How Neutrino Beams Are Made

A muon-neutrino beam coming out of a Fermilab or J-PARC facility looks, from the outside, like nothing at all. The neutrinos emerge from underground, pass through the surface, and travel several hundred kilometres to a far detector site — invisibly, leaving no trace. Even at the production facility itself, no one notices anything unusual: the accelerator hums, the target absorbs energy, the cooling systems work — but the neutrinos that emerge cannot be seen, heard, or measured at the source.

What happens, though, is one of the most complex sequences in modern accelerator physics. A 120-GeV proton beam from the Fermilab Main Injector, or a 30-GeV beam from J-PARC, hits a carefully designed target. The collision produces a spray of secondary particles. Magnetic horns focus the wanted ones, defocus the unwanted ones, and direct the resulting forward-going meson flux into an evacuated tunnel. The mesons decay in flight, producing neutrinos. A hadron absorber at the end of the tunnel removes everything except the neutrinos and the daughter muons. The muons themselves are absorbed in rock further downstream, leaving the neutrinos to propagate in a clean, mono-flavor beam toward the far detector.

This post is about the components of an accelerator neutrino beam — what each part does, why it has to be there, and how the resulting beam ends up with the properties that long-baseline neutrino experiments rely on.

The basic chain

Every accelerator neutrino beam has the same fundamental structure:

1. Proton driver: accelerates protons to high energy
2. Target: produces secondary mesons through proton-nucleus collisions
3. Magnetic horns: focus or defocus charged mesons depending on sign
4. Decay tunnel: an evacuated drift region where mesons decay in flight to produce neutrinos
5. Hadron absorber: stops all remaining hadrons (protons, kaons, neutrons, etc.) and undecayed mesons
6. Muon dump: stops the daughter muons in rock
7. Pointing: the entire beamline is precisely aligned to the far detector

The result, at the far detector site, is a beam of muon neutrinos with a known energy spectrum, known baseline, known timing, and predictable backgrounds.

The proton driver

Modern neutrino beams require proton beams with substantial intensity. The relevant parameters:

Energy: typically 30-120 GeV. Higher energy gives more secondaries and a wider neutrino energy spectrum.

• T2K at J-PARC: 30 GeV protons
• NOvA / DUNE at Fermilab Main Injector: 120 GeV protons
• CNGS at CERN (now decommissioned): 400 GeV protons

Intensity: typically $10^{13}$–$10^{14}$ protons per pulse, with pulse rates of 1-10 Hz. The Fermilab NuMI beam has been operating at approximately 500 kW.

Pulse structure: protons are extracted in short pulses (microsecond-scale) with intervals of seconds between pulses. The pulse structure allows precise timing in the far detector, enabling the experiment to distinguish neutrino events from background by their arrival time.

For DUNE, the proton driver will be the upgraded Main Injector with planned beam power of 1.2 MW. This is approximately 2.5 times the current NuMI intensity and provides the statistical leverage needed for DUNE’s precision measurements.

The target

The target is where the protons hit and produce secondary mesons. Material choice is constrained by:

Atomic number: Lower Z (carbon, beryllium) gives cleaner pion/kaon production with less other-particle contamination. Higher Z (mercury, lead) gives more total particle production but with more electron-induced backgrounds.

Density and length: The target must be long enough to absorb most of the proton beam (typically 1-2 interaction lengths) but not so long that secondary particles re-interact with downstream material.

Thermal handling: A multi-kW proton beam dumped into a small target deposits enormous power. The target must dissipate this heat without melting or fracturing.

Common choices:

• NuMI (Fermilab): segmented graphite target, with internal cooling channels
• T2K (J-PARC): graphite target, water-cooled
• CNGS (CERN): graphite target

Beryllium targets have also been used at higher-energy proton drivers (e.g., the original LBNF target design). The choice depends on the balance between particle production rate, secondary-spectrum cleanness, and thermal handling.

The magnetic horns

After the target, the secondary mesons emerge in a forward-going cone with angular spread of order tens of milliradians. Without focusing, the resulting neutrino beam at the far detector would be too divergent to give a usable flux.

Magnetic horns are toroidal-shape conductors carrying very high current (~300 kA pulsed). The current produces a strong azimuthal magnetic field that focuses charged particles of one sign while defocusing the other sign. The focusing strength is proportional to the charge times the perpendicular momentum.

A typical neutrino beamline uses 2-3 horns in series:

• Horn 1: focuses positive mesons forward (for a ν_μ beam) by reversing the focal length for negatives
• Horn 2 (and 3): refines the focusing, particularly for higher-momentum mesons that the first horn couldn’t fully capture

The current direction can be reversed: with reversed current, the horns focus negative mesons instead of positive, producing a ${\bar{\nu }}_{\mu }$ beam instead of a ${\nu }_{\mu }$ beam. This is how long-baseline experiments measure both neutrino and antineutrino oscillations: by alternating horn polarity between running periods.

The decay tunnel

The decay tunnel is a long, evacuated steel tube where the mesons decay in flight. Typical length: 500-700 metres for NuMI and similar beamlines.

The relevant physics: a pion with energy $E_{\pi }$ in the lab frame has a Lorentz factor $\gamma =E_{\pi }/m_{\pi }{c}^2$. Its decay length is: $L_{\text{decay}}=\gamma c{\tau }_{\pi }\approx 56\text{ metres}\times \frac{E_{\pi }}{1\text{ GeV}}$

For 3 GeV pions, the decay length is approximately 170 metres. For a 500-metre decay tunnel, approximately 95% of the pions will decay before reaching the absorber. The remaining 5% are absorbed and removed from the beam.

The tunnel is evacuated to prevent meson-nuclear interactions that would produce additional secondaries (and additional, unwanted neutrinos). The vacuum is typically maintained at $10^{-3}$ to $10^{-4}$ Torr — adequate to suppress reinteractions without requiring high-vacuum sealing infrastructure.

The hadron absorber

At the end of the decay tunnel, a thick absorber (typically aluminum or steel, ~10 metres thick) stops all remaining hadrons:

• Undecayed pions and kaons (~5% of the original)
• Reinteracted secondaries (additional pions, protons, neutrons, etc.)
• Knock-on particles from the meson-tunnel-wall interactions

The absorber also absorbs much of the gamma-ray and electromagnetic radiation. Heat from the absorber is removed via water cooling.

Downstream of the absorber, only daughter muons and neutrinos remain. The muons themselves are absorbed in the rock further downstream — typical depths of 200-500 metres of granite are sufficient to stop muons with energies up to several GeV.

The neutrino beam composition

After all this, the resulting beam at the far detector is predominantly muon neutrinos with small contaminations:

• ν_μ: 90-95% (the desired component)
• ν̄_μ: 5-10% (from K⁻ decay and other minor sources)
• ν_e: 0.5-1% (from K_e3 decays and muon decays in flight)
• ν̄_e: 0.1-0.3%

The ν_e contamination is particularly important because it constitutes an irreducible background for ν_μ → ν_e appearance measurements. Modern beam designs minimise this contamination through careful tuning of the horn currents and target geometry, but it can never be eliminated completely.

The energy spectrum is approximately log-normal in shape, peaking at a few GeV (depending on the proton-driver energy and horn focusing). The spectrum extends from a few hundred MeV to tens of GeV.

Far-detector targeting

The beamline must be precisely aimed at the far detector. The geometry is solved by:

Geodetic survey: GPS-based determination of the absolute positions of the production point and the far detector. Modern surveys provide approximately 1-metre absolute precision over distances of hundreds of kilometres.

Beam direction: the proton-target axis is set at the appropriate downward angle so that the resulting neutrino beam emerges from the production point on a trajectory that passes through the far detector. For Fermilab to South Dakota (DUNE), the downward angle is approximately 5°.

Calibration: muon-monitor measurements in the absorber region (at variable distance behind the hadron absorber) provide real-time monitoring of the beam direction. Small variations in the proton-beam alignment, target alignment, or horn alignment can shift the resulting neutrino beam direction; the muon monitors detect these shifts and feed back to adjust the alignment.

For NOvA, the baseline is 810 km from Fermilab to Soudan, Minnesota. For DUNE, the baseline is 1300 km from Fermilab to South Dakota. Both beams have geodetic alignment accuracy of approximately 10 metres at the far site — well within the multi-hundred-metre fiducial volume of the detectors.

Different beam designs

Three main accelerator-neutrino beams have been operating in the past two decades:

T2K at J-PARC (Japan, since 2010): 30-GeV proton driver, beam power approximately 500 kW, 295-km baseline to Super-Kamiokande. Optimized for sub-GeV neutrino energies, well-matched to the first oscillation maximum at the chosen baseline.

NuMI at Fermilab (USA, since 2005): 120-GeV proton driver, beam power approximately 500 kW, baseline 810 km to NOvA at Soudan or 730 km to MINOS+ at the same site (decommissioned). Mid-energy beam tuned to first oscillation maximum.

LBNF/DUNE (Fermilab, expected to start operation 2030): 120-GeV proton driver upgraded to 1.2 MW, 1300-km baseline to South Dakota. Designed specifically for the longer baseline and the resulting matter-effect enhancement of CP-violation sensitivity.

CNGS at CERN (operational 2006-2012): 400-GeV proton driver, 730-km baseline to Gran Sasso. Higher-energy beam with substantial ${\nu }_{\tau }$ component for OPERA’s tau-appearance measurement. Decommissioned after OPERA’s tau-appearance result was completed.

What this enables

The beam-based experimental programme is the foundation of modern long-baseline neutrino physics:

• Oscillation parameter measurements: ${\theta }_{23}$, $\Delta m_{31}^2$, ${\theta }_{13}$, ${\delta }_{CP}$ are all measured at long-baseline beam experiments.
• Matter-effect measurements: the matter-induced asymmetry between neutrino and antineutrino oscillation at long-baseline is the primary signature for mass ordering.
• Cross-section measurements: by varying detector position relative to the beam, the neutrino-nucleon cross-section is measured as a function of energy.
• Beyond-Standard-Model searches: at long baselines, deviations from the three-flavor predictions could signal sterile neutrinos, non-standard interactions, or other new physics.

Each measurement requires the beam to be tuned, characterised, and aligned with the far detector to high precision. The technical effort to produce and maintain a working neutrino beamline is itself a major physics-and-engineering programme, paralleling the detector-side effort.

By 2030 DUNE will join the existing T2K and NOvA beams in operation. The combined dataset will provide the most precise neutrino oscillation parameter measurements ever performed, with statistical sensitivities at the 1-5% level on most parameters.

For more on specific long-baseline experiments, see T2K, NOvA and the CP-violation hunt, DUNE: the deep underground experiment, MINOS and precision mass-splitting, and OPERA’s tau appearance result.

The beamline is, in some respects, the unsung hero of modern neutrino physics. The headlines go to the detectors and to the oscillation results. But none of it would work without the careful engineering of the proton-target-horn-tunnel-absorber chain that produces the beam in the first place. Every neutrino detected at the far end of a long-baseline experiment started its life as a high-energy proton at the accelerator — and was shaped, along its journey, by every component of the beamline.