PRUSSIC III -- ALMA and NOEMA survey of dense gas in high-redshift star-forming galaxies

TL;DR

This work presents a large ALMA/NOEMA survey (PRUSSIC) of mid-J dense-gas tracers (HCN, HCO+, HNC) in 11 gravitationally lensed DSFGs at z ≈ 1.6–3.2, detecting 34 transitions and ancillary isotopologue/CN lines. It derives excitation ladders, line ratios, and near-linear dense-gas scaling with FIR, finding DSFGs possess more excited dense gas and shorter dense-gas depletion times (~$τ_{dense} ≈ 24$ Myr) than local galaxies under a canonical α_HCN = 10. By combining line-based dense-gas masses with FIR-based SFRs, the study infers a substantial, source-to-source variation in dense-gas fractions (f_dense) and suggests a modest redshift evolution of the cosmic dense-gas mass density ρ_dense, consistent with the evolution of the total molecular gas density when a small dense-gas fraction is assumed. The analysis also identifies key systematics, including spatial offsets between tracers, differential lensing magnification, and large uncertainties in the HCN-to-dense-gas conversion factor, which can strongly affect inferred gas masses and depletion times. Overall, PRUSSIC demonstrates the feasibility and scientific payoff of systematic dense-gas surveys in the early Universe and provides benchmarks for upcoming cosmological simulations that explicitly treat dense molecular gas.

Abstract

Characterising the relationship between dense gas and star formation is critical for understanding the assembly of galaxies throughout cosmic history. However, due to the faintness of standard dense-gas tracers - HCN, HCO+, and HNC - dense gas in high-redshift galaxies remains largely unexplored. We present ALMA and NOEMA observations targeting HCN/HCO+/HNC (3-2) and (4-3) emission lines in eleven (mostly) gravitationally lensed dusty star-forming galaxies (DSFGs) at redshift z = 1.6--3.2. We detect at least one line in 10 out of 11 galaxies. Altogether, we detect 34 dense-gas transitions, more than quadrupling the number of extant high-redshift detections. Additionally, in two targets, we detect lower-abundance CO isotopologues 13^CO and C^18O, as well as CN emission. We derive excitation coefficients for HCN, HCO+ and HNC in DSFGs, finding them to be systematically higher than those in nearby luminous infrared galaxies. Assuming a canonical dense-mass conversion factor (alpha_HCN = 10), we find that DSFGs have shorter dense- gas depletion times (median 23 Myr) than nearby galaxies (~60 Myr), with a star-forming efficiency per free-fall time of 1-2%, a factor of a few higher than in local galaxies. We find a wide range of dense-gas fractions, with HCN/CO ratios ranging between 0.01 and 0.15. Finally, we put the first constraints on the redshift evolution of the cosmic dense-gas density, which increases by a factor of 7+/-4 between z = 0 and z = 2.5, consistent with the evolution of the cosmic molecular-gas density.

Figures (18)

• Figure 1: NOEMA Band-1/2 spectra for J1202. We detect the HCN, HCO$^+$, HNC (3--2) and (4--3) lines, alongside the CO(3--2) isotopologues and CN(3--2). The HNC(3--2) and (4--3) line profiles differ significantly from those of HCN and HCO$^+$. See Appendix \ref{['app:spectra']} for the spectra of remaining targets.
• Figure 2: NOEMA narrow-band images of the HCN/HCO$^+$/HNC lines for SDP.9, which is detected in all three transitions. For the remaining sources, see Appendix. Contours start at $\pm$2$\sigma$ and increase in steps of 2$\sigma$.
• Figure 3: Comparison of CO, HCN, HCO$^+$ and HNC linewidths in individual galaxies, ordered by increasing CO linewidth. The linewidths of different tracers are generally consistent within 1$\sigma$ uncertainties. Note the large discrepancies between HCN/HCO$^+$ and HNC linewidths in G04v1.40 and J1202.
• Figure 4: Correlation between far-IR luminosity and HCN/HCO$^+$(3--2) and (4--3) luminosities. Orange solid/dashed lines indicate the empirical trends from this work and Zhang2014 with $\pm$1$\sigma$ scatter; orange dotted lines indicate the empirical trend from Nishimura2024. Individual data points show $z=0$ galaxy-averaged Greve2009Papadopoulos2014Zhang2014Nishimura2024 and resolved observations Tan2018, and high-$z$ detections in individual galaxies Riechers2011Canameras2021 and this work (blue). We also show results from spectral stacks Reuter2022Hagimoto2023. Where appropriate, we correct for the lensing magnification. We also plot errorbars on high-z measurements; these are typically smaller than symbol size.
• Figure 5: HCO$^+$/HCN and HNC/CO luminosity ratios in our sample (solid circles). We also include data from Yang2023 for comparison (empty squares). The shaded regions indicate HCO$^+$/HCN and HNC/HNC ratios for the EMPIRE sample Jimenez2019. HCN is the brightest line in most of our targets. SDP.81 shows strong HCO$^+$ enhancement, potentially due to low metallicity Rybak2023. The HCN(3--2) emission line in SDP.11 is unusually bright, potentially due to AGN-driven chemistry.