MicroBooNE 2024: The Verdict on the MiniBooNE Excess

The MiniBooNE experiment ran at Fermilab from 2002 to 2019, accumulating one of the longest data sets in modern accelerator-neutrino physics. Its primary purpose was to test the LSND anomaly — the 1996 result from Los Alamos that had reported an excess of electron-antineutrino-like events in a muon-antineutrino beam at short baseline, suggestive of oscillation-induced appearance through a sterile neutrino at $\Delta m^2\sim 1$ eV².

By 2007, MiniBooNE had reported its first results. There was no excess at the high reconstructed-energy end where LSND-like oscillation should have produced a clean signal. But there was an unexpected excess at low reconstructed energy — events that looked electron-like, with energy peaking around 200-400 MeV, well below the LSND-favored region.

The 2007 excess was about 4$\sigma$ above the expected background. Various interpretations were possible: oscillation through an unconventional sterile-neutrino model, single-photon backgrounds from neutrino-induced processes, mismodelled beam-related background, or detector misreconstruction. The MiniBooNE collaboration’s final 2018 analysis, with the full data set, reaffirmed the excess at greater than 4.7$\sigma$ significance — but the interpretation remained unclear.

The fundamental problem: MiniBooNE could not distinguish electrons from photons. Its mineral-oil Cherenkov detector saw an electromagnetic shower as a single ring of light, regardless of whether the original particle was an electron (charged-current ${\nu }_e$ interaction signature) or a photon (a possible background from neutrino-induced ${\pi }^0$ or $\Delta$-resonance processes). The reconstruction algorithms had to assume one or the other; misclassification was systematic.

To resolve the question, the field needed a detector that could distinguish electrons from photons event-by-event. MicroBooNE, a liquid-argon time-projection chamber positioned in the same Fermilab Booster Neutrino Beam beamline as MiniBooNE, was designed specifically for this purpose. By 2024, after multiple years of data analysis, MicroBooNE has issued what is widely considered the definitive verdict on the MiniBooNE excess.

This post is about MicroBooNE’s analyses, what they found, and what the MiniBooNE saga now looks like as a closed chapter.

The MiniBooNE excess in detail

MiniBooNE’s data sample, accumulated over 17 years of running, contained roughly 1,800 candidate electron-like events in neutrino mode and approximately 200 in antineutrino mode. The expected background from intrinsic beam ${\nu }_e$ contamination, neutral-current ${\pi }^0$ production with one missed photon, and other processes accounted for about 1,200 in neutrino mode. The excess of approximately 460 events at low energy was the anomaly.

The excess had a characteristic energy spectrum: rising from threshold at 200 MeV, peaking around 300 MeV, and falling above 400 MeV. The shape was inconsistent with the LSND-like oscillation hypothesis (which would peak at higher energy) but consistent with at least three other interpretations:

• Oscillation through a sterile neutrino in a non-LSND parameter region
• A photon-like background that peaks at low energy
• Mismodelled neutrino-induced single-photon background processes
• Detector reconstruction artifacts at low energy

The MiniBooNE collaboration reported the excess as a low-energy electron-like effect without committing to a specific interpretation. Theory papers proliferated through the 2010s, proposing dozens of beyond-Standard-Model scenarios that could produce the excess.

Why electron-vs-photon discrimination matters

The fundamental limitation of MiniBooNE was that mineral-oil Cherenkov detection is “topologically blind” to whether the electromagnetic shower originated from an electron or a photon. Both produce a forward-going light cone of approximately the same opening angle and similar light yield per unit energy.

In a liquid-argon TPC, by contrast, the topology is fully resolved. An electron leaves a continuous ionisation trail from the interaction vertex — a single track with no gap. A photon, having no charge, propagates invisibly until it pair-produces ($\gamma \to e^{+}{e}^{-}$) after a few centimetres of argon (radiation length $\sim 14$ cm); the resulting electromagnetic shower then begins, but at a displaced vertex with a clear gap from the original interaction point.

For neutrino-induced events at MiniBooNE-like energies (200-400 MeV), the gap between the interaction vertex and the photon-conversion vertex is typically several centimetres — well above the spatial resolution of a TPC. This is the technical basis for distinguishing the two scenarios.

MicroBooNE’s analyses

MicroBooNE began taking data in 2015 and accumulated approximately $1.3\times 10^{21}$ protons-on-target through its decommissioning in 2021. The detector is an 85-tonne liquid-argon TPC, with the active region arranged as a single large drift volume with three orthogonal planes of wire-readout for three-dimensional event reconstruction.

The collaboration published a series of analyses, each addressing the MiniBooNE excess from a different angle:

2021 — Single-photon background search: MicroBooNE searched for events with a photon-like signature at low energy. The expected rate, if the MiniBooNE excess were due to single-photon backgrounds (e.g., from $\Delta \to N\gamma$ radiative decay), was calculable. The observed photon rate was lower than expected, with the result reported as $0.6\pm 0.4$ standard-deviation tension with the MiniBooNE-as-photon hypothesis.

2022 — Charged-current electron-neutrino appearance search: MicroBooNE searched directly for the electron-neutrino appearance signal that the LSND-MiniBooNE oscillation hypothesis would predict. The observed rate was consistent with the no-oscillation expectation (intrinsic beam ${\nu }_e$ plus expected backgrounds). The exclusion was at >95% C.L. of the parameter space favored by MiniBooNE under the oscillation hypothesis.

2024 — Combined analysis: The collaboration’s June 2024 release combined multiple data sets and analysis channels into a single likelihood. The result excluded the electron-neutrino interpretation of the MiniBooNE excess at >99% C.L., regardless of the assumed sterile-neutrino mass-squared parameter.

The combined message: whatever caused the MiniBooNE low-energy excess, it was not electron-neutrino appearance. The remaining options are misidentified photons from neutrino-induced processes, detector reconstruction issues, or some other background that does not produce a true electron-like signal in argon.

What is causing the MiniBooNE excess?

Without electron-neutrino appearance as the explanation, the field’s working hypothesis is that the MiniBooNE excess is due to misidentified single-photon backgrounds.

Several photon-producing processes are candidates:

• Neutral-current $\Delta$-resonance radiative decay: $\nu +N\to \nu +{\Delta }^{\ast }\to \nu +N+\gamma$. The $\gamma$ produces an electromagnetic shower that mimics an electron in mineral-oil Cherenkov detection. MicroBooNE has searched for this directly and found the rate consistent with Standard-Model predictions but slightly higher than initial estimates — perhaps enough to account for some, though not all, of the MiniBooNE excess.

• Coherent ${\pi }^0$ production with one missed photon: $\nu +A\to \nu +A+{\pi }^0\to \nu +A+2\gamma$, where one of the photons escapes detection or is merged with the other. MiniBooNE included this in its background estimate but the cross-section is uncertain.

• Single-photon production via NC $\nu +N\to \nu +N+\gamma$: Direct radiation of a photon in the neutral-current interaction. Cross-section is small but non-zero, and was historically underestimated.

By 2025-2026, the consensus in the field is that the MiniBooNE excess is most plausibly a combination of these photon-producing backgrounds, with cross-section estimates that were lower than reality. MicroBooNE’s measurements provide direct constraints on these processes, and the resulting picture is consistent with the MiniBooNE excess being approximately 50-80% photon-driven. Residual events may be explained by other unmodelled detector or reconstruction effects.

The key point: the electron-neutrino oscillation interpretation — the original LSND-MiniBooNE narrative — is excluded.

Implications for the LSND saga

The MiniBooNE result was always tied to the broader LSND saga. With MicroBooNE’s 2024 verdict on MiniBooNE specifically:

• LSND remains the original anomaly, never directly repeated, ongoing 30 years after publication.
• MiniBooNE excess is most likely photon background, not appearance.
• Reactor anomaly is most likely flux mismodeling, not oscillation (PROSPECT/STEREO).
• Gallium anomaly remains real but isolated (no broad confirmation).

The combined picture: there is no clean, broad evidence for an LSND-favored sterile neutrino. Each anomaly that originally pointed toward the parameter space has now been “tested away” by a dedicated successor experiment, except for LSND itself (which remains a single-experiment anomaly without dedicated follow-up).

The “eV-scale sterile neutrino” hypothesis is now considered disfavored by most of the community. By 2026, theoretical interest has largely shifted toward keV-scale sterile neutrinos (warm dark matter candidates), GeV-scale right-handed Majorana neutrinos (collider searches), and TeV-scale heavy neutral leptons (seesaw scenarios). The eV-scale window is essentially closed.

What MicroBooNE leaves open

Three things MicroBooNE does not directly address:

The original LSND signal. LSND was an antineutrino-mode experiment at relatively short baseline, and MicroBooNE is a neutrino-mode experiment in a different beam. While the results are inconsistent with a unified LSND-MiniBooNE oscillation interpretation, they do not directly disprove the original LSND signal. A dedicated LSND-style follow-up at a stopped-pion source is in advanced design (the SBN program at Fermilab).

The MicroBooNE excess interpretation in detail. While the electron-neutrino appearance hypothesis is excluded, the photon-based explanation has not been definitively pinned down. The exact cross-sections of NC $\Delta$-radiative decay and other photon-producing processes remain uncertain at the 20-30% level. Improved measurements are needed.

Other parameter spaces. Sterile neutrinos at much larger or much smaller mass-squared values are not constrained by MicroBooNE. Some scenarios (heavy sterile, very-light sterile) might still hide. The LSND saga has been resolved at its specific parameter point, but the broader sterile-neutrino landscape remains open.

The Short-Baseline Neutrino program

Fermilab’s Short-Baseline Neutrino (SBN) program continues the story. Three liquid-argon detectors share the Booster Neutrino Beam:

• SBND (near detector, 110 metres) — commissioning 2024-2025, will collect higher event rates than MicroBooNE.
• MicroBooNE (now decommissioned) — 470 metres baseline.
• ICARUS (far detector, 600 metres) — operating since 2021.

The program is designed to make a definitive search for ${\nu }_{\mu }\to {\nu }_e$ appearance and disappearance over short baselines, with full electron-photon discrimination at all three positions. By 2027-2028 the combined SBN data set should provide the cleanest, most decisive test of the LSND-style sterile-neutrino hypothesis ever performed.

If SBN confirms MicroBooNE’s verdict — no appearance signal — the eV-scale sterile-neutrino hypothesis will be considered fully closed. If by chance an appearance signal does emerge in SBND or ICARUS that wasn’t visible in MicroBooNE, the field would have to revisit. Most physicists expect the former.

A closing chapter

The MiniBooNE excess was, for nearly two decades, one of particle physics’ more persistent puzzles. It generated hundreds of theoretical papers, dozens of follow-up proposals, and substantial experimental effort. Its resolution — through MicroBooNE’s electron-photon-discriminating argon TPC — is a satisfying technical demonstration of how detector physics can resolve interpretive ambiguity.

The lesson is technical: when an anomaly’s interpretation depends on detector systematics that the original experiment cannot resolve, the path forward is to build a successor experiment with the relevant capability. MicroBooNE was specifically designed to discriminate electrons from photons. It did. The original interpretation does not survive. The remaining puzzle (what specifically causes the photon background) is now a tractable measurement question.

This is, in many ways, how particle physics resolves anomalies in the modern era. Not through theoretical reinterpretation alone, but through experimental designs that test the specific assumptions on which the anomaly rested. The MiniBooNE chapter is now closed in the same way the reactor antineutrino anomaly chapter was closed by PROSPECT and STEREO. The field updates its understanding and moves on.

For LSND specifically, the situation remains slightly unresolved — the original signal has not been directly repeated, but every successor experiment that should have reproduced an LSND-like effect has failed to. By 2030, with the SBN program fully analysed, the LSND saga should be definitively closed. The eV-scale sterile-neutrino hypothesis will, by then, have been tested at multiple sites and energies, with consistent null results. The community will have moved on.

The MiniBooNE excess, in retrospect, will be remembered as a particularly clear example of how technological sophistication in detector design can resolve what seemed intractable. The argon TPC’s three-dimensional topology resolution was the key. The verdict is in.