NUCLEUS: Reactor CEνNS Experiment

• NUCLEUS is a reactor coherent elastic neutrino–nucleus scattering experiment using gram-scale cryogenic calorimeters with thresholds near 10 eV to access the fully coherent regime.
• The experiment employs a multi-layered active and passive veto system at the Chooz very-near site to harness high antineutrino flux while suppressing challenging particle backgrounds.
• Commissioning results reveal excellent baseline resolution, though a prominent low-energy excess remains the primary obstacle to achieving precise Standard Model and new-physics measurements.

Searching arXiv for recent NUCLEUS papers and related reactor CEνNS work. NUCLEUS is a reactor coherent elastic neutrino–nucleus scattering experiment designed for deployment at the Chooz nuclear power plant in France, at a “Very-Near-Site” located between two commercial reactor cores. Its defining feature is the use of gram-scale cryogenic calorimeters with thresholds at the $\mathcal O(10\,\mathrm{eV})$ scale, enabling access to reactor CE$\nu$NS in the fully coherent regime and to recoil energies below the inverse-beta-decay threshold domain. Across its proposal, construction, commissioning, and prospective-physics literature, NUCLEUS is presented as a compact, multi-veto, ultra-low-threshold platform for CE$\nu$NS observation, precision electroweak studies at very low momentum transfer, and searches for neutrino magnetic moments, non-standard interactions, exotic neutral currents, sterile-neutrino-related effects, and light mediator scenarios (Angloher et al., 2019, Abele et al., 25 Mar 2026).

1. Physical basis and scientific scope

NUCLEUS targets coherent elastic neutrino–nucleus scattering in the regime where the momentum transfer is smaller than the inverse nuclear radius, typically satisfied for $E_\nu \lesssim 30$ MeV. Reactor antineutrinos have energies up to about $10$ MeV, with mean energy around $1.5$ MeV, so the experiment operates in essentially the fully coherent regime. In that limit, the Standard Model differential cross section is written as

$$\frac{d\sigma}{dE_R}(E_\nu,E_R) = \frac{G_F^2}{4\pi}\,Q_W^2\,F^2(q^2)\,m_N \left(1-\frac{E_R}{E_R^{\max}}\right),$$

with

$$Q_W = N - Z\left(1-4\sin^2\theta_W\right), \qquad E_R^{\max}=\frac{2E_\nu^2}{m_N+2E_\nu}.$$

Since $F(q^2)\simeq 1$ in this regime, CE$\nu$NS benefits from the familiar coherence enhancement, approximately $\nu$0, which makes heavy nuclei especially attractive despite the correspondingly smaller recoil energies (Angloher et al., 2019, collaboration et al., 2022).

This kinematic domain gives NUCLEUS a distinctive physics profile. Unlike inverse beta decay, CE$\nu$1NS has no $\nu$2 MeV threshold, so reactor CE$\nu$3NS can probe the low-energy reactor antineutrino spectrum below the IBD threshold. The experiment is therefore positioned not only as a first-principles measurement of a Standard Model neutral-current process, but also as a low-$\nu$4 probe of $\nu$5, neutrino magnetic moments, non-standard neutrino interactions, and reactor-antineutrino systematics. The same recoil scale also makes NUCLEUS directly relevant to the “neutrino floor” and to low-mass dark-matter detector technology, because CE$\nu$6NS and low-mass dark-matter scattering share a sub-keV nuclear-recoil signature (Angloher et al., 2019, collaboration et al., 2022).

2. Chooz very-near site and source geometry

The Chooz site is central to the experiment’s feasibility. The reactor complex has two N4-type pressurized-water reactors, B-1 and B-2, each at $\nu$7, separated by $\nu$8 m. The NUCLEUS Very-Near-Site is a basement room in an administrative building located between the reactor buildings, with baselines of $\nu$9 m to one core and $\nu$0 m to the other. Different NUCLEUS publications quote the antineutrino flux at the site as about $\nu$1, $\nu$2, and $\nu$3, with the latter value stated for both reactors at nominal power (Angloher et al., 2019, Abele et al., 3 Sep 2025).

The site combines very high flux with very shallow overburden. Muon attenuation measurements with a three-layer plastic-scintillator “cosmic wheel” found that the VNS muon rate is reduced only to about $\nu$4 of the surface rate, corresponding to an average overburden of

$\nu$5

with angular variation from about $\nu$6 to $\nu$7 m.w.e. for zenith angles below $\nu$8. A later Geant4 transport study reproduced this as $\nu$9 and found an omnidirectional muon attenuation factor $E_\nu \lesssim 30$0 (Angloher et al., 2019, Abele et al., 3 Sep 2025).

Fast neutrons are the other defining environmental constraint. Early on-site neutron measurements with EJ-301 liquid scintillators found an energy-independent reduction factor $E_\nu \lesssim 30$1 relative to the surface over $E_\nu \lesssim 30$2 to $E_\nu \lesssim 30$3. A later Bonner-sphere campaign targeted the $E_\nu \lesssim 30$4 MeV component and obtained a surface-to-VNS attenuation factor

$E_\nu \lesssim 30$5

consistent with Geant4 transport through the building. The practical consequence is that NUCLEUS is source-rich but background-limited: the site maximizes CE$E_\nu \lesssim 30$6NS statistics per gram, while cosmic-ray-induced muons and, especially, cosmic-ray-induced neutrons define the dominant experimental challenge (Angloher et al., 2019, Abele et al., 3 Sep 2025).

3. Detector concept, cryogenics, and veto architecture

NUCLEUS is based on cryogenic calorimeters derived from CRESST. The operating principle is calorimetric: $E_\nu \lesssim 30$7 with $E_\nu \lesssim 30$8 the crystal heat capacity at $E_\nu \lesssim 30$9. A $10$0 g $10$1 prototype, a $10$2 crystal instrumented with a transition edge sensor, achieved

$10$3

which underpins the experiment’s threshold strategy. The proposal-era staged program defined NUCLEUS-10g as a discovery phase with a $10$4 g total target and NUCLEUS-1kg as a precision phase aimed at percent-level cross-section measurements. The initial $10$5 g layout used a $10$6 array of CaWO$10$7 crystals totaling $10$8 g and a $10$9 array of Al$1.5$0O$1.5$1 crystals totaling $1.5$2 g, implementing a heavy-target/light-target strategy in which CaWO$1.5$3 served as the signal channel and Al$1.5$4O$1.5$5 as an in situ background monitor (Angloher et al., 2019).

The shielding and veto architecture is explicitly layered. The target is enclosed by a TES-instrumented silicon Inner Veto for surface-event and holder-related rejection; outside it sits a Cryogenic Outer Veto made of HPGe crystals, intended as a $1.5$6-covering active shield against ambient radioactivity, gammas, muons, and part of the neutron background; outside the cryostat is a room-temperature Muon Veto made of plastic scintillator panels. Passive shielding consists of lead for gamma attenuation, borated polyethylene for neutron moderation and capture, and an internal $1.5$7 neutron absorber close to the target. Monte Carlo studies in the proposal literature indicated that the combined inner-plus-outer veto system could suppress backgrounds by $1.5$8 (Angloher et al., 2019, collaboration et al., 2022).

A dedicated muon-veto development program established the compactness–efficiency trade-off required by the shallow site. The final veto concept uses 28 plastic-scintillator panels, each $1.5$9 cm thick, with wavelength-shifting fibers and SiPM readout. Simulations for the full assembly found a geometrical muon-tagging efficiency of $$\frac{d\sigma}{dE_R}(E_\nu,E_R) = \frac{G_F^2}{4\pi}\,Q_W^2\,F^2(q^2)\,m_N \left(1-\frac{E_R}{E_R^{\max}}\right),$$0, an identified muon rate of $$\frac{d\sigma}{dE_R}(E_\nu,E_R) = \frac{G_F^2}{4\pi}\,Q_W^2\,F^2(q^2)\,m_N \left(1-\frac{E_R}{E_R^{\max}}\right),$$1 Hz, and a dead time of $$\frac{d\sigma}{dE_R}(E_\nu,E_R) = \frac{G_F^2}{4\pi}\,Q_W^2\,F^2(q^2)\,m_N \left(1-\frac{E_R}{E_R^{\max}}\right),$$2 for a $$\frac{d\sigma}{dE_R}(E_\nu,E_R) = \frac{G_F^2}{4\pi}\,Q_W^2\,F^2(q^2)\,m_N \left(1-\frac{E_R}{E_R^{\max}}\right),$$3 veto window, satisfying the requirement of $$\frac{d\sigma}{dE_R}(E_\nu,E_R) = \frac{G_F^2}{4\pi}\,Q_W^2\,F^2(q^2)\,m_N \left(1-\frac{E_R}{E_R^{\max}}\right),$$4 identification in a constrained footprint (Wagner et al., 2022).

The current reactor-run outlook emphasizes a CaWO$$\frac{d\sigma}{dE_R}(E_\nu,E_R) = \frac{G_F^2}{4\pi}\,Q_W^2\,F^2(q^2)\,m_N \left(1-\frac{E_R}{E_R^{\max}}\right),$$5-focused deployment. The 2025 commissioning paper reports a minimal detector module built around CaWO$$\frac{d\sigma}{dE_R}(E_\nu,E_R) = \frac{G_F^2}{4\pi}\,Q_W^2\,F^2(q^2)\,m_N \left(1-\frac{E_R}{E_R^{\max}}\right),$$6 and double-TES technology, while the 2026 prospects paper models both a Technical Run and a Physics Run with a total of about $$\frac{d\sigma}{dE_R}(E_\nu,E_R) = \frac{G_F^2}{4\pi}\,Q_W^2\,F^2(q^2)\,m_N \left(1-\frac{E_R}{E_R^{\max}}\right),$$7 g CaWO$$\frac{d\sigma}{dE_R}(E_\nu,E_R) = \frac{G_F^2}{4\pi}\,Q_W^2\,F^2(q^2)\,m_N \left(1-\frac{E_R}{E_R^{\max}}\right),$$8, thresholds of $$\frac{d\sigma}{dE_R}(E_\nu,E_R) = \frac{G_F^2}{4\pi}\,Q_W^2\,F^2(q^2)\,m_N \left(1-\frac{E_R}{E_R^{\max}}\right),$$9 eV and $$Q_W = N - Z\left(1-4\sin^2\theta_W\right), \qquad E_R^{\max}=\frac{2E_\nu^2}{m_N+2E_\nu}.$$0 eV respectively, and an instrumented-holder strategy for low-energy-background control (Abele et al., 4 Aug 2025, Abele et al., 25 Mar 2026).

4. Commissioning results and the low-energy excess

Commissioning at the shallow underground laboratory of the Technical University of Munich constituted the first integrated operation of essential NUCLEUS subsystems. The setup used a BlueFors LD400 cryogen-free dilution refrigerator, two cryogenic target detectors, one $$Q_W = N - Z\left(1-4\sin^2\theta_W\right), \qquad E_R^{\max}=\frac{2E_\nu^2}{m_N+2E_\nu}.$$1 kg HPGe Cryogenic Outer Veto crystal, the external muon veto, and passive shielding. The main run lasted eight weeks, from 1 August 2024 to 27 September 2024, with $$Q_W = N - Z\left(1-4\sin^2\theta_W\right), \qquad E_R^{\max}=\frac{2E_\nu^2}{m_N+2E_\nu}.$$2 h of high-quality $$Q_W = N - Z\left(1-4\sin^2\theta_W\right), \qquad E_R^{\max}=\frac{2E_\nu^2}{m_N+2E_\nu}.$$3 data and $$Q_W = N - Z\left(1-4\sin^2\theta_W\right), \qquad E_R^{\max}=\frac{2E_\nu^2}{m_N+2E_\nu}.$$4 h of CaWO$$Q_W = N - Z\left(1-4\sin^2\theta_W\right), \qquad E_R^{\max}=\frac{2E_\nu^2}{m_N+2E_\nu}.$$5 data, for a total measurement period of $$Q_W = N - Z\left(1-4\sin^2\theta_W\right), \qquad E_R^{\max}=\frac{2E_\nu^2}{m_N+2E_\nu}.$$6 h and an overall duty cycle of $$Q_W = N - Z\left(1-4\sin^2\theta_W\right), \qquad E_R^{\max}=\frac{2E_\nu^2}{m_N+2E_\nu}.$$7 (Abele et al., 4 Aug 2025).

The achieved cryogenic performance was already in the few-eV baseline-resolution range. Reported baseline resolutions were

$$Q_W = N - Z\left(1-4\sin^2\theta_W\right), \qquad E_R^{\max}=\frac{2E_\nu^2}{m_N+2E_\nu}.$$8

for the two $$Q_W = N - Z\left(1-4\sin^2\theta_W\right), \qquad E_R^{\max}=\frac{2E_\nu^2}{m_N+2E_\nu}.$$9 TES channels and the CaWO$F(q^2)\simeq 1$0 detector, respectively. With an offline trigger threshold at $F(q^2)\simeq 1$1, this corresponded to threshold estimates of about $F(q^2)\simeq 1$2–$F(q^2)\simeq 1$3 eV. The external muon veto achieved a measured efficiency

$F(q^2)\simeq 1$4

close to the simulated $F(q^2)\simeq 1$5, while the COV reached a matched-filter noise RMS of $F(q^2)\simeq 1$6, implying a threshold estimate of $F(q^2)\simeq 1$7 at $F(q^2)\simeq 1$8 and $F(q^2)\simeq 1$9 trigger efficiency (Abele et al., 4 Aug 2025).

The most consequential commissioning result was the identification of the low-energy excess as the dominant unresolved obstacle. In the keV range, the target-detector spectra were broadly interpretable in terms of muons, ambient gammas, and material radioactivity, but below $\nu$0 in CaWO$\nu$1 and $\nu$2 in $\nu$3 they showed a sharp low-energy rise not explained by the particle-background simulations. For CaWO$\nu$4, the paper quotes an LEE-induced trigger rate in $\nu$5–$\nu$6 of about $\nu$7, while the expected CE$\nu$8NS signal in the same interval for one nominal Chooz CaWO$\nu$9 detector is $\nu$00. That is the central controversy in NUCLEUS’s present status: particle backgrounds are approaching the design target, but the detector-intrinsic or holder-related sub-keV excess must be suppressed or discriminated before CE$\nu$01NS can be extracted in the intended recoil window (Abele et al., 4 Aug 2025).

5. Particle-background model and shallow-site feasibility

A detailed Geant4-based particle-background study was developed specifically for the Chooz deployment, combining VNS environmental measurements with full detector-and-building transport. The analysis focused on the $\nu$02–$\nu$03 eV nuclear-recoil interval, where most of the reactor CE$\nu$04NS signal is expected and where detector and transport uncertainties are least constrained. The dominant external source classes were atmospheric neutrons, atmospheric muons and muon-induced secondaries, environmental gammas from concrete radioactivity, and material radioactivity in the detector and shielding. The simulation assumed a target threshold of $\nu$05 eV, an Inner Veto threshold of $\nu$06 eV, a Cryogenic Outer Veto threshold of $\nu$07 keV, and a Muon Veto threshold of $\nu$08 MeV (Abele et al., 3 Sep 2025).

The chief result is that the layered shielding concept reduces the total particle background by more than two orders of magnitude, leaving a residual spectrum strongly dominated by cosmic-ray-induced neutrons. In the CE$\nu$09NS region of interest between $\nu$10 and $\nu$11 eV, the total particle background rate in CaWO$\nu$12 is predicted to be

$\nu$13

which the paper states corresponds to a signal-to-background ratio $\nu$14. In the same study’s final event-budget table for CaWO$\nu$15, the residual $\nu$16–$\nu$17 eV background is $\nu$18 mcpd, of which $\nu$19 mcpd comes from atmospheric neutrons, while the corresponding CE$\nu$20NS signal is $\nu$21 mcpd under the stated reactor-operation assumptions (Abele et al., 3 Sep 2025).

This background hierarchy has architectural consequences. Lead is necessary for gamma suppression but must be limited because high-$\nu$22 material enhances muon-induced and neutron-induced secondaries. Borated polyethylene provides the main moderation of external neutrons, the internal $\nu$23 layer captures the moderated low-energy component near the target, and the HPGe COV is not merely auxiliary but central to the overall rejection power, especially for ambient gammas and for additional rejection of muon-induced or neutron-associated activity. The background paper therefore defines a narrow feasibility window: reactor CE$\nu$24NS at Chooz is viable on particle-background grounds, but only if the full passive-and-active shield performs near design and if cosmic-ray-induced neutrons remain the only dominant irreducible contribution (Abele et al., 3 Sep 2025).

6. Sensitivity program, precision reach, and new-physics prospects

Sensitivity projections in the original proposal and in the recent prospects paper describe two distinct eras of NUCLEUS analysis. In the proposal-era multi-target $\nu$25 g design, likelihood studies with a $\nu$26 eV threshold and a flat background benchmark of $\nu$27 projected a $\nu$28 CE$\nu$29NS observation in less than $\nu$30 days. For a deliberately pessimistic non-flat background,

$\nu$31

with

$\nu$32

CaWO$\nu$33 alone had essentially no discriminating power, but simultaneous fitting of CaWO$\nu$34 and Al$\nu$35O$\nu$36 restored an expected preference for background-plus-signal over background-only of more than $\nu$37 after one year. Under the same flat-background assumptions, NUCLEUS-10g was projected to reach statistical precision of about $\nu$38 after one year, while the hypothetical NUCLEUS-1kg Ge upgrade with $\nu$39 eV threshold and $\nu$40 background index was projected to reach $\nu$41 statistical precision after one year (Angloher et al., 2019).

The newer prospects study reframes the program around a Technical Run and a Physics Run at Chooz, both with about $\nu$42 g of CaWO$\nu$43. The Technical Run assumes a $\nu$44 eV threshold and a data-driven low-energy-excess model; because the run is LEE-dominated, no Standard Model CE$\nu$45NS observation is expected, but a combined time-and-energy likelihood using reactor-power variation reaches a median expected $\nu$46 CL sensitivity to about $\nu$47 the Standard Model CE$\nu$48NS rate. The Physics Run assumes a $\nu$49 eV threshold and complete suppression of the LEE. Under that optimistic assumption, one year of data with $\nu$50 overall efficiency gives a projected $\nu$51 observation of CE$\nu$52NS with a statistical precision of about $\nu$53 and about $\nu$54 expected CE$\nu$55NS events (Abele et al., 25 Mar 2026).

Because the recoil scale is so low, the projected physics reach is unusually strong for very-low-$\nu$56 observables. The Physics Run study gives a $\nu$57 interval

$\nu$58

and a $\nu$59 CL sensitivity to the electronic neutrino charge radius

$\nu$60

It also projects leading CE$\nu$61NS-based sensitivity to light vector mediator models in the $\nu$62–$\nu$63 MeV mass range and a $\nu$64 CL limit on the neutrino magnetic moment of

$\nu$65

for the Physics Run, compared with

$\nu$66

for the LEE-dominated Technical Run (Abele et al., 25 Mar 2026).

The limiting issues are correspondingly clear. Proposal-era studies already identified reactor antineutrino flux prediction below the IBD threshold, low-energy calibration, and unknown sub-keV backgrounds as the dominant systematic frontier. Commissioning and background-characterization work narrowed the particle-background problem and validated the shallow-site infrastructure, but they also elevated the low-energy excess from a generic concern to the central unresolved issue. A plausible implication is that NUCLEUS’s decisive transition—from a technically successful shallow-site cryogenic deployment to a reactor CE$\nu$67NS measurement—depends less on keV-to-MeV particle-background suppression, where the experiment is now comparatively mature, than on demonstrating stable, instrumented, low-threshold operation with effective discrimination or elimination of the sub-keV excess (Angloher et al., 2019, Abele et al., 4 Aug 2025, Abele et al., 25 Mar 2026).