Reactor CEvNS: CONUS, CONNIE, and NUCLEUS

When COHERENT first observed coherent elastic neutrino-nucleus scattering (CEvNS) in 2017, the field opened a new window for low-energy neutrino physics. The reaction is, in some respects, fundamental: an antineutrino scatters off an entire atomic nucleus as a coherent unit, with the cross-section enhanced by the square of the neutron number. The signature is a small nuclear recoil with energy of approximately $E_{\nu }^2/(2M_A)$ — for a 30 MeV neutrino on a cesium-iodide nucleus, this is approximately 1 keV.

COHERENT used a stopped-pion neutrino source — proton beam dumps that produce abundant 30-50 MeV neutrinos through pion and muon decay chains. At these energies, the nuclear recoils are detectable with conventional cryogenic detectors.

A complementary frontier is reactor CEvNS. Reactor antineutrinos have typical energies of 1-5 MeV — about a factor of 10 lower than COHERENT’s source. The corresponding nuclear recoils are at sub-keV energies, below the threshold of most detector technologies. To detect reactor CEvNS, experiments must reach detector sensitivities that have not previously been demonstrated for neutrino physics.

Three independent experiments are pursuing this goal: CONUS in Germany, CONNIE in Brazil, and NUCLEUS in France/Italy. Each uses a different detector technology, each is sited at a different reactor, and each has its own strategy for backgrounds, thresholds, and analysis.

By 2024, the field has its first results. CONUS has reported a 3.7σ first observation of reactor CEvNS. NUCLEUS has reported preliminary first events. CONNIE continues to accumulate exposure. The reactor CEvNS programme has begun.

This post is about the three experiments, what they measure, and what they could reveal.

Why reactor CEvNS is interesting

The Standard-Model coherent cross-section is calculable to about 5-10% precision and grows approximately as $E_{\nu }^2{N}^2$ (where $N$ is the neutron number). At reactor energies (~3 MeV average), the cross-section per cesium nucleus is approximately: ${\sigma }_{\text{CEvNS}}(3\text{ MeV})\approx 10^{-40}{\text{ cm}}^2$

This is about $10^4$ times larger than inverse beta decay at the same energy — but the recoil energy is much smaller (sub-keV vs. MeV), making detection technically much harder.

The reasons for pursuing reactor CEvNS specifically:

Lower-energy regime tests Standard Model. The CEvNS cross-section can be parameterised in terms of the weak nuclear form factor, the running of ${\sin }^2{\theta }_W$ at low energies, and the relative weak charges of protons and neutrons. Reactor CEvNS probes a different combination of these than COHERENT, with potentially independent sensitivity to new physics.

Coupling to neutrino magnetic moment. If the neutrino has a magnetic moment, it adds an electromagnetic contribution to the scattering. The magnetic-moment contribution scales differently with neutrino energy than the Standard-Model weak contribution. By measuring CEvNS at multiple energies (reactor + stopped-pion), the magnetic moment can be constrained or measured.

Coupling to non-standard interactions (NSI). NSI couplings can be parameterised in the four-fermion interaction Lagrangian. CEvNS measurements directly probe these couplings, and the reactor-energy regime tests a different kinematic phase space than stopped-pion sources.

Energy harvesting demonstration. CEvNS is the leading candidate for neutrino-energy-harvesting concepts, where the recoiling nucleus’s kinetic energy is converted to useful work. Demonstrating the underlying physics at multiple sources establishes the foundation.

The technical challenge

A reactor antineutrino at 3 MeV scattering coherently off a germanium nucleus (atomic weight 73) produces a maximum recoil kinetic energy of: $T_{\text{max}}=\frac{2E_{\nu }^2}{M_{\text{Ge}}}\approx \frac{2\times (3\text{ MeV}{)}^2}{73\times 940\text{ MeV}}\approx 260\text{ eV}$

The average recoil energy is approximately half this. Detecting an event that deposits 100-200 eV in a detector is far more demanding than the keV-scale recoils at COHERENT.

For comparison:

• A photon from indoor lighting: ~3 eV
• A photon from blackbody at 300 K: ~25 meV
• A typical thermal vibration at room temperature: ~25 meV
• KATRIN’s energy resolution at the tritium endpoint: ~1 eV
• COHERENT’s threshold at CsI: ~5 keV (kinetic energy in the crystal, equivalent to electron-equivalent energy of ~2 keVee)

Reactor CEvNS detectors must operate at the eV-scale energy regime, where signal and noise are governed by phonon physics, defect production, and very subtle calibration effects.

CONUS — germanium at Brokdorf

The CONUS experiment uses point-contact high-purity germanium detectors at the KKB nuclear power plant in Brokdorf, Germany. The reactor is a pressurized water reactor with thermal power of approximately 3.9 GW. CONUS is located in a service building about 17 metres from the reactor core, with limited overburden but substantial passive shielding.

The detector consists of four germanium crystals, each approximately 1 kg, arranged in a hexagonal configuration with anti-coincidence shielding. Each germanium crystal is a point-contact device with energy resolution of approximately 100 eV at sub-keV energies — an order of magnitude better than commercial germanium detectors.

The dominant backgrounds are:

• Cosmic-ray muons (suppressed by the building’s reactor-hall shielding)
• Natural radioactivity (suppressed by ultra-purified materials)
• Cosmogenic activation products (germanium has long-lived cosmogenic isotopes ⁷¹Ge, ⁶⁸Ge)

CONUS reported in 2024 a 3.7σ first observation of reactor CEvNS. The data sample integrated approximately 2 years of running. The reported event rate is consistent with Standard-Model predictions within current uncertainties (which are dominated by background subtraction and energy-calibration systematic effects).

Future running of CONUS-plus (an upgrade) will improve statistics and aim for the 5σ discovery threshold by approximately 2026.

CONNIE — silicon CCDs at Angra

The CONNIE experiment (Coherent Neutrino-Nucleus Interaction Experiment) uses silicon charge-coupled device (CCD) detectors at the Angra-2 reactor in Brazil. The reactor is a 4 GW pressurized water reactor; CONNIE is located approximately 30 metres from the core.

The detector array consists of approximately 16 silicon CCDs, each about 70 cm³. The CCDs have very low intrinsic backgrounds and provide excellent imaging — every detected event includes 3D position information that helps reject backgrounds.

The silicon CCD approach has different strengths than germanium:

• Lower atomic mass means smaller recoil energies but also lower atomic backgrounds
• 3D imaging allows precise spatial localization
• Lower active mass means longer integration times for statistically meaningful samples

CONNIE has accumulated significant exposure but, due to the silicon CCD’s lower active mass per detector volume, the expected CEvNS event rate is smaller than at CONUS. As of 2024, CONNIE has not yet reported a discovery-level observation, but is approaching the threshold with continued operation.

NUCLEUS — cryogenic calorimeters at Chooz

The NUCLEUS experiment uses cryogenic calorimeters at the Chooz nuclear power plant in France/Belgium. The reactors are pressurized water reactors with thermal power of approximately 4 GW each; NUCLEUS is located at approximately 100-metres baseline.

The detector is a stack of cryogenic transition-edge sensors based on CaWO₄ and Al₂O₃ crystals. The crystals are cooled to millikelvin temperatures, where the energy deposited by a nuclear recoil produces a measurable temperature change. The threshold for detecting nuclear recoils in such systems can be as low as 20-30 eV — substantially below germanium’s threshold.

The cryogenic approach is technically demanding (operating large detectors at millikelvin temperatures requires extensive dilution-refrigerator infrastructure) but offers unique sensitivity to the very-low-energy regime. NUCLEUS aims to operate with a total active mass of approximately 10 grams, much smaller than CONUS but with much better energy resolution.

NUCLEUS reported its first commissioning data in 2024, with preliminary evidence for CEvNS events. The full-scale measurement is expected within the next 2-3 years.

Comparison and complementarity

The three experiments are complementary rather than competitive:

Experiment	Detector	Active mass	Energy threshold	Strength
CONUS	HPGe point-contact	~4 kg	~200 eV	High mass, mature technology
CONNIE	Silicon CCDs	~50 g	~50 eV	3D imaging, low backgrounds
NUCLEUS	Cryogenic calorimeter	~10 g	~20 eV	Lowest threshold, best resolution

Each accesses a slightly different recoil-energy regime. Combined, they provide a richer view of the CEvNS process than any single experiment could.

If consistent observations are reported across the three experiments, the result will:

• Confirm the Standard-Model coherent cross-section at reactor energies (independent of COHERENT’s stopped-pion result)
• Provide constraints on neutrino magnetic moments at the $10^{-11}{\mu }_B$ level
• Constrain non-standard interactions at sub-percent precision
• Establish the foundation for next-generation CEvNS experiments

If discrepancies appear between the experiments, the implications could include hints of non-standard physics, systematic effects in one or more experiments, or surprises in the nuclear form factor that have not been anticipated.

What’s next

Several upgrades and new experiments are in development:

CONUS-plus at Leibstadt (Switzerland). A successor to CONUS with larger germanium mass and improved background suppression, designed for 5σ-level statistics. First operations planned for 2026.

CONNIE-2 at Angra. Improved silicon CCD array with better energy resolution and larger active mass.

NUCLEUS expansion. The NUCLEUS detector mass is planned to grow from 10 grams to 100 grams over the next several years, allowing higher-statistics measurements.

RICOCHET at the ILL high-flux reactor in Grenoble. A bolometric germanium detector with sub-100 eV threshold, optimized specifically for the lowest-energy nuclear recoils. Operations are beginning.

Theia and other multi-purpose detectors. Some proposed next-generation neutrino detectors (Theia, Jinping, etc.) include CEvNS-targeted reactor configurations as part of their broader physics programmes.

By 2030, the reactor CEvNS field is expected to have multiple high-statistics measurements at percent precision. The combined data will provide precision tests of the Standard-Model coherent cross-section at the lowest neutrino energies, sensitivity to neutrino magnetic moments at the $10^{-12}{\mu }_B$ level, and substantial constraints on non-standard neutrino interactions.

A new energy regime

The reactor CEvNS programme is opening a previously inaccessible energy regime for neutrino physics. Sub-keV nuclear recoils — events that deposit energies comparable to thermal vibrations in solids — were beyond the reach of any detector technology until very recently. The combination of improved point-contact germanium, cryogenic calorimetry, and silicon CCD imaging has brought these recoils within reach.

The first observations at CONUS (and pending confirmation at NUCLEUS and CONNIE) mark the beginning of a new sub-MeV neutrino-detection paradigm. Future experiments at this energy scale could include:

• CEvNS-based reactor monitoring: detector technology suitable for safeguards and arms-control verification
• Solar neutrino detection at lower thresholds: extending the energy range of solar measurements below current Borexino limits
• Dark-matter detection cross-checks: the same detector technology used for sub-keV nuclear recoils from neutrinos is also relevant for weakly-interacting massive particle searches
• Energy harvesting concepts: practical CEvNS-based energy extraction requires demonstrating the underlying detection at reactor scale

The implications extend well beyond particle physics. CEvNS technology has potential applications in reactor monitoring, dark-matter detection, and (longer term) the conversion of cosmic neutrino energy to useful work. Each of these applications requires the basic detection capability that CONUS, CONNIE, NUCLEUS, and successors are demonstrating.

For more on the original CEvNS observation, see COHERENT at Oak Ridge. For the physics overview, see CEvNS explained. For the energy-harvesting angle, see CEvNS to energy conversion.

The reactor CEvNS programme is, in some respects, neutrino physics’ equivalent of opening a new spectral window in astronomy. We have always known that low-energy reactor antineutrinos exist and interact coherently with nuclei — but we have only just begun to actually observe these events. The information content of those observations, accumulated over the coming decade, will substantially expand our understanding of the neutrino in its lowest-energy interaction regime.