Off-Axis Neutrino Beams: How T2K and NOvA Get a Narrow Spectrum from a Wide One

A typical accelerator neutrino beam starts as a proton beam slamming into a target, producing pions and kaons that decay in flight to give a beam of mostly muon neutrinos. The simplest configuration places the detector directly in line with the parent meson direction. The resulting neutrino flux is broad in energy: pions of different energies all contribute to the on-axis flux, and the resulting spectrum spans from a few hundred MeV up to several GeV.

For a long-baseline oscillation experiment, that breadth is partly a feature and partly a bug. The feature: by measuring the oscillation probability across a broad range of energies, you can map out the full oscillation curve and extract several parameters at once. The bug: a broad spectrum means many neutrinos sit at energies far from the oscillation maximum, where they contribute statistically but not informatively, and where they can produce backgrounds — particularly neutral-current pion events at the high-energy end that can fake electron-neutrino appearance signals.

The off-axis beam technique trades the broad on-axis spectrum for a narrow, peaked one by simply pointing the detector a few degrees off the axis of the parent pion beam. The trick exploits the kinematics of two-body pion decay in a way that turns a continuous pion energy spectrum into a tightly peaked neutrino energy spectrum. This is the technique that T2K pioneered as part of its initial design and that NOvA adopted to chase complementary L/E sensitivity. This post is about why off-axis works, where the narrow peak comes from, and what it changes about the experiments using it.

The kinematics of pion decay

A charged pion at rest decays into a muon and a muon neutrino through the two-body process

${\pi }^{+}\to {\mu }^{+}+{\nu }_{\mu }$

The decay is isotropic in the pion rest frame, with the neutrino carrying a fixed momentum $p_{\nu }^{\ast }=(m_{\pi }^2-m_{\mu }^2)/(2m_{\pi })\approx 30$ MeV.

In the lab frame, a pion travelling with momentum $p_{\pi }$ and Lorentz factor $\gamma =E_{\pi }/m_{\pi }$ produces a neutrino whose energy depends on the angle between the neutrino and the pion’s direction of motion. Working through the Lorentz boost:

$E_{\nu }=\frac{(1-{\beta }^2{)}^{1/2}{p}_{\nu }^{\ast }}{1-\beta \cos {\theta }^{\ast }}\gamma =\frac{p_{\nu }^{\ast }}{1-\beta \cos {\theta }^{\ast }}$

with $\beta =p_{\pi }/E_{\pi }$ and ${\theta }^{\ast }$ the neutrino angle in the rest frame.

When the lab-frame angle $\theta$ between neutrino and pion is small, two effects conspire. The boost factor $\gamma$ multiplies the rest-frame neutrino energy upward; but the angular spread of the pion-rest-frame decays maps to lab-frame angles that are inversely proportional to $\gamma$. The net result, after some algebra, is that the lab-frame neutrino energy at a fixed lab-frame angle $\theta$ becomes nearly independent of the pion energy for high-energy pions:

$E_{\nu }(\theta )\approx \frac{(1-m_{\mu }^2/m_{\pi }^2)}{1+{\gamma }^2{\theta }^2}{E}_{\pi }$

For sufficiently high ${\gamma }^2{\theta }^2$ — that is, sufficiently off-axis — this saturates at a fixed neutrino energy determined by $\theta$ alone:

$E_{\nu }(\theta )\approx \frac{(1-m_{\mu }^2/m_{\pi }^2)m_{\pi }}{2\theta }\approx \frac{30\text{MeV}}{\theta }$

A 2.5-degree off-axis angle gives a neutrino energy near 600 MeV. A 14-milliradian off-axis gives about 2 GeV. The result is essentially independent of the parent pion’s energy, so even a broad pion spectrum produces a sharp neutrino energy peak — exactly what is wanted for an oscillation experiment.

Where the narrowness comes from

The off-axis peak’s narrowness has a clean geometric interpretation. On axis, pions of every energy contribute neutrinos at their boosted endpoint energies, producing a flat-topped flux that extends up to the maximum pion energy. Off axis, only a narrow band of pion energies projects neutrinos in the off-axis direction at all, and that band coincidentally contributes neutrinos at nearly the same energy because higher-energy pions produce smaller-angle neutrinos.

The trade-off is event rate. A given off-axis detector sees only a fraction of the neutrinos that an on-axis detector at the same distance would see. T2K’s 2.5-degree off-axis angle reduces the flux by about a factor of 5 compared to an on-axis configuration at the same baseline. NOvA’s 14-milliradian off-axis reduces it by a comparable factor. The cleaner spectrum and improved oscillation sensitivity at the chosen L/E generally compensate, but the experiments must run longer or use larger detectors to accumulate the same statistics.

What experiments choose with the off-axis angle

The choice of off-axis angle is a strategic decision about which oscillation physics to optimise.

T2K chose 2.5 degrees off-axis with a 295 km baseline from J-PARC to Super-Kamiokande. The resulting peak neutrino energy of about 600 MeV sits at the first oscillation maximum for the atmospheric mass splitting at this baseline: $L/E\approx 500$ km/GeV. This places the experiment in the regime of strong ${\nu }_{\mu }\to {\nu }_e$ appearance, with the CP-violating contribution to the appearance probability maximised. The narrow spectrum also concentrates events near the first oscillation maximum and reduces the high-energy tail that produces neutral-current pion backgrounds.

NOvA chose 14 milliradians off-axis with an 810 km baseline from Fermilab to northern Minnesota. The longer baseline and higher peak energy of about 2 GeV move NOvA into a regime where matter effects are stronger than at T2K and contribute substantially to the asymmetry between neutrino and antineutrino appearance probabilities. The longer baseline gives NOvA more sensitivity to the mass ordering through these matter effects, complementing T2K’s more matter-effect-free CP-sensitive measurement.

Hyper-Kamiokande, T2K’s successor, uses the same J-PARC beam at the same off-axis angle as T2K, but with a far detector roughly ten times larger. The 2.5-degree geometry is preserved because Hyper-K’s physics program continues at the same L/E as T2K’s, with the improved statistics targeting precision CP-violation discovery.

DUNE takes a different approach — an on-axis wide-band beam from Fermilab to South Dakota with a 1300 km baseline. The on-axis configuration gives DUNE a broad spectrum covering both the first and second oscillation maxima, allowing it to map the oscillation pattern across multiple energies simultaneously. The broad spectrum is well-matched to the LArTPC’s excellent particle-identification, which mitigates the higher backgrounds, and the longer baseline brings stronger matter effects that disentangle the mass ordering.

The downsides

The off-axis trick is not without compromises.

The first is the lower event rate per proton-on-target. As noted, the off-axis flux is a factor of 5 or so below the on-axis flux at the same baseline. For statistics-limited measurements this is a real cost.

The second is that the narrow peak means the experiment measures the oscillation probability at essentially one $L/E$. Mapping out the full oscillation curve requires complementary measurements at different baselines or beam configurations. T2K and NOvA together cover two distinct $L/E$ values; DUNE’s on-axis approach covers many at once.

The third is that the peak energy depends on the off-axis angle, so the precise positioning of the detector matters. Both T2K and NOvA carefully verify the detector alignment and account for the angular spread of the neutrino flux at the detector.

Why this design choice persists

Despite the trade-offs, off-axis beams will remain part of the long-baseline toolkit because they offer a clean way to concentrate events at the physics-relevant L/E and to reduce specific backgrounds that on-axis beams suffer. Hyper-K’s continuation of T2K’s 2.5-degree configuration is the clearest signal that the experimental community values the narrow-spectrum advantages enough to design new experiments around them. DUNE’s on-axis choice is a strategic complement, not a refutation — the two approaches probe overlapping but distinct parts of the parameter space, and the precision goals of CP violation discovery and mass-ordering determination require inputs from both.

The off-axis trick also has applications beyond oscillation. The narrow energy spectrum is useful for cross-section measurements, where knowing the incident neutrino energy precisely simplifies the analysis. Several dedicated cross-section experiments — MINERvA’s off-axis runs, T2K’s near-detector measurements — exploit the same principle to measure exclusive interaction channels at fixed energy.

Summary

The off-axis beam technique exploits the two-body kinematics of pion decay to produce a narrow neutrino energy spectrum at the cost of overall flux. A detector placed a few degrees off the parent pion beam axis sees neutrinos at an energy approximately independent of the parent pion energy, set by the off-axis angle through $E_{\nu }\approx 30\text{MeV}/\theta$. T2K’s 2.5-degree geometry produces a sharp peak near 600 MeV at 295 km baseline, optimised for first-oscillation-maximum ${\nu }_{\mu }\to {\nu }_e$ appearance with low background contamination. NOvA’s 14-milliradian configuration peaks at 2 GeV with an 810 km baseline, optimised for stronger matter effects and mass-ordering sensitivity. The technique trades event rate for spectral cleanliness and persists into the next generation through Hyper-Kamiokande, complementing DUNE’s on-axis wide-band approach. Together, off-axis and on-axis beams cover the long-baseline oscillation phase space with overlapping but distinct strengths.