The Mass–Metallicity Relation at z ∼ 1–2 and Its Dependence on the Star Formation Rate

Abstract

We present a new measurement of the gas-phase mass-metallicity relation (MZR) and its dependence on star formation rates (SFRs) at 1.3 < z < 2.3. Our sample comprises 1056 galaxies with a mean redshift of z = 1.9, identified from the Hubble Space Telescope Wide Field Camera 3 (WFC3) grism spectroscopy in the Cosmic Assembly Near-infrared Deep Extragalactic Survey and the WFC3 Infrared Spectroscopic Parallel Survey. This sample is four times larger than previous metallicity surveys at z ∼ 2 and reaches an order of magnitude lower in stellar mass (10 8 M e ). Using stacked spectra, we find that the MZR evolves by 0.3 dex relative to z ∼ 0.1. Additionally, we identify a subset of 49 galaxies with high signal-to-noise (S/N) spectra and redshifts between 1.3 < z < 1.5, where Hα emission is observed along with [O III] and [O II]. With accurate measurements of SFR in these objects, we confirm the existence of a mass-metallicity-SFR (M-Z-SFR) relation at high redshifts. These galaxies show systematic differences from the local M-Z-SFR relation, which vary depending on the adopted measurement of the local relation. However, it remains difficult to ascertain whether these differences could be due to redshift evolution, as the local M-Z-SFR relation is poorly constrained at the masses and SFRs of our sample. Lastly, we reproduced our sample selection in the IllustrisTNG hydrodynamical simulation, demonstrating that our line flux limit lowers the normalization of the simulated MZR by 0.2 dex. We show that the M-Z-SFR relation in IllustrisTNG has an SFR dependence that is too steep by a factor of around 3.

The average dust attenuation curve at z ∼ 1.3 based on HST grism surveys

“…To fit the stacked spectra, we fit a set of Gaussian profiles to the lines in the region of 4400Å < λrest < 7100Å, covering the Hβ, [OIII], Hα+[NII], and [SII] lines. Similar to Henry et al (2021), we find that single Gaussian components are not sufficient to match the observed line profiles and adopt two Gaussian components for each line, one narrow and one broad component. Multiple components can arise due to kinematic differences among ionising sources (e.g., HII vs AGN), but can also occur in grism spectra due to line profiles having a dependence on the spatial distribution of the emitting sources.…”

“…The 3D-HST spectroscopy is reprocessed to be similar to the format utilised for the WISP emission pipeline (Bagley et al in prep. ) and follows the method presented in Henry et al (2021). As a result of this procedure, all 3D-HST spectra are visually inspected and interactively fit in the exact same manner as done for the WISP sample and provides self-consistency between the datasets.…”

“…Galaxy spectra are stacked and fit following similar methods to those described in Henry et al (2021), but modified for stacking based on Hα instead of [OIII]. In brief, we take the The median errorbar on β phot for each survey is indicated (0.51 for WISP, 0.24 for 3D-HST).…”

“…We construct a single bin of 65 galaxies at 1.35 < z < 1.5 in WISP (log M⋆ < 10.2) with G102+G141, providing full coverage from [OII] to Hα, where metallicities can simultaneously be determined with Balmer decrements. Following Henry et al (2021), we use 3).…”

“…They find that their low-metallicity sample (8.2 12+log(O/H) < 8.5) has a steeper curve with no significant 2175Å feature relative to their high-metallicity sample (12+log(O/H) > 8.5). Unfortunately, the low spectral resolution of the HST grism data prevent the use of the [NII]/Hα diagnostic and instead must rely on R23 (e.g., Henry et al 2021), which is only available for a small subset of our sample.…”