Spectral characterization and performance of SPT-SLIM on-chip filterbank spectrometers

TL;DR

Line Intensity Mapping (LIM) seeks three-dimensional maps of the cosmic structure by tracing emission lines, and SPT-SLIM demonstrates on-chip filterbank spectrometers as a pathfinder in the 120-180 GHz window to constrain CO emission. The paper develops a spatial-domain fitting approach to recover unbiased filterbandpasses from a low-resolution on-site Fourier Transform Spectrometer, enabling extraction of central frequencies and bandwidths beyond the instrument's raw spectral resolution. On-sky results show an average spectral resolution of about 34 with band centers downshifted by ~10 GHz, dominated by dielectric losses in SiN rather than design expectations. The work establishes a viable, scalable on-chip spectrometer approach for LIM, identifies dielectric loss as the key bottleneck, and points to material and fabrication improvements needed to approach the designed performance for CO LIM science.

Abstract

The South Pole Telescope Shirokoff Line Intensity Mapper (SPT-SLIM) experiment is a pathfinder for demonstrating the use of on-chip spectrometers for millimeter Line Intensity Mapping. We present spectral bandpass measurements of the SLIM spectrometer channels made on site using a Fourier Transform Spectrometer during SPT-SLIMs first deployment the 2024-2025 austral summer observing season. Through this we demonstrate a technique for measuring the narrow band passes of the SPT-SLIM filterbanks that improves beyond the intrinsic resolution of a Fourier Transform Spectrometer.

Figures (7)

• Figure 1: A schematic diagram of the full SPT-SLIM focal plane robson2024.
• Figure 2: A spatial-domain fit of the interference signal from one of the SLIM detectors (top). The spectrum of the corresponding spectrometer channel measured by the FTS and the spectral profile obtained from the spatial-domain fit are shown in the spectral (Fourier) domain (bottom). Note that the measured profile is smeared to a wider spectral bandpass due to the instrumental line shape of the low-resolution FTS.
• Figure 3: The disagreement between simulated values and measurements of the central frequency (top) and spectral resolution/bandpass (bottom) of a filter extracted from spatial-domain fitting using simulated FTS measurements (see Eqs. \ref{['eq:fitEq']}--\ref{['eq:filterR']}). Simulated measurements were generated with a detector noise model at different signal-to-noise ratios (SNRs) and at different FTS resolving powers with each bin/pixel showing the mean of 20 simulated measurements. In each panel, the ratio of the resolving power of the South Pole FTS at 150 GHz to the $R=100$ simulated filter resolution is marked by a vertical dashed line. The contours in the bottom figure mark uncertainty thresholds in values of $R$ determined from this method determined from within the parameter space.
• Figure 4: The simulated spectral band pass of a filter with and without a modeled asymmetry (see Eq. \ref{['eq:assymProf']}).
• Figure 5: The fitted resolution, $R$, of the on-sky submodules extracted from measurements with the South Pole FTS (see Eq. \ref{['eq:filterR']}). Error bars represent the uncertainty determined from the simulated data shown in Fig. \ref{['fig:simulation']} at the corresponding SNR of the interferogram measured with the FTS.