Sensitivity enhancement techniques for cryogenic calorimeters in the NUCLEUS experiment

This paper presents two complementary sensitivity enhancement techniques for the NUCLEUS experiment's cryogenic calorimeters—operating point optimization and two-dimensional optimum filter analysis—which, when combined, achieved a baseline energy resolution of 2.94 ± 0.05 eV using a CaWO4 detector.

The Gist

Imagine you are trying to hear a single, tiny whisper in a room that is constantly buzzing with the hum of a refrigerator, the wind outside, and people talking. That is essentially what the NUCLEUS experiment is trying to do. They are looking for the faint "whispers" of subatomic particles (specifically, neutrinos bouncing off atomic nuclei) hitting a crystal detector.

To hear these whispers, they use cryogenic calorimeters—basically, ultra-sensitive thermometers made of special crystals (like Calcium Tungstate) that are cooled down to temperatures colder than outer space. When a particle hits the crystal, it creates a tiny ripple of heat (a phonon), which the detector senses.

The problem? The "whisper" is so quiet that the detector's own internal noise often drowns it out. The goal of this paper is to teach the detector how to listen better. The team used two clever tricks to make the detector super-sensitive.

Here is how they did it, explained with some everyday analogies:

Trick #1: Finding the "Sweet Spot" (Operating Point Optimization)

Think of the detector's sensor (called a TES) like a volume knob on a very finicky radio.

• If you turn the knob too low, you can't hear the music (the signal is too weak).
• If you turn it too high, the static (noise) gets so loud it drowns out the music.
• There is a tiny, perfect spot in the middle where the music is clear and the static is low.

In the past, scientists would just guess where that "sweet spot" was by turning the knob and seeing how loud a test pulse sounded. But that's like guessing the radio station by just listening to the static.

The New Method:
Instead of just guessing, the team turned the knob slowly across the entire range, stopping every few minutes to take a "snapshot" of the radio.

1. They injected a known "test sound" (a flash of light from an LED) to see how loud the signal was.
2. They listened to the silence (the background noise) to see how much static was there.
3. They calculated a Signal-to-Noise Ratio (SNR): How much louder is the whisper than the background hum?

By mapping out the entire radio dial, they found the exact mathematical "sweet spot" where the whisper is clearest. They didn't just pick the loudest signal; they picked the spot where the signal is the clearest compared to the noise.

Trick #2: The "Stereo" Effect (2D Optimum Filter)

The detector used in this experiment has two sensors (two microphones) attached to the same crystal. Usually, scientists look at each microphone separately.

The Analogy:
Imagine you are in a noisy room with two friends, Alice and Bob, both trying to hear a whisper.

• Alice hears the whisper, but she also hears the hum of the fridge.
• Bob hears the whisper, but he hears the wind outside.
• If you listen to Alice alone, the fridge hum might make you think the whisper isn't there.
• If you listen to Bob alone, the wind might do the same.

The New Method:
The team created a mathematical algorithm that listens to Alice and Bob simultaneously.

• The "fridge hum" and the "wind" are random and different for each person (uncorrelated noise).
• The "whisper" happens at the exact same time for both (correlated signal).

The algorithm acts like a smart mixer. It says, "Okay, Alice and Bob both heard a sound at the same time. That must be the whisper! But Alice heard a fridge hum that Bob didn't, and Bob heard wind that Alice didn't. Let's cancel out those differences."

By combining the two streams of data and subtracting the unique noise of each sensor, the "whisper" becomes much clearer. It's like using noise-canceling headphones, but for a scientific detector.

The Result: A Crystal Clear Whisper

By using both tricks together:

1. They found the perfect volume knob setting (Operating Point).
2. They used the "Stereo" trick to cancel out the background noise (2D Filter).

They achieved a Baseline Resolution of 2.94 eV.

What does that mean?
In the world of these detectors, 3 eV is incredibly small. It's like being able to hear a single drop of water falling into a bucket from a mile away, even while a jet engine is running nearby.

This is a massive improvement. It means that when the NUCLEUS experiment moves to its final location at a nuclear power plant in 2026, it will be able to detect neutrinos with much lower energy than ever before. This opens the door to discovering new physics and understanding the universe in ways we couldn't before.

In short: They taught a super-cold crystal how to tune its radio dial perfectly and how to use two ears to cancel out the noise, allowing it to hear the faintest whispers of the universe.

Technical Summary

1. Problem Statement

The NUCLEUS experiment aims to detect Coherent Elastic Neutrino-Nucleus Scattering (CE$\nu$NS) using reactor antineutrinos. This detection requires measuring nuclear recoil energies in the range of a few tens of eV or lower. To achieve this, the experiment utilizes phonon-mediated cryogenic calorimeters equipped with Transition-Edge Sensors (TES).

While previous iterations of NUCLEUS detectors achieved baseline energy resolutions (BLR) below 10 eV, further sensitivity improvements are necessary to lower the energy threshold and access the strongest part of the neutrino-induced recoil spectrum. The challenge lies in optimizing the detector's operating conditions and processing the dual-readout data to minimize noise and maximize the Signal-to-Noise Ratio (SNR).

2. Methodology

The authors propose and test two complementary optimization techniques using a CaWO$_4$-based double-TES detector (1.74 g absorber) operated in a dilution refrigerator at the Technical University of Munich (TUM).

A. Operating Point (OP) Optimization via SNR Mapping

• Traditional Approach: Historically, TES operating points were selected manually by maximizing the amplitude of heater-injected pulses. This method ignores detailed noise characteristics.
• New Approach: The authors developed a systematic, automated method to map the SNR across a 2D grid of operating parameters:
  • Parameters: Bias current ($I_B$) ranging from 1.5 to 4.5 $\mu$A and heater voltage ($V_H$).
  • Procedure: "Slow heater sweeps" were performed where the heater power was stepped down. At each step, the system stabilized for 4 minutes.
  • Signal Injection: Two types of pulses were injected:
    1. Saturated Heater Pulses: To monitor the position on the superconducting transition curve.
    2. LED Pulses: 255 nm optical photons delivered via fiber to simulate particle-induced electron recoils. These provide a realistic signal template.
  • Analysis: For each operating point, a pulse template was created from LED events, and a Noise Power Spectrum (NPS) was derived from noise traces. An Optimum Filter (OF) was applied to calculate the SNR as the ratio of the median LED pulse height to the BLR.
  • Selection: A 2D SNR map was generated. The optimal OP was selected to maximize SNR for both TESs simultaneously, accounting for the constraint that they share a single heater.

B. 2D Optimum Filter (2D-OF) Formalism

• Concept: Since the detector features two independent TESs reading the same absorber, the authors applied a matrix-based 2D optimum filter to combine the waveforms.
• Mathematical Framework: The filter combines the Fourier transforms of the two sensor waveforms ($\tilde{v}_1, \tilde{v}_2$) using a noise covariance matrix ($\hat{N}$) and template vectors ($\tilde{S}$).
  • The filter accounts for the noise power spectra of individual sensors ($N_1, N_2$) and their cross-correlation ($c_{12}$).
  • The theoretical resolution gain depends on the noise correlation magnitude ($|\rho|$) and phase ($\theta$). If noise is uncorrelated, precision improves; if correlated with a phase shift, noise subtraction occurs.
• Implementation: The 2D-OF was applied to calibration data (LED pulses) from the best-performing operating point identified in the first step.

3. Key Contributions

1. SNR-Based OP Selection: Introduction of a robust, automated method to select the TES operating point by maximizing SNR rather than just pulse amplitude. This method utilizes LED-induced pulses as a superior proxy for physical particle events compared to heater pulses.
2. 2D Optimum Filter Application: Demonstration of a 2D-OF algorithm that effectively combines dual-readout data, explicitly modeling noise correlations to reduce the baseline noise RMS.
3. Record-Breaking Resolution: Achievement of a sub-3 eV baseline resolution for a NUCLEUS-type calorimeter, setting a new benchmark for the experiment.

4. Results

• Operating Point Optimization:
  • The SNR mapping successfully identified optimal configurations.
  • Configuration B (selected as optimal) yielded the best performance.
  • A strong inverse correlation was observed between SNR and BLR: high SNR configurations resulted in significantly lower energy resolutions.
  • In sub-optimal configurations (D and E), the resolution degraded by factors of ~1.2 to ~2 compared to Configuration B.
• 2D Optimum Filter Performance:
  • Applying the 2D-OF to Configuration B data resulted in a Baseline Resolution (BLR) of $2.94 \pm 0.05$ eV.
  • This represents a ~20% improvement over the standard 1D optimum filter resolutions of the individual sensors ($\sigma_1 \approx 3.81$ eV, $\sigma_2 \approx 3.60$ eV).
  • The experimental gain matched theoretical predictions based on the measured noise correlation ($|\rho| \approx 0.27$).
  • The 2D-OF amplitude estimates were found to be perfectly compatible with standard 1D-OF estimates for valid events.

5. Significance

• Enhanced Physics Reach: Achieving a BLR below 3 eV significantly lowers the energy threshold for the NUCLEUS experiment. This is critical for CE$\nu$NS detection, as the signal rate increases sharply at lower energies.
• Technical Readiness: The results validate the optimization procedures for the upcoming technical run at the Chooz nuclear power plant (scheduled for 2026).
• Scalability: The authors estimate that applying these techniques to earlier commissioning data (using an Al$_2$O$_3$ detector) could have achieved resolutions of ~2.7 eV, suggesting these methods are universally applicable to the NUCLEUS detector array.
• Noise Reduction: The reduction in noise RMS implies a lower trigger threshold and a reduction in "negative triggers" (noise fluctuations mimicking signals), thereby improving the purity of the low-energy data spectrum.

In conclusion, the combination of systematic SNR-based operating point selection and 2D optimum filtering provides a powerful pathway to maximize the sensitivity of cryogenic calorimeters for rare-event searches.