Research Projects

The apparatus pairs a single-channel NaI(Tl) scintillator on a 9266B PMT with a (Lu_1-xY_x)₂SiO₅ array optically coupled to a Hamamatsu H8500 flat-panel multianode PMT (49×49 mm² active area, 8×8 anodes), read out through a CAEN NIM and VME chain: N417 discriminator, 2255A dual timer defining the integration gate, N405 logic unit for the coincidence, V792 QDC and V775N TDC, bridged to the DAQ by a V1718. Since the QDC integrates only negative currents, a per-channel pedestal calibration was performed with a random trigger from a pulse generator; the response is flat below I_ped ≈ 80 (ADC underflow) and linear above ≈ 90, fixing the operating point at I_ped = 100.

The NaI reference was characterised first: operating at 550 V it gives a linear calibration of 2.52±0.01 ADC/keV, with all five photopeaks of ¹³⁷Cs, ⁶⁰Co and ²²Na individually resolved and a relative resolution falling from σ/E ≈ 3.3% at 511 keV to ≈ 2.1% at 1332 keV, consistent with 1/√E scaling. The LYSO detector, at 980 V, calibrates to 2.6±0.1 ADC/keV with a markedly worse resolution of σ/E ≈ 14.8% at 511 keV, as expected from the lower light yield per channel and the intrinsic background; its gain follows an exponential dependence on HV with b = 0.0053±0.0001 V^-1. Summing a 3×3 pin cluster centred on the pin with the largest charge proved a reliable estimator of the total event energy, and four distinct bands appear in the Q_central–Q_neighbours plane (slopes m ≈ 0.36, 1, 2, 3.8), interpreted as interactions occurring at different positions relative to the pin boundaries.

Three background-suppression strategies were then compared. A hardware back-to-back coincidence between the NaI and LYSO pin 36, refined by an energy window around the NaI 511 keV photopeak, is by far the most effective: it resolves the ²²Na photopeak at 1722±6 ADC with no background subtraction required. A topological filter, accepting only events whose (Q_central, Q_neighbours) pair falls inside a region selected from the background-subtracted 2D correlation, recovers the peak at 1736±11 ADC from the single-trigger spectrum — consistent within 1σ, but the region has to be chosen by eye and the method needs high-statistics background runs. A TDC timing cut leaves the spectrum essentially unchanged even at a deliberately radical 0.1σ threshold, because roughly 80% of events fire only the trigger channel. Crucially, neither software method can separate a genuine source photon from a ¹⁷⁶Lu photon of the same energy; they reject accidental contamination only.

A monoenergetic 100 MeV proton pencil beam was fired into a homogeneous G4_WATER phantom in Geant4 11.4, with the depth–dose profile scored in a user stepping action at 0.25 mm and re-binned to 0.5 mm; a 0.1 mm step limit keeps the tracking steps well below the bin width. Each configuration was run with 5×10⁴ primaries, and the reference configuration was repeated with five independent seeds to establish the run-to-run scatter against which any model differences have to be judged.

The reconstructed Bragg curve places the peak at 76.75 mm and the practical range at R₈₀ = 77.53 mm, about 0.4% above the tabulated PSTAR CSDA value of 77.18 mm — a difference consistent with the definitions of the two quantities. The entrance dose of 0.759 MeV/mm agrees with the PSTAR stopping power to about 4%, and the distal fall-off (80% → 20%) is 1.16 mm, confirming that the electromagnetic energy-loss modelling which fixes the range is sound.

Six reference hadronic physics lists were then compared, spanning the Binary, Bertini and Liège cascade families, together with a variant swapping in the option4 electromagnetic constructor. The range proves essentially insensitive to both choices: R₈₀ spans only 0.03 mm across all lists, comparable to the 0.009 mm statistical scatter measured from the five seeds. The hadronic model instead shows up in the energy escaping the phantom, which ranges from 2.04 MeV per proton for the Bertini-based lists to 2.30 MeV for the Binary Cascade and QBBC — a ~10% gap that is several standard deviations wide and therefore a genuine systematic difference between the models, not noise. On this basis QGSP_BIC is the appropriate default at this energy, with QGSP_BIC_HP preferable when thermal-neutron transport or activation is of interest.

Neutrino short-baseline anomalies refer to puzzling observations in neutrino experiments where neutrinos appear to behave differently than our current understanding predicts. These anomalies have sparked immense interest in the scientific community, leading to investigations that may revolutionize our understanding of neutrino physics.

The anomalies arise from experiments involving neutrinos that travel shorter distances, such as those generated in nuclear reactors or accelerators. When observed over short baselines, unexpected deficit neutrino flux or excess in neutrino interactions, particularly electron antineutrinos, has been noticed. These discrepancies challenge the Standard Model of particle physics, indicating potential new physics beyond our current understanding.

One proposed solution to these anomalies involves the existence of sterile neutrinos. Sterile neutrinos are hypothetical particles that don't interact via the weak nuclear force, making them exceedingly difficult to detect. If these sterile neutrinos have a small mass and mix with the standard neutrinos, they could lead to the observed anomalies. The presence of sterile neutrinos could create a new type of neutrino oscillation, where active neutrinos transform into sterile neutrinos over short distances, causing the unexpected experimental results.