Enhanced sensitivity to trace $^{238}$U impurity of sapphire via coincidence neutron activation analysis

TL;DR

This work demonstrates that γγ coincidence counting combined with neutron activation analysis can achieve parts-per-trillion sensitivity for trace ${}^{238}$U and ${}^{232}$Th in sapphire, addressing critical radiopurity needs for low-background detectors. By irradiating a Saint Gobain sapphire sample and performing both single-γ and γγ-coincidence measurements with two high-purity Ge detectors, the study leverages a detailed detector-response model validated against calibration data and a fly ash reference method to control neutron-flux and cross-section uncertainties. The analysis yields a 90% confidence upper limit of ${}^{232}$Th < 0.26 ppt and ${}^{238}$U < 2.3 ppt, representing the most stringent sapphire radiopurity constraints to date and confirming prior computational predictions. The approach provides a robust, background-resilient pathway to assay trace impurities in radiopure materials and is extendable to other substrates in rare-event experiments.

Abstract

Sapphire has mechanical and electrical properties that are advantageous for the construction of internal components of radiation detectors such as time projection chambers and bolometers. However, it has proved difficult to assess its $\rm ^{232}Th$ and $\rm ^{238}U$ content down to the picogram per gram level. This work reports an experimental verification of a computational study that demonstrates $γγ$ coincidence counting, coupled with neutron activation analysis (NAA), can reach ppt sensitivities. Combining results from $γγ$ coincidence counting with those of earlier single-$γ$ counting based NAA shows that a sample of Saint Gobain sapphire has $\rm ^{232}Th$ and $\rm ^{238}U$ concentrations of $<0.26$ ppt and $<2.3$ ppt, respectively; the best constraints on the radiopurity of sapphire.

Figures (7)

• Figure 1: Partial level scheme of $^{239}\text{Pu}$. Energies, transition strengths, spin-parity assignments and half-lives are taken from Ref. Browne2014. The transitions used in this study are indicated.
• Figure 2: Sapphire sample together with a PE counting vial.
• Figure 3: Simulated detector geometry. Detectors separated by 2$\,\text{cm}$. The sapphire is shown inside a sample counting bottle. Because of the size mismatch between sapphire sample and counting bottle, the location and orientation of the sample was not well defined. Simulations were performed for various orientations. The top panel shows the sample leaning diagonally inside the bottle; its most likely orientation. The two calibration source geometries (gray) are shown inside a source holder with different thickness spacers. The sample holder, that sets the separation between the detectors, is not modeled.
• Figure 4: Ratio of calibration source derived detection efficiency to simulation derived detection efficiency for prominent $\gamma$ rays from the scatterless source (see text for a list of nuclides) and from the $^{60}\text{Co}$ source (open squares). Ge-A in orange; Ge-B in blue. Data points are averages over several calibration runs with their RMS errors. Top are for single $\gamma$ rays. Bottom includes coincident pairs of the most prominent $\gamma$ rays from $^{60}\text{Co}$, $^{88}\text{Y}$, and $^{134}\text{Cs}$. Both permutations of the coincidence condition are shown in the lower figure: like pairs of orange points and like pairs of blue points are not independent measurements; the analysis does not depend on the choice of tagging detector.
• Figure 5: Time dependence of the observed count rates for four coincident photopeaks associated with $^{24}\text{Na}$, $^{46}\text{Sc}$, $^{99}\text{Mo}$, and $^{181}\text{Hf}$ present in the sapphire sample. An exponential decay model, both with fixed and free half-life parameter, are shown after a least squares fit to the normalization of the data points.