High-z galaxies with JWST and local analogues -- it is not only star formation

TL;DR

This work analyzes JWST NIRSpec ERO spectra of SMACS 0723 to extend high-z metallicity studies by introducing a robust double-line temperature diagnostic that mitigates flux-calibration biases. It combines empirical Te/abundance methods with Bayesian photoionization modeling (CL01 and G16 grids) to derive gas-phase abundances across z>5 and compares them with local SDSS analogues, finding good agreement with model-based abundances and evidence for AGN activity in at least one high-z source. The study also reveals shock-dominated excitation in several z<5 galaxies and highlights potential biases in the mass-metallicity relation when selecting by auroral line detections, arguing for bias-aware analyses using full photoionization modeling. Altogether, it demonstrates the practicality and importance of Bayesian modeling and local analogues for interpreting JWST-era abundance measurements and cautions against simplistic line-ratio calibrations in the early data regime.

Abstract

I present an analysis of the JWST NIRSpec data of SMACS 0723 released as Early Release Observations. As part of this three new redshifts are provided, bringing the total of reliable redshifts to 14. I propose a modification to the direct abundance determination method that reduces sensitivity to flux calibration uncertainties by a factor of ~3 and show that the resulting abundances are in good agreement with Bayesian photoionization models of the rest-frame optical spectrum. I also show that 6355 is most likely a narrow-line active galactic nucleus (AGN) with $M_* \sim 10^9$ Msun at z=7.66, and argue that 10612 might also have an AGN contribution to its flux through comparison to photoionization models and low-redshift analogues. Under the assumption that the lines come from star-formation I find that the galaxies have gas depletion times of ~$10^7$ years, comparable to similar galaxies locally. I also identify a population of possibly shock-dominated galaxies at z<3 whose near-IR emission lines plausibly come nearly all from shocks and discuss their implications. I close with a discussion of the potential for biases in the determination of the mass-metallicity relation using samples defined by detected [O III]4363 and show using low-z galaxies that this can lead to biases of up to 0.5 dex with a systematic trend with mass.

Figures (14)

• Figure 1: An illustration of the effect of flux correction of the NIRSpec spectrum. In this particular case, total magnitudes are used. The original L3 spectrum, normalised to the F277W flux is shown as a dotted gray line, while the re-normalised spectrum is shown with the solid black line. The galaxy shown here is 3042 and this is a complex source, as discussed in the text, which in part is likely to be the reason for the substantial difference in spectral shape.
• Figure 2: The overlapping galaxies in 3042. The spectrum showed here has the continuum subtracted. There is a very clear redshift of $z=1.9938$ which is demonstrated in the top left panel and the middle row. This presumably corresponds to the object detected in the bands bluest bands (f090w and bluewards) shown in the bottom row. These images are all 5" on the side. In f150w a second galaxy, nearly orthogonal to this stars to appear and by f356w only this galaxy is easily visible. The colour image on the left is constructed from F090W, F150W and F200W and shows this clearly. As discussed in the text, there are two possible redshifts for this source: $z=6.253$ illustrated in the top middle panel, and $z=3.258$ illustrated in the top right panel.
• Figure 3: Left: The relative difference between estimated and true $T_e$ as a function of the amount of dust attenuation applied (see text for details). The solid lines show the results for the five true $T_e$ values shown when applying the standard [Oiii]4363/[Oiii]5007 ratio, while the dashed lines show the same when using the proposed double line ratio, [Oiii]4363/$\mathrm{H}\gamma$/[Oiii]5007/$\mathrm{H}\beta$. Right: the difference in estimated $12+\log \mathrm{O/H}$ as a function of dust attenation for the standard method (solid lines) and the proposed robust method (dashed lines).
• Figure 4: A comparison of literature determinations of the gas-phase oxygen abundance of the five high-z galaxies.
• Figure 5: A comparison of the line ratios for the high-z galaxies compared to the G16 model grid with $M_{\mathrm{upper}}=100\,\mathrm{M}_\odot$ and $n=100\,\mathrm{cm}^{-3}$. The grid lines correspond to the ionization parameter, $\log U$ shown with blueish lines and $\log Z$ in reddish tones.