JUNO First Results — What the 2025 Data Tells Us

The Jiangmen Underground Neutrino Observatory switched on in 2024 and delivered its first oscillation-spectrum measurements in 2025. JUNO is the largest liquid-scintillator detector ever built — twenty kilotons of organic scintillator, 78% photo-coverage, sub-percent energy resolution — and it is positioned at the single most sensitive baseline for mass-ordering determination through vacuum oscillation. Its 2025 results mark the start of a five- to seven-year data-taking campaign expected to deliver the first definitive answer to one of the last open questions in the neutrino sector.

This is what we learned from the first data, and what we should expect next.

The 53-kilometre strategy

JUNO sits in an underground hall 700 metres below ground level in Guangdong Province, southern China. The site is 53 kilometres from two reactor complexes — Yangjiang and Taishan — with a combined thermal power of 35 gigawatts.

The 53 km distance is not arbitrary. It places the detector near the first oscillation maximum of the slow $\Delta m_{21}^2$ “solar-sector” oscillation, which develops over ~60 km for few-MeV reactor antineutrinos. At this baseline the observable ${\bar{\nu }}_e$ spectrum is shaped simultaneously by two frequencies:

• The slow envelope modulation at $\Delta m_{21}^2\approx 7.4\times 10^{-5}$ eV²
• The fast ripple at $\Delta m_{31}^2\approx 2.5\times 10^{-3}$ eV²

The two frequencies interfere. The shape of the interference pattern — specifically, the phase relationship between the two modulations — depends on the sign of $\Delta m_{31}^2$. This is the mass ordering. JUNO’s measurement thus extracts the ordering through a pure vacuum-oscillation spectral analysis, independent of matter effects in the Earth.

Why this matters

Two parallel programmes attack the mass ordering problem. Long-baseline accelerator experiments — DUNE (1,300 km), NOvA (810 km) — rely on matter effects in the Earth’s mantle, which shift neutrino and antineutrino oscillation probabilities asymmetrically in a way that depends on the sign of $\Delta m_{31}^2$ but is also degenerate with the CP-violating phase ${\delta }_{CP}$. Disentangling the two requires high statistics and careful analysis.

JUNO’s vacuum-oscillation approach is degenerate-free. The spectral pattern at 53 km does not depend on ${\delta }_{CP}$, nor on Earth density. The ordering sits there, unambiguously, in the data. The only observational challenge is energy resolution.

The two approaches are therefore beautifully complementary. Consistency would provide an independent confirmation of the mass ordering; inconsistency would point to new physics.

The detector

Inside JUNO’s 35.4-metre-diameter acrylic sphere:

• 20 kilotons of linear-alkylbenzene liquid scintillator with 3 g/L PPO fluor and 15 mg/L wavelength-shifter
• 17,612 large-area (20-inch) photomultipliers
• 25,600 small-area (3-inch) photomultipliers
• Collectively 78% photo-coverage

Around the sphere: a stainless-steel support structure, 43.5 m diameter water Cherenkov veto, and 2,400 top-tracker RPCs for cosmic muon identification.

The target performance: 3% energy resolution at 1 MeV. Achieving this is not easy. Photon statistics limit the resolution to $\sigma /E\propto 1/\sqrt{N_{PE}}$, and reaching 3% means detecting on the order of 1,400 photoelectrons per MeV of event energy. The combination of extremely high scintillator light yield, extremely high photo-coverage, and carefully optimised optical properties is what makes it possible.

The 2025 first-data result

JUNO has been filling and commissioning since 2024, with the first science run beginning later that year. The 2025 release covers approximately one year of running with the majority of the central detector online.

Three headline results emerge:

Detector performance meets specification. The as-built energy resolution matches the 3% design goal within calibration uncertainties. The scintillator optical properties — attenuation, scintillation yield, purity — are all within design tolerances. A dedicated calibration system with 13 radioactive sources spanning 0.3 to 9 MeV has mapped the non-linearity of the energy response.

First oscillation spectrum observed. The measured ${\bar{\nu }}_e$ energy spectrum shows the expected interference pattern: a slow envelope modulation at the $\Delta m_{21}^2$ scale, with a superposed fast ripple at the $\Delta m_{31}^2$ scale. The pattern is statistically consistent with standard three-flavor oscillation using global-fit parameter values. The statistical power is not yet sufficient to distinguish normal from inverted ordering.

No anomalies. The spectrum shows no features suggestive of sterile neutrinos, non-standard interactions, or unexpected reactor flux distortions. The long-standing “5-MeV bump” in reactor antineutrino spectra (observed by Daya Bay and others) is reproduced, consistent with a reactor-flux-modelling effect rather than a beyond-SM signal.

TAO and the systematic strategy

A critical companion experiment, TAO (Taishan Antineutrino Observatory), sits at 30 m from one of the Taishan reactors. TAO is a 2.8-tonne gadolinium-loaded scintillator detector designed to measure the reactor antineutrino flux and spectrum at the source, independent of oscillation.

The combined JUNO + TAO analysis cancels reactor flux-modelling uncertainties that would otherwise limit the oscillation measurement. The design of this two-detector configuration — inspired by the success of the Daya Bay near-far strategy — is central to JUNO’s sensitivity goals.

Timeline

With the full detector operational in 2025 and a target data-taking duration of six to seven years, JUNO should deliver:

• Δm²₂₁ and θ₁₂ at sub-percent precision within two years — more than an order of magnitude better than current values
• Mass ordering at 3–4σ significance within five to six years
• Precision Δm²₃₁ from the fast spectral modulation, complementing accelerator results

Along the way, JUNO will serve as a general-purpose low-energy neutrino observatory: solar neutrino measurements below 1 MeV, atmospheric neutrinos, geoneutrinos at higher statistics than KamLAND or Borexino achieved, and supernova-neutrino sensitivity (about 5,000 events from a galactic core-collapse supernova at 10 kpc).

The broader context

JUNO joins two other flagship neutrino projects in the late 2020s:

• DUNE (USA, first data expected 2028) — long-baseline accelerator experiment for CP violation and mass ordering through matter effects
• Hyper-Kamiokande (Japan, first data 2027) — the 260-kiloton successor to Super-K, also for CP violation and precision oscillation

The three experiments together will close the current generation of oscillation measurements. The three mixing angles, both mass-squared differences, the mass ordering, and the CP-violating phase should all be measured at the percent level by the early 2030s. The three-flavor oscillation framework, whose discovery spanned the late twentieth century, will by then be a precision laboratory.

The first-data moment

For a project that has been in construction for more than a decade, the delivery of a first oscillation spectrum is a milestone in its own right. The JUNO collaboration — a consortium of over 750 physicists from 80 institutions across 18 countries — has assembled what is now one of the most sensitive neutrino detectors ever built. The next five years will be spent letting it do what it was built for.