JADES: the chemical enrichment pattern of distant galaxies -- $α$ enhancement, silicon depletion, and iron enhancement

TL;DR

This study uses JWST/NIRSpec R1000 spectra from 564 galaxies at $z=4$–7 to derive gas-phase abundances of C, O, Ne, Si, Ar, and Fe via direct-$T_e$ methods and ionization-correction factors. By stacking spectra in bins of stellar mass, SFR, sSFR, and UV slope, the authors uncover a chemically young ISM with CCSN-dominated yields, evidenced by low C/O and Si/O ratios that are further reduced by dust depletion. A notable finding is supersolar Fe/O in high-sSFR and blue-$\beta_{UV}$ subsamples, which cannot be easily explained by standard SN channels, suggesting selective Fe enhancement or exotic nucleosynthetic pathways in the early universe. The work also shows Si depletion consistent with rapid silicate-dust production, and it uses local galaxy comparisons and Kobayashi 2020-like chemical-evolution models to contextualize the results, though some abundance patterns remain challenging to reproduce, pointing to gaps in our understanding of early dust and metal enrichment. Overall, this paper provides the first robust, population-wide Si/O and Fe/O measurements at $z>4$ and highlights the complex interplay between stellar yields, star-formation histories, and dust in shaping the chemical evolution of the early cosmos.

Abstract

We present gas-phase abundances of carbon (C), $α$-elements (O, Ne, Si, and Ar) and iron (Fe) obtained from stacked spectra of high-$z$ star-forming galaxies with the deep Near Infrared Spectrograph medium-resolution data from the James Webb Space Telescope Advanced Deep Extragalactic Survey. Our 564 sources at $z=4$--7 have a median stellar mass of $\log(M_{*}/M_{\odot})=8.46$ and a median star-formation rate of $\log(\mathrm{SFR}/M_{\odot}\,\mathrm{yr^{-1}})=0.30$, placing them close to the star-formation main sequence. We find that the stacked spectrum of all our 564 sources has relatively low [C/O]$=-0.70$, moderate [Ne/O]$=-0.09$, and low [Ar/O]$=-0.28$ values at a low gas-phase metallicity of $12+\log(\mathrm{O/H})=7.71$ ($Z\sim 0.1~Z_\odot$), suggesting dominant yields of core-collapse supernovae evolved from massive stars. The detection of a weak SiIII] emission line in our stacked spectrum provides a silicon-to-oxygen abundance ratio of [Si/O]$=-0.63$, which is lower than that of stars in the Milky Way disc and lower than expected by chemical evolution models, suggesting silicon depletion onto dust grains. Likewise, this Si/O value is lower than that we newly derive for two individual $z>6$ galaxies (GN-z11 and RXCJ2248) with negligible dust attenuation. By performing spectral stacking in bins of $M_{*}$, SFR, specific SFR (sSFR), and ultra-violet (UV) continuum slope $β_{\mathrm{UV}}$, we identify [FeIII] line detections in the high-sSFR bin and the blue-$β_{\mathrm{UV}}$ bin, both of which exhibit supersolar Fe/O ratios, while their C/O, Ar/O, and Si/O ratios are comparable to those of the all-sources stack. Our findings support a chemically young gas composition with rapid dust depletion in the general population of high-$z$ star-forming galaxies, while raising the possibility of anomalous, selective Fe/O enhancement at the very early epoch of star formation.

Figures (9)

• Figure 1: R1000 composite spectra of the all-sources stack (black) with the errors (lightblue), where the $y$ axis is in the unit of flux density per wavelength ($f_{\lambda}$) normalised by the peak of [O iii]$\lambda$5007 ([O iii]$_{\mathrm{peak}}$). The grey dotted lines with the black texts show the wavelengths of emission lines that are used to measure nebular properties in Section \ref{['subsec:neb']}.
• Figure 2: Best-fit model (red) of the all-sources stack (black) with its errors (grey). The $y$ axis in each panel is rescaled for visualisation purpose, while the scale for the panel of [Fe iii] is the same as that of H$\gamma$+[O iii]$\lambda$4363. The blue texts indicate detected emission lines, while the yellow text corresponds to the undetected line. All of the emission lines shown here are detected, except for [Fe iii].
• Figure 3: Same as Figure \ref{['fig:bestfit']} but it highlights spectra of the high-sSFR and blue-$\beta_{\mathrm{UV}}$ stacks around [Fe iii]. These stacks are the only spectra whose $S/N$ ratio of [Fe iii] exceeds 3.
• Figure 4: SFR as a function of $M_{*}$. The distribution of all sources at $z=4$--7 in our sample is represented by the red contour, whose levels are 50 and 84 percentiles of the kernel density estimated with the scipy package gaussian_kde. The left panel shows the all-sources stack (red circle), and the others show the low- (orange square), mid- (orange pentagon), and high-value (orange hexagon) stacks with respect to the property indicated at the top of the panel. We plot the measurements of GN-z11 (magenta star) and RXCJ2248 (magenta cross; Section \ref{['subsec:gnz11']}). The small magenta diamonds are the references of $z>4$ individual galaxies (see Appendix \ref{['apsec:ref']} for details). The grey solid line and the cyan dashed line with the shades denote the SFMS based on the $M_{*}$-complete observations Simmonds2025 and the zoom-in radiation hydrodynamics simulations of thesan-zoomMcClymont2025a, respectively (Section \ref{['subsec:fund']} for more details). The median $M_{*}$ and SFR values of our stack are generally closer to the SFMS than most of the individual galaxies.
• Figure 5: Chemical abundance ratios (C/O: top, Ne/O: second top, Ar/O: third top, Si/O: second bottom, Fe/O: bottom) as a function of $M_{*}$ (left), SFR (second left), sSFR (second right), and $\beta_{\mathrm{UV}}$ (right). The small white pentagons are the references of $z>4$ stacks Hu2024Hayes2025. The horizontal dotted line shows the solar abundance Asplund2021. The other symbols are the same as in Fig. \ref{['fig:sfms']}. Our stacks generally show low C/O, moderate Ne/O, marginally low Ar/O, low Si/O, and low Fe/O ratios compared to individual galaxies at $z>4$, with the exception that the high-sSFR stack and the blue-$\beta_{\mathrm{UV}}$ stack exhibit enhanced Fe/O ratios comparable to those of the $z>4$ galaxies.