There and back again? Neutral outflows in z~3.5 quiescent galaxies

Abstract

Neutral gas outflows play a crucial role in the baryon cycle of galaxies, their properties provide key insights into the transition from star formation to quiescence. In this work, we investigate the neutral gas outflow of 23 massive ($M_\star = 10^{10.1-11.6}\,\rm M_\odot$) quiescent galaxies (QGs) at z=2.82--4.61, selected from the JWST NIRSpec (R~1000) and NIRCam program DeepDive. We trace the neutral gas outflows using the NaI Doublet absorption lines and detect excess NaI D in 13/23 (57%) targets, of which 7/23 (30%) show blueshifted absorption with velocity offsets $|Δv|$ >~ 150 km/s. The z ~ 3.5 targets have $Δv$ similar to those of their local counterparts; they are also equivalent when compared in SFR--$Δv$ space. We derive mass outflow rates and identify the most extreme neutral gas outflow rate $\log(\dot M_{\rm out} / \mathrm{M_\odot \, yr}^{-1})=2.68\pm0.27$ beyond the local Universe, coincident with an X-ray AGN. For all NaI D detected systems, the inferred mass outflow rate can, in principle, suppress ongoing star formation; however, the outflows are unlikely to escape their hosts, suggesting recycling on relatively short timescales (~3--180 Myr), depending on the assumed potential and launching radius. All NaI D detected targets occupy the LI(N)ER region of the BPT diagram and/or are X-ray detected, but we find no strong correlation between ongoing AGN and the neutral outflow: 2/4 broad-line/X-ray AGNs are NaI D undetected -- yet, the outflows can be powered by fossil/episodic AGNs, and one broad-line target shows a possible P-Cygni profile that indicates strong outflows. As neutral outflows alone are not able to permanently quench star formation by removing gas in our sample at z ~ 3.5, the presence of gas cycling in and out of massive passive systems may instead be the signature of feedback-regulated quenching-maintenance processes.

Figures (6)

• Figure 1: DeepDiveNai D detected spectra, including tentative detections. The first and third rows show the median-normalized spectra in black and best-fit stellar continuum in red; the second and fourth rows show the continuum-normalized (i.e., flux/continuum) spectra and the MCMC best-fit for Nai D absorption. The spectral S/N of each normalized spectrum is labeled on the lower left, and the target's DD-ID is labeled on the upper right: the cyan labels represent systematic Nai D absorption; the hollow/red labels represent (tentatively) detected blueshifted absorption. The two panels in the lower right show fittings with and without redshifted emission to fit the possible P-Cygni profile of DD-229. Additionally, DD-134 is "Jekyll" of the merging pair "Jekyll & Hyde" schreiberJekyllHydeQuiescence2018; DD-185 and DD-196 are reported by itoMergingPairMassive2025 as a merging pair with ids 61167 and 61168, respectively.
• Figure 2: DeepDive QGs plotted on the redshift--stellar mass plane. The Nai D blueshift (tentatively) detected targets are in (hollow) red circles, and the systemic targets are in cyan squares. The Nai D excess absorption non-detections are plotted as gray triangles. No targets are detected below a stellar mass threshold of $10^{10.5}M_\odot$. The median S/N ratios for galaxies above and below this threshold are indicated.
• Figure 3: $\mathrm{SFR_{100\,Myr}}$ as a function of stellar mass and Nai D (or Mgii for the gray markers) velocity offset. The red hollow/filled circles are the blueshifted Nai D (tentatively) detected DD targets itoMergingPairMassive2025, and the cyan squares are the systemic/redshifted Nai D targets. The big orange and small yellow triangles represent the local quiescent (stacked with SFR upper limits) and post-starburst samples in sunEvolutionGasFlows2024, respectively. The brown big triangle represents the recently quenched source with extreme outflow in sunExtremeNeutralOutflow2025. The green diamonds represent the Nai D blueshifted sample from daviesJWSTRevealsWidespread2024. The blue stars are the two recently quenched high-z galaxies reported by valentinoGasOutflowsTwo2025. The gray contours represent the homogenized literature compilation of Mgii outflows across redshifts davisExtendingDynamicRange2023. The location of two $z\sim3$ recently quenched sources in manExquisitelyDeepView2021 is marked by gray squares. On the left panel, we plot the star-forming main sequences popessoMainSequenceStarforming2023 at $z=3.7$ (median DeepDive redshift), $z=1$, and $z=0$ as light red/gray/orange lines, respectively.
• Figure 4: The Nai D velocity offset $\Delta v$ as a function of stellar mass (Left), and post-burst age (time passed since 90% of $M_\star$ formed, Right). The blueshifted Nai D (tentatively) detected DeepDive targets are plotted as hollow/red circles, and the systemic targets as cyan squares. As a comparison, sunEvolutionGasFlows2024 local PSBs are plotted as yellow triangles. The median stack of the DeepDive targets (excluding the merging pair, including the systemic targets) and of the sunEvolutionGasFlows2024 PSBs (only those in the same mass and age bin as DeepDive targets) are marked with colored crosses. The DeepDive targets show similar $\Delta v$ as their local counterparts, and there is no clear trend in either of the figures, given the limited sample size.
• Figure 5: Left: The $\rm SFR_{100Myr}$ vs. the mass outflow rate. All the DeepDive targets show higher mass outflow rates than the ongoing SFR, implying the neutral gas outflow is able to suppress the ongoing star formation. Right: The dynamical mass escape velocity at $r_\mathrm{eff}$ vs. the outflow velocity. Most DeepDive targets show an outflow velocity smaller than the escape velocity, meaning that the blueshifted neutral gas is not likely to escape the galaxy. Three targets show $v_{\rm out}> v_{\rm esc}$, two of which are the itoMergingPairMassive2025 merging pair, and the other one is the X-ray source DD-236 with an extreme mass outflow rate (Section \ref{['sec: mass outflow rate']}).