In-Situ Performance of FBK VUV-HD3 and HPK VUV4 SiPMs in the LoLX Liquid Xenon Detector

TL;DR

This paper presents an in-situ comparison of FBK VUV-HD3 and HPK VUV4 SiPMs in the LoLX LXe detector, using external gamma sources to quantify relative PDE. By coupling Geant4–NEST–Chroma simulations with an angular-dependent PDE model that includes surface shadowing, the authors show HPK devices detect only about $0.62$–$0.67$ of the light that FBK devices do in LXe, a deficit larger than vacuum PDE predictions. The study demonstrates that geometry and surface effects, not just intrinsic sensor efficiency, govern effective PDE in LXe, and that accurate detector predictions require full optical transport modeling. These results have direct implications for sensor selection and the design of future large-scale LXe detectors, where absolute PDE calibration and shadowing corrections will be essential. The work also outlines paths toward absolute PDE measurements using internal calibration sources and windowless HPK devices.

Abstract

Silicon Photomultipliers (SiPMs) are a critical technology for the next generation of rare-event search experiments using liquid xenon (LXe). While two VUV-sensitive SiPMs are available, comprehensive in-situ studies are needed to inform detector design and compare device response. This work presents a direct comparison of Fondazione Bruno Kessler (FBK) VUV-HD3 and Hamamatsu (HPK) VUV4 SiPMs operated simultaneously within the Light-only Liquid Xenon (LoLX) detector. Using data collected with gamma sources placed outside the detector, we characterized the relative performance of these photosensors. Our analysis reveals that under these operating conditions, the HPK SiPMs are 33-38% less efficient than the FBK devices, a larger difference than predicted by standard PDE models in vacuum measurement. We show that this discrepancy is resolved by our angular and wavelength dependent PDE model incorporating surface shadowing effects into our optical simulation, which then accurately reproduces the experimental data. This finding has significant implications for the selection and implementation of photosensors in future large-scale LXe detectors.

Figures (6)

• Figure 1: Three-dimensional CAD rendering of the LoLX 2 detector’s photosensor configuration. The 4 cm cubic active volume is filled with liquid xenon and instrumented with three types of photosensors: Hamamatsu VUV4 SiPMs (blue), FBK VUV HD3 SiPMs (yellow), and a centrally located Hamamatsu R8520-406 SEL photomultiplier tube (PMT, red) at the top. The SiPM tiles on the south and west faces are hidden in this figure for visibility.
• Figure 2: An example pulse fitting for saturation correction in SiPM waveforms from gamma-source data. The measured waveform (blue) exhibits saturation at the digitizer's dynamic-range limit (-1V); the fitted reconstruction (orange) extends beyond the saturation threshold to recover the estimated true pulse shape. The saturated portion of the pulse is reconstructed using the fitted function (orange shading), while the unsaturated portions are integrated from the original waveform (green shading).
• Figure 3: 2D histograms of the charge per unit area recorded by HPK VUV4 versus FBK HD3 SiPMs for (a) ${}^{133}\text{Ba}$ and (b) ${}^{137}\text{Cs}$ gamma interactions at an overvoltage of 3V on each SiPM. The color scale represents event counts in each bin. Red lines indicate the ODR linear fit, yielding slopes of $0.62^{+0.03}_{-0.04}$ for ${}^{133}\text{Ba}$ and $0.67^{+0.03}_{-0.05}$ for ${}^{137}\text{Cs}$.
• Figure 4: Flowchart of the Bernoulli trial photon detection scheme for SiPM simulation in Chroma. The decision tree illustrates the sequential probabilistic processes governing photon interactions with SiPMs, including the fill factor $FF$, wavelength and angular dependent transmission through oxide layers into bulk $T(\lambda, \theta)$, specular reflection $R_{Sp}(\lambda, \theta)$, diffuse reflection $R_D$ from inactive regions, and internal photon detection efficiency $iPDE(\lambda, OV)$. Orange text indicates individual trial probabilities at each decision point, while blue text shows the total probability for reaching each final state.
• Figure 5: Angular dependence of optical parameters for the FBK HD3 and HPK VUV4 SiPMs, modeled at the xenon scintillation wavelength ($\lambda\sim175\,$nm) and an over-voltage of $V_{\mathrm{OV}}=3\,$V. The plot shows the transmission, $T(\theta)$ and the resulting PDE as a function of the incident angle, $\theta$, which is measured with respect to the normal to the SiPM surface. The PDE represents the product of the terms in Eq. \ref{['eq:PDE']}.