GeSparK Detector: Coincidence Spectroscopy

• GeSparK is a low-background detector that couples an HPGe spectrometer with a liquid scintillator for time-coincidence alpha-beta/gamma detection.
• It employs a six-panel muon veto to suppress cosmic muon backgrounds, achieving 92.69% reduction in the 20–3000 keV range and near 99% efficiency for high-energy events.
• The optimized design enhances sensitivity through improved minimum detectable activities using LS pulse-shape discrimination and tailored passive shielding.

Searching arXiv for the specified paper and closely related GeSparK references. arXiv search query: (Barresi et al., 17 Feb 2025) GeSparK detector muon veto alpha-beta/gamma coincidence detector GeSparK is a composite low-background detector that couples a high-purity germanium spectrometer with a liquid scintillator cell operated in time coincidence, so that an alpha or beta interaction in the liquid scintillator is selected only when it is temporally correlated with a gamma deposition in the HPGe detector. In the 2025 implementation study, the detector’s defining instrumental development is a six-panel active muon veto designed to suppress the dominant residual background from cosmic muons and muon-induced showers in the passive lead and copper shielding. The reported result is a background reduction of 92.69(3)% in the 20–3000 keV range for LS–HPGe coincidences, with 98.87(1)% reduction for HPGe-saturated events above 3000 keV, which are almost entirely muon-induced (Barresi et al., 17 Feb 2025).

1. Detector concept and coincidence principle

GeSparK is configured around an HPGe–LS coincidence architecture rather than a standalone germanium spectrometer. The term “alpha-beta/gamma coincidence” denotes the selection of events in which an alpha or beta interaction in the LS is temporally correlated with a gamma deposition in the HPGe. This permits tagging of decay signatures with characteristic time correlations, including alpha–gamma and beta–gamma cascades, and strongly suppresses environmental and cosmogenic backgrounds that are not time-correlated (Barresi et al., 17 Feb 2025).

The HPGe element is an Ortec coaxial p-type detector with 38% relative efficiency and 1.8 keV FWHM at 1332 keV. Its end-cap has a thin carbon window to enhance low-energy photon transmission. The LS is hosted in a cylindrical PTFE container with 2 mm bottom thickness, capped by PMMA and coupled to a low-background 3-inch Hamamatsu PMT. Up to 200 mL of sample is mixed 1:1 with Ultima Gold AB. The passive shielding is asymmetric by subsystem: at the bottom and around the HPGe and LS there are 10 cm of copper surrounded by 15 cm of lead; around the PMT the shield is 5 mm Cu plus 10 cm Pb; the top has 5 cm Cu plus 10 cm Pb.

Subsystem	Specification
HPGe	Ortec coaxial p-type; 38% relative efficiency; 1.8 keV FWHM at 1332 keV
LS cell	Cylindrical PTFE container, 2 mm bottom; PMMA cap; low-background 3-inch Hamamatsu PMT
Sample loading	Up to 200 mL mixed 1:1 with Ultima Gold AB
Passive shielding	10 cm Cu + 15 cm Pb around HPGe/LS; 5 mm Cu + 10 cm Pb around PMT; 5 cm Cu + 10 cm Pb above

This configuration is designed for coincidence spectroscopy rather than for conventional singles counting. A plausible implication is that GeSparK’s instrumental sensitivity depends as much on rejection of time-uncorrelated backgrounds as on intrinsic full-energy-peak efficiency.

2. Background regime and the role of cosmic muons

Despite the coincidence technique, the dominant residual background before installation of the final veto is attributed to cosmic muons. In GeSparK, muons and their showers in the Pb/Cu shielding produce broad, featureless energy depositions and secondary gamma/e± backgrounds that can mimic radioactivity. The study therefore treats the shield not only as a suppressor of external gamma fluxes but also as the principal site of muon-induced secondary production (Barresi et al., 17 Feb 2025).

The logic of the veto analysis is expressed through a small set of relationships. For counts $N_{\mathrm{before}}$ and $N_{\mathrm{after}}$ in a defined energy window, the background reduction fraction is

$$R = 1 - \frac{N_{\mathrm{after}}}{N_{\mathrm{before}}}$$

or, equivalently,

$$R = \frac{N_{\mathrm{before}} - N_{\mathrm{after}}}{N_{\mathrm{before}}}\times 100\%.$$

For muons, the veto efficiency is

$$\epsilon_{\mathrm{veto}} = \frac{N_{\mathrm{tagged\;muons}}}{N_{\mathrm{total\;muons}}}.$$

The study also states the accidental veto rate estimate for uncorrelated rates $R_{\mathrm{scint}}$ and $R_{\mathrm{HPGe}}$ with coincidence gate $\Delta t$,

$$r_{\mathrm{acc}} \approx 2\,R_{\mathrm{scint}}\,R_{\mathrm{HPGe}}\,\Delta t,$$

and the live-time correction for veto dead time,

$L = T_{\mathrm{run}} - \sum_i t_{\mathrm{veto},i}.$

No measured accidental rates or dead-time numbers are reported. Instead, the detector geometry is optimized so that four veto panels are placed inside the shielding, reducing the scintillator event rate due to external radioactivity and thereby suppressing accidental vetoes and minimizing dead time. This is important for interpreting the final result: the veto is designed not only for high tagging efficiency, but also for low parasitic trigger rate.

3. Muon-veto geometry and optimization pathway

The final muon-veto system consists of six plastic scintillator detectors positioned around the HPGe and LS detectors. Two detectors are placed above the detector, and four are installed inside the passive shielding between the inner copper and outer lead layers, surrounding the sides of the HPGe/LS region. The study identifies two explicit advantages of this internal placement: it reduces the required panel area because the external lateral Pb surfaces are otherwise too large, and it suppresses the trigger rate from external radioactivity impinging on externally mounted panels. This “inside-Pb/outside-Cu” topology is described as providing near-4$N_{\mathrm{after}}$0 coverage of trajectories traversing the sensitive region and the Pb/Cu shields (Barresi et al., 17 Feb 2025).

The development path proceeded through earlier and intermediate configurations. The original GeSparK had two top veto panels of 80×40×5 cm, producing a 53.2(3)% reduction of the total background rate. A later test configuration used two large top panels of 1000×500×50 mm and four external side panels of 300×420×50 mm to benchmark simulation against data. In that benchmark, measured tagging efficiencies were 53.2(3)% for “Top,” 36.3(5)% for “2 Side,” 70.7(3)% for “Top+2 Side,” and 81.7(1)% for “Top+4 Side,” consistent with Monte Carlo expectations using 6–10 GeV angular inputs.

The final six-panel geometry was simulated to yield a potential muon tagging efficiency of 99.4(1)%. The passive shielding geometry was unchanged except for added lead layers on top to further suppress external flux. The paper illustrates the placements through CAD-based models and installation photographs, but it does not specify the scintillator material or the optical readout hardware of the veto panels. That omission is a genuine instrumentation limitation in the published implementation record.

4. Readout, coincidence logic, and simulation framework

GeSparK uses coincidence and anti-coincidence at the DAQ level. The LS PMT signal is split and digitized into two dynamic ranges with an NI PXI-5153 oscilloscope at 1 GHz and 8 bits, enabling low-threshold alpha/beta detection while preserving headroom for high-energy pulses. The LS channel provides the acquisition trigger, and HPGe signals are recorded in coincidence. Muon-veto signals are then applied in anti-coincidence to tag or reject events in the LS–HPGe coincidence stream. The anti-coincidence selection is described conceptually, but the paper does not report the specific thresholds, time windows, or veto gate widths used in hardware or firmware (Barresi et al., 17 Feb 2025).

Muon-flux characterization was carried out in a laboratory located at the third underground floor. The angular distribution was measured with a two-panel muon telescope consisting of two 300×420×50 mm scintillators placed 30 cm apart on a rotatable frame and scanned along planes parallel and perpendicular to the building. The data follow $N_{\mathrm{after}}$1 with small deviations attributed to asymmetric absorption. Because the telescope integrates over an acceptance window, the measured intensity was modeled as

$N_{\mathrm{after}}$2

with Monte Carlo studies giving $N_{\mathrm{after}}$3 or $N_{\mathrm{after}}$4.

The simulation framework is Geant4-based “Genesis.” It models the full detector geometry, with the HPGe crystal and internals implemented via text geometry and the Pb/Cu shields, LN dewar/cold finger, and scintillator support frames implemented via CAD. Primaries were generated on a hemispherical surface of radius 1.5 m around the detector. Muon energies from 1–200 GeV were used, with sea-level distributions and angular PDFs including $N_{\mathrm{after}}$5 and energy-dependent angular distributions from literature. Offline veto selection in Monte Carlo required panel coincidences according to the studied configuration, mirroring the experimental logic. The simulated LS–HPGe coincidence background spectrum without veto reproduced the measured muon energy deposition in the HPGe, including the absence of spectral lines and the prominent 511 keV annihilation feature, with only minor discrepancies in the very low-energy region.

5. Measured performance and sensitivity gains

After installation of the final veto, a dedicated background measurement with no LS sample yielded a background reduction of 92.69(3)% in the 20–3000 keV energy range for LS–HPGe coincidences with the muon veto applied. For HPGe-saturated events above 3000 keV, which are almost entirely muon-induced, the measured reduction was 98.87(1)%, in excellent agreement with the 99.4(1)% Monte Carlo prediction within modeling and angular-distribution uncertainties (Barresi et al., 17 Feb 2025).

Several rate changes quantify the practical impact of the new configuration. In the 40–2700 keV interval, the total LS–HPGe coincidence rate increased from 47.63(0.13) mHz in the previous configuration to 75.6(0.1) mHz because the dual-range acquisition enabled a lower LS threshold; once the veto was applied, the rate fell to 5.58(0.03) mHz, corresponding to a 4.4× reduction relative to the earlier top-only veto. The 511 keV annihilation peak in coincidence decreased from 0.58(0.01) mHz to 0.025(0.002) mHz, an 8.6× improvement. Rates in regions of interest around prominent lines also fell strongly; for example, the 238.5 keV ROI dropped from 213(5) $N_{\mathrm{after}}$6Hz to 17.7(1.4) $N_{\mathrm{after}}$7Hz.

The LS also provides alpha/beta pulse-shape discrimination. Selecting alpha-like LS pulses yields a cleaner HPGe spectrum, and combining alpha-selection with the muon veto removes events misidentified as alpha-like that originate from muon activity. Sensitivity was evaluated with the Currie formalism at 95% CL for a 100 g sample measured for 31 d. The minimum detectable activity is written as

$N_{\mathrm{after}}$8

where $N_{\mathrm{after}}$9 is the background counts in a $$R = 1 - \frac{N_{\mathrm{after}}}{N_{\mathrm{before}}}$$0 ROI, $$R = 1 - \frac{N_{\mathrm{after}}}{N_{\mathrm{before}}}$$1 is full-energy-peak efficiency, $$R = 1 - \frac{N_{\mathrm{after}}}{N_{\mathrm{before}}}$$2 is branching ratio, $$R = 1 - \frac{N_{\mathrm{after}}}{N_{\mathrm{before}}}$$3 is sample mass in kg, and $$R = 1 - \frac{N_{\mathrm{after}}}{N_{\mathrm{before}}}$$4 is live time in s. The minimum detectable concentration is

$$R = 1 - \frac{N_{\mathrm{after}}}{N_{\mathrm{before}}}$$5

with $$R = 1 - \frac{N_{\mathrm{after}}}{N_{\mathrm{before}}}$$6 the decay constant, $$R = 1 - \frac{N_{\mathrm{after}}}{N_{\mathrm{before}}}$$7 Avogadro’s number, and $$R = 1 - \frac{N_{\mathrm{after}}}{N_{\mathrm{before}}}$$8 atomic mass.

Reported MDA improvements include 214Pb at 352 keV from 26 to 11 mBq/kg, 208Tl at 583 keV from 43 to 21 mBq/kg, 60Co at 1173 keV from 14 to 4.0 mBq/kg, and 60Co at 1332 keV from 15 to 5.4 mBq/kg. In the 235U chain, alpha-tagged lines improved strongly, with 235U at 143.7 keV decreasing from 9.7 to 2.0 mBq/kg. MDCs follow correspondingly; one example given is 60Co down to $$R = 1 - \frac{N_{\mathrm{after}}}{N_{\mathrm{before}}}$$9 g/g. These results indicate that the performance gain arises from the combination of muon-veto suppression and LS alpha/beta selection rather than from the veto alone.

6. Residual background floor, interpretation, and scope

The difference between the near-unit simulated muon-tagging efficiency and the measured 92.69(3)% reduction over 20–3000 keV is explicitly interpreted in the study as a change in background composition, not as a failure of the veto. Once muons are strongly suppressed, non-muon backgrounds become relatively more important. Residual backgrounds after veto are attributed largely to environmental radioactivity and intrinsic detector or shield materials, which set the ultimate floor when the muon contribution is reduced to the percent level (Barresi et al., 17 Feb 2025).

The paper identifies the principal limitation of conventional HPGe muon vetoes as the practice of placing panels only above the lead castle, which gives poor angular coverage and high trigger rates from external radioactivity. GeSparK’s inside-shield topology addresses both limitations simultaneously. The strong suppression of the 511 keV feature and the agreement between measured and simulated HPGe energy deposition reinforce the conclusion that the pre-veto coincidence spectrum is muon-dominated.

Future improvements are discussed in terms of residual-background-limited applications, including long-lived neutron activation products such as 233Pa in NAA. The suggested directions are refined modeling of underground angular distributions, incremental shielding upgrades, and improved PSD and thresholds. These are presented as further optimization steps rather than as changes to the detector’s basic concept.

Within the sources considered here, the formal name GeSparK refers to this HPGe–LS alpha-beta/gamma coincidence detector. Other papers in the broader detector literature describe unrelated devices, and in some cases the term does not appear at all; for example, the Horizon-T implementation is formally called the “glass detector” or “Cherenkov glass detector,” and the spaceborne scintillator device is consistently referred to as the SSPD (Duspayev et al., 2017, Simhony et al., 2024). This suggests that GeSparK should be understood, in the strict instrumental sense, as the low-background coincidence spectrometer and its six-panel inside-shield muon-veto implementation rather than as a generic label for spark- or scintillation-based detectors.