at $z\approx 5$.

This transition matters because it changes the neutrino’s gravitational behaviour. Relativistic neutrinos act like radiation: they free-stream out of forming density fluctuations. Non-relativistic neutrinos can in principle cluster — but their initial relativistic motion has already carried them across substantial distances, smoothing out any density fluctuations at smaller scales.

The net effect: massive neutrinos suppress the power spectrum of cosmic matter density fluctuations at small scales (below a characteristic free-streaming scale that depends on $m_{\nu }$). The suppression is small but calculable: roughly $\Delta P/P\approx -8f_{\nu }$ at scales below the free-streaming length, where $f_{\nu }={\Omega }_{\nu }/{\Omega }_m$ is the fraction of total matter density in neutrinos.

What we measure

The cosmological neutrino-mass measurement is therefore essentially a measurement of structure formation as a function of scale. If the matter power spectrum is suppressed at small scales relative to a no-neutrino-mass universe, that’s the signature.

Two complementary observational programmes provide the data.

CMB anisotropy measurements (Planck, ACT, SPT, Simons Observatory, CMB-S4). The cosmic microwave background carries information about the universe at the time of recombination. Neutrino masses affect the CMB through their contribution to the radiation density, the integrated Sachs-Wolfe effect (a low-redshift signature), and the lensing of the CMB by intervening matter. The most powerful single CMB constraint on $\sum m_{\nu }$ comes from CMB lensing — the weak gravitational deflection of CMB photons by intervening dark matter and structure, which sees the integrated lensing power.

Galaxy redshift surveys (BOSS, eBOSS, DESI, Euclid, LSST). These surveys measure the three-dimensional distribution of galaxies, which traces the underlying matter distribution. Massive neutrinos affect the matter power spectrum at small scales, which is observable in the galaxy correlation function or its Fourier transform.

The combined analysis: take CMB + galaxy clustering + local-universe constraints on the Hubble constant and other cosmological parameters; fit a multi-parameter model including $\sum m_{\nu }$ as a free parameter; extract the constraint.

The current best published bound, from Planck 2018 + BOSS + Lyman-alpha forest: $\sum m_{\nu }<0.12\text{ eV}(95\%\text{ C.L., }\Lambda \text{CDM})$

This is conservative — model extensions can relax the bound to about 0.5 eV.

Comparing with the lower bound

Oscillation experiments measure mass-squared differences. From the differences plus an unknown lightest mass $m_0$, the sum can be computed:

Normal ordering ($m_1<m_2<m_3$): $\sum m_{\nu }=m_0+\sqrt{m_0^2+\Delta m_{21}^2}+\sqrt{m_0^2+|\Delta m_{31}^2|}$ With $m_0=0$, this gives $\sum m_{\nu }\approx 0.058$ eV.

Inverted ordering ($m_3<m_1<m_2$): $\sum m_{\nu }=m_0+\sqrt{m_0^2+|\Delta m_{31}^2|}+\sqrt{m_0^2+|\Delta m_{31}^2|-\Delta m_{21}^2}$ With $m_0=0$, this gives $\sum m_{\nu }\approx 0.099$ eV.

So the minimum allowed sum is 0.06 eV (normal) or 0.10 eV (inverted). The current cosmological bound of 0.12 eV is starting to cross these floors. Inverted ordering is mildly disfavored at the 1σ level. Normal ordering with $m_0\gtrsim 0.04$ eV is excluded.

The plot to remember: a horizontal line at $\sum m_{\nu }=0.12$ eV (current upper bound), descending bars representing the IO and NO floors, and an arrow toward the future surveys’ sensitivity at ~0.04 eV. The IO floor is just barely below the current bound; the NO floor is well below.

What ΛCDM assumes

The bound depends on the assumed cosmological model. Standard ΛCDM uses:

• General relativity for gravity
• Cold dark matter (collisionless, non-relativistic)
• Dark energy as a cosmological constant
• Three light neutrinos with assumed degenerate or hierarchical masses
• Standard inflation-generated initial perturbations

Each of these is well-tested but not unique. Modifications relax the bound:

Extra relativistic species (e.g., a fourth sterile neutrino or non-Standard-Model dark radiation). The CMB measurement of $N_{\text{eff}}$ provides the constraint. Adding extra species shifts the inferred $\sum m_{\nu }$.

Time-varying dark energy (quintessence, phantom dark energy). Modifies the late-time expansion history, which affects how cosmological measurements decompose into different cosmological parameters.

Modified gravity. If general relativity is wrong on cosmological scales, the relationship between matter clustering and neutrino mass would change.

Curvature. A non-flat universe shifts the inferred matter content.

In practical terms: the 0.12 eV bound is robust for standard ΛCDM, but cosmology cannot definitively distinguish “neutrinos have mass below 0.12 eV” from “the cosmological model has additional physics that we attribute to high effective $\sum m_{\nu }$”.

This is the principal limitation of the cosmological approach. It is also the principal advantage of complementary laboratory measurements (KATRIN, Project 8, $0\nu \beta \beta$): they don’t depend on cosmological assumptions.

The future programme

Several next-generation surveys will tighten the cosmological bound dramatically over the next decade.

DESI (Dark Energy Spectroscopic Instrument) — Mayall Telescope at Kitt Peak. Five-year survey of 35 million galaxies and quasars. Operational since 2021; first data releases through 2025-2026 will improve the $\sum m_{\nu }$ bound by perhaps 30%.

Euclid (European Space Agency satellite) — launched 2023. Six-year survey of weak lensing and galaxy clustering across $\sim 14,000$ deg² of sky. Will independently constrain $\sum m_{\nu }$ at the few-percent level.

LSST / Vera Rubin Observatory — Chile. Ten-year photometric survey of the southern sky. Galaxy clustering and weak lensing measurements will provide independent constraints with different systematics.

Simons Observatory + CMB-S4 — South Pole and Atacama. New CMB experiments with vastly improved sensitivity to CMB lensing. Direct constraints on $\sum m_{\nu }$ at $\sim 0.04$ eV precision are expected.

Combining these by 2030 should give: $\sigma (\sum m_{\nu })\sim 0.02\text{ eV}$ At this precision, normal ordering’s lower bound (0.06 eV) is detectable at $3\sigma$. Inverted ordering’s bound (0.10 eV) is detectable at $5\sigma$.

If the cosmological bound at 2030 finds $\sum m_{\nu }\approx 0.06$ eV (consistent with NO minimum) or $\sum m_{\nu }\approx 0.10$ eV (consistent with IO minimum), it will be the first laboratory-independent positive measurement of an absolute neutrino mass scale. If it finds zero — consistent with ultra-light $m_{\nu }\to 0$ in the lightest mass eigenstate — it will rule out the inverted ordering definitively.

Tension as a possibility

A subtle but important point: cosmology and laboratory measurements could conflict. The current Planck + BOSS bound favors $\sum m_{\nu }<0.12$ eV with mild preference for ordering parameters that match the global oscillation fit. KATRIN’s bound of 0.45 eV is consistent. There is no tension in 2026.

But the cosmological precision will improve faster than KATRIN’s. By 2027-2028, cosmology may report $\sum m_{\nu }=0.07\pm 0.02$ eV (a 3σ detection of normal-ordering minimum). KATRIN will still be at 0.2 eV upper bound. The two would be technically compatible — cosmology measures the sum, KATRIN measures the kinematic $m_{\nu }$ which is a different combination — but the comparison would be revealing.

If KATRIN later reaches 0.04 eV and finds a positive signal (Project 8 era), the comparison with cosmology becomes critical. Discrepancies would suggest either new physics in neutrino mass (e.g., neutrino decay into other species after decoupling, modifying the cosmological imprint) or new physics in cosmology (extra species, modified expansion).

The next decade is structured to converge on a consistent picture. By 2030-2035, the absolute neutrino mass should be known either through cosmology, laboratory, or both. The current 0.12 eV bound is the threshold at which the question stops being “is the mass small enough?” and becomes “what is the mass, exactly?”.

A perspective

For most of physics, “fundamental constants” are measured directly in laboratories. The neutron’s mass, the electron’s charge, the speed of light, $G_F$ — all are direct measurements at controllable apparatus.

The neutrino’s absolute mass may be different. The first credible positive measurement may come from cosmology — a measurement of the universe’s matter clustering, indirectly inferring the cumulative gravitational effect of trillions of trillions of neutrinos that became non-relativistic billions of years ago.

This is unusual. It places neutrino mass alongside dark matter and dark energy as a fundamental physical quantity that we measure not in a laboratory but in the structure of the cosmos itself. It also gives the cosmological measurement a particular intellectual character: tightly coupled to gravity, structure formation, and the expansion history of the universe.

Whether the first positive measurement comes from cosmology or from KATRIN/Project 8 is, at this point, an open question. The race between the two approaches is the most interesting development in absolute-mass physics over the next 5 years. Whichever finishes first, the other will follow. By 2030, the neutrino mass — perhaps 0.06 eV, perhaps 0.10 eV — will be a known number rather than an upper bound.

That alone will be a milestone.