REVIEW 3 major objections 5 minor 1 cited by
Fluorescence emission of the JUNO liquid scintillator
T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash
Pith's one-line read The fluorescence of the JUNO liquid scintillator is described by four exponential components with a common rise time, and the measured alpha/beta timing difference supports 99.5% alpha rejection at 99% beta acceptance.
desk verdict Credible new timing parameters for JUNO's scintillator, but the unquantified finite-window tail correction is the main thing a referee should push on. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The central object is the four-exponential fluorescence model $F_{\mathrm{fit}}(t) = N \sum_{i=1}^{4} \frac{q_i}{\tau_i - \tau_r}\left(e^{-(t-t_0)/\tau_i} - e^{-(t-t_0)/\tau_r}\right)\Theta(t-t_0)$, where $\tau_r$ is a common rise time, $\tau_i$ and $q_i$ are the decay times and relative weights of the four components, and $\Theta$ is the step function marking the start of the pulse. The model is convolved analytically with the instrument response function (the system's time response to an instantaneous flash), which is itself represented as a sum of seven Gaussians fitted to a 405-nm laser pulse scattered in pure LAB; a finite-window correction accounts for the 1600-ns acquisition truncating the slowest component. This machinery separates the fast nanosecond component from the slow delayed-fluorescence tail and attributes the $\alpha$/$\beta$ difference to the increased weight of the slower components for $\alpha$ excitation.
What would settle it
Measure the instrument response function at several wavelengths inside the emission band (for example 430 nm and 480 nm) and with the full JUNO scintillator in the cuvette instead of pure LAB, then refit the four-exponential model with each IRF; if the fitted $\tau_i$ move by more than the quoted statistical errors (about 1% for the fast components) or the weights $q_3+q_4$ change by more than a few percent, the fixed seven-Gaussian IRF is not representative and the reported parameters are biased.
Extended reading notes
Core claim
On its own terms, the paper establishes that the JUNO liquid scintillator's fluorescence time distribution is well fitted by a sum of four exponentials sharing one rise time, after convolution with an instrument response function modeled as seven Gaussians. For $\alpha$ excitation the decay times are $\tau = (4.06, 17.29, 91.3, 598)\,\mathrm{ns}$ with weights $(54.0, 23.7, 14.1, 8.9)\%$; for $\beta$ excitation they are $\tau = (3.86, 14.52, 81.3, 570)\,\mathrm{ns}$ with weights $(67.8, 20.0, 7.5, 5.0)\%$. The slower components carry a larger weight for alphas, which is the physical handle for pulse-shape discrimination. Using these profiles in a toy Monte Carlo at 1500 photoelectrons (about 1 MeV), a tail-to-total cut achieves a separation merit factor $D = 3.7$, corresponding to 99.5% $\alpha$ rejection at more than 99% $\beta$ acceptance.
Load-bearing premise
The instrument response function, measured with a 405-nanometer laser scattering off a diffuser in pure LAB and held fixed during the fluorescence fit, is assumed to be identical for the real scintillation wavelengths and for the entire 1600-nanosecond window; if the response varies with wavelength or with the full scintillator mixture, the reported decay times and weights shift beyond the quoted errors.
Editorial extensions
If this is right
- JUNO's full Monte Carlo can adopt these $\tau_i$ and $q_i$ as fixed scintillation timing parameters, making simulated events match the measured single-photon time structure.
- An alpha/beta separation merit factor of $D=3.7$ at 1500 photoelectrons means a simple tail-to-total cut already reaches 99.5% alpha rejection at more than 99% beta acceptance for roughly 1 MeV events in the small-sample geometry.
- Because a pseudocumene-based scintillator measured in the same setup separates alpha and beta better, while JUNO collects about three times more photoelectrons per unit energy, the two detectors should end up with comparable alpha/beta discrimination at similar energies.
- The measured emission spectrum peaking at 400 and 420 nm matches the quantum efficiency of JUNO's large PMTs, supporting the light-collection assumptions behind energy-resolution predictions.
- The absorption length is inferred to be at least 18 m at wavelengths above 430 nm, so attenuation at those wavelengths will not dominate light loss in the 20-kton detector.
Reading between the lines
- Because the rise time is a single effective parameter common to all four components, a natural testable extension is to measure the time profile through narrow-band emission filters; if the energy transfer from LAB to PPO to Bis-MSB is wavelength-dependent, the fitted rise time and fast decay time should shift with wavelength, which the paper's Cherenkov crosscheck does not probe.
- Interpolating between the measured alpha and beta parameter sets could yield a continuous $\mathrm{d}E/\mathrm{d}x$-dependent timing parameterization for other particles (protons, heavier ions), which JUNO background simulations would need but the paper does not provide.
- The small-sample PSD estimate omits detector-scale light propagation, absorption, re-emission, and scattering; if those effects smear the time profile, the full-detector alpha rejection at fixed beta acceptance could be lower than 99.5%, with JUNO's roughly threefold higher photoelectron statistics serving as the compensating factor — a trade-off the paper leaves for the full Monte Carlo.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper reports optical characterization of the JUNO liquid scintillator (LAB + 2.5 g/L PPO + 3 mg/L Bis-MSB): emission spectrum from a spectrofluorimeter, absorption length from a spectrophotometer, and fluorescence time profiles measured with a custom TCSPC setup for alpha (244Cm) and beta (60Co) excitation. The timing analysis models the decay as a sum of four exponentials with a common rise time (Eq. 5.1), convolved with an instrument response function measured with a 405 nm pulsed laser and represented by seven Gaussians. The best-fit parameters are given in Table 2, with tau1 about 4 ns and tau4 about 570-600 ns; alpha excitation produces a slower tail than beta. The authors use the measured profiles in a toy Monte Carlo to estimate alpha/beta discrimination via a tail-to-total ratio, claiming 99.5% alpha rejection at 99% beta acceptance for 1 MeV events.
Significance. If correct, Table 2 provides the actual scintillation timing parameters for the JUNO liquid scintillator, which are needed for position/energy reconstruction and alpha/beta discrimination in JUNO. The paper has notable strengths: the IRF is measured and fixed before the fluorescence fit; the four-exponential model is compared with three- and five-exponential alternatives; the Cherenkov emission is used as a crosscheck of the IRF main-peak width; and the PSD estimate is a clearly described toy-MC realization of the measured profiles rather than a circular claim. The main deliverable is a parameter table with sub-nanosecond precision on the fast component, which is valuable for the JUNO Monte Carlo.
major comments (3)
- [Sec. 5.3, Eq. (5.2), Table 2] The tail correction assumes that the fourth exponential is the slowest component and that it continues unchanged beyond t_w = 1600 ns. Since tau4 is about 600 ns, roughly 9% of that component lies outside the window; any additional slower component with tau of order microseconds and sub-percent weight would be almost linear over the window and would be absorbed into tau4 and q4 (and, through the fit, into tau3 and q3). The five-exponential test reported in the same section ('railed to zero') is not a discriminating test for such a component because its in-window contribution is small. The errors in Table 2 are statistical only, so this bias is unquantified, and it propagates directly to the Sec. 5.4 PSD tail-to-total prediction. Please quantify this systematic, e.g., by fitting with a fifth exponential with a prior, by generating pseudo-data with a slow component and refitting with the four-exponential model, or by extending the acquisition window.
- [Sec. 5.1.1, Figs. 7-8, Table 2] The IRF is measured at 405 nm with pure LAB and the IRF parameters are fixed during the fluorescence fits. The Cherenkov crosscheck validates only the FWHM of the main peak; it does not validate the afterpulse structure (e.g., the bump at about 103 ns), which lies in a region where the scintillation profile changes slope and therefore has a strong influence on the fitted tau_i and q_i. The statement that the IRF does not change with wavelength is supported only by a Gaussian FWHM comparison at >530 nm, not by a full-shape comparison. Please either measure the full IRF shape at representative wavelengths or perform a sensitivity study in which the IRF parameters are varied within plausible ranges and the resulting changes in Table 2 are reported as a systematic uncertainty.
- [Sec. 5.3, Eq. (5.2)] The notation in Eq. (5.2) is ambiguous about whether the q_i entering the right-hand side are the fitted (in-window) weights or the final corrected weights. If the q_i (i=1..3) are the final fractions that sum to 1 with q4, then the formula gives tilde q4 = q4/[1-exp(-(t_w-t0)/tau4)], which is larger than q4 and is inconsistent with Table 2. If instead they are the pre-correction fitted values, the text should say so explicitly and give the derivation. Please clarify what quantity is reported in Table 2 and how Eq. (5.2) is used in the fit.
minor comments (5)
- [Sec. 5.1.1] The sentence 'By using a toy Monte-Carlo we also verified that the seven Gaussian model ensures a description of the fluorescence time distributions with an accuracy on the order of ~ %.' contains an incomplete number; it should read '~1%' or give the actual value.
- [Sec. 5.3] The statement that the four-exponential model is 'the best choice' would be easier to evaluate if the chi2 or goodness-of-fit values for the three-, four-, and five-exponential fits were reported.
- [Eq. (5.4)] The convolution is written with N_i and N_j, but these quantities are not defined; please define the normalization factors or use a clearer notation.
- [Table 2] Please state explicitly in the caption that the quoted uncertainties are statistical only, and consider adding a systematic error row or a systematic budget in the text.
- [Sec. 4] The phrase 'Figure 4 shows the error-weighted mean trend with 5σ error bar' is unusual; please clarify whether the error bars are 5σ or whether the text means something else.
Circularity Check
No significant circularity: the fitted fluorescence parameters come from fresh TCSPC data with an independently measured IRF, and the PSD estimate is a Monte Carlo realization of those measured profiles.
full rationale
The paper's central deliverable, Table 2, is obtained by fitting Eq. (5.1) to newly acquired TCSPC data, convolving the four-exponential model with an IRF whose seven Gaussian parameters were fixed from an independent 405-nm laser measurement (Sec. 5.1.1). Sec. 5.3 states explicitly that 'the IRF parameters were fixed during the fitting procedure of the fluorescence curve to the value obtained by analyzing the IRF curve', so the response function is calibrated before the physics fit rather than determined by it. The tau_i and q_i in Table 2 are free parameters of this fit, not inputs. The four-exponential functional form is adopted from prior literature (ref. [20], which shares some authors with the present paper), but the paper independently tests three- and five-exponential alternatives and reports that the fifth component 'railed to zero'; the model choice is therefore not established solely by self-citation. The PSD result in Sec. 5.4 is a toy Monte-Carlo realization 'following the measured profiles of figure 9', i.e., an applied calculation, not a hidden fitted result. Stated limitations — the finite 1600 ns window with about 9.1% of the fourth component uncollected (Sec. 5.3, Eq. 5.2) and the small-sample caveat that light propagation effects could worsen PSD in full JUNO (Sec. 5.4) — are model assumptions and scope disclaimers, not circular reductions. No equation in the paper defines a quantity in terms of the quantity it claims to predict, and no fitted parameter is renamed as an independent prediction. The paper is self-contained as a measurement paper; the central derivation chain is not circular.
Assumptions & free parameters
free parameters (19)
- tau_1 (alpha) =
4.06 +/- 0.01 ns
- tau_2 (alpha) =
17.29 +/- 0.17 ns
- tau_3 (alpha) =
91.3 +/- 1.1 ns
- tau_4 (alpha) =
598 +/- 6 ns
- q_1 (alpha) =
54.04 +/- 0.17 %
- q_2 (alpha) =
23.70 +/- 0.14 %
- q_3 (alpha) =
14.06 +/- 0.09 %
- q_4 (alpha) =
8.92 +/- 0.26 %
- tau_1 (beta) =
3.86 +/- 0.02 ns
- tau_2 (beta) =
14.52 +/- 0.25 ns
- tau_3 (beta) =
81.3 +/- 1.8 ns
- tau_4 (beta) =
570 +/- 10 ns
- q_1 (beta) =
67.79 +/- 0.35 %
- q_2 (beta) =
20.01 +/- 0.27 %
- q_3 (beta) =
7.53 +/- 0.11 %
- q_4 (beta) =
5.03 +/- 0.50 %
- tau_r (common rise time) =
not quoted
- IRF seven-Gaussian parameters =
not quoted
- t_cut =
70 ns
assumptions (7)
- domain assumption Scintillation time profile is a sum of four exponentials with a common rise time.
- domain assumption IRF parameters are fixed from a separate laser measurement and remain valid for all fluorescence wavelengths.
- domain assumption The weakly coupled PMT detects at most one photon per event (single-photon counting condition).
- domain assumption Nitrogen bubbling removes dissolved oxygen, so oxygen quenching is negligible.
- domain assumption Hexane reference behaves optically like the scintillator for absorbance measurements.
- domain assumption Cherenkov light is emitted instantaneously and its contribution in the 350-550 nm fluorescence band is less than 1%.
- domain assumption Absorbance is purely due to absorption in the selected range 0.04 < A < 1.3; scattering and fluorescence are negligible.
Cite this review
Pith. "Pith review of Fluorescence emission of the JUNO liquid scintillator." pith.science (2026). https://pith.science/paper/UINFQATR
@misc{pith2026250109988,
author = {Pith},
title = {Pith review of: Fluorescence emission of the JUNO liquid scintillator},
year = {2026},
howpublished = {\url{https://pith.science/paper/UINFQATR}},
note = {Machine review of arXiv:2501.09988}
}
read the original abstract
JUNO is a huge neutrino detector that will use 20 kton of organic liquid scintillator as its detection medium. The scintillator is a mixture of linear alkyl benzene (LAB), 2.5 g/L of 2,5-diphenyloxazole (PPO) and 3 mg/L of 1,4-Bis(2-methylstyryl)benzene (Bis-MSB). The main goal of JUNO is to determine the Neutrino Mass Ordering [1, 2, 3]. In order to achieve this purpose, good energy and position reconstruction is required, hence a complete understanding of the optical characteristics of the liquid scintillator is mandatory. In this paper we present the measurements on the JUNO scintillator emission spectrum, absorption length and fluorescence time distribution performed respectively with a spectrofluorimeter, a spectrophotometer and a custom made setup
Forward citations
Cited by 1 Pith paper
-
First Demonstration of a Hybrid Cherenkov and Scintillation Detector in a Proof-of-Principle Axion Search at a Beam Dump
First event-by-event Cherenkov separation from sub-MeV electrons in liquid argon enables a proof-of-principle ALP search excluding new parameter space despite no observed excess.
Reference graph
Works this paper leans on
-
[1]
The JUNO collaboration, JUNO physics and detector, Progress in Particle and Nuclear Physics, 123(2022) – 15 –
work page 2022
-
[2]
JUNO sensitivity to 7Be, pep, and CNO solar neutrinos, JCAP, 10(2023)
work page 2023
-
[3]
The JUNO collaboration, Real-time monitoring for the next core-collapse supernova in JUNO, JCAP, 01(2024)
work page 2024
-
[4]
Kumar Jason and Sandick Pearl, Searching for Dark Matter Annihilation to Monoenergetic Neutrinos with Liquid Scintillation Detectors, JCAP, 06(2015)
work page 2015
-
[5]
S. Appel et al. ,Improved Measurement of Solar Neutrinos from the Carbon-Nitrogen-Oxygen Cycle by Borexino and Its Implications for the Standard Solar Model, Phys. Rev. Lett., 129 (25) (2022)
work page 2022
-
[6]
D. Adey et al., Measurement of the Electron Antineutrino Oscillation with 1958 Days of Operation at Daya Bay, Phys. Rev. Lett., 121 (24) (2018)
work page 2018
-
[7]
Q. R. Ahmad et al., Direct evidence for neutrino flavor transformation from neutral current interactions in the Sudbury Neutrino Observatory, Phys. Rev. Lett., 89 (2002)
work page 2002
-
[8]
M. Agostini et al., First Simultaneous Precision Spectroscopy of pp, 7Be, and pep Solar Neutrinos with Borexino Phase-II, Phys. Rev. D 100, (8) (2019)
work page 2019
Show all 28 references
-
[9]
S. N. Ahmed et al., Measurement of the total active B-8 so- lar neutrino flux at the Sudbury Neutrino Observatory with en- hanced neutral current sensitivity, Phys. Rev. Lett., 92 (2004)
2004
-
[10]
Basilico et al., Optimized𝛼/𝛽 pulse shape discrimination in Borexino, Phys
D. Basilico et al., Optimized𝛼/𝛽 pulse shape discrimination in Borexino, Phys. Rev. D, 109 (2024)
2024
-
[11]
Landini, et al., Distillation and gas stripping purification plants for the JUNO liquid scintillator, NIM A, 1069 (2024)
C. Landini, et al., Distillation and gas stripping purification plants for the JUNO liquid scintillator, NIM A, 1069 (2024)
2024
-
[12]
Zhang, et al., Refractive index in the JUNO liquid scintillator, NIM A, 1068 ( 2024)
H. Zhang, et al., Refractive index in the JUNO liquid scintillator, NIM A, 1068 ( 2024)
2024
-
[13]
Yu, et al., Improved measurement of the light attenuation length of a high-quality linear alkylbenzene for the JUNO experiment, Journal of Applied Physics, 134 (24) (2023)
G. Yu, et al., Improved measurement of the light attenuation length of a high-quality linear alkylbenzene for the JUNO experiment, Journal of Applied Physics, 134 (24) (2023)
2023
-
[14]
G. F. Knoll, Radiation detection and measurement, 4th ed., WILEY, 2010
2010
-
[15]
Production of a gadolinium-loaded liquid scintillator for the Daya Bay reactor neutrino experiment, NIM A, 763 (2014)
2014
-
[16]
Park, et al., Production and optical properties of Gd-loaded liquid scintillator for the RENO neutrino detector, NIM A, 707 (2013)
J. Park, et al., Production and optical properties of Gd-loaded liquid scintillator for the RENO neutrino detector, NIM A, 707 (2013)
2013
-
[17]
SNO+ collaboration, The SNO+ experiment, Journal of Instrumentation, 16 (08) (2021)
2021
-
[18]
Barberio, et al., Simulation and background characterisation of the SABRE South experiment, EPJ C, 83 (9) (2023)
E. Barberio, et al., Simulation and background characterisation of the SABRE South experiment, EPJ C, 83 (9) (2023)
2023
-
[19]
Zhang, Z
Y. Zhang, Z. Yu, X. Li, Z. Deng, L. Wen, A complete optical model for liquid-scintillator detectors, NIM A, 967 (2020)
2020
-
[20]
Lombardi, F
P. Lombardi, F. Ortica, G. Ranucci, A. Romani, Decay time and pulse shape discrimination of liquid scintillators based on novel solvents, NIM A, 701 (2013)
2013
-
[21]
Abusleme et al., Prediction of Energy Resolution in the JUNO Experiment, Chinese Physics C 49, (1) (2025)
A. Abusleme et al., Prediction of Energy Resolution in the JUNO Experiment, Chinese Physics C 49, (1) (2025)
2025
-
[22]
dissertation), Ph.D
L.J.Wen, (ph.d. dissertation), Ph.D. thesis (2010)
2010
-
[23]
Yin et al., Precise measurement of attenuation length of the juno liquid scintillator, Radiation Detection Technol- ogy and Methods, 4 (3) (2020)
X.-W. Yin et al., Precise measurement of attenuation length of the juno liquid scintillator, Radiation Detection Technol- ogy and Methods, 4 (3) (2020)
2020
-
[24]
Rompel et al., Palm - precision attenuation length measurement in liquid scintillators (2024), arXiv:2412.15855 – 16 –
V. Rompel et al., Palm - precision attenuation length measurement in liquid scintillators (2024), arXiv:2412.15855 – 16 –
2024 arXiv
-
[25]
Optical attenuation measurements in metal-loaded liquid scintillators with a long-pathlength photometer, NIM A, 637 (1) (2011)
2011
-
[26]
Chen ,et al
H.T. Chen ,et al. Aging of LAB-based liquid scintillator in stainless steel containers, CPC, 39 (2014)
2014
-
[27]
Cichon et al., Scintillation decay-time constants for alpha particles and electrons in liquid xenon, Review of Scientific Instruments, 93 (11) (2022)
D. Cichon et al., Scintillation decay-time constants for alpha particles and electrons in liquid xenon, Review of Scientific Instruments, 93 (11) (2022)
2022
-
[28]
H. T. J. Steiger et al. Development, characterization and production of a novel water- based liquid scintillator based on the surfactant triton x-100 (2024), arXiv:2405.05743 – 17 –
2024 arXiv
Reviewed August 10, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.