arxiv: 2604.00176 · v1 · submitted 2026-03-31 · ⚛️ physics.ins-det · hep-ex

Recognition: unknown

Design and performance of a large-area scintillator-based chamber for the MID subsystem of ALICE 3

Ruben Alfaro Molina , Juan Carlos Cabanillas Noris , Edmundo Garc\'ia Solis , Laura Helena Gonz\'alez Trueba , Varlen Grabski , Gerardo Herrera Corral , Jes\'us Eduardo Mu\~noz M\'endez , Ildefonso Le\'on Monz\'on

show 9 more authors

Antonio Ortiz Antonio Paz Ian P\'erez Garc\'ia Ricardo Rodr\'iguez Pineda Solangel Rojas Torres Guillermo Tejeda Mu\~noz Paola Vargas Torres Victor V\'azquez Campos Yael Antonio V\'asquez Beltran

Authors on Pith no claims yet

Pith reviewed 2026-05-14 20:56 UTC · model grok-4.3

classification ⚛️ physics.ins-det hep-ex

keywords scintillator chambermuon identifierALICE 3machine learningbeam testsilicon photomultiplierpion rejectionabsorber length

0 comments

The pith

A two-layer scintillator chamber for ALICE 3 muon identification achieves over 99 percent muon efficiency using machine learning on beam data.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper presents the design and beam-test performance of a two-layer scintillator chamber intended for the muon identifier in the ALICE 3 upgrade. Each layer contains 24 bars read out by silicon photomultipliers, arranged orthogonally to create 4 by 4 cm cells. A machine learning algorithm trained on half the muon and pion beam data achieves over 99 percent muon detection efficiency when requiring a signal in at least one layer. The same algorithm applied to pion data shows that the fake-muon rate falls exponentially with increasing iron absorber length, characterized by a slope of 18.79 cm. These results indicate that the chamber meets key requirements for reliable muon identification in future heavy-ion collision experiments.

Core claim

The central discovery is that the scintillator-based chamber, consisting of two orthogonal layers of wavelength-shifting fiber readout bars, delivers a muon identification efficiency exceeding 99 percent using a machine learning classifier on 3 GeV/c beam data, while the misidentification rate for pions is described by an exponential function of absorber thickness with a slope parameter of 18.79 cm.

What carries the argument

Machine learning classifier trained on 50 percent of muon signal and pion background beam data to separate particles based on detection patterns across the two orthogonal scintillator layers.

If this is right

The chamber design supports scaling to the full MID subsystem for the ALICE 3 upgrade.
Muon efficiency remains high in the OR detection mode across tested absorber lengths.
Fake-muon rates can be tuned by selecting iron absorber thickness according to the measured exponential dependence.
The results validate the mechanical structure and readout for use in high-rate heavy-ion environments.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The exponential slope parameter may correspond to the effective pion interaction length in iron and could be used to predict performance with alternative absorber materials.
This ML approach for layer-based coincidence could be adapted to improve pion rejection in other scintillator detectors at accelerator facilities.
Integration with tracking detectors in ALICE 3 might further reduce backgrounds in physics measurements beyond what the MID alone provides.
Validation at beam energies closer to those in ALICE 3 collisions would test whether the quoted efficiencies hold under realistic conditions.

Load-bearing premise

The 50 percent split of the beam data is representative of the particle rates and backgrounds expected in the final ALICE 3 environment without significant overfitting.

What would settle it

Repeating the test with full dataset statistics or higher-energy beams and finding muon efficiency below 99 percent or a fake-muon exponential slope deviating from 18.79 cm would falsify the performance claims.

Figures

Figures reproduced from arXiv: 2604.00176 by Antonio Ortiz, Antonio Paz, Edmundo Garc\'ia Solis, Gerardo Herrera Corral, Guillermo Tejeda Mu\~noz, Ian P\'erez Garc\'ia, Ildefonso Le\'on Monz\'on, Jes\'us Eduardo Mu\~noz M\'endez, Juan Carlos Cabanillas Noris, Laura Helena Gonz\'alez Trueba, Paola Vargas Torres, Ricardo Rodr\'iguez Pineda, Ruben Alfaro Molina, Solangel Rojas Torres, Varlen Grabski, Victor V\'azquez Campos, Yael Antonio V\'asquez Beltran.

**Figure 2.** Figure 2: (Left) Assembly steps for the bars from left to right: from a bar without a fiber to the bar already wrapped with the [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** Figure 3: Mechanical structure for the 1 × 1m2 prototype, where N2 and N4 are the scintillator layers separated by N3 by 1 cm. 3 [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗

**Figure 4.** Figure 4: Complete design for the mechanical structure, showing how each bar is arranged in its place, with the SiPM protection [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗

**Figure 5.** Figure 5: Simulation of the mechanical stress for the top and bottom aluminum cover (left), and for the intermediate aluminum [PITH_FULL_IMAGE:figures/full_fig_p004_5.png] view at source ↗

**Figure 6.** Figure 6: The trigger signals were processed with a NIM leading-edge discriminator (CAEN N840). The output [PITH_FULL_IMAGE:figures/full_fig_p004_6.png] view at source ↗

**Figure 6.** Figure 6: Experimental setup. The trigger scintillators are labeled as Trigger 1, ..., Trigger 5. The beam travels from right to left [PITH_FULL_IMAGE:figures/full_fig_p005_6.png] view at source ↗

**Figure 7.** Figure 7: Input variables used for training: hit multiplicity (upper-left plot), fired channel (upper-right plot), charge (bottom-left [PITH_FULL_IMAGE:figures/full_fig_p006_7.png] view at source ↗

**Figure 8.** Figure 8: Geometrical acceptance times efficiency as a function of the BDT output. Results for all the absorber lengths are shown. [PITH_FULL_IMAGE:figures/full_fig_p007_8.png] view at source ↗

**Figure 9.** Figure 9: Hit-position distribution measured for the 3 GeV/ [PITH_FULL_IMAGE:figures/full_fig_p007_9.png] view at source ↗

**Figure 10.** Figure 10: Muon-candidate efficiency as a function of absorber length (left). Results for the pion-enriched (red round markers) and [PITH_FULL_IMAGE:figures/full_fig_p008_10.png] view at source ↗

**Figure 11.** Figure 11: Data-driven background efficiency as a function of absorber length. The 2025-test-beam data are compared with the [PITH_FULL_IMAGE:figures/full_fig_p009_11.png] view at source ↗

read the original abstract

This paper reports on the design and construction of a chamber for the muon identifier detector (MID) of the ALICE 3 upgrade project. The chamber consists of two sensitive layers separated by a 1 cm air gap. Each layer holds 24 scintillator bars ($1\times4\times100$ cm$^3$) manufactured by FNAL-NICADD. The bars are equipped with Kuraray wavelength shifting fibers and the readout is provided by a silicon photomultiplier from Hamamatsu. The bars in the second layer are orthogonal to the bars in the first layer, thus providing an overlapping cell size of 4$\times$4 cm$^{2}$. The bar assembly as well as the design of the mechanical structure is described. The design of the chamber is close to that considered in the ALICE 3 letter of intent. The chamber was tested at the CERN T10 beamline using 3 GeV/$c$ pion-enriched and muon beams. The chamber was placed behind an iron absorber, with different absorber lengths considered in the test. The muon identification is performed using a Machine Learning algorithm, which was trained and tested using muon (signal) and pion (background) data (50% of the available statistics). The trained ML algorithm was applied to muon data, yielding a muon efficiency above 99% for the OR condition (detection in either layer 1 or 2). The implementation in the pion-beam data gives the fake-muon efficiency as a function of the absorber length that is well described by an exponential function with a slope parameter of 18.79 cm. The next steps towards finalizing the optimization are outlined.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This delivers the first beam-test numbers for the ALICE 3 MID scintillator prototype, with solid mechanical details and data-backed efficiencies, but the ML classifier's 50% split leaves its robustness to real conditions unproven.

read the letter

The paper's main contribution is the beam-test validation of a two-layer scintillator chamber built to the ALICE 3 letter-of-intent geometry. The authors describe the assembly of 24 bars per layer, the orthogonal orientation giving 4x4 cm cells, the Kuraray fibers, Hamamatsu SiPM readout, and the mechanical frame. They then report results from the CERN T10 line using 3 GeV/c muon and pion beams behind iron absorbers of varying thickness. The ML tagger, trained on half the data, reaches above 99% muon efficiency in the OR of the two layers, while the pion fake rate falls exponentially with absorber length and is fitted to a slope of 18.79 cm. These are the first published performance figures for this exact layout, and the numbers come straight from the recorded events rather than from simulation alone. The construction section is clear and practical, which is the real value for anyone who might build similar detectors. The soft spot is the ML part. A single 50% train-test split on monoenergetic beam data does not test whether the same features and thresholds will hold under the broader momentum spectrum, higher occupancy, and different background mix expected in ALICE 3. No uncertainties are quoted on the slope, and there is no cross-validation or discussion of possible beam-specific artifacts. That gap is real but not fatal for an engineering note. This work is aimed at the ALICE 3 instrumentation group and others developing scintillator muon identifiers. It supplies concrete test-beam benchmarks that the collaboration can use, even if further studies will be needed before final design. I would send it to peer review. The data are independent of the claims, the design choices are documented, and the results are useful for the upgrade even with the limited ML validation.

Referee Report

1 major / 1 minor

Summary. The paper describes the design, construction, and beam-test performance of a scintillator-based chamber prototype for the muon identifier (MID) subsystem of ALICE 3. The chamber comprises two orthogonal layers of 24 scintillator bars (1×4×100 cm³) read out by Kuraray WLS fibers and Hamamatsu SiPMs, yielding 4×4 cm² cells. Tested at CERN T10 with 3 GeV/c pion and muon beams behind variable-length iron absorbers, a machine-learning classifier trained on 50% of the data achieves >99% muon efficiency for the OR of the two layers; the fake-muon efficiency in pion data is reported to follow an exponential dependence on absorber length with slope 18.79 cm.

Significance. If the reported efficiencies hold, the work supplies direct experimental validation of a scintillator technology close to the ALICE 3 LoI baseline, demonstrating high muon efficiency and an exponentially suppressible fake rate. The beam-test data and ML implementation constitute concrete, reproducible performance benchmarks that strengthen the case for this detector concept in the final ALICE 3 environment.

major comments (1)

[Machine-learning performance evaluation] Machine-learning section: the muon efficiency (>99% for OR) and the exponential fit to fake-muon efficiency (slope 18.79 cm) are extracted from a single 50% train/test split of the 3 GeV/c beam sample. The manuscript must specify the algorithm, input features, hyper-parameter choices, and any cross-validation or uncertainty estimation performed; without these, the robustness of the quoted numbers against statistical fluctuations or beam-specific artifacts cannot be assessed.

minor comments (1)

[Abstract] The abstract states the slope parameter 18.79 cm but omits fit quality metrics or uncertainties; these should be reported in the main text and referenced from the abstract.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the positive assessment of our manuscript and the recommendation for minor revision. We address the single major comment below and will incorporate the requested details in the revised version.

read point-by-point responses

Referee: [Machine-learning performance evaluation] Machine-learning section: the muon efficiency (>99% for OR) and the exponential fit to fake-muon efficiency (slope 18.79 cm) are extracted from a single 50% train/test split of the 3 GeV/c beam sample. The manuscript must specify the algorithm, input features, hyper-parameter choices, and any cross-validation or uncertainty estimation performed; without these, the robustness of the quoted numbers against statistical fluctuations or beam-specific artifacts cannot be assessed.

Authors: We agree that additional technical details on the machine-learning implementation are required for full reproducibility and to allow readers to evaluate robustness. In the revised manuscript we will add a dedicated paragraph (or subsection) specifying the algorithm, the complete list of input features, the hyper-parameter selection procedure, and the cross-validation/uncertainty estimation method used. These clarifications will be inserted without changing the quoted performance figures. revision: yes

Circularity Check

0 steps flagged

No circularity: efficiencies extracted from held-out beam-test data with no self-referential reduction

full rationale

The paper's central results are muon efficiency (>99% for OR) and fake-muon efficiency (exponential slope 18.79 cm) obtained by training an ML classifier on 50% of the 3 GeV/c pion/muon beam data and evaluating on the remaining 50%. These are direct empirical measurements on independent test samples; no equations or derivations reduce the quoted numbers to parameters fitted from the same outputs. No self-citations, uniqueness theorems, or ansatzes are invoked to close the argument. The chain is self-contained against external beam-test benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is a pure experimental instrumentation paper. No theoretical free parameters, axioms, or invented entities are introduced; performance numbers rest on measured light yields, standard photomultiplier response, and standard ML classification on labeled beam data.

pith-pipeline@v0.9.0 · 5716 in / 1340 out tokens · 29662 ms · 2026-05-14T20:56:06.747350+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

10 extracted references · 9 canonical work pages · 2 internal anchors

[1]

Acharya, et al., The ALICE experiment: a journey through QCD, Eur

S. Acharya, et al., The ALICE experiment: a journey through QCD, Eur. Phys. J. C 84 (8) (2024) 813. arXiv:2211.04384,doi:10.1140/epjc/s10052-024-12935-y

work page doi:10.1140/epjc/s10052-024-12935-y 2024
[2]

R. Bala, I. Bautista, J. Bielcikova, A. Ortiz, Heavy-ion physics at the LHC: Review of Run I results, Int. J. Mod. Phys. E 25 (07) (2016) 1642006.arXiv:1605.03939,doi:10.1142/S0218301316420064

work page internal anchor Pith review Pith/arXiv arXiv doi:10.1142/s0218301316420064 2016
[3]

Letter of intent for ALICE 3: A next-generation heavy-ion experiment at the LHC, CERN-LHCC-2022-009, LHCC-I-038 (11 2022).arXiv:2211.02491

work page arXiv 2022
[4]

Dainese, A

A. Dainese, A. Di Mauro, Scoping document for the ALICE 3 detector, Tech. rep., CERN, Geneva (2025). URLhttps://cds.cern.ch/record/2925455

work page arXiv 2025
[5]

J. E. Muñoz Méndez, et al., ML-based muon identification using a FNAL-NICADD scintillator chamber for the MID subsystem of ALICE 3, JINST 20 (09) (2025) P09015.arXiv:2507.02817,doi:10.1088/ 1748-0221/20/09/P09015. 9

work page arXiv 2025
[6]

O. A. Grachov, et al., Study of new FNAL-NICADD extruded scintillator as active media of large EMCal of ALICE at LHC, in: 11th International Conference on Calorimetry in High-Energy Physics (Calor 2004), 2004, pp. 49–54.arXiv:physics/0405028,doi:10.1142/9789812701978_0007

work page internal anchor Pith review Pith/arXiv arXiv doi:10.1142/9789812701978_0007 2004
[7]

R. Alfaro, et al., Characterisation of plastic scintillator paddles and lightweight MWPCs for the MID subsystem of ALICE 3, JINST 19 (04) (2024) T04006.arXiv:2401.04630,doi:10.1088/1748-0221/19/ 04/T04006

work page doi:10.1088/1748-0221/19/ 2024
[8]

Pérez Rodríguez, Mecánica del MID (Identificador de Muones), Proyecto ALICE3 CERN, Tech

J. Pérez Rodríguez, Mecánica del MID (Identificador de Muones), Proyecto ALICE3 CERN, Tech. rep., Instituto de Física, UNAM (2025)

2025
[9]

van Dijk, et al., Particle production and identification for the T10 secondary beamline of the CERN East Area, Nucl

M. van Dijk, et al., Particle production and identification for the T10 secondary beamline of the CERN East Area, Nucl. Instrum. Meth. B 569 (2025) 165907.arXiv:2507.02567,doi:10.1016/j.nimb.2025.165907

work page doi:10.1016/j.nimb.2025.165907 2025
[10]

Hocker, et al., TMVA - Toolkit for Multivariate Data Analysis with ROOT: Users guide

A. Hocker, et al., TMVA - Toolkit for Multivariate Data Analysis with ROOT: Users guide. TMVA - Toolkit for Multivariate Data Analysis, Tech. rep., CERN, Geneva, tMVA-v4 Users Guide: 135 pages, 19 figures, numerous code examples and references (2007). URLhttps://cds.cern.ch/record/1019880 10

work page arXiv 2007