Data sorting modes of phoswich detector array

The different data-sorting modes of the phoswich detector array PARIS used for detecting high-energy (4$-$10 MeV) $γ$ rays are investigated. The characteristics including time resolution, energy resolution and detection efficiency under various modes are studied. The present study shows that PARIS has capabilities of rejecting escape and pileup events when used for decay spectroscopy. Notably, the methods presented in this work refer specifically to the $β$-decay experiment of $^{80g+m}$Ga conducted with three PARIS clusters comprising 27 phoswich detectors, rather than to a general report on the PARIS array or its overall performance for in-beam spectroscopy. Compared with the 2"$\times$2"$\times$2" LaBr$_3$(Ce) detector (Ciemała et al., 2009), even in individual mode, PARIS provides significant suppression of single- and double-escape peaks and reduces background via vetoing function of the outer-volume NaI(Tl) crystals. In contrast to the common approach of adding back the energies in LaBr$_3$(Ce) and NaI(Tl) to increase the detection efficiency of the full-energy peak, using NaI(Tl) as a veto shield provides a superior trade-off for applications where spectral purity is essential. Employing add-back analysis within each cluster of nine phoswiches or between all phoswiches could enhance full-energy peak efficiency and further suppress escape peaks and background. Applying a multiplicity condition provides a further suppression but simultaneously lowers the statistics of full-energy peaks.

💡 Research Summary

The manuscript presents a comprehensive experimental evaluation of the data‑sorting strategies and performance characteristics of the PARIS (Photon Array for studies with Radioactive Ion and Stable beams) phoswich detector array when used to detect high‑energy γ‑rays in the 4–10 MeV range. Each detector in the array consists of a 2 inch × 2 inch × 2 inch LaBr₃(Ce) (or CeBr₃) crystal optically coupled to a 2 inch × 2 inch × 6 inch NaI(Tl) crystal, both read out by a single 8‑stage 46 mm Hamamatsu PMT. The authors operated three clusters (27 phoswiches) in a β‑decay experiment of ⁸⁰g+mGa and recorded data with the trigger‑less FASTER digital acquisition system, which provides two charge integrals per event: a short gate (120 ns) capturing the fast LaBr₃ component and a long gate (820 ns) covering the slower NaI(Tl) tail.

By plotting the ratio Q_long/Q_short for each event, the authors identify distinct regions corresponding to (A) full energy deposition in LaBr₃ (ratio ≈ 1), (D) full deposition in NaI(Tl) (ratio > 2), (E/F) partial energy sharing between the two crystals (ratio between 1 and 2), and (G) pile‑up events where more than one γ‑ray arrives within the 820 ns window (ratio 9–12). This separation enables a pulse‑shape analysis (PSA) that distinguishes genuine LaBr₃ signals from NaI(Tl) signals, and also isolates pile‑up and background contributions.

The authors calibrate the short‑gate and long‑gate charges separately using standard γ‑ray sources, converting them into energies E_LaBr₃ and E_NaI(Tl). Spectra can then be built in several ways: (i) LaBr₃‑only (E_NaI = 0), (ii) NaI(Tl)‑only (E_LaBr₃ = 0), (iii) internal add‑back (E_LaBr₃ + E_NaI) within each detector, and (iv) add‑back across detectors. While add‑back increases full‑energy peak efficiency, it degrades energy resolution because events with incomplete energy deposition in NaI(Tl) broaden the peak.

Time alignment is performed using a digital constant‑fraction discriminator (CFD) on the FASTER system. The authors align each detector’s LaBr₃ and NaI(Tl) channels separately, using the time difference between a β‑particle detected in a surrounding plastic scintillator and the γ‑ray in the phoswich. The measured LaBr₃ timing resolution (FWHM) for the 659.2 keV and 1083.6 keV cascade γ‑rays is 437 ± 49 ps, corresponding to σ ≈ 185 ps, which is close to the design goal of 250 ps at 1 MeV. NaI(Tl) timing is about 700 ps, and the HPGe detectors in the same setup exhibit the expected slower response. The good timing performance enables fast‑timing lifetime measurements and time‑of‑flight neutron/γ discrimination.

Four data‑sorting modes are investigated:

1. Individual mode – each phoswich is treated as an independent detector; NaI(Tl) signals are used as an anti‑coincidence (veto) to suppress escape and pile‑up events.

2. Cluster add‑back – energies from the nine detectors within a single cluster are summed, improving full‑energy peak efficiency by ~30–50 % while retaining some escape‑peak suppression.

3. Global add‑back – energies from all 27 detectors are summed, maximizing detection efficiency (up to a factor of ~2) but allowing escape peaks and background to re‑appear.

4. TAS‑like (Total Absorption Spectroscopy) mode – the entire array is treated as a single calorimeter, achieving the highest total absorption efficiency at the cost of substantial resolution loss.

The authors quantify the trade‑offs. Using NaI(Tl) as a veto in Individual mode reduces single‑ and double‑escape peaks by more than 70 % and lowers the continuum background by a factor of three compared with a simple add‑back approach. Cluster add‑back raises the full‑energy peak efficiency by ~1.3–1.5× while still keeping escape‑peak suppression superior to a pure add‑back of LaBr₃ alone. Global add‑back further raises efficiency but re‑introduces escape structures, making it suitable only when absolute efficiency is the primary concern.

Applying a multiplicity condition (requiring coincidences among multiple detectors) provides additional background suppression, yet it reduces the statistics of the full‑energy peaks by roughly 30 % because many events are discarded. The authors recommend a flexible analysis pipeline: for experiments where spectral purity is paramount (e.g., precise branching‑ratio or strength measurements), the Individual mode with NaI(Tl) veto is optimal; for studies where high efficiency is needed (e.g., weak β‑delayed γ‑ray branches), cluster add‑back offers a balanced solution; and for very low‑intensity beams where every count matters, global add‑back may be justified despite the increased background.

The paper also emphasizes that the presented results are specific to the ⁸⁰g+mGa β‑decay experiment with three clusters; they are not a general performance benchmark for the entire PARIS array in in‑beam spectroscopy. Nevertheless, the demonstrated capability of the NaI(Tl) outer layer to act as an effective veto distinguishes PARIS from conventional 2 inch × 2 inch × 2 inch LaBr₃(Ce) detectors, which lack such built‑in background suppression.

In conclusion, the study validates that the PARIS phoswich array provides excellent timing (≈ 440 ps), good energy resolution, and, most importantly, a powerful mechanism for rejecting escape and pile‑up events through NaI(Tl) anti‑coincidence. By judiciously selecting data‑sorting modes and, if needed, imposing multiplicity cuts, users can tailor the balance between detection efficiency and spectral purity to the specific demands of high‑energy γ‑ray spectroscopy, β‑decay studies, and nuclear structure investigations.