NEW-MUSIC: Next-Gen Multi-Band Sub/millimeter Camera

• The paper presents a breakthrough multi-band KID polarimeter with a 2.4-octave range, achieving background-limited sensitivity and sub-arcminute resolution.
• It details innovative hierarchical phased-array antennas, integrated on-chip filterbanks, and broadband meta-AR structures that optimize beam coupling and reduce optical losses.
• The instrument supports diverse applications including time-domain astrophysics, SZ effect mapping in clusters, and redshift-complete extragalactic surveys.

The Next-generation Extended Wavelength Multi-band Sub/millimeter Inductance Camera (NEW-MUSIC) is a large-format, multi-band, kinetic-inductance-detector (KID)-based polarimeter designed to populate the focal plane of the Leighton Chajnantor Telescope (LCT) with simultaneous far-infrared, millimeter, and submillimeter imaging and spectroscopy across a 2.4-octave range (80–420 GHz) (Golwala et al., 3 Sep 2024, Huang et al., 12 Nov 2025). This instrument leverages hierarchical, phased-array slot-dipole antennas, photolithographic on-chip filters, and advanced lumped-element KIDs, integrating them with wideband antireflection (AR) meta-surfaces and rapid polarimetric modulation for background-limited sensitivity and sub-arcminute angular resolution. The wide simultaneous spectral coverage supports time-domain astrophysics, Sunyaev–Zeldovich (SZ) science in clusters and galaxy halos, and polarization mapping of dust emission in star-forming regions.

1. Optical and Spectral Architecture

NEW-MUSIC implements a six-band spectral camera covering 80–420 GHz (wavelengths from 3.8 to 0.7 mm), mapping a 14′ diameter field of view on the LCT (Golwala et al., 3 Sep 2024, Huang et al., 12 Nov 2025). The simultaneous six-band structure enables time-resolved spectral energy distributions (SEDs) and precision polarimetry across a dynamic range of 1:5.25 (2.4 octaves). Table 1 summarizes key band parameters:

Band	Center freq. (GHz)	Band edges (GHz)	Δf/f₀	Science driver
B1	90	77.5–106	0.32	SZ/low-z clusters
B2	150	133.5–172.5	0.26	Dust in star formation
B3	230	201–246	0.20	High-z dusty galaxies
B4	290	270–310	0.14	Mid-band continuum/SZE
B5	350	335–360	0.07	Narrow atmospheric window
B6	400	390–411	0.05	Highest-frequency channel

Fractional bandwidths, Δν/ν, are matched to key atmospheric windows and science requirements. The optical design employs hierarchical, back-illuminated slot-dipole arrays, re-imaging optics, and a cold Lyot stop at 4 K. The beamwidth for each band is set by the relation θ ≃ 1.22 λ/D, where D = 10.4 m (LCT aperture). Field re-imaging to the cold focal plane uses a silicon relay lens system with AR treatment (Golwala et al., 3 Sep 2024, Kovács et al., 4 Jan 2025).

2. Hierarchical Phased-Array Antenna and Filterbank Systems

Each focal-plane “pixel” begins with a 16×16 slot-dipole Nb antenna array (slot length 1.664 mm, width 18 μm), back-illuminated through high-resistivity Si (Huang et al., 12 Nov 2025). The slot array is hierarchically summed using microstrip trees—Levels 0, 1, and 2—so that the effective aperture and pitch scale with wavelength, maintaining near-optimal coupling and angular resolution across the spectral span.

Summed signals from the antenna are routed through staged, photolithographic on-chip filterbanks. These include 5th-order Chebyshev low-pass filters (LPF) and 3rd/5th-order Chebyshev bandpass filters (BPF), distributed at each hierarchy level. The filterbank design uses the following prototype transformations:

$$L_k = \frac{Z_0\,g_k}{\omega_c}, \qquad C_k = \frac{g_k}{Z_0\,\omega_c}$$

for LPFs with cutoff ω_c, and analogous transforms for BPFs. Transmission lines at the interfaces employ impedance tapers (Z₀ = 37–54 Ω) to maintain S₁₁ < –20 dB return loss. Typical filterbank insertion loss is <1 dB, with band isolation >20 dB (Huang et al., 12 Nov 2025).

Integrated beamforming produces a main lobe FWHM scaling as λ/(16 × p), with p = slot pitch (104 μm). Simulated infinite-array efficiency exceeds 80% throughout the 80–420 GHz range, with ≤1% main-beam ellipticity (Huang et al., 12 Nov 2025). This array–filter architecture supports simultaneous, on-chip selection of six spectral bands per spatial pixel.

3. Detector Implementation: MS-PPC-LEKIDs

NEW-MUSIC employs microstrip-coupled, parallel-plate capacitor, lumped-element kinetic inductance detectors (MS-PPC-LEKIDs)—a focal-plane architecture that decouples the optical/resonant circuit design and enables high fill-factor, low-noise, and background-limited operation (Golwala et al., 3 Sep 2024, Huang et al., 12 Nov 2025). Each detector incorporates an Al or AlMn meander inductor, sandwiched between Nb ground and top wiring planes, and parallel-plate capacitors using low-loss α-Si:H dielectric (tan δ ≈ 10⁻⁵).

Key equations governing detector resonance and sensitivity include:

$f_r = \frac{1}{2\pi\sqrt{LC}}$

$$\mathrm{NEP}_\gamma^2 = 2h\nu P_{\rm load} + 2 \frac{P_{\rm load}^2}{\Delta\nu}$$

$$\mathrm{NEP}_{\rm GR} = \frac{2\Delta}{\eta_{\rm pb} \sqrt{P_{\rm load}/(RV)}}$$

where L, C are the total inductance and capacitance; Δ is the SC gap; η_pb the pair-breaking efficiency; R the recombination constant; and V the inductor volume.

Coupling from microstrip to KID is accomplished via a dedicated capacitor (C_c), with resonator coupling Q governed by:

$Q_c = \frac{\omega_r L_{\rm KID}}{R_0 C_c^2}$

Typical f_r is 1–2 GHz, with internal Q_i > 10⁵ at 100 mK. Simulated NEP is ≈10⁻¹⁸ W/√Hz, matching background-limited expectations.

4. Optical Coupling, Antireflection Engineering, and System Efficiency

A three- or four-layer etched-silicon AR meta-structure is applied to the back-illuminated focal plane, reducing reflection losses to <1% over the full 80–420 GHz band (Golwala et al., 3 Sep 2024). The effective refractive index profile is chosen to match the free-space–to–silicon impedance, implementing moth‐eye or quarter-wave matching principles:

$t = \frac{\lambda}{4n_{\rm layer}}$

$R = \left(\frac{n_1 - n_2}{n_1 + n_2}\right)^2$

where t is the thickness of the matching layer. System-level coupling efficiency is determined as

$$\eta_{\rm opt} = \frac{P_{\rm absorbed}}{P_{\rm incident}}$$

Combining antenna, filter, AR, and microstrip losses, end-to-end optical efficiency per pixel is 75–90% (Huang et al., 12 Nov 2025). The array fill factor and beam-matching efficiency are optimized by adjusting pixel pitch d and focal ratio F/# according to

$$\eta_{\rm fill} \approx \left(\frac{d}{F/\#\,\lambda}\right)^2$$

$$\eta_{\rm coupling} = \frac{\left|\iint A(\theta,\phi) B^*(\theta,\phi) d\Omega\right|^2}{\iint |A|^2 d\Omega \iint |B|^2 d\Omega}$$

where A and B are the antenna and telescope beam patterns.

5. Readout, Multiplexing, and Focal-Plane Scaling

The superconducting KID arrays are frequency-domain multiplexed, supporting O(10³–10⁴) detectors per readout bandwidth of a few GHz (Janssen et al., 2014, Kovács et al., 4 Jan 2025, Duan et al., 2020). Room-temperature ROACH- or RFSoC-based systems with ADC/DAC bandwidths of 1–2 GHz enable simultaneous acquisition of all spectral and spatial channels. At the observed NEP, channel crosstalk is suppressed below –20 dB by microstrip layout and ground fencing (Huang et al., 12 Nov 2025).

To scale from prototype to full science focal planes, tiling of detector wafers and multi-train cryogenic wiring are employed (Kovács et al., 4 Jan 2025, Golwala et al., 3 Sep 2024). The graduated rollout strategy begins with a 25% focal plane using the legacy MUSIC cryostat and optics, with subsequent quadrupling for full 1,600-pixel wide-area surveys. Pixel counts per band (full instrument) are: B1–B2 (64), B3–B4 (256), B5–B6 (1024) (Golwala et al., 3 Sep 2024).

6. Key Science Drivers and Performance Metrics

NEW-MUSIC’s science goals are enabled by the combination of broad simultaneous spectral coverage, background-limited sensitivity, and polarized focal-plane mapping:

• Six-band imaging polarimetry: SED and Stokes Q,U per beam for mapping dust and synchrotron sources.
• Thermal and kinematic SZ effect mapping in clusters: ΔI(ν) = I₀ y f(ν), permitting precise constraints on electron pressure and velocities in clusters.
• Time-domain studies: Capture rapid transients and variable sources with broadband SED tracking.
• Cosmic star-formation: Volumetric spectroscopic mapping via molecular lines and dust emission, with survey yields on 10-m telescopes reaching ≈23,000 galaxies/year at 1.2 mm (Kovács et al., 4 Jan 2025).

7. Technological Innovations and Development Trajectory

• MS-PPC-LEKID architecture: Decoupling resonance and optical coupling, allowing suppression of TLS noise and flexible band assignment (Golwala et al., 3 Sep 2024).
• Hierarchical phased arrays: Efficient wideband beam synthesis and scalable focal-plane tiling (Huang et al., 12 Nov 2025).
• Integrated filterbanks: Compact on-chip filtering with precise band definition and minimal loss.
• Meta-AR structures: Broadband silicon AR for maximized system throughput.
• Modular, open-source collaboration: All hardware, firmware, and software for NEW-MUSIC development is managed via shared, revision-controlled repositories, with forking and collaborative improvement as a guiding principle (Kovács et al., 4 Jan 2025).
• Phased deployment: Initial quarter-focal-plane science runs, extended to full-scale surveys as detector production and readout mature (Golwala et al., 3 Sep 2024).

This technological and organizational model is intended to maximize survey speed, minimize confusion and systematic error, and enable rapid progress in time-domain astrophysics, SZ cosmology, and polarized mapping of star-formation regions. A plausible implication is that NEW-MUSIC provides the first practical path to simultaneous, confusion-limited, redshift-complete extragalactic surveys in the sub/millimeter regime, and supports rapid, target-of-opportunity response for transient phenomena across broad spectral bands.