COHERENT at Oak Ridge: Confirming CEvNS

Forty-three years is a long wait between prediction and observation. The Higgs boson waited 48 years — a cautionary benchmark. Gravitational waves waited a full century. Coherent elastic neutrino-nucleus scattering, predicted by Daniel Freedman in 1974 and directly observed in 2017, fits comfortably into this pattern of Standard Model processes whose detection lagged their theoretical foundations by decades.

This is the story of how CEvNS was finally caught, by a 14.6-kilogram scintillator crystal in a utility corridor in Tennessee.

The prediction

Freedman’s 1974 paper was a response to the 1973 discovery of weak neutral currents at the Gargamelle bubble chamber at CERN. The $Z^0$ boson mediates neutrino scattering on any charged fermion — electrons, quarks, or, through the quarks, entire nuclei. For neutrinos at MeV energies, Freedman observed, the momentum transfer to the nucleus is so small that all nucleons scatter in phase.

The amplitudes add, not the probabilities. The cross-section scales as the square of the neutron number N (the proton contribution being suppressed by a factor of ${\sin }^2{\theta }_W$-related physics). For heavy nuclei with $N\sim 70$, this gives a coherent enhancement of roughly $5000\times$ over incoherent nucleon-level scattering.

The problem Freedman himself flagged: the only observable signal is a nuclear recoil of a few keV. Detectors of the 1970s had thresholds of MeV or higher. “It is an act of hubris”, Freedman wrote, “which only an experimentalist could commit to predict its detection.”

The detection challenge

The difficulty is kinematic. A neutrino of energy $E$ scattering elastically off a nucleus of mass $M$ produces a recoil with maximum kinetic energy $T_{\max }=\frac{2E^2}{M+2E}$ For a 30 MeV neutrino on a CsI nucleus of ~130 atomic-mass units, $T_{\max }\approx 14$ keV. The typical recoil is lower. Detecting keV-scale nuclear recoils demands detectors with thresholds of a few hundred electron-volts in the deposited signal — well below the natural background of most detector materials.

The technology caught up only in the 2010s, largely through direct-dark-matter-detection R&D. Germanium bolometers, silicon charge-coupled devices, and sodium-iodide scintillators all reached the required sensitivity.

The Spallation Neutron Source

COHERENT’s decision to work at Oak Ridge National Laboratory’s Spallation Neutron Source was driven by three factors: high neutrino flux, well-characterised neutrino spectrum, and — uniquely among intense accelerator-neutrino sources — short pulsed timing.

The SNS is a 1-gigaelectron-volt proton accelerator running at 1.4 MW. Each 600 ns proton pulse striking the mercury target produces a flux of stopped pions whose decay yields a well-known neutrino composition:

• Prompt (within 1 μs): monoenergetic ${\nu }_{\mu }$ at 30 MeV from ${\pi }^{+}\to {\mu }^{+}{\nu }_{\mu }$
• Delayed (exponential with 2.2 μs lifetime): ${\nu }_e$ and ${\bar{\nu }}_{\mu }$ with well-calculated spectra up to ~50 MeV from the subsequent muon decay

The pulsed structure provides the key experimental handle. Beam-correlated backgrounds in the narrow prompt window are easy to characterise. Beam-unrelated backgrounds — radioactive decays, cosmic-ray-induced events — are averaged over the full 60 Hz cycle and subtracted statistically. The signal-to-background ratio is orders of magnitude better than at any continuous-beam neutrino source.

Neutrino Alley

The COHERENT collaboration deployed its first detectors in “Neutrino Alley”, a basement utility corridor about 20 metres from the SNS target. The corridor provides about 12 metres-water-equivalent of concrete shielding — sufficient to suppress fast neutrons and cosmic-ray backgrounds, without requiring excavation of a dedicated underground hall.

Several detector technologies were deployed in parallel or in sequence:

• CsI[Na] (14.6 kg single crystal) — used for the 2017 first observation
• Liquid argon (24 kg CENNS-10) — different $N$, tests the $N^2$ scaling
• Germanium (multiple kg) — lower threshold, extended energy range
• NaI[Tl] (multi-tonne scale) — future expansion with both Na and I targets

The 2017 observation

COHERENT announced its first result in Science in August 2017. Based on 308 live days of CsI[Na] exposure, the collaboration reported 134 observed CEvNS events over an expected background of about 40. The statistical significance was 6.7σ — well above the 5σ threshold for discovery.

The measured cross-section matched Standard Model expectations within about 10%, with the uncertainty dominated by the nuclear quenching factor (the fraction of recoil energy converted to detectable scintillation light). Subsequent dedicated quenching-factor measurements tightened the comparison.

Multi-target confirmation

A single measurement is never enough. COHERENT extended its programme with liquid argon and germanium targets, testing the predicted $N^2$ scaling across very different atomic numbers:

• Ar: $N=22$
• Ge: $N=42$
• CsI: $N\approx 74$ (averaged over Cs and I)

Within experimental uncertainties, the cross-section ratios matched the $N^2$ scaling. The coherent interpretation was solidly supported, not just the existence of the process.

Reactor CEvNS

In parallel, a community of reactor-based experiments has been pursuing CEvNS with the cleaner antineutrino flux from commercial reactors. CONUS (at Brokdorf and Leibstadt, Germany/Switzerland), CONNIE (Angra, Brazil), Dresden-II (USA), RED-100 (Russia), and NUCLEUS (France) all deploy detectors with sub-keV thresholds at 10–100 metre standoff from reactor cores.

Reactor CEvNS has one clean advantage over accelerator CEvNS: the ${\bar{\nu }}_e$ flux is better characterised and the energy spectrum is softer, probing lower momentum transfers where the coherent approximation is cleanest. The disadvantage is the still-lower threshold requirement and the absence of pulsed-beam timing.

As of early 2026, reactor-CEvNS observations have been reported but not at the same significance as the COHERENT SNS measurement. The field is active and competitive.

Physics beyond the Standard Model

CEvNS is a Standard Model process, but its cross-section and spectrum are sensitive to a variety of beyond-Standard-Model possibilities:

• Non-standard interactions (NSI) — effective four-fermion operators beyond the Standard Model would modify the effective neutrino-quark couplings
• Neutrino magnetic moment — a finite magnetic moment contributes an electromagnetic CEvNS term with different energy dependence
• Sterile neutrinos — would modify the effective flux at the detector if mixing occurs over the short baseline
• Light $Z^{\prime }$ bosons — new gauge mediators with masses below GeV would enhance or suppress CEvNS

COHERENT’s measurements already set competitive limits in each of these directions, and reactor CEvNS experiments extend the reach to lower energies.

The applied connection

CEvNS is also the fundamental process underlying applied research on converting components of the invisible radiation spectrum into usable energy. The Master Equation framework for neutrinovoltaic conversion treats the effective cross-section ${\sigma }_{\text{eff}}(E)$ for the neutrino channel as fundamentally the CEvNS cross-section. Verification of the process at COHERENT in 2017 was therefore a prerequisite for positioning neutrinovoltaic work on peer-reviewable physical foundations.

Outlook

COHERENT itself is continuing data-taking with expanded detector masses and new target materials. The measured cross-section will be constrained at the few-percent level across the 2020s, turning CEvNS into a precision Standard Model observable. Reactor-CEvNS will deliver complementary, lower-momentum-transfer results in the same timeframe.

The forty-three-year gap between Freedman’s prediction and the 2017 observation is, in retrospect, not a comment on the difficulty of the physics — the theory was clear from the start — but on the difficulty of the experiment. The direct-dark-matter R&D community ultimately delivered the technology that made the detection possible. In particle physics, as often, the enabling instrument arrives from an unrelated physics programme.