How do polyurethane scintillators perform for large-area particle detectors?

A low-dopant PU matrix for future ALICE detectors

The paper asks whether a polyurethane (PU) host matrix, doped with the primary fluor 2,5-diphenyloxazole (PPO) and the secondary fluor 1,4-bis(5-phenyloxazol-2-yl)benzene (POPOP), can serve large-area high-energy-physics detectors. The stated motivation is future hardware: the ALICE 3 Muon Identifier (MID) and the upgrade of the ALICE Fast Interaction Trigger (FIT).

The design choice that shapes every later result is dopant concentration. Typical PU-based scintillator formulations reported in the literature use up to 20–30 wt% of the primary dye, and the paper notes that high dopant concentrations have been linked to faster aging and greater susceptibility to radiation damage. This study therefore deliberately works at low loadings, with PPO screened from 0 to 5 wt% and POPOP from 0.016 to 0.070 wt%, to test whether radiation hardness can be improved without giving up light output or timing.

Three measurements define response, spectrum, and speed

A series of compositions was fabricated and characterized with three complementary probes. Minimum-ionizing particles (MIP) provided the scintillation light-output measurement: charge spectra were collected with a Hamamatsu H3177-51 PMT and a CAEN DT5720A digitizer using a 200 ns integration window, with the extracted MIP charge normalized to sample thickness. Optical spectroscopy supplied transmittance and photoluminescence emission spectra, and time-correlated single-photon counting supplied the time-resolved photoluminescence (TRPL) decays.

Two measurement details matter for reading the numbers. Transmittance was taken with an integrating sphere in double-beam mode, which also collects part of the diffusely scattered light, so the values may differ from a conventional setup. And because the emission band has finite width, the paper reports an emission-weighted effective transmittance, Teff, alongside the transmittance at the emission maximum, T(λem). These two quantities are not interchangeable, and the paper shows they do not track each other monotonically.

PPO sets light output while POPOP shifts the spectrum

The clearest composition trend is in the MIP response. Relative light output rises markedly with PPO concentration across the investigated range, and the paper identifies P50S030 (5 wt% PPO, 0.030 wt% POPOP) as the sample with the largest thickness-normalized MIP response, which then serves as the 100% reference for all other relative light-output values.

POPOP behaves differently. Varying POPOP at fixed PPO produces weaker and non-monotonic changes, and within the P30 to P50 series the highest POPOP loading never corresponds to the highest response. The emission spectra reinforce this division of labor: undoped P0S0 emits at about 294 nm, while doped samples are dominated by POPOP-related emission at roughly 419–423 nm, with a weaker residual PPO feature at about 358–363 nm.

The paper explicitly warns against one tempting shortcut. The PPO-to-POPOP spectral ratio should not be read directly as an energy-transfer efficiency, because the measured spectra can contain contributions from both non-radiative energy transfer and radiative reabsorption. Improved spectral conversion also does not automatically mean a better detected signal: Teff shows no direct correspondence with relative MIP response.

Doping speeds up decay, but composition barely moves it

Timing separates the doped samples sharply from the undoped matrix. Undoped P0S0 has an effective decay time of 4.10 ± 0.01 ns, while every doped sample stays within 1.59–1.75 ns despite the wide range of PPO and POPOP concentrations tested. Introducing the PPO–POPOP fluor system, not tuning it, is what produces the fast response.

Within the doped set the dependence is weak and not strictly monotonic. The clearest variation appears in the P50 series, where τeff rises from 1.62 ns for P50S030 to 1.75 ns for P50S060 before falling to 1.69 ns for P50S070. Because large changes in MIP response come with only modest changes in τeff, the paper concludes that timing adds little additional constraint over this composition space, and that P50S030 offers the most favorable measured combination of light output and fast response among the samples studied. The authors state this is not a global optimum, since PPO above 5 wt% was not investigated.

Meter-scale bars work with one fiber and SiPM readout

The best-performing formulation was scaled into 20 cm and 1 m scintillator bars operated with a single embedded wavelength-shifting fiber and SiPM readout. The detected response falls with propagation distance, but the MIP signal remains distinguishable from the pedestal at 800 mm, which the paper presents as evidence that the material suits meter-scale light collection.

The comparison to other systems needs care. The photoelectron response is of the same order as reported for other wavelength-shifting-fiber plastic-scintillator systems, but absolute values cannot be compared directly because geometry, number of fibers, optical coupling, and readout configuration all differ. The paper's own reading of this is a coupled optimization strategy: fluor composition and light-collection architecture should be tuned together rather than independently.

Proton damage is severe but partly reversible

Proton irradiation at the CERN IRRAD facility produced systematic degradation of both optical and scintillation responses as exposure increased. Immediately after irradiation, light-output retention was 22%, 16%, and 13% for the low-, intermediate-, and high-exposure samples. After approximately two months, retention rose to 69%, 50%, and 34% respectively.

The mechanism the paper infers is not fluor-emission damage alone. Radiation-induced absorption peaks at 405–407 nm, overlapping the POPOP emission region, and the induced absorption coefficient at the 420 nm emission maximum rises from 3.94 to 6.35 and 8.17 cm⁻¹ across the three samples, with the emission-weighted value rising from 2.77 to 4.51 and 5.77 cm⁻¹. The authors therefore attribute the light-output loss to a combination of modified excitation and emission processes plus increased attenuation of POPOP-region photons during optical transport.

Recovery is substantial but incomplete, and the recovered fraction shrinks as exposure grows, which the paper reads as accumulation of persistent radiation-induced degradation alongside a large transient component.

The 10% loss threshold shifts when recovery counts

The paper converts its retention data into an approximate exposure scale for a 10% light-output reduction. Immediately after irradiation, linear interpolation gives about 0.13 Mrad and a local exponential assumption gives about 0.07 Mrad. If the recovered response is used instead, the same estimates rise to roughly 0.32 Mrad and 0.28 Mrad.

These are order-of-magnitude estimates only, and the paper is explicit about why. The immediate response over the measured 1–10 Mrad range is strongly non-linear and cannot be described by a single exponential attenuation law; no measurements were taken between the pristine state and the lowest irradiation level; and the absorbed dose in the scintillator was not independently determined. Dedicated measurements in the tens to a few hundred kRad range are needed to establish the actual onset of transient and persistent degradation.

Where P50S030 fits and where it does not

Within the composition space actually measured, P50S030 is the paper's priority: it gives the highest thickness-normalized MIP response, a 1.62 ns effective decay time, and demonstrated operation in a 1 m bar with a single wavelength-shifting fiber and SiPM readout. The paper's own framing is that this combination of fast timing, efficient MIP detection, scalable bar geometry, and substantial post-irradiation recovery is compatible with the main performance requirements for large-area high-energy-physics detectors such as the ALICE 3 MID.

Two boundaries define how far that conclusion travels. First, PPO above 5 wt% was never tested, so P50S030 is a measured optimum within the studied range rather than a global one. Second, the radiation tolerance relevant to the MID environment has not been quantified, because the low-dose region was not sampled and the absorbed dose was not independently measured. The paper also points to a concrete comparison worth pursuing: polystyrene formulations co-doped with naphthalene and p-terphenyl retained 79–83% of light output immediately after a 10 Mrad dose and recovered to about 92–97%, far above the 13–22% immediate retention and 34–69% recovery seen here. That gap, not the headline recovery figure, is the most useful signal for deciding whether low-dopant PU is the right path for a given detector.

Sources used in this answer