GA-NIFS: high prevalence of dusty and metal-enriched outflows in massive and luminous star-forming galaxies at $z\sim3-9$

TL;DR

This study uses JWST/NIRSpec IFU to spatially resolve ionized outflows in 40 regions across 15 massive star-forming galaxies at $z\sim3-9$, targeting systems with no clear AGN. Through one- and two-component Gaussian fits to emission lines (notably [O III] $\lambda5007$ and H$\alpha$), the authors map outflow kinematics, dust extinction, and gas-phase metallicities, finding that outflows are common but generally do not unbind gas from their hosts ($v_{\rm out}<v_{\rm esc}$; $\eta\lesssim1$ in most regions). They show that outflowing gas is more dust-attenuated by $\sim0.59$ mag and enriched by $\sim0.13$ dex in oxygen compared to the host ISM, consistent with redistribution of dust and metals within galaxies. A significant positive correlation between outflow velocity and SFR emerges when including prior high-$z$ studies, indicating a universal scaling of ionized outflow speeds with star-formation activity across cosmic time. Overall, ejective feedback from ionized outflows appears inefficient at suppressing star formation in these massive high-$z$ systems but plays a meaningful role in dust and metal redistribution, contributing to the chemical and dust evolution of early galaxies.

Abstract

We present a search for and characterization of ionized outflows in 15 star-forming systems at $z\sim3-9$ with no evidence of Active Galactic Nuclei (AGN), observed with JWST/NIRSpec IFU as part of the GA-NIFS program. The targets often show satellites and complex substructure, from which we isolate 40 galaxies/regions. The sample probes the high-mass end of the galaxy population, with most sources having log$_{10}$~(M$_\star$/M$_\odot$)=$9.5-11$, extending previous studies on high-z star formation driven outflows that mainly focused on lower-mass galaxies. Using the [OIII]5007 and H$α$ emission lines, we identify broad kinematic components consistent with galactic outflows in 14 galaxies/regions. We find that the outflowing gas is more dust attenuated (by $A_{\rm V}$$\sim0.59$ mag on average) and metal-enriched (0.13 dex) than the interstellar medium (ISM) of the host galaxies, but its velocities are insufficient to escape the galaxies and reach the circumgalactic medium, suggesting that outflows mainly redistribute dust and metals around their hosts. The outflows identified in this study display velocity dispersions within $σ_{\rm out}=130-340$~km~s$^{-1}$ and outflow velocities $v_{\rm out}=170-600$~km~s$^{-1}$, and, when combined with less luminous and less massive star-forming galaxies from previous works, reveal a statistically significant correlation between $v_{\rm out}$ and star formation rate (SFR). The typically low mass-loading factors ($η=\dot{M}_{\rm out}/SFR$$\leq1$, in 9 out of 14 the outflows) indicate that these outflows do not strongly suppress star formation. Overall, our results suggest that ejective feedback via ionized outflows is inefficient in massive, luminous star-forming galaxies within the first 2 Gyr of the Universe.

Figures (37)

• Figure 1: Maps of integrated fluxes in $[\mathrm{O}\textsc{ iii}]\,\lambda 5007$ for all the galaxy systems in our sample, with the exception of HFLS3$\_$g1 for which we show integrated $\mathrm{H}\alpha$ fluxes. Orange apertures mark small regions at the peaks of emission of the different (sub-) systems to constrain the dominant ionization mechanisms (see Figure \ref{['fig:integrated_line_diagnostics']} and Section \ref{['subsec:diagnostics']}). Pink apertures mark the larger regions used to study their extended emission. Blue contours correspond to the continuum emission at 2500-3000Å rest-frame at different levels, starting from $3\times$ the standard deviation of the background. Note that for HFLS3$\_$g1 at z=3.5, these contours also include the emission from the z=6.3 system.
• Figure 2: Emission-line diagnostics for the galaxies in our sample obtained for the apertures highlighted in Figure \ref{['fig:integrated_oiii_maps']}. Left: BPT-NII and BPT-SII diagrams baldwinClassificationParametersEmissionline1981aveilleuxSpectralClassificationEmissionLine1987. Dashed kauffmannHostGalaxiesActive2003b and dotted black lines kewleyTheoreticalModelingStarburst2001b demarcate the separation between ionization coming from star formation and AGN activity based on studies of local galaxies. The dashed green lines in both plots have been recently proposed by scholtzJADESLargePopulation2025 to conservatively separating AGN galaxies (see Sect. \ref{['subsec:diagnostics']} for details). Right: Alternative diagnostic diagrams using $[\mathrm{O}\textsc{ iii}]\,\lambda 4363$ from mazzolariNewAGNDiagnostic2024
• Figure 3: Stellar mass vs total SFR for the individual star forming galaxies in our sample, color coded by redshift of the source. The values displayed are estimated from the large apertures highlighted in pink in Figure \ref{['fig:integrated_oiii_maps']}. Solid lines and shaded regions correspond to the star formation main sequence at different redshift bins and the associated scatter from clarkeStarformingMainSequence2024a. For comparison we show the sample of star forming galaxies with outflows at $3<z<9$ from carnianiJADESIncidenceRate2024 and xuStellarAGNDrivenOutflows2025a.
• Figure 4: Velocity dispersion maps of the broad component ($\sigma_{\rm broad}$) for all galaxies in our sample. The regions identified as outflows are highlighted in red. No outflows are identified in SPT0311-58 and MACSJ1149-JD1. Blue contours correspond to the stellar continuum emission, as in Fig. \ref{['fig:integrated_oiii_maps']}.
• Figure 5: Comparison of the nebular dust attenuation (left) and gas metallicity (right) for the narrow and broad components in the galaxy regions in our sample identified to host ionized outflows. Symbols are color-coded according to the redshift of the galaxies. The black dashed lines show the one-to-one relation.