Weak Evolution of Cosmic Atomic Hydrogen over the Past 4.5 Billion Years

Technical Summary: Weak Evolution of Cosmic Atomic Hydrogen over the Past 4.5 Billion Years

Problem Statement
The cosmic star formation rate density (CSFRD) has declined sharply toward the present day. While this decline is widely linked to the observed decrease in the cosmic molecular gas ($H_2$) density, the role of the more extended atomic hydrogen ($H\text{\textsc{i}}$) reservoir remains uncertain. Previous measurements of the cosmic $H\text{\textsc{i}}$ density ($\Omega_{H\text{\textsc{i}}}$) have relied on Damped Ly$\alpha$ (DLA) absorbers or spectral line stacking from 21-cm surveys. However, DLA measurements probe high-column-density cross-sections, while existing 21-cm emission studies at intermediate redshifts ($z > 0.1$) suffer from limited survey areas, small sample sizes, and significant cosmic variance. Consequently, there is a lack of self-consistent, wide-area 21-cm emission measurements to determine if the late-time decline in CSFRD is accompanied by a comparable depletion of the galaxy-associated $H\text{\textsc{i}}$ reservoir.

Methodology
This study leverages a unique synergy between the Five-hundred-meter Aperture Spherical radio Telescope (FAST) and the Dark Energy Spectroscopic Instrument (DESI) to measure $\Omega_{H\text{\textsc{i}}}$ over $0 < z < 0.41$ (the past 4.5 Gyr).

• Data Sources: The analysis combines $H\text{\textsc{i}}$ spectra from the FAST All-Sky $H\text{\textsc{i}}$ survey (FASHI) with optical spectroscopy from the DESI Bright Galaxy Survey (BGS). This cross-match yields a statistically representative sample of $\sim2.5$ million galaxies covering $\sim12,000$ deg$^2$.
• Stacking Technique: Due to the intrinsic faintness of the 21-cm line at these redshifts, individual detections are not feasible. The authors employ spectral line stacking of $r$-band magnitude-limited galaxies. Galaxies are grouped into bins of stellar mass ($M_*$) to derive the average $H\text{\textsc{i}}$ gas fraction, $\langle f_{H\text{\textsc{i}}}(M_*) \rangle \equiv \langle M_{H\text{\textsc{i}}}/M_* \rangle$.
• Redshift Intervals: To mitigate Radio Frequency Interference (RFI), the analysis focuses on four specific redshift intervals: $0 < z < 0.09$ and $0.25 < z < 0.41$, with mean redshifts of 0.033, 0.069, 0.281, and 0.358.
• Systematic Corrections:
  • Confusion Correction: Given FAST's large beam size ($\sim2.9$ arcmin), a "clean sample" approach was adopted. Galaxies with neighbors within specific angular and velocity separations were excluded to prevent signal contamination, removing $\sim33.4\%$ of the original sample.
  • Luminosity Bias: To correct for the flux-limited nature of the BGS sample (which biases toward luminous, gas-rich galaxies), the authors developed a method to model the dependence of $f_{H\text{\textsc{i}}}$ on both stellar mass and $r$-band luminosity ($f_{H\text{\textsc{i}}}(L_r|M_*)$). This analytical correction allows for the recovery of the intrinsic $f_{H\text{\textsc{i}}}$ distribution across the full mass range.
  • Forward Modeling: A mock catalogue based on the IllustrisTNG100-1 simulation was constructed to quantify systematic uncertainties arising from sample selection, fiber assignment, and residual confusion. This forward model provided conservative estimates of systematic offsets.

Key Results

• Stellar Mass Dependence: The average $H\text{\textsc{i}}$ gas fraction decreases systematically with increasing stellar mass, well-described by a broken power-law form rather than a single power-law "main sequence."
• Weak Evolution: At fixed stellar mass, the average $H\text{\textsc{i}}$ gas fraction evolves by less than 0.2 dex from $z=0.41$ to the present day. This near-invariance holds across the entire probed mass range ($10^9 M_\odot$ to $10^{11.5} M_\odot$), indicating that the weak evolution is a population-wide trend rather than a result of compensating changes in different galaxy sub-populations.
• Cosmic $H\text{\textsc{i}}$ Density ($\Omega_{H\text{\textsc{i}}}$):
  • The raw data indicates a decrease in $\Omega_{H\text{\textsc{i}}}$ by a factor of $1.35 \pm 0.10$ over the past 4.5 Gyr.
  • After applying conservative systematic corrections derived from the forward model, the inferred decline is reduced to a factor of $1.12 \pm 0.10$.
  • This measured decline is significantly weaker than the decline in CSFRD, which decreases by a factor of 2.46 over the same period.
• Comparison with Models: The results are consistent with the SIMBA hydrodynamical simulation (which predicts a modest decline) but contradict the TNG simulation (which predicts an approximately constant $\Omega_{H\text{\textsc{i}}}$). The measurements are also systematically higher and up to three times more precise at $z > 0.25$ compared to previous 21-cm emission studies.

Significance and Claims
The paper claims that these quantitative differences rule out the rapid depletion of galaxy $H\text{\textsc{i}}$ as the primary driver of the late-time CSFRD decline. Instead, the results suggest that the decline in star formation is driven by a reduced conversion of cold atomic gas into molecular gas and stars, rather than a depletion of the atomic reservoir itself.

The weak evolution of $\Omega_{H\text{\textsc{i}}}$ and $f_{H\text{\textsc{i}}}(M_*)$ implies that the late-time $H\text{\textsc{i}}$ reservoir remains close to a regulated balance (governed by accretion, recycling, and feedback), even as the throughput into star formation decreases. The authors posit that the decline in gas supply acts as an upstream driver, but its impact is mediated through the baryon cycle—specifically affecting the conversion to molecular gas and subsequent star formation—rather than appearing as a strong depletion of $H\text{\textsc{i}}$.

This work provides a stringent benchmark for galaxy formation models, requiring them to reproduce a scenario where the atomic gas reservoir remains nearly invariant while molecular and star-forming components evolve strongly. The study establishes that the decline in cosmological gas supply does not produce a strong mass-dependent depletion of $H\text{\textsc{i}}$, offering new constraints on gas accretion, phase conversion, and star-formation regulation mechanisms.