 | (left) Experimental setup as implemented in GEANT4. Scintillation light, produced when \textalpha\ particles deposit energy in LAr, is wavelength-shifted to visible light at the PEN walls. Visible photons reflect on the surfaces until they are either absorbed or detected by the PMT. The \textalpha\ source is moved radially and vertically to different positions for characterizing foil uniformity. (middle) Picture of the hexagonal cage partially lined with PEN and reflector foils. (right) Installation of the cage in the 2-tonne LAr dewar at CERN. |
 | Traces of \textalpha-events, triggering the DAQ (left) or in the post trigger window (middle). The energy of the \textalpha\ event is estimated by the area of pulses in the blue shaded time window. The green-shaded region indicates the prompt time window. (right) SPE spectrum with model fit. |
 | Events, occurring at the trigger time (left) or in the post-trigger window (right), recorded over the span of 1 hour in Fprompt vs.~TotalPE space. Different populations are identified: \textbf{A} - \nuc{Am}{241} \textalpha's. \textbf{B} - \nuc{Am}{241} \textalpha's with pileup. \textbf{C} - CR \textmuon's. \textbf{D} - Unkown population but most likely candidate is protons from inelastic interaction of CR \textmuon's and neutrons. \textbf{E} - Short bursts of light inconsistent with scintillation; several mechanisms create such $\mathcal{O}$(10~PE) pulses, such as Cherenkov light produced in the PMT cathode glass, PMT afterpulsing, and ionizing particles (from CR's or \nuc{Ar}{39} beta decays) interacting in the PEN foil. \fprompt\ for these events is close to 1, and the band stretching toward lower values is due to pile-up. \textbf{A$^{\prime}$} - \nuc{Am}{241} \textalpha's in post-trigger window of \textbf{E}. |
 | Example energy spectra recorded for (left)~\nuc{Am}{241} \textalpha's and (middle)~CR secondary \textmuon's. (right) LAr pulse shape from \textmuon\ events. |
 | (left) Evolution of LY (after correcting for triplet lifetime variation) and the triplet lifetime. Data points in brackets were taken during periods of high noise rates. (right) Measured LY and the MC model expectation for different \textalpha\ source positions. |