PROSPECT Reactor Antineutrino Experiment

• PROSPECT is a reactor antineutrino experiment that uses a segmented 6Li-doped liquid scintillator to measure the 235U spectrum and search for sterile neutrinos.
• It employs a movable detector positioned 7–20 m from HFIR, enabling precision spectral and oscillation analyses that reduce reactor flux model dependencies.
• Innovative segmentation, calibration, and background rejection techniques in PROSPECT advance detector technology and enhance reactor antineutrino research.

PROSPECT, the Precision Reactor Oscillation and Spectrum Experiment, is a short-baseline reactor antineutrino program centered on measurements at the High Flux Isotope Reactor (HFIR) at Oak Ridge National Laboratory. Its defining objectives are a precision measurement of the $^{235}\mathrm{U}$ reactor antineutrino spectrum and a search for short-baseline $\bar{\nu}_e$ disappearance as a signature of eV-scale sterile neutrinos, using a segmented $^{6}\mathrm{Li}$-loaded liquid scintillator detector at baselines of a few to a few tens of meters from a compact highly enriched uranium reactor core (Norcini, 2015, Surukuchi, 2019).

1. Scientific aims and anomaly context

PROSPECT was conceived in response to two linked problems in reactor neutrino physics. First, global comparisons of measured reactor $\bar{\nu}_e$ rates with modern flux predictions revealed a deficit of roughly $5\%$ or about $6\%$, depending on the summary, usually termed the reactor antineutrino anomaly (Norcini, 2015, Surukuchi, 2019). Second, high-precision $\theta_{13}$ experiments at low-enriched uranium reactors observed spectral distortions relative to the Huber–Mueller prediction, especially an excess in the $4\text{–}7$ MeV or $5\text{–}7$ MeV region (Norcini, 2015, Surukuchi, 2019).

At PROSPECT baselines, matter effects and standard three-flavor oscillations are negligible, so a sterile-neutrino interpretation can be parameterized in a two-flavor or 3+1 short-baseline form. The relevant disappearance probability is written as

$$P_{ee} \simeq 1 - \sin^2 2\theta_{14}\,\sin^2\!\left(1.27\,\Delta m^2_{41}\frac{L}{E_\nu}\right),$$

or equivalently with $\bar{\nu}_e$0 and $\bar{\nu}_e$1 in an effective two-flavor approximation (Norcini, 2015, Surukuchi, 2019). PROSPECT was therefore optimized for $\bar{\nu}_e$2 values corresponding to $\bar{\nu}_e$3, with design studies emphasizing baselines of $\bar{\nu}_e$4 m and antineutrino energies of $\bar{\nu}_e$5 MeV (Norcini, 2015).

The choice of HFIR is central to the program. HFIR is a compact highly enriched uranium research reactor whose fission fraction from $\bar{\nu}_e$6 remains $\bar{\nu}_e$7 throughout each cycle, making it an almost pure $\bar{\nu}_e$8 source rather than the mixed-isotope source characteristic of commercial low-enriched uranium reactors (Surukuchi, 2019). This gives PROSPECT an unusually direct handle on isotope-specific spectral modeling and on whether the spectral anomaly can be attributed primarily to $\bar{\nu}_e$9.

2. Reactor source and detector configuration

Early PROSPECT design studies described a staged program. Phase I centered on a movable antineutrino detector at $^{6}\mathrm{Li}$0 m from the HFIR core, while Phase II proposed a larger detector at $^{6}\mathrm{Li}$1 m or $^{6}\mathrm{Li}$2 m to expand $^{6}\mathrm{Li}$3 coverage and improve sensitivity (Ashenfelter et al., 2015, Norcini, 2015). In construction and operation, the collaboration deployed a single optically segmented detector near HFIR, with the detector center about $^{6}\mathrm{Li}$4 m from the core center and an internal baseline span of approximately $^{6}\mathrm{Li}$5 m in the first oscillation analyses (Ashenfelter et al., 2018).

The deployed PROSPECT Antineutrino Detector used about $^{6}\mathrm{Li}$6 tons of $^{6}\mathrm{Li}$7-doped liquid scintillator in an $^{6}\mathrm{Li}$8 array of 154 optical segments (Surukuchi, 2019, Ashenfelter et al., 2018). Segment dimensions were reported as $^{6}\mathrm{Li}$9 cm in length with a $\bar{\nu}_e$0 cross section in the detector and calibration-system papers, and as $\bar{\nu}_e$1 cm with a $\bar{\nu}_e$2 cross section in the spectrum paper, reflecting the dimensions used in those descriptions (Collaboration et al., 2019, Surukuchi, 2019). Each segment was read out at both ends by photomultipliers, giving segment identification and longitudinal position information from charge or timing asymmetry (Surukuchi, 2019).

The experiment operated with minimal overburden. The first oscillation result explicitly described the detector as operating under less than $\bar{\nu}_e$3 meter water equivalent overburden, and later summaries likewise treated the installation as effectively on-surface (Ashenfelter et al., 2018, Andriamirado et al., 2024). Consequently, the apparatus combined segmentation with a substantial shielding package. The deployed detector used water bricks, borated polyethylene, lead around the detector, and an additional lead wall adjacent to the reactor pool to reduce reactor-related $\bar{\nu}_e$4-ray backgrounds (Surukuchi, 2019). Earlier design studies also emphasized passive shielding of borated polyethylene and lead, guided by HFIR background surveys and Monte Carlo studies targeting a signal-to-background ratio above unity (Norcini, 2015).

3. Segmented $\bar{\nu}_e$5 scintillator instrumentation

PROSPECT’s active medium is a $\bar{\nu}_e$6-loaded liquid scintillator based on EJ-309, chosen for its pulse-shape-discrimination capability and its ability to tag neutrons through the capture reaction

$\bar{\nu}_e$7

Because the triton and alpha are heavily ionizing, the visible delayed signal is strongly quenched and appears as a narrow peak around $\bar{\nu}_e$8, often quoted near $\bar{\nu}_e$9 or $5\%$0 (Norcini, 2015, Ashenfelter et al., 2018).

A substantial R&D program preceded construction. In small-volume tests, Li-loaded EJ-309 achieved a light yield of about $5\%$1 photons/MeV, and the PROSPECT-0.1 prototype measured a pulse-shape-discrimination figure of merit of $5\%$2 in the neutron-capture region (Norcini, 2015). Full-scale prototype segments then established detector-level optical performance. The PROSPECT-20 segment prototype, with Li-loaded EJ-309 and internal reflectors, measured $5\%$3 photoelectrons/MeV, meeting the requirement for an energy resolution of about $5\%$4 (Norcini, 2015). The later two-segment PROSPECT-50 prototype reported $5\%$5 photoelectrons/MeV, $5\%$6 at $5\%$7 MeV, an effective scintillation attenuation length of $5\%$8 cm, and a neutron capture time of $5\%$9 for a $6\%$0 by-mass $6\%$1 concentration (Ashenfelter et al., 2018).

The internal optical segmentation is implemented by a low-mass optical grid. PROSPECT’s optical grid defined a $6\%$2 array of optically isolated segments using reflective separators with a carbon-fiber backbone, specular reflector film, and FEP encapsulation, held in place by 3D-printed PLA support rods (Collaboration et al., 2019). This subsystem was designed to maintain high reflectivity in the $6\%$3 nm band while contributing only about $6\%$4 to $6\%$5 of the mass in the active volume (Collaboration et al., 2019). The grid is not merely mechanical; it sets segment boundaries, limits optical cross-talk, and feeds directly into the geometry and optical transport used in PROSPECT analyses (Collaboration et al., 2019).

Calibration was correspondingly elaborate. The source calibration system used 35 PTFE tubes arranged in a $6\%$6 grid, allowing deployment of $6\%$7, $6\%$8, $6\%$9, $\theta_{13}$0, and AmBe sources through the detector (Collaboration et al., 2019). Bench tests showed source-position reproducibility within $\theta_{13}$1 mm and absolute positioning to about $\theta_{13}$2 cm, while combined fits to multiple $\theta_{13}$3-ray features, the $\theta_{13}$4 $\theta_{13}$5 spectrum, and the $\theta_{13}$6 MeV $\theta_{13}$7-H capture line produced data–Monte Carlo agreement at the less-than-$\theta_{13}$8 level (Collaboration et al., 2019).

4. Inverse beta decay reconstruction and background suppression

PROSPECT detects reactor antineutrinos through inverse beta decay,

$\theta_{13}$9

The prompt signal is the positron kinetic energy plus annihilation $4\text{–}7$0 rays; the delayed signal is the thermalized neutron captured on $4\text{–}7$1 after tens of microseconds (Norcini, 2015, Surukuchi, 2019). In the spectral-analysis convention, the antineutrino energy is approximately related to the prompt energy by $4\text{–}7$2, although PROSPECT generally compared prompt spectra after folding through detector response rather than unfolding event-by-event neutrino energies (Surukuchi, 2019).

The detector’s segmentation is integral to event identification. Prompt and delayed signals are required to occur in the same segment or nearby segments, with the delayed capture localized by the compact $4\text{–}7$3 signature (Norcini, 2015, Surukuchi, 2019). In the first oscillation result, the delayed cluster energy window was $4\text{–}7$4, the prompt window was $4\text{–}7$5, and the prompt–delayed time separation was required to satisfy $4\text{–}7$6 (Ashenfelter et al., 2018). Prompt clusters had to be gamma-like in PSD, while delayed clusters were required to be strongly neutron-like (Ashenfelter et al., 2018).

The shallow-depth environment makes background rejection a defining technical issue. Reactor-related $4\text{–}7$7 rays, cosmogenic fast neutrons, multiple-neutron showers, and accidental prompt–delayed coincidences all contribute (Norcini, 2015). PROSPECT addressed these backgrounds through multi-layer shielding, PSD in both prompt and delayed channels, timing windows matched to the $4\text{–}7$8 capture time, multiplicity and topology cuts, fiducialization of outer segments, and direct reactor-off background measurement (Norcini, 2015, Surukuchi, 2019). In the 2019 $4\text{–}7$9 spectrum measurement, 40.3 days of reactor-on and 37.8 days of reactor-off data yielded $5\text{–}7$0 IBD events after subtraction of accidentals and pressure-corrected correlated backgrounds, with a signal-to-background ratio of $5\text{–}7$1 (Surukuchi, 2019). This established that an on-surface segmented detector could achieve precise antineutrino spectroscopy under HFIR conditions.

5. Oscillation searches and sterile-neutrino limits

PROSPECT’s oscillation program is explicitly reactor-model independent. Rather than fitting absolute flux deficits, the collaboration compared prompt-energy spectra across multiple detector baselines, searching for the relative $5\text{–}7$2-dependent distortions predicted by sterile-neutrino oscillations (Ashenfelter et al., 2018, Andriamirado et al., 2024). This design strongly suppresses dependence on the absolute HFIR flux model and on common spectral-shape uncertainties.

The first oscillation result used 33 live days of reactor-on data at nominal $5\text{–}7$3, yielding $5\text{–}7$4 detected IBD events (Ashenfelter et al., 2018). PROSPECT showed that reactor antineutrinos could be detected at $5\text{–}7$5 statistical significance within two hours of on-surface reactor-on data taking, and the relative spectral comparison constrained significant portions of previously allowed sterile-neutrino parameter space at $5\text{–}7$6 confidence level (Ashenfelter et al., 2018). In that first result, the best fit of the Reactor Antineutrino Anomaly was disfavored at $5\text{–}7$7 (Ashenfelter et al., 2018).

The final PROSPECT-I oscillation search used the complete 2018 data set with a multi-period selection designed to recover exposure from segments that became only partially functional over time (Andriamirado et al., 2024). The analysis used 95.6 days of reactor-on and 73.1 days of reactor-off data and obtained $5\text{–}7$8 IBD events after background subtraction (Andriamirado et al., 2024). Inverse beta decay positron spectra from six different reactor–detector distance ranges were found to be statistically consistent with one another, as expected in the absence of sterile-neutrino oscillations (Andriamirado et al., 2024).

The final result significantly sharpened PROSPECT’s role in the short-baseline landscape. It excluded at $5\text{–}7$9 confidence level sterile-neutrino parameter space above $$P_{ee} \simeq 1 - \sin^2 2\theta_{14}\,\sin^2\!\left(1.27\,\Delta m^2_{41}\frac{L}{E_\nu}\right),$$0 that had been previously unexplored by terrestrial experiments, including all parameter space below $$P_{ee} \simeq 1 - \sin^2 2\theta_{14}\,\sin^2\!\left(1.27\,\Delta m^2_{41}\frac{L}{E_\nu}\right),$$1 suggested by the strengthened Gallium Anomaly (Andriamirado et al., 2024). The best-fit point claimed by the Neutrino-4 reactor experiment was ruled out at more than five standard deviations (Andriamirado et al., 2024). This outcome did not eliminate all possible sterile-neutrino constructions, but it substantially narrowed the viability of minimal 3+1 $$P_{ee} \simeq 1 - \sin^2 2\theta_{14}\,\sin^2\!\left(1.27\,\Delta m^2_{41}\frac{L}{E_\nu}\right),$$2-disappearance explanations for the reactor and gallium anomalies.

6. $$P_{ee} \simeq 1 - \sin^2 2\theta_{14}\,\sin^2\!\left(1.27\,\Delta m^2_{41}\frac{L}{E_\nu}\right),$$3 spectrum measurement, interpretation, and legacy

PROSPECT’s second major deliverable is a direct $$P_{ee} \simeq 1 - \sin^2 2\theta_{14}\,\sin^2\!\left(1.27\,\Delta m^2_{41}\frac{L}{E_\nu}\right),$$4-dominated spectrum measurement from an HEU reactor. The 2019 spectrum analysis used the same segmented detector and calibration framework to compare the measured prompt-energy spectrum with a Huber-based $$P_{ee} \simeq 1 - \sin^2 2\theta_{14}\,\sin^2\!\left(1.27\,\Delta m^2_{41}\frac{L}{E_\nu}\right),$$5 prediction supplemented by non-fission isotope and non-equilibrium contributions (Surukuchi, 2019). The comparison was intentionally shape-only, with the overall normalization floated to isolate spectral disagreement from absolute flux normalization (Surukuchi, 2019).

The result showed that the Huber-based $$P_{ee} \simeq 1 - \sin^2 2\theta_{14}\,\sin^2\!\left(1.27\,\Delta m^2_{41}\frac{L}{E_\nu}\right),$$6 prediction is a poor overall description of the measured PROSPECT spectrum shape. The shape-only fit yielded

$$P_{ee} \simeq 1 - \sin^2 2\theta_{14}\,\sin^2\!\left(1.27\,\Delta m^2_{41}\frac{L}{E_\nu}\right),$$7

corresponding to a one-sided $$P_{ee} \simeq 1 - \sin^2 2\theta_{14}\,\sin^2\!\left(1.27\,\Delta m^2_{41}\frac{L}{E_\nu}\right),$$8-value of $$P_{ee} \simeq 1 - \sin^2 2\theta_{14}\,\sin^2\!\left(1.27\,\Delta m^2_{41}\frac{L}{E_\nu}\right),$$9 (Surukuchi, 2019). PROSPECT then tested an ad hoc model in which the local deviation observed by Daya Bay in the $\bar{\nu}_e$00 MeV region was transferred to $\bar{\nu}_e$01 with a free normalization $\bar{\nu}_e$02. The best fit was $\bar{\nu}_e$03, while the extreme hypothesis that $\bar{\nu}_e$04 alone accounts for the full Daya Bay bump, corresponding to $\bar{\nu}_e$05, was disfavored at $\bar{\nu}_e$06 (Surukuchi, 2019). The measurement therefore indicates both that the Huber $\bar{\nu}_e$07 shape is not fully consistent with HEU data and that a purely $\bar{\nu}_e$08-only explanation of the entire low-enriched-uranium bump is disfavored (Surukuchi, 2019).

Taken together, the oscillation and spectrum programs place PROSPECT at the intersection of neutrino phenomenology, detector instrumentation, and reactor-spectrum modeling. It demonstrated precise reactor antineutrino detection in an aboveground segmented detector with good energy resolution and controlled backgrounds, produced strong limits on eV-scale sterile neutrinos, and delivered a modern high-statistics $\bar{\nu}_e$09 spectrum benchmark (Andriamirado et al., 2021). A proposed successor, PROSPECT-II, was motivated by the remaining parameter space above $\bar{\nu}_e$10, the strengthened Gallium Anomaly, and the continuing need for higher-precision $\bar{\nu}_e$11 spectral and flux measurements; the collaboration proposed an upgraded detector with impactful physics in as little as one calendar year of data (Andriamirado et al., 2022). This suggests that PROSPECT’s enduring significance lies not only in the constraints already obtained, but also in having established a detector and analysis paradigm for short-baseline reactor antineutrino physics under challenging experimental conditions (Andriamirado et al., 2022).