JUNO: Settling the Mass Ordering From 52 Kilometres Away
There are three neutrino mass states, and we know the gaps between them but not their sequence: two are close together, one sits apart, and whether that lone state is the heaviest or the lightest has been an open question for two decades. Two experiments are now attacking it from opposite directions. DUNE will use the way matter in the Earth distorts oscillation over 1,300 kilometres. JUNO does it without matter effects at all, by reading an interference pattern in reactor neutrinos that have travelled just 52 kilometres - which demands an energy resolution no detector of this size has ever achieved.
Why 52 Kilometres Is a Designed Number
Reactor cores produce electron antineutrinos in enormous quantities, with a well-understood energy spectrum, and they do it continuously. That makes them an unusually convenient source: no accelerator, no beam time, and a flux that can be cross-checked against the reactors' thermal output.
As those antineutrinos travel, they oscillate at two different rates simultaneously, set by the two independent mass-squared differences. One is slow and one is fast. At short distances only the fast one has had time to act; at very long distances the pattern washes out. At around 52 kilometres both are in play at once, and the fast oscillation appears as a fine ripple riding on top of the slow one.
The position of that ripple relative to the slow oscillation depends on the ordering of the mass states. Reading it is therefore a direct measurement of the ordering - but only if the detector can resolve the ripple, which is a question of energy resolution rather than of size.
This is why JUNO sits where it does. Two power plants, Yangjiang and Taishan, are both roughly 52 kilometres away, which is close to the optimum for this measurement and is unlikely to be a coincidence in the site selection.
Resolution as the Engineering Target
JUNO's design specification is an energy resolution of about 3 percent at one megaelectronvolt - roughly twice as good as any large scintillator detector before it. Everything about the construction follows from that single number.
Resolution depends on collecting as many scintillation photons as possible from each event. So the acrylic sphere is enormous but the photosensor coverage is denser than usual: 17,612 large photomultiplier tubes of 20 inches, supplemented by 25,600 small three-inch tubes that fill the gaps between them and provide an independent calibration. Together they cover roughly 78 percent of the surface, against about 40 percent for Super-Kamiokande.
The scintillator itself, based on linear alkylbenzene, has to be transparent enough that light from the far side of a 35-metre sphere still arrives, and radiopure enough that natural decay does not swamp the signal. Both requirements are at the limit of what is achievable at this volume.
The dual photosensor system also solves a subtler problem. Large tubes saturate on bright events; small tubes count individual photons cleanly. Having both means the detector can calibrate itself across its full dynamic range instead of relying on external models.
The Observatory in Numbers
JUNO completed filling and began data taking in August 2025.
- Location: Jiangmen, Guangdong Province, China, roughly 700 metres underground
- Target: 20,000 tonnes of linear alkylbenzene liquid scintillator
- Vessel: acrylic sphere 35.4 metres in diameter, the largest ever constructed
- Photosensors: 17,612 twenty-inch and 25,600 three-inch photomultiplier tubes, about 78 percent coverage
- Baseline: approximately 52 kilometres from the Yangjiang and Taishan nuclear power plants
- Design energy resolution: about 3 percent at 1 MeV
- Collaboration: around 700 members from roughly 74 institutions in 17 countries
Two Routes to One Answer
The mass ordering can be approached in two fundamentally different ways, and it is worth understanding why both are being built.
DUNE relies on the matter effect: over 1,300 kilometres of rock, electrons in the Earth modify how electron neutrinos oscillate, and the sign of that modification reveals the ordering. The method is powerful but it depends on modelling the Earth's density profile and on the value of other oscillation parameters.
JUNO relies on vacuum oscillation only. At 52 kilometres the matter effect is negligible, so the measurement is a nearly pure quantum interference reading, independent of Earth models and largely independent of the CP-violating phase that complicates accelerator experiments.
If the two agree, the answer is settled in a way neither could establish alone. If they disagree, that is more interesting still. This is the same logic that made the Sudbury and Super-Kamiokande results convincing: independent systematics converging is what turns a measurement into knowledge.
What Else It Will Measure, and What It Will Not
The same instrument will measure three oscillation parameters to better than one percent, which is roughly an order of magnitude better than the current state of the art, and it will observe solar neutrinos, geoneutrinos, atmospheric neutrinos and any galactic supernova. Its scintillator technology puts it in the same family as Borexino, at seventy times the mass.
It will not measure the absolute neutrino mass. Like every oscillation experiment it is sensitive to differences and ordering, not to scale - that remains the domain of direct measurements such as KATRIN.
And the reactor source deserves a note, because it is sometimes read as evidence that neutrino flux is easy to use. A nuclear power plant produces something like 10^20 antineutrinos per second, of which JUNO - 20,000 tonnes, 52 kilometres away, under 700 metres of rock - expects to record on the order of a few dozen per day. The gap between those two numbers is the whole reason detectors like this exist.
That gap is also why claims about ambient neutrino flux need care. The question in neutrinovoltaic research is not whether reactor or solar neutrinos can be captured for power; JUNO shows exactly what capturing them requires. It concerns whether ambient radiation and thermal fluctuations at an engineered material surface can drive a measurable current, which this detector does not address in either direction.
Frequently asked questions
What is the neutrino mass ordering?
There are three mass states. Two lie close together and one sits apart, and the open question is whether the separate state is the heaviest or the lightest. The gaps are measured; the sequence is not.
Why put the detector 52 kilometres from the reactors?
Because at that distance the two oscillation frequencies interfere most informatively - the faster oscillation appears as a fine ripple on the slower one, and the position of that ripple reveals the ordering.
How does JUNO differ from DUNE on the same question?
DUNE uses the matter effect over 1,300 kilometres of rock, which depends on Earth density models. JUNO uses vacuum oscillation only, so its result is independent of those models. Agreement between two independent methods is what would settle the question.
Why does it need so many photomultiplier tubes?
Because the measurement is limited by energy resolution, not by size, and resolution depends on collecting as many scintillation photons as possible. JUNO reaches roughly 78 percent photocathode coverage, against about 40 percent for Super-Kamiokande.
Does using reactor neutrinos mean the flux is easy to exploit?
No. A reactor emits on the order of 10^20 antineutrinos per second; JUNO, with 20,000 tonnes of scintillator under 700 metres of rock, expects a few dozen detections per day. The distance between those numbers is why the detector has to be this large.
