Neutrino Atlas
EN
  1. Learn
  2. Neutrino beams and long-baseline experiments

Chapter 2: Sources and detection · Lesson 2.5

Neutrino beams and long-baseline experiments

Open

An accelerator turns protons into a controlled beam of muon neutrinos or antineutrinos and sends it hundreds of kilometres through the Earth. Comparing the beam at a near and a far detector measures oscillation with a known source, energy and distance.

Beginner

Making neutrinos on purpose

The Sun, the atmosphere and reactors deliver neutrinos for free, but nobody can switch them on and off or choose their energy. A particle accelerator can. The recipe for an accelerator neutrino beam has four steps. First, protons are accelerated to tens of billions of electronvolts and fired into a target, usually a rod of graphite. The collisions create showers of short-lived particles called pions and kaons. Second, powerful pulsed magnets called magnetic horns gather the charged pions and steer them forward, much as a lens gathers light into a beam. Third, the pions fly through a decay tunnel hundreds of metres long and decay into muons and muon neutrinos. Fourth, a thick block of steel and concrete, followed by earth and rock, stops everything except the neutrinos.

The idea came from Bruno Pontecorvo and Melvin Schwartz around 1960. In 1962 the first such beam at Brookhaven showed that muon neutrinos are a different kind from electron neutrinos1. Simon van der Meer invented the horn in 1961.

Neutrinos or antineutrinos at the flip of a switch

A horn focuses particles of only one electric charge. Positive pions give neutrinos, negative pions give antineutrinos. By reversing the direction of the electric current in the horns, physicists can therefore choose a beam of mostly neutrinos or mostly antineutrinos. Comparing how the two kinds change on their journey tests whether matter and antimatter behave differently, one of the big open questions described in the lesson “Mixing and CP violation”.

Two detectors, before and after

Most long-baseline experiments use two detectors. A near detector a few hundred metres from the target measures the beam before any oscillation has taken place. A far detector hundreds of kilometres away measures it after the journey. Comparing the two is like weighing a parcel before and after shipping: many uncertainties about the beam itself drop out, and what remains is the change that happened on the way.

Looking slightly to the side

A detector placed exactly in the centre of the beam sees neutrinos of many energies. A few degrees to the side it sees fewer neutrinos, but almost all with the same energy. By choosing this off-axis angle, physicists tune the energy to the distance at which oscillation is strongest.

Experiments over the years

K2K sent the first long-baseline beam 250 kilometres from Tsukuba to Super-Kamiokande from 1999 and confirmed the oscillation seen in atmospheric neutrinos2. MINOS sent neutrinos 735 kilometres from Fermilab near Chicago to a mine in Minnesota3. T2K has sent its beam 295 kilometres across Japan since 2010, and in 2011 it saw the first signs of muon neutrinos turning into electron neutrinos. NOvA sends neutrinos 810 kilometres from Fermilab to northern Minnesota. In 2025 the two published their first joint analysis5.

The next generation

DUNE will send neutrinos 1,300 kilometres from Fermilab to the Sanford Underground Research Facility in South Dakota, to four detectors filled with about 17,000 tonnes of liquid argon each, 1.5 kilometres underground. Hyper-Kamiokande, a water tank about eight times larger than Super-Kamiokande, will receive the upgraded Japanese beam from about 2028. Together they should reveal which neutrino is the heaviest and whether neutrinos and antineutrinos oscillate differently.

Quiz

Test your knowledge

4 questions on this level · Print as worksheet

  1. 1What do the magnetic horns in a beamline do?
  2. 2How do physicists switch from a neutrino beam to an antineutrino beam?
  3. 3Why do most long-baseline experiments use a near detector?
  4. 4How far will DUNE send its neutrinos?

Intermediate

Anatomy of a beamline

  • Proton driver: 30 GeV protons from the J-PARC Main Ring for T2K, 120 GeV protons from the Fermilab Main Injector for the NuMI beam. Beam power approaches 1 MW; intensity is counted in protons on target.
  • Target: a graphite rod roughly a metre long, in which most protons interact and produce pions and kaons.
  • Horns: pulsed currents of a few hundred kiloamperes create a toroidal field that focuses mesons of one charge. Horn polarity selects a νμ beam (forward current) or a ν̄μ beam (reverse current).
  • Decay volume: about 96 m at T2K, 675 m for NuMI and about 194 m for LBNF, matched to the decay length of the pions of interest.
  • Absorber and rock: stop the remaining hadrons and range out the muons; muon monitors behind the absorber check the beam direction pulse by pulse.

No beam is flavour-pure. Muon and kaon decays add an intrinsic νe component of order one percent, and some mesons of the wrong sign leak through. Both are backgrounds for the search for νe appearance.

The off-axis trick

In the two-body decay π → μ ν, the neutrino energy seen at a fixed angle to the beam depends only weakly on the pion energy. A detector off the beam axis therefore sees a narrow energy peak. T2K sits 2.5° off-axis, giving a peak near 0.6 GeV at the first oscillation maximum for 295 km; NOvA sits 14 mrad off the NuMI axis, giving about 2 GeV for 810 km. The narrow band also removes high-energy neutrinos that would produce background events. DUNE instead uses an on-axis wide-band beam to cover the first and second maxima.

Near and far

The near detector measures the product of flux and cross-section before oscillation. T2K's ND280 sits 280 m from the target; NOvA's 300-tonne near detector, about 1 km away, uses the same technology as its 14 kt far detector, as MINOS did. Near-to-far extrapolation then predicts the unoscillated far spectrum. What does not cancel are differences in neutrino–nucleus interactions between νe and νμ, and between neutrinos and antineutrinos, which today form the leading systematic uncertainty.

Milestones

  • K2K (1999–2004): 12 GeV KEK protons, 250 km; 112 events observed against 158 expected without oscillation, Δm² ≈ 2.8 × 10⁻³ eV²2.
  • MINOS (2005–2016): NuMI, 735 km to a 5.4 kt magnetised steel detector in the Soudan mine; precise Δm²32 and separate ν and ν̄ samples3.
  • OPERA (2008–2012): CERN's CNGS beam to Gran Sasso, 730 km; tau-neutrino appearance established in 2015.
  • T2K (since 2010): νe appearance hinted in 2011 and established in 2013; first constraints on δCP in 20204.
  • NOvA (since 2014): 810 km, νe and ν̄e appearance with the largest matter effect of current beams; joint fit with T2K in 20255.

What the beams measure

νμ disappearance fixes θ23 and |Δm²32|. νe appearance depends on θ13, on the octant of θ23, on δCP and, through matter effects, on the mass ordering. Running in both horn polarities separates these. The 2025 joint fit measured Δm²32 to better than 2 % but found no clear preference for either ordering.

Quiz

Test your knowledge

4 questions on this level · Print as worksheet

  1. 1Which off-axis angle and peak energy does T2K use?
  2. 2What did K2K observe at Super-Kamiokande?
  3. 3What is today's leading systematic uncertainty for beam experiments?
  4. 4Which measurement fixes θ23 and |Δm²32|?

Expert

Beam kinematics

Eν = (1 − mμ² / mπ²) Eπ / (1 + γ²θ²) ≈ 0.43 Eπ / (1 + γ²θ²)Neutrino energy from π → μ ν decay in flight at angle θ to the pion direction, γ = Eπ / mπ

At fixed θ, Eν is maximal for γθ = 1, which gives Eν,max ≈ 0.43 mπ / (2θ) ≈ 30 MeV / θ. For T2K (θ = 2.5° ≈ 0.044 rad) this is about 0.7 GeV, with the flux peaking near 0.6 GeV; for NOvA (14 mrad) about 2 GeV. Two-body kaon decays, with 1 − mμ²/mK² ≈ 0.95, populate the high-energy tail, and Ke3 and muon decays supply the intrinsic νe and ν̄e. In reverse-horn mode the wrong-sign νμ fraction is larger, because neutrino cross-sections are two to three times those of antineutrinos.

Oscillation probabilities

P(νμ → νμ) ≈ 1 − sin²2θ23 sin²(1.27 Δm²32 L / E)Leading-order disappearance; Δm² in eV², L in km, E in GeV
Emax ≈ 2.54 Δm²31 L / πEnergy (GeV) of the first oscillation maximum: about 0.6 GeV for 295 km, 1.6 GeV for 810 km, 2.6 GeV for 1,300 km

Appearance is led by sin²θ23 sin²2θ13 sin²(1.27 Δm²_31 L/E), corrected by a CP-odd term proportional to sin δCP that changes the rate by tens of percent; the full expansion is given in the lesson “Mixing and CP violation”. In the crust, matter effects rescale the leading term:

Pmat ≈ Pvac × (1 ± 2E / ER), ER = Δm²31 / (2√2 GF ne) ≈ 11 GeVUpper sign for neutrinos in normal ordering; crust density ρ ≈ 2.8 g/cm³

This gives about ±10 % at T2K, ±30 % at NOvA and ±45 % at DUNE. The ordering-induced asymmetry is then comparable to or larger than the δCP effect, which is why longer baselines break the ordering–δCP degeneracy that limits T2K and NOvA.

Near-to-far extrapolation

Nfar(E) ≈ Nnear(E) × [Φfar(E) / Φnear(E)] × (Mfar / Mnear) × P(E)Flux ratio from beam simulation; near the decay pipe the source is extended, so the ratio depends on energy

The ratio method cancels most of the flux and cross-section normalisation, but not differences in nuclear effects between targets, flavours and ν versus ν̄. Multi-nucleon (2p2h) scattering and final-state interactions bias energy reconstruction by tens of MeV, especially for T2K's kinematic method on water; NOvA and DUNE reconstruct calorimetrically. Water (T2K, Hyper-K), hydrocarbon (NOvA) and argon (DUNE) targets give complementary systematics. DUNE's near detector complex is designed to move part of its apparatus off-axis, so that a set of narrower spectra measured at different angles can be combined to calibrate the mapping from true to reconstructed energy.

Results and outlook

The 2025 T2K–NOvA joint fit gives Δm²32 = 2.43 (+0.04, −0.03) × 10⁻³ eV² (normal) or 2.48 (+0.03, −0.04) × 10⁻³ eV² (inverted), no strong ordering preference and, for inverted ordering, CP conservation disfavoured at the level of evidence5. DUNE plans a 1.2 MW wide-band beam, upgradable to 2.4 MW, and four 17 kt liquid-argon modules, aiming at 5σ ordering for all δCP and 5σ CP-violation discovery for half the δCP range6. Hyper-Kamiokande, with 188 kt fiducial water and a 1.3 MW J-PARC beam, plus an intermediate water Cherenkov detector, starts around 2028; see CP-violation sensitivity.

Quiz

Test your knowledge

4 questions on this level · Print as worksheet

  1. 1Using Eν,max ≈ 30 MeV / θ, what peak energy does NOvA's 14 mrad off-axis angle give?
  2. 2Where is the first oscillation maximum for a 1,300 km baseline?
  3. 3How large is the matter-induced change in appearance probability at DUNE?
  4. 4What does the near-to-far ratio method fail to cancel?

Open questions

Current beams have not yet determined the mass ordering, the octant of θ23 or whether δCP violates CP symmetry; T2K and NOvA alone are limited by statistics and by the degeneracy between ordering and δCP.

The precision that DUNE and Hyper-Kamiokande need depends on modelling neutrino–nucleus interactions in water and argon at the percent level, which is not yet achieved.

Terms to explore

People

On the timeline

  1. 1960Proposals for accelerator neutrino beams
  2. 1962Brookhaven experiment discovers the muon neutrino
  3. 1999K2K starts the first long-baseline experiment
  4. 2011T2K finds hints of electron-neutrino appearance
  5. 2025NOvA and T2K publish a joint analysis
  6. 2026DUNE begins installing detector support structures

References

  1. G. Danby et al., “Observation of high-energy neutrino reactions and the existence of two kinds of neutrinos”, Phys. Rev. Lett. 9, 36 (1962), doi:10.1103/PhysRevLett.9.36
  2. M. H. Ahn et al. (K2K), “Measurement of neutrino oscillation by the K2K experiment”, Phys. Rev. D 74, 072003 (2006), arXiv:hep-ex/0606032, doi:10.1103/PhysRevD.74.072003
  3. D. G. Michael et al. (MINOS), “Observation of muon neutrino disappearance with the MINOS detectors in the NuMI neutrino beam”, Phys. Rev. Lett. 97, 191801 (2006), arXiv:hep-ex/0607088, doi:10.1103/PhysRevLett.97.191801
  4. K. Abe et al. (T2K), “Constraint on the matter–antimatter symmetry-violating phase in neutrino oscillations”, Nature 580, 339 (2020), doi:10.1038/s41586-020-2177-0
  5. NOvA and T2K Collaborations, “Joint neutrino oscillation analysis from the T2K and NOvA experiments”, Nature 646, 818 (2025), doi:10.1038/s41586-025-09599-3
  6. B. Abi et al. (DUNE), “Deep Underground Neutrino Experiment (DUNE) Far Detector Technical Design Report, Volume II: DUNE Physics”, arXiv:2002.03005 (2020), arXiv:2002.03005

Last updated: · Editorial standard

Cite this page

Text

Neutrino Atlas (2026). “Neutrino beams and long-baseline experiments”. Retrieved 22 November 2026, from https://neutrino-atlas.com/learn/neutrino-beams-and-long-baselines/

BibTeX

@misc{neutrinoatlas2026inobeamsandlongbaselines,
  title        = {Neutrino beams and long-baseline experiments},
  author       = {{Neutrino Atlas}},
  year         = {2026},
  url          = {https://neutrino-atlas.com/learn/neutrino-beams-and-long-baselines/},
  note         = {Accessed 2026-11-22},
  language     = {english}
}