Low-energy single-electron detector with sub-micron resolution

TL;DR

The study introduces a YAG:Ce scintillator–based single-electron detector with optical readout that achieves sub-micron spatial resolution ($\sim$ $1\,\mu\mathrm{m}$) across $17$–$30\,\mathrm{keV}$ and near-unity single-electron classification efficiency and purity. By combining detailed Monte Carlo simulations of electron trajectories with a photon-counting CMOS readout and robust image processing, the authors demonstrate high-fidelity electron detection and a PSF of $\sim1\,\mu\mathrm{m}$ (FWHM) that, when accounting for walk-off, yields a net spatial resolution of $\sim2.3\,\mu\mathrm{m}$ at $30\,\mathrm{keV}$. They further show diffraction experiments at a sample–screen distance of $380\,\mu\mathrm{m}$, approaching electron mean free paths in air, enabling atmospheric-diffraction studies and potential table-top, time-resolved, or miniature diffraction setups. The work suggests significant practical impact for high-precision electron imaging and diffraction in compact, low-energy regimes, with avenues to further improve light yield and temporal performance through alternative scintillators and faster detectors.

Abstract

Single-electron detectors are a key component of electron microscopes and advanced electron optics experiments. We present a YAG:Ce scintillator-based single-electron detector with a spatial resolution of 1 micrometer at an electron energy of 17 keV. Single-electron detection events are identified with an efficiency and purity larger than 0.8 at an electron energy of 17 keV, reaching 0.96 at 30 keV. We show that the detector enables electron diffraction studies with a sample-detector distance comparable to the mean free path of electrons at atmospheric pressure, potentially enabling atmospheric electron diffraction studies.

Figures (6)

• Figure 1: (a) Schematic of the setup: an incident electron $e^-$ hits the YAG:Ce scintillator, where a part of its kinetic energy is converted into luminescent photon emission. The resulting signal is relayed by an infinity-corrected optical system onto the CMOS camera sensor. (b) Zoomed-in illustration of the electron trajectory in the YAG:Ce scintillator (black solid line) and the emitted photons (green arrows): $r_{\mathrm{in}}$ and $r_{\mathrm{c}}$ denote the coordinates of the electron incidence point and of the center of deposited energy, respectively. The distance between them is $d_{\mathrm{ic}}$. (c) Radius $\sigma_G$ enclosing $68\%$ of the emitted photons (purple line, round markers), and median distance $\tilde{d}_{\mathrm{ic}}$ between $r_{\mathrm{in}}$ and $r_{\mathrm{c}}$ (blue line, square markers), both shown as a function of the electrons' initial kinetic energy.
• Figure 2: Data analysis for $30\,keV$ electrons: (a) Zoom-in of a square subregion of the raw data frame; (b) Same subregion after applying a Gaussian filter. (c) purity $p$ versus efficiency $\eta$ curve, with the maximum $F_{1}$-score indicated. (d) Histogram of photon counts within a circle of diameter $d_{\mathrm{ref}}$, centered on local maxima in the Gaussian-filtered image. The histogram is fit with a linear combination of Log-Normal and Normal distributions. The confusion matrix for binary classification is calculated with a threshold $T_{\mathrm{rm}}=17\,\mathrm{counts}$, which optimizes the purity and is used to select events for computing the average PSF. (e) Cross section of the average PSF (S, green line) with corresponding FWHM = 1.04 $\mu$m, and the average background (B, blue line).
• Figure 3: (a) Measured FWHM of the PSF as a function of energy. (b) Average photon counts per detected event as a function of energy (blue line, round markers) and linear fit (orange line). The linear fit has a slope of 0.67 photons/keV and an intercept of 4.8 photons. (c) purity-efficiency curves for electron energies between 17 keV and 30 keV. (d) Best $F_{\mathrm{1}}$-score and corresponding purity $p$ and efficiency $\eta$ as a function of electron energy.
• Figure 4: (a) Diffraction pattern of a [001]-oriented single gold crystal, obtained with a 30 keV electron beam. Scale bar: 15 in the detection plane. (b) Measurement scheme: The gold crystal (Au) is on a 300 mesh gold TEM grid (omitted in sketch) and placed on top of the YAG:Ce scintillator with a plastic spacer in between, leading to a sample-screen distance of $L= 380\mu m$.
• Figure 5: Monte Carlo simulation: (a) and (b) normalized x-y and y-z projections of the spatial distribution of energy deposited by the electrons on the scintillator. Electron trajectories used in the computation are aligned with respect to their centre of deposited energy $r_{\mathrm{coel}} = (x_{\mathrm{c}},y_{\mathrm{c}})$. (c) Distributions of $d_{\mathrm{ic}}$ (distance of the center of deposited energy from the entrance point) for $30 \, keV$ electrons (blue, right) and $20 \, keV$ electrons (orange, left). The dotted lines mark the medians of each distribution.