MINOS: The First Precision Measurement of Atmospheric Mass Splitting

In the late 1990s, Super-Kamiokande’s measurement of the atmospheric neutrino deficit established that muon neutrinos oscillate during long-baseline propagation. The result was a generational milestone — but it left several open questions that atmospheric neutrinos alone could not resolve. The ratio of muon to electron neutrinos was indirect, the energy spectrum was broad, and the source flux was uncertain at the 20% level.

To convert “discovery” into “precision physics”, what was needed was a controlled accelerator-based experiment with a known beam and a dedicated detector. The Main Injector Neutrino Oscillation Search — MINOS — was the experiment that delivered this transition. Operating from 2005 to 2016 with a 735-km baseline from Fermilab in Illinois to the Soudan mine in Minnesota, MINOS measured the atmospheric mass splitting to 3% precision and the atmospheric mixing angle to comparable precision. Equally importantly, it tested whether neutrinos and antineutrinos oscillate at the same rate — a key probe of CPT invariance — and established the framework for the next generation of accelerator experiments.

This post covers MINOS as a case study in the transition from discovery to precision in neutrino physics.

The experimental design

MINOS consisted of two functionally identical iron-and-scintillator tracking calorimeter detectors:

• Near detector (980 tons), 1 km from the Fermilab beam target
• Far detector (5,400 tons), 735 km from the source, deep in the Soudan iron mine in Minnesota

Both detectors were heavily magnetised — a uniform 1.5 T toroidal field — making them tracking spectrometers as well as calorimeters. Charged-current ${\nu }_{\mu }$ events produced muons whose curvature in the magnetic field gave the muon’s charge (positive for ${\bar{\nu }}_{\mu }$, negative for ${\nu }_{\mu }$) and momentum.

The signature: a ${\nu }_{\mu }\to {\nu }_{\mu }$ event would produce a muon track from charged-current scattering. The number of such tracks at the far detector compared to the near detector, after accounting for $1/r^2$ flux drop, gave the survival probability — and thus the oscillation parameters $\Delta m_{atm}^2$ and ${\sin }^2(2{\theta }_{atm})$.

The beam was the NuMI beam at Fermilab (Neutrinos at the Main Injector). A 120-GeV proton beam from the Main Injector struck a graphite target, producing a focused secondary beam of pions and kaons that decayed to muon neutrinos. The beam was tunable: it could be configured for low-energy ($\sim 3$ GeV peak), medium-energy ($\sim 6$ GeV peak), or high-energy operation. MINOS ran primarily in low-energy mode, where the first oscillation maximum sat near the beam peak energy.

The polarity of the focusing horn could be reversed to produce an antineutrino-dominated beam. This let MINOS run separate ${\nu }_{\mu }$ and ${\bar{\nu }}_{\mu }$ campaigns — a distinctive feature.

The headline measurement

By the end of its primary running phase (2005-2012), MINOS had measured: $|\Delta m_{atm}^2|=(2.41\pm 0.10)\times 10^{-3}{\text{ eV}}^2$ ${\sin }^2(2{\theta }_{23})>0.95\text{ at 90\% C.L.}$

The mass-splitting precision was approximately 4% — comparable to the best Super-Kamiokande measurement, but with the controlled beam systematics. The mixing angle was consistent with maximal mixing.

By 2014, with extended MINOS+ running, the precision improved to about 3% on $\Delta m^2$ and similar precision on the angle. This is comparable to current global-fit averages and is consistent with subsequent measurements from T2K and NOvA.

CPT and antineutrino comparison

A unique contribution of MINOS was its dedicated antineutrino programme. By switching the focusing-horn polarity, the experiment could compare: $P({\nu }_{\mu }\to {\nu }_{\mu })\text{vs.}P({\bar{\nu }}_{\mu }\to {\bar{\nu }}_{\mu })$ If neutrinos and antineutrinos have different oscillation parameters, this would be a violation of the CPT theorem — one of the deepest assumptions of quantum field theory. Pre-MINOS, this had not been tested at the level of percent precision.

MINOS’s antineutrino mode produced approximately 25% of the statistics of its neutrino mode (because of the lower antineutrino interaction cross-section). This was sufficient to constrain $|\Delta m_{atm}^2(\bar{\nu })-\Delta m_{atm}^2(\nu )|<0.5\times 10^{-3}{\text{ eV}}^2\text{ at 90\% C.L.}$ consistent with zero, supporting CPT invariance.

A 2010 preliminary analysis had shown a tension between $\nu$ and $\bar{\nu }$ best-fit parameters at about $2\sigma$ — briefly raising hopes (or fears) of CPT violation. With more data and improved analysis, the tension reduced to consistency. The result is now considered a textbook example of the importance of patience in precision physics: anomalies at $2\sigma$ usually disappear with more data.

The “atmospheric” angle

MINOS’s measurement of ${\theta }_{23}$ near maximal — ${\sin }^2(2{\theta }_{23})>0.95$ — was specifically the measurement that motivated subsequent experiments to focus on the octant question. With ${\theta }_{23}$ very close to $\pi /4$, both ${\theta }_{23}<\pi /4$ (“lower octant”, ${\sin }^2{\theta }_{23}\approx 0.45$) and ${\theta }_{23}>\pi /4$ (“upper octant”, ${\sin }^2{\theta }_{23}\approx 0.55$) gave essentially identical ${\sin }^2(2{\theta }_{23})$. Distinguishing the two became the next-generation challenge for T2K, NOvA, and DUNE.

MINOS itself had only weak octant sensitivity. The data slightly preferred lower octant, but the significance was below $2\sigma$. The question is still not definitively resolved as of 2026.

Other contributions

MINOS made several other measurements during its 11-year run:

Sterile-neutrino searches: MINOS examined the spectrum of muon-neutrino disappearance for shape distortions characteristic of mixing with a fourth, sterile, neutrino state. The combined MINOS / MINOS+ analysis excluded substantial regions of the LSND-favored sterile-neutrino parameter space, providing independent constraints on the most popular new-physics interpretation of the LSND anomaly.

Atmospheric muon-neutrino flux: With the deep underground detector and substantial run time, MINOS recorded a large sample of cosmic-ray atmospheric neutrinos as a calibration sample. This provided independent measurements of atmospheric oscillation parameters that complemented the beam-based measurement.

Search for neutrino-nucleon interactions: Detailed MINOS data on charged-current and neutral-current cross-sections at GeV energies on iron targets contributed to nuclear-physics inputs that are now used in DUNE and other experiments.

Legacy

MINOS’s contribution to neutrino physics is best understood as bridging two eras.

Before MINOS: Atmospheric oscillation was a discovered phenomenon with order-of-magnitude precision on its parameters. Subsequent measurements were happening (K2K in Japan, others), but no single experiment had achieved few-percent precision on the atmospheric mass splitting.

After MINOS: $\Delta m_{atm}^2$ was measured to 3% precision. CPT invariance in neutrino oscillation was tested at the percent level. The framework for combining accelerator measurements with reactor measurements was established. The next-generation experiments — T2K, NOvA, future DUNE — could be designed knowing that 3% atmospheric precision was the starting point.

MINOS also demonstrated, in practical terms, that a magnetised tracking calorimeter at deep underground depth was a viable architecture for long-baseline oscillation. Its 5,400-ton iron-scintillator detector was, until recently, the largest off-axis-magnet neutrino detector ever built. The architecture proved less competitive than later water Cherenkov (Hyper-K) and liquid argon (DUNE) for the next-generation experiments, but the systematic experience MINOS built up — particularly around beam flux uncertainty, detector calibration, and event reconstruction — fed directly into successor designs.

Decommissioning

MINOS+‘s extended programme ran until June 2016. The Fermilab NuMI beam was reconfigured for the upcoming NOvA experiment, which had started taking data in 2014 and which uses a different detector technology (liquid scintillator) and a longer 810-km baseline. The Soudan iron-mine site, which had hosted MINOS for 11 years, was decommissioned. The mine itself closed for safety reasons in 2018.

The MINOS physics analyses continued to be published into the early 2020s as final cross-checks and refinements. The collaboration has now disbanded, but its members — many hundreds across dozens of institutions — distributed into NOvA, T2K, DUNE, and other current programmes. The expertise built up over MINOS’s eleven years is the institutional foundation on which the next decade of accelerator-neutrino physics is being built.

When the 2015 Nobel Prize was awarded for atmospheric oscillation discovery, the citation specifically referenced “the discovery of neutrino oscillations”. MINOS was not part of the discovery — Super-Kamiokande was. But MINOS was a key part of the precision programme that turned the discovery into a quantitative theory, and it deserves a prominent place in any history of neutrino oscillations.