Towards low-cost lead screening with transmission XRF

Abstract

Human exposure to lead (Pb) is a global health concern, yet existing technologies for detecting lead in our environment remain prohibitively expensive for widespread deployment. Here we present a new concept towards lead screening using X-ray fluorescence (XRF) in an unconventional geometry we coin transmission XRF in which the sample is placed between the source and detector. For cost reduction, we then show that $^{241}$Am found in ionizing smoke detectors is spectrally suitable for Pb L-shell XRF generation and can thus replace X-ray tubes used in conventional XRF devices. Exploring soil screening as the first application, we demonstrate with Monte Carlo simulations that a configuration with 7$\times$ $^{241}$Am sources and a standard silicon drift detector can enable screening-relevant detection limits (100 ppm Pb) in soil within practical measurement times (<30 min). We believe this concept opens a route toward low-cost and scalable XRF instrumentation for democratizing lead screening across a wide range of samples.

Figures (5)

• Figure 1: Transmission XRF concept.a, Conventional backscatter geometry in a portable device including an X-ray tube, filter, and detector. b, Our proposed transmission geometry with the sample sandwiched between the $^{241}$Am source and the detector. c, $^{241}$Am emission spectrum showing Np, Au, and Ag peaks (blue) as well as Pb L-shell data overlaid (red). Note that the $^{241}$Am sample in smoke detectors contains both Au and Ag from the manufacturing process belanger1979environmental, explaining the line emissions. Np L-shell lines arise from XRF excitation of $^{237}$Np (decay product of $^{241}$Am) by the $^{241}$Am gamma emissions (mainly 59.5 and 26.3 keV). d, Simulated 10-hour $^{241}$Am exposure on a soil sample, comparing backscatter and transmission geometries for 100 ppm Pb in the soil, demonstrating superior Pb detection statistics for the transmission geometry.
• Figure 2: Finding the optimal soil thickness.a, United States Environmental Protection Agency (EPA) guidelines for soil lead levels, updated in 2024. b, Soil sample preparation, where we identify the optimal thickness at 100 ppm Pb through a 10-hour $^{241}$Am simulated exposure (cf. Fig. \ref{['fig:1']}d). c, Simulated 10-hour spectra for 1.5, 2.5, and 3.5 mm soil thickness showing Pb L$\alpha$ and L$\beta$ peaks (red) and background (blue). d, Integrated XRF signal (red) and background counts (blue) as a function of soil thickness, showing that the sample acts as a background filter. e, Signal-to-noise ratio (SNR) as a function of soil thickness for 100 ppm Pb. Results for lower (10 ppm) and higher (1000 ppm) concentrations can be found in Fig. \ref{['suppl:fig_optimal_thickness']}.
• Figure 3: Measurement time for soil screening.a, Optimized arrangement with 7$\times$$^{241}$Am sources and a 50 mm$^2$ silicon drift detector. b, Simulated detector spectra at example timepoints and Pb concentrations. c, Signal-to-noise ratio versus measurement time for the L-shell Pb XRF with the expected $\sqrt{t}$ dependence. d, Detection criterion schematic. e, Limit of detection versus measurement time for 1$\times$ and 7$\times$$^{241}$Am configurations, with uncertainties in the measurement time calculated according to Suppl. Sect. \ref{['suppl:sec_sigma_t']}. For derivations of the fit equations in c and e, see Suppl. Sect. \ref{['suppl:sec_LOD_derivation']}.
• Figure 4: Requirements on the detector.a, Baseline detector configuration with 100% efficiency ($\eta$), 150 eV resolution ($\Delta E$), and 50 mm$^2$ area ($A$), similar to silicon drift detectors commonly found in commercial handheld XRF devices. b, Measurement time scaling factors (shown as $\times$) for reaching a limit of detection (LOD) of 100 and 200 ppm Pb when independently varying detection efficiency (25--100%), energy resolution (150--500 eV), and active area (12.5--50 mm$^2$).
• Figure S1: Optimal sample thickness for different lead concentrations.a, Transmission geometry schematic with variable thickness. b, Integration window definition for Pb L$\alpha$ and L$\beta$ peaks. c, Photon counts and signal-to-noise ratio versus thickness for 1000 ppm, 100 ppm, and 10 ppm Pb concentrations.