Sensitivity enhancement techniques for cryogenic calorimeters in the NUCLEUS experiment

Abstract

Phonon-mediated cryogenic calorimeters find application in rare event searches due to their intrinsically low energy threshold. Achieving the best sensitivity for this kind of detectors is crucial for signal identification, leading to various optimization techniques. In this work, we present two complementary methods to increase the sensitivity of cryogenic detectors read out by transition-edge sensors, developed and tested in the context of the NUCLEUS experiment. The first procedure maps the signal-to-noise ratio of the device across a wide range of operating points, to identify the configuration with maximal sensitivity to be used during data taking. The second method exploits the double readout of the detector, combining the information on different channels with a two-dimensional optimum filter analysis that effectively lowers the energy threshold. With both techniques at the same time, we obtained a baseline resolution of 2.94 $\pm$ 0.05 (stat) eV using a CaWO4 based detector, achieving a promising result in view of the first run of NUCLEUS at the experimental site.

Figures (8)

• Figure 1: Schematic of the used CaWO4 double-TES detector. The absorber crystal measures $11.5\times 5\times 5~\mathrm{mm^3}$. The depicted TES size is not to scale. The TES readout circuit is explicitly shown for the left TES but also present for the right one. The TES acts as a temperature dependent resistor $R_T$ and is operated in parallel with a SQUID and a shunt resistance $R_S = \mathrm{40}~ \Omega$. Resistance changes in the TES lead to a different branching of the bias current $I_B$, which is picked up (and amplified) by the SQUID and converted into a voltage signal. Only one heater is operated during measurements to heat up the whole crystal, while the other one is left unconnected.
• Figure 2: Detector channel signal response of part of a heater sweep for $3.0~\mu\mathrm{A}$ fixed bias current. The heater power was decreased every 4 minutes. For every heater power step, the baseline level in the detector channel also drops due to the temperature decrease. A zoomed inset shows the artificially injected heater pulses, which are sent every $15$ s to track the depth of the OP on the transition curve. Additionally, LED pulses are sent every $15$ s, which serve as signal reference pulses. They grow in amplitude the steeper the TES transition curve gets. The other visible pulses mostly originate from the $^{55}$Fe X-ray source and smaller injected heater pulses.
• Figure 3: Waveforms on the two sensors (TES 1 and TES 2) following the same energy deposition in the absorber cube, together with the filtered signal in the time domain, obtained from Eq. \ref{['eq:v_filtered']}.
• Figure 4: Expected resolution of the two-dimensional OF $\sigma_{\text{2D}}$, computed from Eq. \ref{['eq:reso_2d']} in the case of two identical sensors, as a function of the absolute value of the noise correlation $|\rho|$, for different values of the phase of the correlation $\theta$. The noise correlation is assumed constant in all the frequency range, and the resolution is normalized with respect to the one-dimensional optimum filter resolution $\sigma_{\text{1D}}$, obtained from Eq. \ref{['eq:reso_1d']}. The expected resolution of the NUCLEUS detector employed in this work, given the measured mean values of $|\rho|$ and $\theta$, is marked. This value should be considered as a coarse estimate since the non-trivial frequency dependence of the correlation was not taken into account.
• Figure 5: SNR curves obtained from LED pulses for the CaWO4 double-TES detector. The colors represent the different applied bias currents during the sweeps. Based on this, five different configurations, labeled A-E and marked with diamonds, were chosen to record calibration data on. For configuration D, the selected OP of TES 1 could not be re-established simultaneously with that of TES 2 during calibration data taking. The position labeled D* indicates the originally expected OP. In practice, the realized OP of TES 1 is shifted to the right toward lower SNR values.