Neutral Atomic Hydrogen in a Star-forming Galaxy 7 Billion Years Ago

TL;DR

This work reports the first direct detection of HI $21\rm\,cm$ emission from a typical star-forming galaxy at intermediate redshift ($z=0.725$) using gravitational lensing and the MeerKAT array. The Dragon Arc harbors $M_{ m HI}=10^{9.66^{+0.16}_{-0.19}}\,M_ot$ with $\text{FWHM}=205^{+66}_{-48}$ km s$^{-1}$, yielding a depletion time $t_{\rm dep,HI}=1.16^{+0.98}_{-0.64}$ Gyr, shorter than local analogs and implying rapid gas replenishment is required to sustain star formation. The results align with $z\sim0.7$ scaling relations and, when compared to SIMBA simulations, reveal distinct roles for radiative AGN winds and jets in quenching galaxies, underscoring the importance of feedback in regulating the cosmic HI reservoir. This demonstration shows that lensing plus modern cm-wave facilities can extend direct HI studies to typical galaxies beyond the local universe, enabling statistical samples to trace the evolution of atomic gas with cosmic time.

Abstract

Neutral atomic hydrogen (HI) constitutes a key phase of the cosmic baryon cycle, bridging the ionised circumgalactic medium and the star-forming molecular gas. Yet, nearly 75 years after its discovery, direct views of HI through its 21cm emission line remain largely confined to the nearby Universe. Indirect measurements and statistical analyses indicate little evolution in the comoving HI density over the past 10 billion years, in stark contrast to the order-of-magnitude decline in the cosmic star-formation rate density over the same epoch. Resolving this tension requires direct HI measurements in individual, representative galaxies at earlier times. Here we report a detection of HI 21cm emission from the Dragon Arc, a gravitationally lensed main-sequence star-forming galaxy at $z=0.725$, observed 6.6 billion years in the past with the MeerKAT radio telescope. The inferred intrinsic HI mass, $M_{\rm HI}=10^{9.66^{+0.16}_{-0.19}}\,\mathrm{M}_\odot$, and velocity width of $205^{+66}_{-48}\,\mathrm{km\,s^{-1}}$ are consistent with expectations from scaling relations for local star-forming galaxies. The resulting HI depletion time of $1.2^{+1.0}_{-0.6}\,\mathrm{Gyr}$ is significantly shorter than the $\sim5-10$ Gyr, measured locally for comparable galaxies. This indicates that the galaxy must rapidly replenish its atomic gas reservoir to remain on the star-forming main sequence. This detection demonstrates that strong gravitational lensing, combined with modern cm-wave facilities, can now reveal the HI reservoirs of typical galaxies well beyond the local Universe, opening a new path toward statistical samples that will directly trace the evolution of the cosmic atomic gas supply.

Figures (8)

• Figure 1: Multiband HST image of the Dragon Arc. Overlaid on the multiband image (F814W, F606W, F435W) we show H i contours at $134$, $143$, and $150\,\rm{Jy}\,\rm{Hz}\,\rm{beam}^{-1}$ ($4$, $4.25$, and $4.5\,\sigma$) from the total intensity H i map generated from $4\times266\,\rm{kHz}$ channels. A 31" scale bar (bottom left) represents the semi-major axis of the smoothed total intensity map restoring beam.
• Figure 2: MeerKAT neutral hydrogen (HI) emission spectrum towards the Dragon Arc. The H i centred at $z=0.7252$ is shown in blue. The thin red vertical line represents the optical spectroscopic redshift with the shaded red band corresponding to the velocity dispersion of the $[\rm{O}\,{II}]$ emission linePatricio_2018. The median posterior of a single Gaussian model is shown in black. Grey vertical bars indicate the RMS of the flux distributions of 500 randomly placed identical apertures in each channel.
• Figure 3: HI mass as a function of redshift for direct detections. The Dragon Arc detected in this work is shown in red. The VLA detectionFernandez_2016 is shown by a yellow triangle, the FASTXi_2024 galaxies are shown with purple diamonds, and the MIGHTEEJarvis_2025 detections are shown with green circles. Shaded regions are defined by the 1 and 99 percentiles of the integrated fluxes of the samples for the respective surveys.
• Figure 4: Observational constraints on the HI gas depletion time in main-sequence galaxies across redshift space, compared with the SIMBA simulation. Results from spectral-line stacking are shown with triangle, square, and circle markersBera_2019Chowdhury_2020Chowdhury_2021. The Dragon Arc result reported here is indicated with a star. We calculate a mean HI depletion time and HI mass for local SFMS galaxies from the xGASS sampleSaintonge_2017Catinella_2018. SFMS galaxies from the flagship SIMBA simulation, $(100\,h^{-1}\,\rm{Mpc})^3$ box, are plotted with points. All points and markers are colourised by redshift.
• Figure 5: Scaling relation comparison between local observations and the SIMBA cosmological hydrodynamical simulations.H i-to-stellar mass fraction (left) and SFR (right) as a function of stellar mass. Cyan and teal contours represent SIMBA main sequence galaxies at $z=0$ and $z=0.7$ respectively, where the former are the individually tracked descendants of the latter. The dashed purple contours represent galaxies in the xGASS ($z\sim0$) sampleCatinella_2018. The red star indicates the estimated intrinsic properties of the Dragon Arc. Contours enclose $95\,\%$ of the data.