THRILS -- The High-(Redshift+Ionization) Line Search: Program Description & Redshift Catalog

TL;DR

THRILS delivers the deepest rest-frame optical spectroscopy to date for high-$z$ galaxies in the CEERS/EGS field using two JWST/NIRSpec pointings with the G395M grating ($>\!8$ hr exposures). By combining an extensive ancillary catalog with MPT-driven target weighting, it yields 89 robust $z_{spec}$ measurements and characterizes 81 non-detections, providing explicit detection thresholds and illustrating the enhanced line sensitivity for high-ionization features, BLAGN, and late-time star-formation history indicators. The program demonstrates three core science goals: testing IMF/top-heavy scenarios via high-ionization lines, constraining the nature of puzzling BLAGN/LRDs through faint ISM lines, and recovering SFHs of massive $z>4$ galaxies through deblended emission lines and Balmer features. Collectively, THRILS informs IMF evolution, black-hole growth, and early mass assembly, highlighting the value of deep, high-resolution spectroscopy for understanding galaxy formation in the first billion years of the universe.

Abstract

To date, many spectroscopic confirmations of z>7 galaxies have been obtained using JWST/NIRSpec prism observations, with most of their physical properties inferred from these observations and corresponding imaging. What is needed are higher-resolution spectra at deeper depths to study these sources in detail. We present The High-(Redshift+Ionization) Line Search (THRILS) program: deep (>8 hr) observations in two pointings of JWST/NIRSpec G395M spectroscopy to 1) probe high ionization spectral features in z>8 galaxies that are indicative of top-heavy initial mass functions or growing massive black holes, 2) search for accreting supermassive black holes in typical galaxies at z~4-9 through broad Balmer line emission, and 3) probe the stellar-mass growth histories of massive galaxies. We include spectroscopic redshift measurements for 89 sources from the THRILS data, as well as a detection threshold for the full and half depth integration times of the program.

Figures (6)

• Figure 1: The two THRILS NIRSpec pointings (green) overlaid on the HST/F160W mosaic image from CANDELS grogin11koekemoer11, with the CEERS NIRCam coverage outlined in purple bagley22bfinkelstein25. The parallel NIRCam imaging from NIRSpec DDT program 2750 haro23 is shown in dark purple.
• Figure 2: Spectroscopic vs photometric redshift. (top) $z_{spec}$ vs $z_{phot}$ for all 89 sources with measured redshifts in THRILS. The robust redshift measurements (grade = 3) are shown as blue circles, while the lower confidence measurements (grade $<3$) are shown as diamonds. We annotate all $z_{phot}$ derived from HST-only data with red borders. (bottom) Residual of $z_{spec} - z_{phot}$, zoomed into sources around zero for clarity.
• Figure 3: Detections and magnitude depths. Plot of the source magnitude at 4.5 $\mu$m from JWST/NIRCam F444W (or Spitzer Ch2 when not observed with JWST) vs redshift for the THRILS survey. (left) The THRILS primary science targets, with full integration time for each pointing ($>$30 ks). The THRILS spectroscopic redshifts ($z_{spec}$) are shown in blue, with high confidence (circles, grade = 3) and lower confidence (diamonds, grade = 1--2) redshifts. For the primary targets that yielded non-detections (orange squares), we use their photometric redshift ($z_{phot}$) as the displayed redshift. (right) The THRILS filler targets or contaminating sources, which had half the integration time (or less) per pointing. The THRILS spectroscopic redshifts are again shown in blue. The non-detected sources are separated into half integration (orange, $\sim$15 ks; three dithers) and less than half (red, $<$12 ks; one or two dithers only). The grey shaded regions show the NIRSpec detection threshold for non-EELGs in both panels, highlighting the depth achieved at both exposure times and advocating for more deep NIRSpec programs.
• Figure 4: Spectrum of a source from THRILS Science Case 1. THRILS-19512 is a $z=8.7146$ galaxy initially observed with NIRSpec by the CEERS program Finkelstein.2024. The 2D spectrum from THRILS is shown across the top, and the 7-filter color image from CEERS is shown in the inset to the left with the THRILS NIRSpec shutter overlaid. The publicly available 1D CEERS spectrum (light blue, Arrabal Haro et al. in prep) has an exposure time of 52.1 min, while the THRILS optimally extracted 1D spectrum (dark blue) has an exposure time of 8.85 hours. Inset plots highlight the detections of He i$\lambda$3890 and H$\delta$$\lambda$4103, which were not possible in the shallower data.
• Figure 5: Spectrum of an AGN from THRILS Science Case 2. THRILS-46155 is a BLAGN at $z=3.5179$ that was first observed for 47.7 min by the RUBIES program degraaff25. The 2D spectrum from THRILS (8.4 hr) is shown across the top, and the 7-filter color image from CEERS is shown in the inset to the right with the THRILS NIRSpec shutter overlaid. The publicly-available RUBIES data (light green) detects the broad component of H$\alpha$$\lambda$6564 taylor24, but the deeper THRILS data (dark green) is needed to measure the fainter [S ii] $\lambda$$\lambda$6718,6733 and [S iii] $\lambda$$\lambda$9069,9531 doublet lines shown in the insets.