MNRAS 000, 1–20 (2020) Preprint 2 November 2020 Compiled using MNRAS LATEX style file v3.

The Evolution of Gas-Phase Metallicity and Resolved Abundances in

Star-forming Galaxies at 𝑧 ≈ 0.6 – 1.8

S. Gillman ,1,2,3★ A. L. Tiley ,1,4 A. M. Swinbank ,1 U. Dudzevičiūtė ,1

R. M. Sharples ,1,5 Ian Smail ,1 C. M. Harrison ,6 Andrew J. Bunker ,7,8
Martin Bureau ,7 M. Cirasuolo ,9 Georgios E. Magdis ,2,3,10,11
arXiv:2010.15847v1 [astro-ph.GA] 29 Oct 2020

Trevor Mendel 12 and John P. Stott 13

1 Centre for Extragalactic Astronomy, Durham University, South Road, Durham, DH1 3LE UK
2 Cosmic Dawn Center (DAWN), Copenhagen, Denmark
3 DTU-Space, Technical University of Denmark, Elektrovej 327, DK-2800 Kgs. Lyngby, Denmark
4 International Centre for Radio Astronomy Research, University of Western Australia, 35 Stirling Highway, Crawley, WA, Australia
5 Centre for Advanced Instrumentation, Durham University, South Road, Durham DH1 3LE UK
6 School of Mathematics, Statistics and Physics, Newcastle University, Newcastle upon Tyne NE1 7RU, UK
7 Sub-department of Astrophysics, Department of Physics, University of Oxford, Denys Wilkinson Building, Keble Road, Oxford OX1 3RH, UK
8 Kavli Institute for the Physics and Mathematics of the Universe (WPI), The University of Tokyo, Kashiwa, Chiba 277-8583, Japan
9 European Southern Observatory, Karl-Schwarzschild-Str 2, D-86748 Garching b. München, Germany
10 University of Copenhagen, Lyngbyvej 2, DK-2100 Copenhagen Ø, Denmark
11 Institute for Astronomy, Astrophysics, Space Applications and Remote Sensing, National Observatory of Athens, GR-15236 Athens, Greece
12 Research School of Astronomy and Astrophysics, Australian National University, Canberra, ACT 2611, Australia
13 Department of Physics, Lancaster University, Bailrigg, Lancaster LA1 4YB, UK

Accepted 2020 October 28. Received 2020 October 19; in original form 2020 August 28

ABSTRACT
We present an analysis of the chemical abundance properties of ≈650 star-forming galaxies at
𝑧 ≈ 0.6 – 1.8. Using integral-field observations from the 𝐾 - band Multi-Object Spectrograph
(KMOS), we quantify the [Nii]/H𝛼 emission-line ratio, a proxy for the gas-phase Oxygen
abundance within the interstellar medium. We define the stellar mass – metallicity relation at
𝑧 ≈ 0.6 – 1.0 and 𝑧 ≈ 1.2 – 1.8 and analyse the correlation between the scatter in the relation and
fundamental galaxy properties (e.g. H𝛼 star-formation rate, H𝛼 specific star-formation rate,
rotation dominance, stellar continuum half-light radius and Hubble-type morphology). We
find that for a given stellar mass, more highly star-forming, larger and irregular galaxies have
lower gas-phase metallicities, which may be attributable to their lower surface mass densities
and the higher gas fractions of irregular systems. We measure the radial dependence of gas-
phase metallicity in the galaxies, establishing a median, beam smearing-corrected, metallicity
gradient of Δ𝑍/Δ𝑅= 0.002 ± 0.004 dex kpc−1 , indicating on average there is no significant
dependence on radius. The metallicity gradient of a galaxy is independent of its rest-frame
optical morphology, whilst correlating with its stellar mass and specific star-formation rate, in
agreement with an inside-out model of galaxy evolution, as well as its rotation dominance. We
quantify the evolution of metallicity gradients, comparing the distribution of Δ𝑍/Δ𝑅 in our
sample with numerical simulations and observations at 𝑧 ≈ 0 – 3. Galaxies in our sample exhibit
flatter metallicity gradients than local star-forming galaxies, in agreement with numerical
models in which stellar feedback plays a crucial role redistributing metals.
Key words: galaxies: abundances – galaxies: high-reshift – galaxies: kinematics and dynamics

1 INTRODUCTION

The ejection of metals into the interstellar medium via winds from
★ E-mail: [email protected] massive stars in the asymptotic giant branch phase or via supernovae

© 2020 The Authors

2 S. Gillman et al.
acts to increase the chemical abundance in star-forming galaxies. different regions of the non-evolving fundamental plane, given the
This influx of metals is mediated by inflows, outflows and cold higher average star formation at earlier cosmic times (Cresci et al.
gas accretion (e.g. Köppen & Edmunds 1999; Calura & Matteucci 2019). To confirm this result, metallicity measurements of large
2006; Erb 2008; Steidel et al. 2010; Davé et al. 2012; Lu et al. samples of high-redshift galaxies are required, which has recently
2015; Sanders et al. 2015; Anglés-Alcázar et al. 2017; Christensen been made possible with the advent of high-redshift multi-object
et al. 2018; Sanders et al. 2018). Empirically constraining the com- spectroscopy.
plex interplay between these secular processes and their imprint on Using the MOSFIRE Deep Evolution Field (MOSDEF) Sur-
the chemical abundance properties of galaxies is crucial to fully vey, Sanders et al. (2015, 2018) demonstrated the presence of the
constrain the baryon cycle. relation between stellar mass, metallicity and star – formation rate
In the local Universe, the correlation between a galaxy’s stellar in 260 galaxies at 𝑧 ≈ 2.3. However the star-forming galaxies in
mass (𝑀∗ ) and its integrated gas-phase metallicity (𝑍), the mass – the MOSDEF survey exhibit ≈ 0.1 dex lower metallicity than lo-
metallicity relation, has been well studied (e.g. Lequeux et al. 1979; cal galaxies, at fixed stellar mass and star-formation rate. Sanders
Tremonti et al. 2004). Higher stellar mass star-forming galaxies have et al. (2018) concludes that there is not a fundamental plane that
been shown to exhibit higher gas-phase metallicities at 𝑧 ≈ 0 with can simultaneously match the properties of star-forming galaxies in
2/5
𝑍 ∝ 𝑀∗ up to 𝑀∗ ∼ 1010 M  , above which the relation saturates both the local and distant Universes. This implies that the metal-
to a constant metallicity. This strong correlation between stellar licity properties of the interstellar medium of star-forming galaxies
mass and metallicity is believed to be a consequence of supernovae evolves with cosmic time.
driven winds and outflows which remove the metal-rich gas from As well as the galaxy integrated metallicity, the distribution of
the interstellar medium (e.g. Tremonti et al. 2004; Tumlinson et al. metals within a galaxy provides insights into the influence of star
2011). In lower stellar mass galaxies, with shallower potential wells, formation, gas accretion, mergers and feedback, that all play a key
a larger fraction of this material is removed from the galaxy resulting role in defining the evolution of galaxies. In the local Universe most
in an overall lower metallicity (e.g. Arimoto & Yoshii 1987; Garnett isolated galaxies exhibit negative log-linear abundance gradients
2002; Brooks et al. 2007; Chisholm et al. 2018). Higher stellar mass (i.e metallicity decreasing with radius). Higher gas-phase metallic-
galaxies are also believed to evolve more rapidly at higher redshifts ities are observed in the central regions of the galaxies where star
and have therefore converted more of their pristine gas into stars and formation is most prevalent and supernovae enrich the surrounding
metals, resulting in higher gas-phase metallicities (e.g. Maiolino interstellar medium (e.g. Sánchez et al. 2014; Kaplan et al. 2016;
et al. 2008; Somerville & Davé 2015; Sanders et al. 2018). Poetrodjojo et al. 2018). The observed negative gradients are pre-
The star-formation rate and gas fraction of a galaxy have also dicted from inside-out theories for the growth of galaxy discs (e.g.
been linked to its offset from the mass – metallicity relation, with Boissier & Prantzos 1999) and are well modelled in hydrodynam-
suggestions of the existence of a fundamental metallicity plane. ical simulations such as Evolution and Assembly of GaLaxies and
In this plane, more highly star-forming galaxies, at a given stellar their Environments (eagle) (e.g. De Rossi et al. 2017; Collacchioni
mass, have lower gas-phase metallicities (e.g. Mannucci et al. 2010; et al. 2020).
Sánchez et al. 2017b; Sánchez-Menguiano et al. 2019). Surveys of At high redshift the picture is much less clear, with various
galaxies in the local Universe, such as Calar Alto Legacy Integral observational studies reporting a range of metallicity gradients in
Field Area Survey (CALIFA; Sánchez et al. 2012) and Mapping isolated star-forming galaxies (e.g. Cresci et al. 2010; Jones et al.
Nearby Galaxies at Apache Point Observatory (MaNGA; Bundy 2010; Swinbank et al. 2012; Jones et al. 2013; Stott et al. 2014;
et al. 2015), which observed thousands of galaxies out to 𝑧 ≈ 0.03, Leethochawalit et al. 2016; Wuyts et al. 2016; Molina et al. 2017;
have shown a correlation between galaxy star-formation rate and Wang et al. 2017; Curti et al. 2020b) as well as in simulations
offset from the mass – metallicity relation (e.g. Sánchez Almeida & (e.g Pilkington et al. 2012; Anglés-Alcázar et al. 2014; Finlator
Sánchez-Menguiano 2019; Cresci et al. 2019). 2017; Sillero et al. 2017; Collacchioni et al. 2020; Hemler et al.
Attempts have been made to define the mass – metallicity rela- 2020). In this paper, we exploit a large sample of high-redshift
tion (and plane) in the distant Universe. The ratio of strong optical star-forming galaxies with spatially-resolved metallicity profiles.
nebular metal emissions lines to the Balmer series (e.g. [Nii]/H𝛼) is Utilizing the strong-line calibration of the [Nii]/H𝛼 ratio, we present
commonly used to infer the gas-phase metallicity of galaxies at high an analysis of the gas-phase metallicity properties of ≈ 650 star-
redshift due to their spectral proximity, making them insensitive to forming galaxies in the redshift range 𝑧 ≈ 0.6 – 1.8. We explore the
dust and observable from ground-based facilities (e.g. Yabe et al. correlations between galaxy morphology, dynamics and position on
2015; Wuyts et al. 2016; Förster Schreiber et al. 2018; Curti et al. the mass – metallicity relation, as well as the metallicity profile of
2020b). the galaxies.
The ratio is often expressed as an Oxygen abundance relative to This paper is organized as follows. In Section 2 we describe
Hydrogen, as Oxygen is generally the most abundant heavy element the observations and sample selection. We define the procedure
by mass and therefore provides a proxy for the metallicity of the used to correct for the galaxies velocity fields and de-redshift the
galaxy (Pettini & Pagel 2004). Using these strong-line calibrations integral-field data, from which we extract the gas-phase metallic-
of optical emission lines, many recent studies have shown that the ities. In Section 3 we present the mass – metallicity relation and
mass – metallicity relation evolves above 𝑧 = 1, with higher-redshift fundamental metallicity plane of the sample. We analyse the radial
galaxies having lower metallicities at a given stellar mass (e.g. Erb dependence of metallicity and its correlation with galaxy dynam-
et al. 2006; Maiolino et al. 2008; Zahid et al. 2011; Stott et al. 2013; ics, morphology and redshift in Section 4, before presenting our
Yabe et al. 2015; Wuyts et al. 2016; Sanders et al. 2018). conclusions in Section 5.
Despite evolution in the normalisation of the mass – metallicity A nine-year Wilkinson Microwave Anisotropy Probe (Hin-
relation, it has been suggested that the fundamental plane of stellar shaw et al. 2013) cosmology is used throughout this work with
mass, gas-phase metallicity and star-formation rate does not evolve ΩΛ = 0.721, Ω𝑚 = 0.279 and 𝐻0 = 70 km s−1 Mpc−1 . In this cos-
with cosmic time. For example Mannucci et al. (2011) proposed that mology an angular resolution of 0.70 arcsecond (the median full
the evolution in the mass – metallicity relation comes from sampling width half maximum (FWHM) of the seeing in our data) corre-

MNRAS 000, 1–20 (2020)

 Gas-Phase Metallicity and Resolved Abundances at 𝑧 ≈ 0.6 – 1.8 3

Rest-frame wavelength ( m) Rest-frame wavelength ( m)

0.65128 0.65628 0.66128 0.65128 0.65628 0.66128
1.2 1.2
KGES 198 Compact [NII]/H = 0.27 Observed frame
Rest frame KGES 31 Disc [NII]/H = 0.13 Observed frame
Rest frame
COSMOS 1.0
obs= 166.0 kms 1
CDFS 1.0
obs= 286.0 kms 1
z 1.6 rest= 66.0 kms 1 z 1.39 rest= 114.0 kms 1

Normalised Flux

Normalised Flux
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0.0 0.0
WFC col 0.5" WFC col 0.5"
0.2 1.7007 1.7057 1.7107 0.2 1.563 1.568 1.573
0.65128 0.65628 0.66128 0.65128 0.65628 0.66128
1.2 1.2
KGES 60 Disc [NII]/H = 0.38 Observed frame
Rest frame KROSS 156 Disc [NII]/H = 0.23 Observed frame
Rest frame
CDFS 1.0
obs= 158.0 kms 1
COSMOS 1.0
obs= 196.0 kms 1
z 1.5 rest= 60.0 kms 1 z 0.93 rest= 64.0 kms 1
Normalised Flux

Normalised Flux
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0.0 0.0
WFC col 0.5" WFC col 0.5"
0.2 1.63435 1.63935 1.64435 0.2 1.25964 1.26464 1.26964
0.65128 0.65628 0.66128 0.65128 0.65628 0.66128
1.2 1.2
KROSS 152 Disc [NII]/H = 0.45 Observed frame
Rest frame KGES 234 Disc [NII]/H = 0.2 Observed frame
Rest frame
COSMOS 1.0
obs= 289.0 kms 1
COSMOS 1.0
obs= 275.0 kms 1
z 0.89 rest= 108.0 kms 1 z 1.4 rest= 114.0 kms 1
Normalised Flux

Normalised Flux
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0.0 0.0
WFC col 0.5" WFC col 0.5"
0.2 1.23635 1.24135 1.24635 0.2 1.57285 1.57785 1.58285
Observed-frame wavelength ( m) Observed-frame wavelength ( m)
Figure 1. Example 𝐻 𝑆𝑇 images and integrated spectra of galaxies in our sample. For each galaxy we show a WFC3 three-colour image composed of F105W,
F125W and F160W images. The semi-major axis of the galaxy (orange line) and stellar continuum centre (black filled circle) are indicated, as well as the
morphological class if available from the Huertas-Company et al. (2015) classification. The galaxy integrated spectra from the observed KMOS data cube
(grey) and de-redshifted rest-frame cube with bulk motions removed (blue) are also shown. The H𝛼 and [Nii] emission lines are indicated (vertical pink dashed
lines) and we overlay the spectral fit to the de-redshifted and corrected emission lines (crimson line). The [Nii]/H𝛼 ratio of each galaxy, as well as the velocity
dispersion of the spectra are indicated in the top left. The examples show a range of [Nii]/H𝛼 ratio from galaxies with varying morphologies and redshifts.

sponds to a physical scale of 5.5 kpc at a redshift 𝑧 = 0.91 (the spaxels. KMOS utilizes image slices to produce data cubes with
median redshift of our data). All quoted magnitudes are in the AB wavelength coverages of 0.8 – 1.0, 1.0 – 1.4, 1.4 – 1.9, 1.9 – 2.5 or
system and stellar masses are calculated assuming a Chabrier initial 1.5 – 2.5 𝜇m in the 𝐼 𝑍, 𝑌 𝐽, 𝐻, 𝐾 or 𝐻𝐾 band respectively. In the
mass function (IMF) (Chabrier 2003). following sections we provide an overview of the KROSS and KGES
surveys.

2 SAMPLE SELECTION AND ANALYSIS

2.1 KROSS
To provide statistically meaningful conclusions about the metallic-
ities of galaxies in the distant Universe, we utilize 644 star-forming KROSS is a KMOS survey of 795 typical star-forming galaxies in
galaxies drawn from two large 𝐾-band multi-object spectrograph the redshift range 𝑧 ≈ 0.6 – 1.0 selected from the Extended 𝐶ℎ𝑎𝑛𝑑𝑟𝑎
(KMOS; Sharples et al. 2004, 2013) programmes at 𝑧 ≈ 0.6 – 1.8. Deep Field South (E-CDFS; Giacconi et al. 2001), Cosmological
The galaxies in our sample are drawn from the KMOS Redshift Evolution Survey (COSMOS; Scoville et al. 2007), UKIRT Infrared
One Spectroscopic Survey (KROSS; Stott et al. 2016; Harrison Deep Sky Survey (UKIDSS) Ultra-Deep Survey (UDS; Lawrence
et al. 2017) at 𝑧 ≈ 0.6 – 1.0 (448 galaxies) and the KMOS Galaxy et al. 2007) and the SA22 (Lilly et al. 1991; Steidel et al. 1998)
Evolution Survey (KGES; Gillman et al. 2019, Tiley et al. in prep.) extragalactic fields. The majority of the sample galaxies are se-
at 𝑧 ≈ 1.2 – 1.8 (196 galaxies). lected using archival spectroscopic redshifts, 25 per cent being H𝛼
All the galaxies in the sample were observed using KMOS, narrow-band emitters at 𝑧 = 0.84 from the High Redshift Emission
a multi-object spectrograph mounted on the Nasmyth focus of the Line Survey (HiZELS) and Canada-France HiZELS (CF-HIZELS)
8-m class UT1 telescope at the VLT, Chile. It has 24 individual surveys (Sobral et al. 2013, 2015). A full description of the KROSS
integral-field units that patrol a 7.2 arcminute diameter field, each survey and derived galaxy properties is given in Stott et al. (2016)
with a 2.8 × 2.8 arcsecond2 field of view and 0.2 × 0.2 arcsecond2 and Harrison et al. (2017).

MNRAS 000, 1–20 (2020)