ANNIE: Fermilab's small tank with an outsized job for the next generation of neutrino detectors

Most accelerator neutrino experiments are giants — beam lines hundreds of kilometers long, detectors carrying tens of thousands of tons of mass. ANNIE is not. Sitting just a few hundred meters from Fermilab’s Booster Neutrino Beam target station, the Accelerator Neutrino Neutron Interaction Experiment fits inside a single repurposed hall and carries only a few tens of tons of water in its central tank. Its job is correspondingly narrow, but the data it produces feed directly into the planning of much larger detectors. ANNIE is studying, with a precision no one has matched before, how often a neutrino interaction in water produces a neutron in the final state.

Why neutron multiplicity matters

Water-Cherenkov detectors — Super-Kamiokande, the future Hyper-Kamiokande, and at much higher energies IceCube — see neutrinos by catching the blue Cherenkov cone of the charged lepton that the interaction produces. What happens on the hadronic side of the event is mostly invisible to the optics. Out of that hadronic mess come protons, pions, and crucially neutrons.

The neutron count is not a curiosity. A neutrino-induced charged-current interaction on a nucleus tends to produce zero or one neutron in the final state; an antineutrino-induced interaction tends to produce more. If you can tag those neutrons on an event-by-event basis, you can statistically separate neutrinos from antineutrinos in the same detector — a critical capability for any long-baseline experiment trying to extract CP violation from the difference between the two. Right now, every prediction of how often Super-K or Hyper-K should see a given neutron multiplicity in a beam event rests on neutrino-generator codes (GENIE, NuWro, NEUT) that have been benchmarked mostly against scattering off heavier targets like carbon and argon. Water, and specifically the oxygen-16 in water, is comparatively under-measured.

Neutrons also matter for supernova physics. When the next galactic core collapse arrives, water detectors will see a flash of mostly electron antineutrinos via inverse beta decay, but also a smaller flux of all flavors via neutrino-oxygen reactions on the oxygen in the tank. The signal channels are partly distinguished by the number of neutrons each produces, and the predictions are tied to nuclear models that have rarely been tested under controlled beam conditions in the relevant energy range. ANNIE is essentially calibrating those models in advance.

Gadolinium and picosecond photodetectors

Catching a single MeV-scale neutron inside a water tank is not easy. ANNIE does it the same way Super-Kamiokande now does — by dissolving a small amount of gadolinium sulfate into the water. A free neutron thermalizes in microseconds and is then captured by a gadolinium nucleus, which de-excites by emitting a cascade of gamma rays totalling about 8 MeV. That cascade is bright enough to be seen by the surrounding photomultiplier tubes. The gadolinium-loading technique was demonstrated at the Super-K scale and has propagated to several other experiments; ANNIE uses a higher concentration to push the capture efficiency upward and characterise it under controlled beam conditions.

The detector also serves as one of the first physics deployments of LAPPDs, Large-Area Picosecond Photodetectors. These flat micro-channel-plate devices were developed by the LAPPD Collaboration around Argonne and produce photon timing on the order of tens of picoseconds — roughly an order of magnitude better than conventional PMTs. With timing that sharp, ANNIE can reconstruct the vertex of a neutrino interaction inside the tank to within centimeters, which is what makes such a small fiducial volume usable. The same technology is on the wish list for several next-generation neutrino telescopes, so ANNIE doubles as an in-beam stress test for it.

Results so far and what comes next

ANNIE took its first full physics data in 2022 and 2023 on the Booster Neutrino Beam, which delivers a flux peaked around 800 MeV — almost exactly the relevant energy range for T2K, Hyper-K, and the lower-energy parts of the DUNE program. The collaboration has reported preliminary measurements of neutron multiplicity distributions and cross-section ratios in conference talks and on the arXiv, with the first peer-reviewed measurements of neutron multiplicity following charged-current muon-neutrino interactions on water appearing in the past two years. A follow-up dataset is being collected with a higher gadolinium concentration and a fuller LAPPD array.

Whether the final number lands a few percent above or below what current event generators predict is, by itself, a fairly narrow piece of physics. But every percent of systematic uncertainty that gets pinned down by ANNIE shows up later as a faster path to a CP-violation discovery in a detector a thousand times larger, and as a more believable supernova reconstruction the next time a star in the Milky Way collapses. The big experiments need small ones to teach them how the water reacts. ANNIE is one of them — and it is starting to pay back the loan.