Description
Abstract
The Coherent Neutrino-Nucleus Interaction Experiment (CONNIE) employs low-noise, fully depleted charge-coupled devices (CCDs) to detect low-energy recoils from coherent elastic neutrino-nucleus scattering (CEνNS) of reactor antineutrinos in silicon, providing a window into physics beyond the Standard Model. The experiment operates approximately 30 meters from the core of the 3.8 GW Angra-2 nuclear reactor in Rio de Janeiro, Brazil. From 2016 to 2020, CONNIE conducted two data-taking runs with 8 scientific CCDs, setting limits on CEνNS and placing competitive constraints on nonstandard neutrino interactions with low-mass vector and scalar mediators. In 2021, the experiment was upgraded with two Skipper-CCD sensors, which utilize multiple nondestructive readouts to achieve sub-electron noise, enabling the detection of individual electrons. This breakthrough lowered the experiment’s energy threshold to an unprecedented 15 eV, making CONNIE the first to use Skipper-CCDs for reactor neutrino detection. Results from this run set new limits on CEνNS and neutrino interactions with light vector mediators and introduced a novel dark matter (DM) search via diurnal modulation, leading to constraints on DM-electron scattering. Additionally, a search for relativistic millicharged particles produced in reactors established world-leading limits, building on the exceptional low-threshold capabilities of Skipper-CCDs. These recent achievements highlight the sensitivity and potential of Skipper-CCDs, reinforcing the need for increased detector mass. As a first step, in 2024, the detector was upgraded with a Multi-Chip-Module containing 16 Skipper-CCDs. This work also outlines future plans to further expand the detector mass and enhance its physics reach.
Scientific Context
The process of Coherent Elastic Neutrino-Nucleus Scattering (CEνNS) was first predicted within the Standard Model (SM) over four decades ago [1, 2], following the discovery of neutral-current neutrino interactions. This process benefits from a coherent enhancement of the elastic scattering cross-section when the incident neutrino energy is low enough, allowing the interaction amplitudes of all nucleons in the nucleus to add constructively [1]. The energy threshold for coherence depends on the target nucleus, and for silicon, it is satisfied for Eν < 60 MeV. The CEνNS cross-section for ~1 MeV neutrinos on silicon is approximately 10⁻⁴² cm² [3]. However, its detection remained challenging until recent years due to the extremely low energy deposition in nuclear recoils, typically below 15 keV for most materials. The first experimental observation of CEνNS was achieved in 2017 by the COHERENT collaboration [4], enabled by advances in detector technology and the intense neutrino flux available at the Spallation Neutron Source (SNS) at Oak Ridge National Laboratory.
CEνNS opens a new avenue for studying low-energy neutrino interactions, making it a powerful tool for probing physics beyond the SM [5, 6]. It has also been recognized as an important background for future dark matter (DM) searches, as CEνNS interactions from solar, atmospheric, and diffuse supernova neutrinos will become increasingly relevant with the next generation of ultra-sensitive detectors [7]. Recent results from the XENONnT and PandaX collaborations [8] provided the first evidence of low-energy nuclear recoils from solar neutrinos, highlighting the need for direct CEνNS measurements in controlled neutrino experiments to properly model and mitigate this background in DM searches. Additionally, anomalies observed in reactor and short-baseline neutrino experiments have led to speculation about the existence of sterile neutrinos [9], prompting several ongoing experimental efforts to investigate these anomalies [10, 11]. Since CEνNS cross-sections are independent of neutrino flavor and accessible at low energies, they provide an ideal framework for studying potential sterile neutrino oscillations at extremely short baselines [12–15].
Beyond sterile neutrinos, CEνNS offers a unique way to explore non-standard neutrino interactions and properties predicted in various SM extensions, including neutrino millicharge [16–18]. Some models propose that neutrino-nucleus scattering is mediated by a new light boson, leading to an enhanced cross-section at low energies. In scenarios where neutrinos possess an anomalous magnetic moment, the event rate could increase by several orders of magnitude [19]. The implications of CEνNS extend beyond particle physics. In astrophysics, understanding neutrino interactions at MeV energies is crucial for modeling energy transport in supernovae, where current uncertainties remain a limiting factor in the development of new theoretical models [20]. Furthermore, interest in using neutrinos for nuclear reactor monitoring has grown in recent years, with CEνNS offering a potential method for this application [21–23].
Objectives
The Coherent Neutrino-Nucleus Interaction Experiment (CONNIE) aims to detect the coherent elastic scattering of reactor neutrinos off silicon nuclei using charge-coupled devices (CCDs) and to explore physics beyond the Standard Model (BSM). The first step toward this goal is to upgrade the experiment by significantly increasing the Skipper-CCD mass—by approximately a factor of 30—through the implementation of Multi-Chip-Module (MCM) technology developed by the Oscura experiment, which integrates an array of 16 Skipper-CCDs. This enhanced detector is being tested and commissioned using low-energy neutrinos from the Angra 2 reactor in Brazil. In the medium to long term, the objective is to establish a large-scale reactor neutrino experiment by deploying up to 1–10 kg of Skipper-CCDs in a new laboratory near a nuclear reactor, which could be Angra 2 or another suitable facility worldwide.
Methodology
Nuclear reactors are an intense source of low-energy neutrinos from fission, with a flux of approximately 1020 v cm-2 s-1 MeV-1 for reactors with a thermal power of about 109 W. Commercial power reactors provide a nearly constant flux, modulated by their fuel cycle, typically with one month of shutdown per year. These neutrinos have an energy spectrum peaking at ~1 MeV, producing silicon nuclear recoils below the keV scale, significantly lower than neutrinos from spallation sources, making detection more challenging. Searching for CEνNS in reactor experiments extends the reach of BSM physics into the low-energy neutrino sector, with sensitivity to models that are only accessible at these energies.
CONNIE [24, 25] uses low-noise, fully depleted charge-coupled devices (CCDs) [26, 27] to detect low-energy recoils from CEνNS interactions between reactor antineutrinos and silicon nuclei [3]. The experiment operates in a portable laboratory (NuLab) located 30 m from the Angra 2 nuclear reactor core, a 3.8 GW pressurized water reactor that has been in commercial operation since 2000. This facility also hosts a water-based neutrino detector, the Neutrinos Angra experiment [23]. The reactor produces a total neutrino flux of 1.21 × 1020 v s-1 [3], with a flux density of 7.8 × 1012 v cm-2 s-1 at the detector site. The experiment is remotely operated, with continuous monitoring and logging of operational parameters. CONNIE’s engineering run in 2014–2015 [28] was followed by the installation of 14 scientific CCDs in 2016.
The standard CCDs used by CONNIE from 2016 to 2020 were developed in collaboration with LBNL Micro Systems Labs [29]. Each sensor consists of a square array of 16 million pixels, each measuring 15 × 15 μm. These detectors are an evolution of fully depleted thick CCDs originally designed for astronomical instruments like DECam [30] and DESI [26], with increased thickness (675 μm) to enhance sensitivity. Full depletion is achieved using high-resistivity (10 kΩ-cm) silicon wafers. To minimize thermally generated dark current, sensors are cooled to 100 K and operate under vacuum (10−7 torr). The detectors are housed in a copper box inside a vacuum vessel, with passive shielding comprising a 15 cm lead layer sandwiched between two 30 cm polyethylene layers for photon and neutron shielding.
For CEνNS detection, the key parameter is the silicon recoil energy. However, the measured energy corresponds to the ionization signal, which is only a fraction of the total recoil energy. The conversion is determined by the quenching factor, which represents the ionization efficiency of nuclear recoils. The latest Skipper-CCD data analysis employed a recent silicon quenching factor model [31], based on Lindhard theory [32] and the latest measurements [34], covering the low-energy range relevant to CONNIE. Future experiments using larger Skipper-CCD masses will require new quenching factor measurements at these energies.
CEνNS signals are searched by applying neutrino event selection criteria to data from reactor-on and off periods, then subtracting their spectra. A statistically significant excess would allow a cross-section measurement, while the absence of a signal leads to a 95% confidence level (CL) upper limit on the CEνNS event rate, which can be translated into BSM physics constraints, particularly on nonstandard neutrino interactions.
The analysis of CONNIE’s 2016–2018 data [24] considered 8 CCDs (47.6 g active mass) with a total exposure of 3.7 kg-days. Subtracting reactor-on and off event rates yielded no significant excess. The results [24] established a 95% CL upper limit at a factor of 40 above the SM prediction for deposited energies of 0.1 keVee, or recoil energies of 1 keV. This was the first CEνNS search at a nuclear reactor reaching such low recoil energies, achieving a detection threshold an order of magnitude lower than the 20 keV threshold used in the first CEνNS detection by the COHERENT experiment [4].
The achieved low threshold enabled constraints on nonstandard interaction models, which at the time were competitive with COHERENT’s bounds. For instance, models with a light mediator [19], which predict a significant event rate increase at low energies, were strongly constrained by CONNIE data. The 2016–2018 results [24] placed world-leading constraints [34] on simplified BSM extensions with light mediators, setting limits on vector mediator masses MZ′ < 10 MeV and scalar mediator masses Mϕ < 30 MeV. These results demonstrated the power of combining a high flux of low-energy reactor antineutrinos with a low-threshold detector to explore new physics via CEνNS. In the 2019–2020 run, an improved readout scheme using hardware binning reduced the detection threshold to 50 eV. The analysis of 2.2 kg-days of data with enhanced techniques set expected (observed) CEνNS limits at 34 (66) times the SM expectation [25].
In 2021, CONNIE upgraded its detector with two Skipper-CCDs, a novel sensor technology enabling multiple nondestructive readouts of pixel charges, reducing noise to sub-electron levels and allowing single-electron counting. This upgrade was part of a broader R&D effort, led by CONNIE members at Fermilab, to develop Skipper-CCDs for next-generation experiments. So far, Skipper-CCDs have been used in dark matter experiments such as SENSEI and DAMIC-M, which face different backgrounds and analysis challenges. During the 2021–2023 Skipper-CCD run, a new analysis chain was developed, significantly reducing background levels by leveraging the sub-electron noise (0.15 e−) to identify defective pixels and artificial events.
Readiness and expected challenges
The CONNIE-Skipper 2021-2023 run marked a major milestone in the experiment, achieving the lowest energy threshold among all CEνNS searches. The implementation of Skipper-CCDs allowed for a significant reduction in readout noise and an improved detection threshold, reaching a record low of 15 eV. Over this period, a total exposure of 18.4 g-days was collected, enabling the first detailed analysis of low-energy events using these sensors in a reactor neutrino experiment.
The analysis of this dataset revealed reactor-on and reactor-off event rates compatible with zero, as expected due to the small detector mass. Nevertheless, the data extended CEνNS sensitivity to lower energies, where higher event rates are predicted. The obtained 95% confidence level (CL) limit stands at 76 times the Standard Model expectation, comparable to the previous result achieved with three orders of magnitude larger exposure. Additionally, the data refined constraints on new light vector mediators, improved limits on dark matter-electron scattering by three orders of magnitude compared to previous surface-level experiments, and enabled the first study of diurnal modulation at CONNIE.
Furthermore, CONNIE carried out a search for relativistic millicharged particles, which could be produced in reactors and interact electromagnetically in the detector. The interaction probability in silicon is enhanced at ionization energies of 10–25 eV due to plasmon excitation. A statistical analysis, conducted in collaboration with the Atucha-II experiment [41], resulted in individual and combined exclusion limits on the mass and charge fraction of millicharged particles. The combined analysis provided a more robust constraint, mitigating systematic uncertainties from each experiment separately, and extended exclusion limits down to ε = 1.4 × 10⁻⁶ for masses as low as 1 eV.
Building upon the success of this run, the continued operation of Skipper-CCDs from 2021 to 2023 allowed for further improvements in energy sensitivity and background suppression. The refined analysis of this extended dataset confirmed that CONNIE had achieved the lowest energy threshold among all CEνNS experiments, while also improving efficiency in the low-energy range. Using an updated quenching factor by Sarkis [31], the expected event rate in the lowest-energy bin (15 to 215 eV) was calculated to be 29.3 events keV-1 kg-1 day-1. Simultaneously, the background event rate was reduced to approximately 4000 events keV-1 kg-1 day-1, consistent with previous measurements by the experiment. With these parameters, a 1 kg detector would allow for a CEνNS measurement at 90% CL with approximately 200 days of reactor-on exposure and 50 days of reactor-off exposure.
These prospects can be further improved by increasing the neutrino flux at the detector by moving the experiment closer to the core of the Angra 2 reactor. Relocating to a position 15 m from the reactor core would quadruple the neutrino flux, significantly reducing the exposure time required to measure CEνNS. Under these conditions, a 1 kg detector could reach a 90% CL measurement in just 13 days of reactor-on exposure. Additionally, as demonstrated by the CONUS experiment [36], positioning the detector under the reactor dome could further suppress cosmic backgrounds, further optimizing the experimental sensitivity.
Currently, the experiment has a new detector with 16 Skipper-CCDs, totaling 8 g of active mass, installed in May 2024. These state-of-the-art sensors are mounted in a compact arrangement on a single board, the Multi Chip Module (MCM), designed for the Oscura experiment [36,37], which will operate with 10 kg of Skipper-CCDs. An Oscura array with 16 MCMs, called a SuperModule, will have approximately 130 g of active mass, enhancing the potential for detecting neutrinos and dark matter in the ultra-low-energy range. With just one SuperModule of Skipper-CCDs, installed at 15 m from the nuclear reactor under the same operating conditions as 2021-2023, a CEνNS measurement would be achieved in 100 days.
Negotiations with Eletronuclear are underway to explore the feasibility of relocating CONNIE to a position under the reactor dome. Such a move would require the development and implementation of new laboratory infrastructure to support the detectors and other necessary instrumentation, as well as obtaining authorization from the nuclear power plant for the installation of a larger experimental setup.
One of the key experimental challenges ahead is obtaining a reliable quenching factor measurement for low energies, which will be accessible for the first time using Skipper-CCDs. The CONNIE collaboration is actively working to reduce the uncertainty on the quenching factor, in collaboration with other teams using silicon targets for nuclear recoil detection [33, 38–40]. In parallel, theoretical efforts are also underway to refine the modeling of the quenching factor, incorporating updated theoretical frameworks based on Lindhard theory [30, 31].
Another crucial challenge is the precise measurement of background contamination levels in the laboratory, including contributions from neutrons, natural radioactivity of detector components, and cosmic muons. To address this, a detailed characterization of the cosmic muon background is being conducted by monitoring the evolution of the cosmic rate and its angular distribution. This effort is complemented by a collaboration with the Neutrinos Angra experiment [23], which continuously tracks the cosmic rate inside the laboratory. Additionally, plans are in place to perform neutron background measurements at NuLab using dedicated neutron detectors from the Instituto de Radioproteção e Dosimetria (IRD) of the Brazilian National Commission for Nuclear Energy (CNEN).
Beyond CONNIE, there is a broader initiative in Latin America to develop the next-generation reactor neutrino experiment using Skipper-CCDs. The technology demonstrated in current CCD and Skipper-CCD experiments, including CONNIE, DAMIC [26], and SENSEI [34, 39, 40], scales efficiently to 100 g. However, reaching larger detector masses will require integrating thousands of Skipper-CCD sensors, necessitating significant engineering developments. Key challenges include designing compact, low-noise readout electronics and developing robust packaging solutions to ensure high yield and thermal stability.
Members of the CONNIE collaboration are currently leading the Oscura DOE program to develop a 10 kg Skipper-CCD experiment for dark matter detection. The technological advances made for Oscura will directly benefit future neutrino experiments with Skipper-CCDs. In fact, CONNIE has already taken a major step in this direction by installing an MCM, becoming the first experiment to implement this recently developed technology for Oscura. This milestone once again highlights the strong synergy between different collaborations and the various research groups working on the advancement and application of Skipper-CCD technology.
Timeline
The planned operation of the current CONNIE detector, with a Multi-Chip-Module (MCM) of 16 Skipper-CCDs and 8 g mass, will continue at least until the next reactor shutdown in Janeiro 2026 and ideally beyond. This run is essential to test the detector’s performance, measure background levels, and refine data analysis techniques.
Starting in 2026, we plan to increase the detector mass to ~100 g by adding more MCMs and upgrading the cryogenic system at CONNIE NuLab. In parallel, we aim to explore relocating the detector under the reactor dome, improving both neutrino flux and shielding. The feasibility of installing a larger detector in a dedicated lab inside the dome will depend on further developments.
By 2027/8, we expect to reach Standard Model (SM) sensitivity for CEνNS. Between 2027 and 2029, the goal is to scale up to 1-10 kg of Skipper-CCDs, significantly enhancing sensitivity. By the 2030s, a multi-kilogram detector at an optimized location could achieve a reach of 0.01 times the SM prediction, enabling precision measurements and new physics searches.
The experiment’s reach is expressed as the ratio of the predicted upper 95% limit to the expected SM CEνNS rate. The current limit is 40 times the SM expectation, with a projected sensitivity reaching the SM prediction by 2027/8.
Final remarks
CEvNS is a powerful probe for both Standard Model physics and potential new physics, offering insights into neutrino properties, nuclear structure, and beyond. Its precise measurement can test fundamental interactions, constrain new physics scenarios, and contribute to astrophysical studies such as supernova dynamics. Skipper-CCD technology has demonstrated the lowest energy thresholds achieved so far for reactor neutrino detection, opening new experimental possibilities. The CONNIE experiment has successfully used this technology, setting a precedent for future advancements. Developing an experiment with a few kilograms of Skipper-CCD detectors will be a crucial step toward precisely measuring the CEvNS cross-section in silicon and exploring signals of physics beyond the Standard Model, further expanding our understanding of neutrino interactions and weak force detection.
References
[1] D. Z. Freedman, Phys. Rev. D 9, 1389 (1974).
[2] V.B. Kopeliovich and L.L. Frankfurt, I, JETP Lett. 19 (1974) 145-147, Pisma Zh.Eksp.Teor.Fiz. 19 (1974) 236-239.
[3] G. Fernandez Moroni, J. Estrada, E. E. Paolini, G. Cancelo, J. Tiffenberg, and J. Molina, Phys. Rev. D 91, 072001 (2015), arXiv:1405.5761 [physics.ins-det].
[4] D. Akimov et al. (COHERENT Collaboration), Science 357 (2017), no. 6356 1123–1126, [arXiv:1708.01294].
[5] T. Ohlsson, Reports on Progress in Physics 76, 044201 (2013), arXiv:1209.2710 [hep-ph].
[6] P. S. Bhupal Dev et al, arXiv:1907.00991.
[7] A. J. Anderson, J. M. Conrad, E. Figueroa-Feliciano, K. Scholberg, and J. Spitz, Phys. Rev. D 84, 013008 (2011), arXiv:1103.4894 [hep-ph].
[8] Zihao Bo et al. (PandaX collaboration), arXiv: 2407.10892 [hep-ex].
[9] K. N. Abazajian et al, arXiv:1204.5379 (2012).
[10] R. Acciarri et al, arXiv:1503.01520 (2015).
[11] S. Gariazzo, C. Giunti, M. Laveder, and Y. F. Li, JHEP 06, 135, arXiv:1703.00860 [hep-ph].
[12] J. A. Formaggio, E. Figueroa-Feliciano, and A. J. Anderson, Phys. Rev. D 85, 013009 (2012), arXiv:1107.3512 [hep- ph].
[13] C. Blanco, D. Hooper, and P. Machado, arXiv:1901.08094 (2019).
[14] T. S. Kosmas, D. K. Papoulias, M. Tortola, and J. W. F. Valle, Phys. Rev. D 96, 063013 (2017), arXiv:1703.00054 [hep- ph].
[15] B. Caas, E. Garcs, O. Miranda, and A. Parada, Physics Letters B 776, 451 (2018).
[16] P. B. Denton, Y. Farzan, and I. M. Shoemaker,, Journal of High Energy Physics 7, 37 (2018).
[17] O. G. Miranda, D. K. Papoulias, M. Tortola, and J. W. F. Valle,, JHEP 2019, 103 (2019).
[18] A. Parada, arXiv:1907.04942 [hep-ph].
[19] R. Harnik, J. Kopp, and P. A. N. Machado, JCAP 1207, 026, arXiv:1202.6073 [hep-ph].
[20] C. Horowitz, K. Coakley, and D. McKinsey, Phys. Rev. D 68, 023005 (2003), arXiv:astro-ph/0302071 [astro-ph].
[21] P. S. Barbeau, J. I. Collar, J. Miyamoto, and I. Shipsey, IEEE Transactions on Nuclear Science 50, 1285 (2003).
[22] C. Hagmann and A. Bernstein, IEEE Transactions on Nuclear Science 51, 2151 (2004).
[23] H. P. Lima et al, Journal of Instrumentation 14 (6), P06010, arXiv:1812.11604 [physics.ins-det].
[24] A. Aguilar-Arevalo et al. (CONNIE collaboration), Phys. Rev. D 100, 092005, [arXiv:1906.02200].
[25] A. Aguilar-Arevalo et al. (CONNIE collaboration), JHEP 05 (2022) 017, arXiv: 2110.13033.
[26] S. E. Holland, D. E. Groom, N. P. Palaio, R. J. Stover, and M. Wei, IEEE Transactions on Electron Devices 50 (Jan, 2003) 225–238.
[27] J. Janesick, Scientific Charge-Coupled Devices, Press Monograph Series (Society of Photo Optical, 2001).
[28] A. Aguilar-Arevalo, et al. (CONNIE collaboration), JINST 11, P07024, arXiv:1604.01343 [physics.ins-det].
[29] LBNL micro systems laboratory, http://engineering. lbl.gov/microsystems-laboratory/.
[30] B. Flaugher et al. (DECam collaboration), Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 10702 (2018) p. 107021F, arXiv:1807.09287 [astro-ph.IM].
[31] Y. Sarkis, A. Aguilar-Arevalo, and J. C. D’Olivo, Phys. Rev. A 107, 062811 (2023), arXiv:2209.04503 [physics.atom-ph].
[32] U. L. James F. Ziegler, J. P. Biersack, The Stopping and Range of Ions in Solids (Pergamon, 1985).
[33] M. F. Albakry et al. (SuperCDMS collaboration), Phys. Rev. Lett. 131, 091801 (2023), arXiv:2303.02196 [physics.ins-det].
[34] A. Aguilar-Arevalo et al. (CONNIE collaboration), JHEP 04 (2020) 054, arXiv:1910.04951.
[35] J. Hakenmüller et al., Eur. Phys. J. C 79, 699 (2019).
[36] Oscura collaboration, arXiv:2202.10518 [astro-ph.IM].
[37] B. Cervantes-Vergara et al. (Oscura collaboration), JINST 18 (2023) 08, P08016.
[38] A. Aguilar-Arevalo et al. (DAMIC collaboration), Phys. Rev. D94, 082006 (2016), arXiv:1607.07410 [astro-ph.CO].
[39] O. Abramoff et al. (SENSEI collaboration), Phys. Rev. Lett. 122, 161801 (2019), arXiv:1901.10478 [hep-ex].
[40] M. Crisler, R. Essig, J. Estrada, G. Fernandez, J. Tiffenberg, M. Sofo-Haro, T. Volansky, and T.-T. Yu (SENSEI collaboration), Phys. Rev. Lett. 121, 061803 (2018), arXiv:1804.00088 [hep-ex].
[41] E. Depaoli, et al, arXiv:2401.07885 [hep-ex].
