nuSTORM: Neutrinos from Stored Muons

• nuSTORM is a stored-muon neutrino source that achieves percent-level flux precision by monitoring muon decay kinematics in situ.
• The facility supports precise neutrino–nucleus cross-section measurements and short-baseline oscillation experiments, critical for DUNE and Hyper-K.
• It serves as both a standalone physics instrument and a technology demonstrator, paving the way for future Neutrino Factories and Muon Colliders.

Neutrinos from Stored Muons (nuSTORM) is a stored-muon neutrino-source concept in which beams of $\nu_\mu/\bar{\nu}_\mu$ and $\nu_e/\bar{\nu}_e$ are generated by the decay of muons circulating in a dedicated racetrack storage ring. Its defining feature is that the parent muon momentum, population, and decay geometry are monitored in situ, so the resulting neutrino flavor content and energy spectrum are calculable from muon-decay kinematics with percent-level normalization uncertainty. nuSTORM was formulated simultaneously as a precision neutrino–nucleus scattering facility, a decisive short-baseline sterile-neutrino experiment, and an entry-level stored-muon accelerator facility relevant to later Neutrino Factory and Muon Collider programs (Ruso et al., 2022).

1. Concept, beam content, and scientific rationale

nuSTORM produces neutrino beams from the decays

$$\mu^+ \to e^+ + \nu_e + \bar{\nu}_\mu,\qquad \mu^- \to e^- + \bar{\nu}_e + \nu_\mu .$$

For $\mu^+$ running, the beam contains $\nu_e$ and $\bar{\nu}_\mu$; for $\mu^-$ running, it contains $\bar{\nu}_e$ and $\nu_\mu$. Because the source process is purely leptonic, the flavor composition and the energy–angle spectra are fixed by decay kinematics rather than by hadron-production and focusing uncertainties, which are the dominant limitation in conventional pion-decay-in-flight beams (Adey et al., 2013).

The principal motivation is precision control of flux and interaction systematics in the few-GeV regime relevant to DUNE and Hyper-K. The 2020 Update of the European Strategy for Particle Physics recommended that muon-beam R&D be treated as a high-priority future initiative and that a program be developed to determine the neutrino cross-sections needed to extract the full physics reach of DUNE and Hyper-K; the nuSTORM and ENUBET collaborations subsequently began a joint, five-year R&D program in that context (Ruso et al., 2022). The same facility concept also addresses the longstanding short-baseline anomalies associated with LSND, MiniBooNE, reactor-flux reevaluations, and gallium-source measurements by providing appearance and disappearance channels with a precisely known initial state (Kyberd et al., 2012).

A persistent distinction in the literature is that nuSTORM is not a full Neutrino Factory. It uses the same stored-muon principle, but at lower stored-muon momentum, without requiring ionization cooling for baseline operation, and with a physics program centered on cross sections and short-baseline oscillations rather than long-baseline CP-violation measurements (Kaplan, 2013). This suggests a staged interpretation: nuSTORM is simultaneously a standalone physics facility and a technology demonstrator for later, more demanding muon-accelerator systems.

2. Muon-decay kinematics and flux prediction

The neutrino spectra follow the Michel distributions. In the muon rest frame, with $x \equiv 2E_\nu^\ast/m_\mu$, the normalized spectra for unpolarized muons are

$\nu_e/\bar{\nu}_e$0

with the corresponding antineutrino spectra obtained by charge conjugation (Adey et al., 2013). Including polarization $\nu_e/\bar{\nu}_e$1, the double-differential rate has the schematic form

$\nu_e/\bar{\nu}_e$2

so ring polarimetry constrains subtle flux-shape systematics when residual polarization is non-negligible (Ruso et al., 2022).

The lab-frame neutrino energy is obtained by a Lorentz boost,

$\nu_e/\bar{\nu}_e$3

with $\nu_e/\bar{\nu}_e$4 and $\nu_e/\bar{\nu}_e$5 determined by the stored-muon momentum (Ruso et al., 2022). For the canonical nuSTORM stored-muon momentum $\nu_e/\bar{\nu}_e$6, $\nu_e/\bar{\nu}_e$7, the forward cone is confined to an opening angle of order $\nu_e/\bar{\nu}_e$8, and forward neutrino energies extend to approximately $\nu_e/\bar{\nu}_e$9 (Adey et al., 2013). This is why the reference design naturally populates the $$\mu^+ \to e^+ + \nu_e + \bar{\nu}_\mu,\qquad \mu^- \to e^- + \bar{\nu}_e + \nu_\mu .$$0–$$\mu^+ \to e^+ + \nu_e + \bar{\nu}_\mu,\qquad \mu^- \to e^- + \bar{\nu}_e + \nu_\mu .$$1 region targeted by long-baseline appearance and disappearance analyses.

For a detector downstream of a production straight, the flux is the line integral over decay positions along that straight. A commonly used approximation is

$$\mu^+ \to e^+ + \nu_e + \bar{\nu}_\mu,\qquad \mu^- \to e^- + \bar{\nu}_e + \nu_\mu .$$2

where $$\mu^+ \to e^+ + \nu_e + \bar{\nu}_\mu,\qquad \mu^- \to e^- + \bar{\nu}_e + \nu_\mu .$$3 is the number of decays in the production straight and $$\mu^+ \to e^+ + \nu_e + \bar{\nu}_\mu,\qquad \mu^- \to e^- + \bar{\nu}_e + \nu_\mu .$$4 encodes the boosted decay spectrum and acceptance (Ruso et al., 2022). The normalization is anchored by the number of stored muons, their momentum distribution, and the surveyed ring geometry, with beam current transformers, BPMs, fast toroids, wall-current monitors, and decay-electron diagnostics identified as the relevant instrumentation in different studies (Ruso et al., 9 May 2025).

At the precision goals of nuSTORM, radiative corrections to muon decay are not negligible. Complete $$\mu^+ \to e^+ + \nu_e + \bar{\nu}_\mu,\qquad \mu^- \to e^- + \bar{\nu}_e + \nu_\mu .$$5 QED corrections modify the expected muon-decay neutrino flux near the peak by $$\mu^+ \to e^+ + \nu_e + \bar{\nu}_\mu,\qquad \mu^- \to e^- + \bar{\nu}_e + \nu_\mu .$$6–$$\mu^+ \to e^+ + \nu_e + \bar{\nu}_\mu,\qquad \mu^- \to e^- + \bar{\nu}_e + \nu_\mu .$$7 permille, which is explicitly at the level of next-generation precision goals, while electron-mass effects at $$\mu^+ \to e^+ + \nu_e + \bar{\nu}_\mu,\qquad \mu^- \to e^- + \bar{\nu}_e + \nu_\mu .$$8 are much smaller and can be neglected for permille-level flux modeling (Tomalak, 2021). A common misconception is therefore that muon-decay beams are “theory exact” once the stored current is known; the literature instead treats precise flux prediction as a combined problem of electroweak decay theory, ring instrumentation, geometry, and detector acceptance.

3. Storage-ring realizations and accelerator design

The canonical nuSTORM facility uses stochastic injection: protons on target produce pions, a sign-selected pion beam of order $$\mu^+ \to e^+ + \nu_e + \bar{\nu}_\mu,\qquad \mu^- \to e^- + \bar{\nu}_e + \nu_\mu .$$9 is transported into a long straight, and pions that decay there produce muons near the ring reference momentum, which are captured without fast kickers (Adey et al., 2013). The baseline stored-muon momentum in the original proposal and expression of interest is $\mu^+$0 with momentum acceptance of approximately $\mu^+$1, while later CERN-oriented implementations and FFAG studies broaden the accepted momentum range to $\mu^+$2 or $\mu^+$3 in specific lattices (Adey et al., 2013).

Two principal ring families were developed. The FODO racetrack design uses large-aperture magnets and compact combined-function arcs; one superconducting-arc version has circumference $\mu^+$4, production straight $\mu^+$5, and non-production straight $\mu^+$6, with tunes $\mu^+$7 before sextupole correction (Liu et al., 2017). In this design, the Orbit Combination Section creates large dispersion for stochastic injection; orbit separation between $\mu^+$8 pions and $\mu^+$9 stored muons is reported at about $\nu_e$0 initially and later around $\nu_e$1 at the injection point (Liu et al., 2017). Genetic-algorithm sextupole optimization improved 100-turn survival from $\nu_e$2 to $\nu_e$3 in the superconducting-arc design (Liu et al., 2017).

The alternative racetrack FFAG designs were motivated by broader momentum bite and near-zero chromaticity. One FFAG racetrack has circumference $\nu_e$4, straight-section length $\nu_e$5, central stored-muon momentum $\nu_e$6, and momentum acceptance $\nu_e$7 (Lagrange et al., 2018). Another triplet-focusing racetrack scaling FFAG uses circumference $\nu_e$8, straight sections $\nu_e$9, tunes $\bar{\nu}_\mu$0, and stored-muon momentum range $\bar{\nu}_\mu$1–$\bar{\nu}_\mu$2, corresponding to $\bar{\nu}_\mu$3 around $\bar{\nu}_\mu$4 (Lagrange et al., 2018). In the FFAG straight cells, the vertical field on the median plane follows

$\bar{\nu}_\mu$5

while the circular scaling-FFAG arcs follow

$\bar{\nu}_\mu$6

which is the mechanism used to preserve tune stability over broad momentum bands (Lagrange et al., 2018).

The production straight is central to the neutrino source. Its large $\bar{\nu}_\mu$7-functions reduce the divergence of the stored muon beam, and the useful decay fraction is set by the straight-to-circumference ratio. In the proposal geometry, $\bar{\nu}_\mu$8 and $\bar{\nu}_\mu$9, giving $\mu^-$0 for the detector-pointing straight (Adey et al., 2013). At $\mu^-$1, the lab-frame muon lifetime is about $\mu^-$2, corresponding to thousands of turns in a ring of order $\mu^-$3 circumference and enabling $\mu^-$4 useful decays in the production straight over the run or per operational year, depending on the study and normalization convention (Ruso et al., 9 May 2025).

Later implementation studies generalized the concept beyond the canonical $\mu^-$5 ring. A CERN-oriented design describes storage of muons with central momentum between $\mu^-$6 and $\mu^-$7 and momentum spread of $\mu^-$8, explicitly to cover the kinematic range of interest to DUNE and Hyper-K (Ruso et al., 9 May 2025). That extension does not replace the earlier reference design; it widens the operational envelope of the stored-muon concept.

4. Detectors and the neutrino-interaction program

The near-detector system is conceived as a high-resolution, magnetized tracking and calorimetry complex located at a baseline of order $\mu^-$9. The proposal literature describes HIRESMNU-like concepts, straw-tube or similar fine-grained trackers, electromagnetic and hadronic calorimetry, and downstream muon spectrometers, with fiducial masses of order $\bar{\nu}_e$0–$\bar{\nu}_e$1 and interchangeable target modules including C, O, Ar, and Fe (Adey et al., 2013). A central design premise is that the same beam simultaneously provides high-statistics $\bar{\nu}_e$2, $\bar{\nu}_e$3, $\bar{\nu}_e$4, and $\bar{\nu}_e$5 samples with comparable spectral control, which is not achievable in conventional horn-focused beams.

The cross-section extraction is written as

$\bar{\nu}_e$6

where $\bar{\nu}_e$7 is the measured yield in an energy bin, $\bar{\nu}_e$8 is the predicted flux from muon decays, $\bar{\nu}_e$9 is the number of scattering centers, and $\nu_\mu$0 is the efficiency including detector response and unfolding (Ruso et al., 2022). The target program spans charged-current and neutral-current channels, quasielastic scattering, resonance production, DIS, coherent processes, multi-nucleon effects, and final-state correlations over the few-hundred-MeV to several-GeV regime (Ruso et al., 2022).

The scientific importance of the near detector is not merely statistical. With $\nu_\mu$1 useful muon decays and near-detector masses $\nu_\mu$2–$\nu_\mu$3, the event samples are so large that key channels become systematically limited; the advantage of nuSTORM is that the flux normalization and flavor composition are known at the percent level, so nuclear-model and detector effects can be isolated by comparing targets and exclusive topologies (Ruso et al., 2022). This is the basis for the often-quoted goal of percent-level cross sections in all four flavors and precise $\nu_\mu$4 ratios for appearance analyses.

Representative event yields were made explicit in later short-baseline analyses. For a $\nu_\mu$5 MIND-like far detector at $\nu_\mu$6, $\nu_\mu$7 POT over 10 years, and $\nu_\mu$8 running, the post-efficiency unoscillated yields used in sensitivity studies are $\nu_\mu$9 for $x \equiv 2E_\nu^\ast/m_\mu$0 CC, $x \equiv 2E_\nu^\ast/m_\mu$1 for $x \equiv 2E_\nu^\ast/m_\mu$2 CC, $x \equiv 2E_\nu^\ast/m_\mu$3 for $x \equiv 2E_\nu^\ast/m_\mu$4 NC, $x \equiv 2E_\nu^\ast/m_\mu$5 for $x \equiv 2E_\nu^\ast/m_\mu$6 NC, and $x \equiv 2E_\nu^\ast/m_\mu$7 for $x \equiv 2E_\nu^\ast/m_\mu$8 CC appearance (Chakraborty et al., 2020). These are oscillation-study numbers rather than general facility rates, but they illustrate the different statistical regimes: disappearance and cross-section channels are high-statistics, while appearance channels are background-sensitive and detector-signature driven.

A common misconception is that nuSTORM removes interaction-model systematics by construction. The facility instead removes the dominant flux ambiguity and provides flavor-pure beams; detector calibration, energy scale, nuclear effects, and unfolding remain central, and the literature repeatedly emphasizes multi-target comparisons, final-state correlation measurements, redundancy in technology, and calibration beams as the mitigation strategy (Ruso et al., 2022).

5. Oscillation physics and beyond-the-Standard-Model searches

The short-baseline oscillation program is optimized around baselines of order $x \equiv 2E_\nu^\ast/m_\mu$9–$\nu_e/\bar{\nu}_e$00 for the near detector and $\nu_e/\bar{\nu}_e$01–$\nu_e/\bar{\nu}_e$02 for the far detector, matching $\nu_e/\bar{\nu}_e$03 to $\nu_e/\bar{\nu}_e$04–$\nu_e/\bar{\nu}_e$05 (Ruso et al., 2022). In the two-flavor approximation used in sensitivity projections,

$\nu_e/\bar{\nu}_e$06

The flagship appearance channel is wrong-sign muon production in a magnetized detector. For stored $\nu_e/\bar{\nu}_e$07, the unoscillated beam contains $\nu_e/\bar{\nu}_e$08 and $\nu_e/\bar{\nu}_e$09, so $\nu_e/\bar{\nu}_e$10 appears as a $\nu_e/\bar{\nu}_e$11 signal over a right-sign $\nu_e/\bar{\nu}_e$12 background (Tunnell, 2012).

The earliest detailed wrong-sign muon studies concluded that, with $\nu_e/\bar{\nu}_e$13, a $\nu_e/\bar{\nu}_e$14 magnetized far detector at $\nu_e/\bar{\nu}_e$15, and useful muon decays in the $\nu_e/\bar{\nu}_e$16–$\nu_e/\bar{\nu}_e$17 range depending on the ring option, nuSTORM could test the LSND/MiniBooNE anomaly region at approximately $\nu_e/\bar{\nu}_e$18 sensitivity in appearance (Tunnell, 2012). A later full sterile-neutrino analysis with GENIE, GEANT4, migration matrices, and multivariate event selection reported, for $\nu_e/\bar{\nu}_e$19 useful decays and a representative point $\nu_e/\bar{\nu}_e$20, a selected appearance sample of 73 signal events over 6 background events (Adey et al., 2014). The same study gave anticipated appearance systematics of $\nu_e/\bar{\nu}_e$21 on signal normalization and $\nu_e/\bar{\nu}_e$22 on background normalization, while showing that the $\nu_e/\bar{\nu}_e$23 conclusion remained robust even under inflated $\nu_e/\bar{\nu}_e$24 and $\nu_e/\bar{\nu}_e$25 assumptions (Adey et al., 2014).

Subsequent work extended the short-baseline program beyond the minimal $\nu_e/\bar{\nu}_e$26 framework. In a 3+1 sterile-neutrino analysis using GLoBES, a $\nu_e/\bar{\nu}_e$27 magnetized iron–scintillator detector, $\nu_e/\bar{\nu}_e$28, Gaussian energy smearing $\nu_e/\bar{\nu}_e$29, and $\nu_e/\bar{\nu}_e$30 bins, combining CC and NC channels was shown to be essential because NC depletion directly constrains active-to-sterile transitions (Chakraborty et al., 2020). That study found sensitivity to $\nu_e/\bar{\nu}_e$31 down to $\nu_e/\bar{\nu}_e$32 at $\nu_e/\bar{\nu}_e$33 CL near $\nu_e/\bar{\nu}_e$34, and bounds in the $\nu_e/\bar{\nu}_e$35–$\nu_e/\bar{\nu}_e$36 plane of roughly $\nu_e/\bar{\nu}_e$37, $\nu_e/\bar{\nu}_e$38 at $\nu_e/\bar{\nu}_e$39 CL for $\nu_e/\bar{\nu}_e$40 and $\nu_e/\bar{\nu}_e$41 (Chakraborty et al., 2020).

The same analysis treated non-unitarity of the mixing matrix through the triangular parametrization

$\nu_e/\bar{\nu}_e$42

and obtained $\nu_e/\bar{\nu}_e$43 sensitivities of $\nu_e/\bar{\nu}_e$44, $\nu_e/\bar{\nu}_e$45 at $\nu_e/\bar{\nu}_e$46, and $\nu_e/\bar{\nu}_e$47 when CC and NC information are combined (Chakraborty et al., 2020). A distinct line of work examined production decoherence in 3+1 and 3+2 sterile models, concluding that source-size decoherence becomes relevant only for $\nu_e/\bar{\nu}_e$48 in the nuSTORM geometry and that short-baseline CP violation in 3+2 can be identified at significance up to $\nu_e/\bar{\nu}_e$49 in favorable regions, particularly when both $\nu_e/\bar{\nu}_e$50 and $\nu_e/\bar{\nu}_e$51 running are used (Ballett et al., 2017).

Later near-detector-only studies widened the BSM scope. Using a $\nu_e/\bar{\nu}_e$52 LArTPC-like detector $\nu_e/\bar{\nu}_e$53 downstream, $\nu_e/\bar{\nu}_e$54 POT, and idealized $\nu_e/\bar{\nu}_e$55 efficiency, one study reported a low-$\nu_e/\bar{\nu}_e$56 weak-mixing-angle sensitivity of

$\nu_e/\bar{\nu}_e$57

from neutrino–electron scattering, and trident yields on argon of 173 coherent plus 29 diffractive $\nu_e/\bar{\nu}_e$58 events, 107 coherent plus 5 diffractive $\nu_e/\bar{\nu}_e$59 events, and 14 coherent plus 9 diffractive $\nu_e/\bar{\nu}_e$60 events (Franklin et al., 10 Sep 2025). The same study treated sterile-neutrino disappearance, large extra dimensions, lepton-flavor violation in pion decay, and heavy axion-like particles in kaon decay. These results extend rather than redefine the core nuSTORM mission: the stored-muon beam is a general precision probe once the flux is known at the percent level.

6. Role in the muon-accelerator roadmap, implementation status, and challenges

nuSTORM occupies a defined position in the broader muon-accelerator roadmap. It was explicitly framed in U.S. and European planning as the first stage of a staged program leading from precision stored-muon neutrino beams to Neutrino Factories and ultimately Muon Colliders (Delahaye et al., 2013). The reason is technical as much as scientific: the facility demonstrates pion production and capture, stochastic injection, large-acceptance muon storage, precision beam monitoring, and detector systems for sign-selected neutrino beams, but does so without requiring the highest-risk ingredients of later facilities, most notably ionization cooling in the baseline configuration (Kaplan, 2013).

This does not mean that cooling is absent from the nuSTORM context. The 2022 status summary explicitly links nuSTORM to a proposed six-dimensional ionization-cooling demonstrator intended to reduce emittance in $\nu_e/\bar{\nu}_e$61 and to build on the experimental validation of ionization cooling by MICE (Ruso et al., 2022). The standard transverse ionization-cooling equation and wedge-based longitudinal manipulation are part of that programmatic extension, although the cooling experiment is conceptually adjacent to, rather than identical with, the core neutrino source.

Implementation studies have been carried out at both Fermilab and CERN. The Fermilab proposal used a $\nu_e/\bar{\nu}_e$62 Main Injector proton driver, $\nu_e/\bar{\nu}_e$63 POT per operational year, a racetrack ring with $\nu_e/\bar{\nu}_e$64 and $\nu_e/\bar{\nu}_e$65, a near detector at $\nu_e/\bar{\nu}_e$66, and a $\nu_e/\bar{\nu}_e$67–$\nu_e/\bar{\nu}_e$68 SuperBIND far detector at $\nu_e/\bar{\nu}_e$69 (Adey et al., 2013). The 2025 CERN implementation summary instead emphasizes existing proton-beam infrastructure at CERN and a storage-ring concept capable of $\nu_e/\bar{\nu}_e$70–$\nu_e/\bar{\nu}_e$71 muons with $\nu_e/\bar{\nu}_e$72 momentum spread, with a stated neutrino-flux precision of $\nu_e/\bar{\nu}_e$73 or better from storage-ring instrumentation (Ruso et al., 9 May 2025). These should be understood as site-dependent realizations of the same stored-muon principle.

The technical risks recur across the design literature. They include high-power target survivability, efficient pion capture in horn or solenoid systems, optimization of stochastic injection and muon capture acceptance, absolute calibration of muon current and polarization monitors, detector energy-scale and acceptance systematics, and control of nuclear-model uncertainties through target comparison and final-state measurements (Ruso et al., 2022). In accelerator terms, the main design trade-offs are between chromatic control and momentum acceptance, between large-aperture conventional optics and more elaborate FFAG solutions, and between simplicity of operation and maximum stored-muon yield (Liu et al., 2017).

A final misconception addressed repeatedly in the literature is that nuSTORM is merely a sterile-neutrino experiment with an accelerator R&D appendix. The proposal history shows the opposite structure: from the LOI onward, the facility was defined by a dual mission—definitive $\nu_e/\bar{\nu}_e$74 and $\nu_e/\bar{\nu}_e$75 cross-section measurements with percent-level precision and decisive short-baseline oscillation tests—while simultaneously constituting the first practical stored-muon accelerator facility (Kyberd et al., 2012). Its encyclopedic significance lies precisely in that conjunction: it is both a physics instrument for the DUNE/Hyper-K era and a bridge from present neutrino beams to future muon-based accelerators.