Performance of a SuperCDMS HVeV Detector with Sub-eV Energy Resolution and Single Charge-sensitivity

TL;DR

This work demonstrates a substantial advancement in superconducting silicon HVeV detectors, achieving a sub-eV baseline energy resolution of 612 meV and single-charge sensitivity by lowering the TES Tc to ~40 mK. Through two independent facilities, it validates a TES-based noise model, constrains phonon collection efficiency to about 45–61%, and reveals a consistent photon-energy loss of ~0.81 eV per charge excitation, likely tied to surface-band effects. The study combines detailed QET characterization, noise analysis, and optical calibration to refine energy scales and understand trapping and surface phenomena, which together enable improved event discrimination and near-quantum-limited phonon readout while addressing energy-scale systematics. The findings have significant implications for low-energy phonon backgrounds, band-structure-related energy losses, and the deployment of high-sensitivity detectors in dark matter and neutrino experiments. The work also shows no strong evidence for intrinsic correlated phonon noise, highlighting the importance of vibration isolation in controlling backgrounds for next-gen cryogenic detectors.

Abstract

We present a detailed characterization of a new generation of athermal-phonon single-charge sensitive Si HVeV detectors, the best of which achieved 612 meV $\pm$ 4 meV baseline resolution. Our sub-eV energy resolution enables precise measurements of single-photon events and reveal consistent energy losses of 0.81 eV $\pm$ 0.03 eV per charge excitation across two facilities. We demonstrate that the noise for these detectors is well described using a standard Transition Edge Sensor noise model. We also place upper bounds on the nominal phonon collection efficiency of 45\%, establishing these detectors as the most efficient athermal phonon detectors to date, limited only by intrinsic limitations of quasiparticle generation.

Figures (11)

• Figure 1: Left The SLAC mounting scheme for a single device of the NFC design clamped in a copper housing. Right The mounting scheme used at CUTE for two devices (NFC design on the left and NFH on the right) clamped in a similar housing. Both detector housings have an interlocking ridge to create an improved light-tight connection to a lid or other housing (in the case of a stacked payload geometry). The Si detectors shown are 1 cm x 1 cm x 4 mm thick.
• Figure 2: Measured TES bias power $P_0$ = $P_{\text{TES}}$ as a function of TES operating points, defined in terms of relative TES resistance. The shaded regions correspond to $\pm$ 1$\sigma$ bands bracketing the data. The error bands are constructed by propagating statistical uncertainties from the measured TES baseline currents and systematic uncertainties dominated by measurements of the shunt resistor in the TES circuit. A fully normal QET channel corresponds to $R_n=100\%$.
• Figure 3: Top Left Constant current TES fall time as a function of the TES bias point (resistance of the TES in steady-state operation here expressed as a percentage of the normal resistance) extracted from fits to the TES complex admittance. Top right Isothermal TES logarithmic current sensitivity $\beta = \frac{d(logR)}{d(logI)}|_T$ as a function of TES bias point. Bottom Fall time of the delta function response in the small signal limit as a function of TES bias point. The points show true measurements made at each facility. The dashed line extrapolates the curve below where data are available from each facility, to compare our predictions with our measured values shown in Table \ref{['tab: basic-params']}, confirming that the detectors at both facilities are thermally at lower transition points than expected from the applied bias currents (see text).
• Figure 4: Left: A comparison of the measured Noise Equivalent Power for a single QET channel at the two test facilities (SLAC and CUTE). The left panel compares the lowest-noise spectrum measured at each of the two facilities along with their corresponding best fit noise models, represented by a quadrature sum of the four noise components described in Section \ref{['sub_sec:noise_performance']}. The red spectrum was computed using SLAC data acquired when the pulse-tube (PT) of the cryostat was turned off, to minimize vibrational noise coupling into the detector. Contributions from environmental noise at 60 Hz (and its harmonics), as well as digital communication noise in the kHz regime are evident. Work to mitigate these noise sources is ongoing. Right: Total (purple), decorrelated (red) and correlated (yellow) noise spectra obtained at SLAC with the PT cooler turned on. The dominant correlated noise contribution seen below $\sim$300 Hz is due to the PT. Mechanical resonances associated with the cryostat are also evident in the spectra near 3 kHz. They do not appear in the PT off spectrum shown in the left panel.
• Figure 5: Left: Cross-talk-subtracted LED energy spectrum (black) from a data set taken at CUTE with 2.0 eV photons. Error bars for the y-axis represent the standard deviation of a Poisson distribution with the mean number of counts equal to that of a given bin. The fit shown in red corresponds to the detector response model described in this work. The energy distribution for events consistent with noise is shaded in blue and has a characteristic width consistent with the resolution returned by the detector response model fit. Right: The energy spectrum of LED events at SLAC with 3.4 eV photons. Note that the jagged features on the peaks are due to a combination of surface trapping and recombination effects, as well as the inequality between charge production and photon number. They are visible here (and not the left panel) because the collected photon energy is sufficiently large compared to the resolution, as discussed in Sec. \ref{['photon_energy_loss']}.