The AURORA Survey: The Mass -- Metallicity and Fundamental Metallicity Relations at $z \sim 2.3$ Based Purely on Direct $T_e$ Metallicities

TL;DR

This work delivers direct $T_e$-based constraints on the Mass--Metallicity Relation (MZR) and Fundamental Metallicity Relation (FMR) at cosmic noon ($z\sim2.3$) using 34 galaxies from the AURORA JWST/NIRSpec survey, including dual-zone $T_e$ measurements from [O III]4363 and [O II]7320,7331. The authors derive precise stellar masses and SFRs via SED fitting with nebular emission and a metallicity-dependent SFR calibration, revealing a MZR with slope $\gamma=0.27\pm0.04$, normalization $12+\log(O/H)=8.44\pm0.04$ at $M_\star=10^{10}\,M_\odot$, and intrinsic scatter $0.10\pm0.02$ dex; the $z\sim2$ FMR is consistent with $z\sim0$ calibrations within $\sim0.1$ dex. Comparisons to six simulations show none reproduce the observed $z\sim2$ MZR normalization evolution, highlighting gaps in chemical-enrichment and feedback modeling at cosmic noon, though selection effects may bias low-Hydrogen measurements toward higher SFR. The results imply smooth secular chemical evolution already in place by $z\sim2$ and demonstrate JWST’s power to anchor simulations with direct metallicity measurements across a broad parameter space, paving the way for larger, less-biased samples.

Abstract

We present new constraints on the Mass -- Metallicity (MZR) and Fundamental Metallicity Relations (FMR) using a sample of 34 galaxies at $1.38\leq~z\leq~3.5$ (median $z=2.28$). These galaxies have direct $T_e$ measurements from [O\sc{iii}]4363Å~and/or [O\sc{ii}]7320,7331Å~auroral emission lines detected with \textit{JWST}/NIRSpec as part of the AURORA survey. The detection of both oxygen auroral lines allows for dual-zone direct $T_e$ measurements and expands the dynamic range in $12+\log\mathrm{(O/H)}$ (7.68 to 8.65 dex), stellar mass ($10^{8}$ to $10^{10.4}$ M$_\odot$), and star-formation rate ($1$ to $100$ M$_\odot$ yr$^{-1}$) compared to previous direct $T_e$ studies of the high-redshift MZR and FMR. We characterize the $z\sim2$ MZR and find a slope of $0.27\pm0.04$ and normalization of $12+\log\mathrm{(O/H)} = 8.44\pm0.04$ at $10^{10}$ M$_\odot$ with an intrinsic scatter of 0.10 dex, consistent with past strong-line MZR measurements. Comparisons with $z\sim2$ predictions from six simulations reveal that none reproduce our observed MZR normalization evolution between $z\sim0$ and $z\sim2$. This discrepancy suggests current models do not fully capture the chemical enrichment and feedback processes occurring at cosmic noon. However, all 34 galaxies are on or above the star-forming main sequence such that our sample may be biased towards lower $12+\log\mathrm{(O/H)}$ if the FMR persists at $z\sim2$. Correcting for this selection effect would increase O/H by $\approx0.1$ dex at 10$^{9.3}$ M$_\odot$ (the median mass of our sample) bringing our MZR into better agreement with that of \texttt{TNG}. Lastly, we find our $z\sim2.3$ sample is consistent with the $z\sim0$ FMR within 0.1 dex in O/H, indicating that the smooth secular mechanisms regulating chemical enrichment, star formation, stellar mass, and outflows were in place at cosmic noon.

Figures (4)

• Figure 1: SFR vs. stellar mass for AURORA sources, shown in squares color-coded by redshift. Overlaid are the Clarke2024 star-forming main sequence relations based on SFR(H$\alpha$). Offsets from the Clarke2024 relations are shown in the bottom panel. AURORA sources lie on and above the main sequence, reaching as high as $31$ times the typical SFR at fixed mass.
• Figure 2: Mass metallicity relation for AURORA galaxies at $z \sim 2.3$ using only direct $T_e$ metallicities. Open circles show individual galaxies color-coded based on whether [O iii]4363Å-only, [O ii]7320,7331Å-only, or both oxygen auroral lines were detected. Our best-fit model (blue solid line) is shown along with the best-fit parameters and the darker (lighter) shading corresponds to the $1\sigma$ ($2\sigma$) confidence regions. Weighted mean values for equally populated stellar mass bins are shown for visualization purposes, but not used in fitting.
• Figure 3: Fundamental Metallicity Relation with two different $\alpha$ that minimizes the scatter in $12+\log(\mathrm{O/H})$ at $z \sim 0$. Our measurements show how combining both [O iii] and [O ii] auroral lines results in a $\sim 1.5$ dex coverage in $\mu$ from direct $T_e$ measurements. We find our measurements are in strong agreement with the Sanders2021 FMR (right) with only a deviation at $\mu > 8.7$ although still within $1\sigma$ errors. The left panel shows the comparison with the Curti2020 FMR where we find agreement within $1\sigma$ but $\sim 0.07$ dex systematically lower $12+\log(\mathrm{O/H})$ at fixed $\mu$ which is within the typical uncertainties of strong-line calibrations. Overall, this suggests that the same physical processes regulating star formation, stellar mass build-up, and chemical enrichment at $z \sim 0$ is in place at cosmic noon.
• Figure 4: Left: Our best-fit Mass -- Metallicity Relationship compared to past direct $T_e$Curti2020Ly2016Sanders2020 and strong-line calibration Henry2021Sanders2021Revalski2024Jain2025Stanton2025 MZR measurements. We find our MZR is in strong agreement with past MZR studies. We also include our $0.19 \times \Delta\log_{10}\mathrm{SFR}$ corrected MZR that takes into account the bias towards high SFR, low mass systems. This MZR is found to result in increased $12+\log(\mathrm{O/H})$ towards low-mass relative to past MZR studies. However, the level of correction needed is uncertain. Right: Comparison of our MZR to different simulation predictions compiled by Garcia2025 including FIREMa2016 and NewHorizonDubois2021. Only Illustris and NewHorizon match in terms of normalization but have a steeper slope. All other simulations are found to have widely different normalizations but slopes consistent with our MZR measurement.