X-ray characterization of fully-depleted p-channel Skipper-CCDs for the DarkNESS mission

The Dark matter Nanosatellite Equipped with Skipper Sensors (DarkNESS) mission is a 6U CubeSat designed to search for X-ray lines from decaying dark matter using Skipper-CCDs. Thick, fully-depleted p-channel Skipper-CCDs provide low readout noise and high quantum efficiency for 1-10 keV X-rays, but their X-ray performance has not yet been demonstrated in the space environment. DarkNESS will operate in low-Earth orbit, where trapped protons induce displacement damage in the sensor that increases charge-transfer inefficiency and degrades the X-ray energy resolution. This work measures the X-ray line response of Skipper-CCDs before and after proton irradiation and quantifies the associated degradation. A sensor was exposed to 217 MeV protons at a fluence of 8.4 x 10^10 protons cm^-2, corresponding to a displacement-damage dose more than an order of magnitude above the three-year expectation for representative mid-inclination and Sun-synchronous low-Earth orbits. A 55Fe source was used to compare the energy resolution of the beam-exposed quadrant to adjacent unexposed quadrants and a non-irradiated reference sensor. These measurements provide a quantitative assessment of radiation-induced spectral degradation in Skipper-CCDs and enable an estimate of the end-of-life X-ray energy resolution expected for DarkNESS operation in low-Earth orbit.

💡 Research Summary

The Dark matter Nanosatellite Equipped with Skipper Sensors (DarkNESS) mission is a 6U CubeSat that will carry four thick, fully‑depleted p‑channel Skipper CCDs to search for narrow X‑ray lines from decaying dark‑matter particles in the 1–10 keV band. Skipper CCDs achieve sub‑electron readout noise by repeatedly non‑destructively sampling each pixel with a floating‑gate amplifier; averaging N samples reduces the noise as 1/√N. Their large depletion depth (up to 725 µm) gives >99 % quantum efficiency at 10 keV, making them ideal for low‑background X‑ray spectroscopy.

The key unknown for a space mission is how displacement damage from trapped protons will affect the CCDs’ spectral performance. To address this, the authors irradiated a prototype Skipper CCD with a 217 MeV proton beam at a fluence of 8.4 × 10¹⁰ protons cm⁻². This corresponds to a non‑ionizing energy loss (NIEL) dose of 1.6 × 10⁸ MeV g⁻¹ (Si), more than ten times the expected three‑year dose for representative low‑Earth orbits (LEO): an International Space Station‑like orbit (410 km, 51.6° inclination) and a sun‑synchronous orbit (500 km, 97.0°). The sensor was mounted in a 2 × 2 array so that only one quadrant (BI4) received the full beam, an adjacent quadrant (IA2) served as an on‑chip reference, and a separate non‑irradiated CCD provided a baseline (NI2).

Measurements were performed in a vacuum chamber at 153 K. A ⁵⁵Fe source produced Mn‑Kα (5.895 keV) and Kβ (6.49 keV) X‑rays. Each raw image contained ten non‑destructive samples per pixel; these were averaged to suppress readout noise (≈3.5 e⁻ rms in non‑irradiated quadrants, ≈4.7 e⁻ rms in the irradiated quadrant). Overscan pixels were used to estimate the noise robustly (median ± MAD, outlier masking). After row‑wise and column‑wise baseline subtraction, events were identified with a two‑threshold connected‑component algorithm (seed at 9σ, split at 3σ). Clusters were classified by topology (single‑pixel, 2‑split, L‑split, C‑split, extended) to separate genuine X‑ray events from charge‑transfer‑inefficiency (CTI) trails and cosmic‑ray tracks.

Spectra were built from the energies of validated clusters. The Mn‑Kα and Kβ peaks were fitted simultaneously with a double‑Gaussian model whose centroid ratio and relative amplitudes were fixed to the known values. The Gaussian width gave the instrumental full‑width at half‑maximum (FWHM) at 5.895 keV; the centroid separation served as an energy‑scale check. The variance was decomposed into contributions from Fano statistics, electronic read noise, and a residual term attributed to radiation‑induced charge‑transfer losses.

Results show that the non‑irradiated quadrant (NI2) achieves an FWHM of ~130 eV, consistent with expectations for a low‑noise Skipper CCD. The irradiated quadrant (BI4) displays a broader line (≈180–200 eV) when the full image is used, and the measured Kβ–Kα separation is distorted (≈623 eV instead of the theoretical 595 eV). This distortion disappears when only the first 100 rows (closest to the serial register) are analyzed, yielding a separation of 600.7 eV and an FWHM near 150 eV. The improvement indicates that the dominant degradation originates from CTI: traps created by displacement damage capture charge during vertical transfer, producing “C‑split” trails and broadening the energy distribution. Quadrant‑to‑quadrant gain variations of 10–15 % were also observed, a known effect of readout‑amp mismatches.

Using the measured displacement‑damage dose as an anchor, the authors folded the AP‑9 trapped‑proton spectra for the two representative LEO environments. The resulting equivalent doses correspond to ≈1.5 × 10⁷ MeV g⁻¹ (ISS) and ≈3 × 10⁷ MeV g⁻¹ (sun‑synchronous) over three years. Propagating the CTI‑induced broadening through a simple model predicts an end‑of‑life energy resolution of 150–170 eV (FWHM) at 5.9 keV for DarkNESS. This resolution is sufficient to keep the analysis window narrow enough that background counts remain low, preserving the sensitivity to faint, narrow X‑ray lines from dark‑matter decay.

In conclusion, thick p‑channel Skipper CCDs retain excellent X‑ray spectroscopic performance after exposure to proton fluences far exceeding the expected LEO dose. The primary radiation effect is an increase in CTI, which can be mitigated by careful event selection (e.g., limiting the number of vertical transfers) and by applying CTI correction algorithms. The study validates the suitability of Skipper CCDs for the DarkNESS mission and demonstrates that the technology is ready for space‑based low‑background X‑ray spectroscopy. Future work will focus on in‑orbit calibration, long‑term annealing studies, and refined CTI correction to further improve the achievable energy resolution.