🔭 astrophysics

Detectors for CLASS-W2: The second 90 GHz telescope of the Cosmology Large Angular Scale Surveyor

The paper reports the successful fabrication, testing, and commissioning of the CLASS-W2 receiver, a second 90 GHz telescope featuring a 296-detector TES bolometer array that achieved a 94% yield, suppressed high-frequency "blue-leak" radiation with a low-pass filter, and increased the CLASS experiment's 90 GHz mapping speed by 41%.

Original authors: John W. Appel, Kyuyoung Bae, Charles L. Bennett, Michael K. Brewer, Sarah Marie Bruno, Carol Yan Yan Chan, Joseph Cleary, Sumit Dahal, Jullianna Denes Couto, Kevin L. Denis, Shannon M. Duff, Joseph R.

Published 2026-06-26

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: John W. Appel, Kyuyoung Bae, Charles L. Bennett, Michael K. Brewer, Sarah Marie Bruno, Carol Yan Yan Chan, Joseph Cleary, Sumit Dahal, Jullianna Denes Couto, Kevin L. Denis, Shannon M. Duff, Joseph R. Eimer, Thomas Essinger-Hileman, Naina Gupta, Johannes Hubmayr, Gregory Jaehnig, John Karakla, Matthew Koc, Jeff Van Lanen, Yunyang Li, Michael J. Link, Tammy Lucas, Tobias Marriage, Carolina Morales Perez, Matthew A. Petroff, Caleigh Ryan, Rui Shi, Deniz A. N. Valle, Edward J. Wollack

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the universe is a giant, ancient room filled with a faint, ghostly afterglow from its birth. This "Cosmic Microwave Background" (CMB) is like the residual heat from a fire that went out billions of years ago. Scientists want to take a picture of this room to understand how the universe began and how it grew up. However, this picture is incredibly faint, and the "camera" needs to be extremely sensitive to see the subtle patterns (polarization) that tell the story of the universe's earliest moments.

This paper describes the construction and testing of a new, super-sensitive camera lens for a project called CLASS (Cosmology Large Angular Scale Surveyor). Specifically, it details the second 90 GHz telescope (called CLASS-W2) and its "retina"—the detector array that actually captures the light.

Here is the story of how they built it, the problems they found, and how they fixed them, explained in everyday terms.

1. The Mission: Taking a Better Picture

The first telescope (CLASS-W1) was already doing great work, but the scientists wanted to take the picture faster and clearer. To do this, they built a second telescope (CLASS-W2) to work alongside the first one. Think of it like adding a second high-speed camera to a sports broadcast; you can now capture twice as much action in the same amount of time.

This new telescope uses a special type of sensor called a Transition Edge Sensor (TES). Imagine these sensors as tiny, ultra-sensitive thermometers. They are so sensitive that they can detect the tiniest drop in temperature caused by a single photon (a particle of light) hitting them. To work, these thermometers must be kept colder than outer space itself—near absolute zero.

2. The "Retina": A Grid of Tiny Thermometers

The heart of this telescope is a focal plane made of 296 detectors.

The Layout: These detectors are arranged in four modules (like four panels on a screen).
The Design: Each module has 37 "feedhorns" (little metal funnels that catch the light). Each funnel splits the light into two directions (polarizations) and sends it to two tiny thermometers.
The Result: This creates a massive grid of 296 eyes, all looking at the sky at the same time.

3. The "Blue Leak" Problem: A Draft in the Window

When they first turned on the telescope in 2025, they noticed a problem. The thermometers were getting warmer than they should be, even when looking at empty space.

The Analogy: Imagine you are trying to listen to a whisper in a quiet room, but someone keeps opening the window and letting in a loud, high-pitched wind. You can't hear the whisper because the wind is drowning it out.

The Culprit: The "wind" was high-frequency radiation (called "blue-leak") coming from the warm parts of the telescope and the atmosphere. This radiation wasn't supposed to reach the detectors, but it was sneaking through the gaps and hitting the thermometers directly.
The Fix: In early 2026, the team installed a special "metal-mesh filter" (a Low-Pass Filter) in front of the detectors.
The Metaphor: Think of this filter as a sieve or a colander. It lets the specific "soup" of light they want (the 90 GHz signal) pass through, but it blocks the "hot water" (the high-frequency blue-leak) from splashing onto the thermometers.
The Result: The filter successfully blocked the unwanted heat, reducing the "noise" on the detectors by a significant amount.

4. How Well Does It Work?

After fixing the "blue leak" and testing everything in the lab and on the sky, here is what they found:

Uniformity: The 296 detectors are remarkably identical. It's like baking a batch of 296 cookies where every single one is the exact same size and temperature. This is crucial because if some detectors were "louder" than others, the final picture would be distorted.
Speed: The detectors react incredibly fast (in less than a millisecond). This allows the telescope to scan the sky quickly without blurring the image.
Efficiency: About 37% of the light that enters the telescope actually makes it to the detectors and is measured. While this might sound low, it's actually very good for this type of instrument.
Success Rate: 94% of the detectors worked perfectly. This is a huge improvement over the first telescope, which had a lower success rate.

5. The Bottom Line: A Faster, Clearer View

By adding this new telescope and fixing the "blue leak" issue, the CLASS project has boosted its ability to map the universe by 41%.

Before: They could map a certain area of the sky in a certain amount of time.
After: They can map that same area 41% faster, or map a much larger area in the same time.

The paper concludes that while there is still a tiny bit of "static" (noise) left in the system—likely due to the inherent limits of the materials or tiny amounts of stray light—the new system is a massive success. It is now ready to help scientists solve mysteries about how the universe was born and how it evolved, specifically by measuring the faint polarization of the Cosmic Microwave Background.

In short: They built a second, super-sensitive camera for the universe, found a draft letting in too much heat, plugged the hole with a special filter, and now the camera is snapping pictures of the cosmos faster and clearer than ever before.

Technical Summary: Detectors for CLASS-W2

Problem and Context
The Cosmology Large Angular Scale Surveyor (CLASS) aims to measure Cosmic Microwave Background (CMB) polarization anisotropy on the largest angular scales to constrain the optical depth to reionization ( $\tau_{reion}$ ), a parameter currently poorly constrained by space-based measurements. To increase mapping speed and sensitivity at the 90 GHz frequency band—where the CMB signal is strongest relative to foregrounds—CLASS deployed a second telescope, CLASS-W2. This paper focuses on the design, characterization, and performance of the CLASS-W2 detector focal plane, which utilizes Transition Edge Sensor (TES) bolometers fabricated at NIST-Boulder. A specific challenge addressed was the presence of "blue-leak" radiation (high-frequency out-of-band power) that coupled directly to the TES islands, increasing optical loading and noise.

Methodology
The study involved the fabrication, laboratory testing, and on-sky commissioning of a four-module detector array (296 detectors) mounted on a 50 mK stage within a dilution refrigerator.

Detector Design: The focal plane consists of feedhorn-coupled TES bolometers. Each of the four modules contains 37 feedhorns, with each feedhorn coupling orthogonal polarizations to two TES bolometers via a superconducting planar ortho-mode transducer (OMT). The detectors utilize AlMn TESs on SiN membranes with gold-plated SiAl baseplates for thermal anchoring.
Characterization:
- Laboratory: Dark I-V curves were acquired across a range of bath temperatures ( $T_b$ ) to extract thermal parameters ( $T_c$ , $\kappa$ , $n$ , $G$ ) and responsivity.
- Optical: Bandpass measurements were conducted using a Martin-Puplett Fourier Transform Spectrometer (FTS). On-sky observations of Jupiter and the Moon were used to measure beam profiles, optical efficiency, and optical loading.
- Mitigation: To address "blue-leak" power, a free-space metal-mesh low-pass filter (LPF) with a 157 GHz cutoff was installed in front of the focal plane in February 2026. Performance was compared between the "no-LPF" (2025) and "LPF" (2026) configurations.
Analysis: Noise Equivalent Power (NEP) and Noise Equivalent Temperature (NET) were calculated by combining measured NEP with optical efficiency and bandpass data. Excess noise was quantified by comparing measured NEP against models including photon, phonon, and readout noise.

Key Results

Detector Uniformity and Yield: The array achieved a 94% detector yield. TES properties were highly uniform, with an array average critical temperature ( $T_c$ ) of $184 \pm 3$ mK, thermal conductance ( $G$ ) of $460 \pm 47$ pW K $^{-1}$ , and normal resistance ( $R_n$ ) of $7.8 \pm 0.3$ m $\Omega$ .
Bandpass and Efficiency: The measured bandpass has a center frequency of 95.2 GHz with a 28.3 GHz bandwidth. The end-to-end optical efficiency is 0.37 on average (ranging from 0.31 to 0.44 across modules).
Optical Loading and Blue-Leak:
- Without the LPF, dark detectors observed $\sim$ 1.1 pW of blue-leak power, indicating direct coupling of high-frequency radiation to the TES islands.
- Installation of the LPF reduced this blue-leak power to $\sim$ 0.2 pW, a suppression of 0.9 pW.
- The mean in-band optical load on the detectors is 3.3 pW, corresponding to an antenna temperature of $\sim$ 23 K.
Noise Performance:
- The array achieved a Noise Equivalent Temperature (NET) of 16 $\mu$ K $\sqrt{s}$ for the four-module array.
- The installation of the LPF reduced the per-detector NEP $^2$ by $(13 \text{ aW}\sqrt{s})^2$ .
- Despite the LPF, a residual excess noise remains, comparable to the detector phonon noise, potentially due to intrinsic detector noise or residual out-of-band power.
Speed: The electro-thermal feedback speeds up the thermal time constant by a factor of $\sim$ 10, resulting in an optical time constant of 0.67 ms, well below the 16 ms requirement for 10 Hz modulation.

Significance and Claims
The paper claims that the deployment of the CLASS-W2 detector array has successfully enhanced the CLASS 90 GHz mapping speed by 41%. This improvement is attributed to the high detector yield (94% vs. 66% for the previous CLASS-W1 generation), the uniformity of the AlMn TES properties, and the faster detector time constants. The study validates the "blue-leak" model, demonstrating that high-frequency radiation couples directly to TES islands and that a metal-mesh low-pass filter effectively mitigates this noise source. While the achieved $T_c$ (184 mK) is higher than the target (160 mK), resulting in higher saturation power and slightly elevated NEP, the system is photon-noise limited and ready for further optimization in future fabrication runs scheduled for 2027. The work establishes a robust platform for continued ground-based CMB polarization measurements to refine constraints on reionization.

1. The Mission: Taking a Better Picture

2. The "Retina": A Grid of Tiny Thermometers

3. The "Blue Leak" Problem: A Draft in the Window

4. How Well Does It Work?

5. The Bottom Line: A Faster, Clearer View

More like this