1D Kinematics from stars and ionized gas at $z\sim0.8$ from the LEGA-C spectroscopic survey of massive galaxies

TL;DR

The study assesses whether ionized-gas velocity dispersions can substitute stellar dispersions for dynamical studies of massive galaxies at z~0.8 using LEGA-C DR2. It employs joint pPXF modeling of continuum and emission lines to measure integrated σ'_g,int and σ'_star,int across ~800 galaxies, comparing them to quantify scatter and biases. The main result is an excellent average agreement but a notable intrinsic scatter (~0.13 dex) that propagates to ~0.24 dex uncertainty in dynamical masses and biases in gas-based scaling relations when emission-line selection is strong. The work demonstrates that gas kinematics can recover scaling relations with caution, emphasizing the roles of inclination and selection effects for high-redshift dynamical studies.

Abstract

We present a comparison of the observed, spatially integrated stellar and ionized gas velocity dispersions of $\sim1000$ massive ($\log M_{\star}/M_{\odot}\gtrsim\,10.3$) galaxies in the Large Early Galaxy Astrophysics Census (LEGA-C) survey at $0.6\lesssim\,z\lesssim1.0$. The high $S/N\sim20{\rmÅ^{-1}}$ afforded by 20 hour VLT/VIMOS spectra allows for joint modeling of the stellar continuum and emission lines in all galaxies, spanning the full range of galaxy colors and morphologies. These observed integrated velocity dispersions (denoted as $σ'_{g, int}$ and $σ'_{\star, int}$) are related to the intrinsic velocity dispersions of ionized gas or stars, but also include rotational motions through beam smearing and spectral extraction. We find good average agreement between observed velocity dispersions, with $\langle\log(σ'_{g, int}/σ'_{\star, int})\rangle=-0.003$. This result does not depend strongly on stellar population, structural properties, or alignment with respect to the slit. However, in all regimes we find significant scatter between $σ'_{g, int}$ and $σ'_{\star, int}$, with an overall scatter of 0.13 dex of which 0.05 dex is due to observational uncertainties. For an individual galaxy, the scatter between $σ'_{g, int}$ and $σ'_{\star, int}$ translates to an additional uncertainty of $\sim0.24\rm{dex}$ on dynamical mass derived from $σ'_{g, int}$, on top of measurement errors and uncertainties from Virial constant or size estimates. We measure the $z\sim0.8$ stellar mass Faber-Jackson relation and demonstrate that emission line widths can be used to measure scaling relations. However, these relations will exhibit increased scatter and slopes that are artificially steepened by selecting on subsets of galaxies with progressively brighter emission lines.

Figures (5)

• Figure 1: The LEGA-C sample in star formation rate versus stellar mass with the whitaker:12b relation (blue band, left panel) and in effective radius versus stellar mass and star-forming and quiescent relations from wel:14 (blue and red bands, right panel). All galaxies with successful dynamical fits are indicated by circles, where symbols are colored by the S/N of their brightest emission line. Targets with poorly fit spectra are included as crosses.
• Figure 2: LEGA-C spectra of a star-forming (ID:166550, top) and quiescent galaxy (ID:109827, bottom). All galaxies are fit with a combination of stellar population templates (red) and gaussian emission lines (blue) to model ionized gas lines (with combined fit, purple). The ACS F814W images with vertical LEGA-C slit are shown in right insets.
• Figure 3: Gas versus stellar observed velocity dispersion measurements for galaxies with detected emission lines in the LEGA-C survey. Galaxies are indicated by open circles and 49 galaxies with X-ray detections are highlighted by red stars. The two measures of galaxy kinematics agree for the population, with significant scatter (overall 0.13 dex). X-ray AGN, for which emission line widths are likely to be sensitive to the central engine in addition to the galaxy potential, are indeed offset to higher $\sigma'_{g, int}$ than $\sigma'_{\star, int}$ but do not account for all outliers.
• Figure 4: Ratios between observed velocity dispersions as a function of $\sigma'_{\star, int}$ and $\sigma'_{g, int}$ (a and b), maximum emission line S/N (c), stellar populations (d and e), slit alignment (f), and structural (g - i) properties. Blue lines and grey bands indicate running average and scatter. Pearson and Spearman correlation coefficients are noted in each panel.
• Figure 5: Mass Faber-Jackson relation measured from absorption ($\sigma'_{\star, int}$, left panel) and emission ($\sigma'_{g, int}$, center and right panels). Open/colored and black symbols indicate galaxies with and without emission lines. Best-fit linear relations with $\sigma'_{\star, int}$ (solid line for all galaxies and dashed for only those with emission lines) and $\sigma'_{g, int}$ (thick red line) are shown in the left and center panels. The right panel reproduces the center panel with symbols colored by emission line S/N along with best-fit relations for subsamples above a range of S/N thresholds (red and orange lines). The Mass FJ relations are very similar for the full population, indicating that either measure can be used to estimate galaxy dynamics, however the $\sigma'_{\star, int}$ relation is much tighter and somewhat shallower, with a vertical scatter of 0.11 dex relative to 0.17 dex measured from $\sigma'_{g, int}$. This increased scatter translates to an additional uncertainty of $\sim0.2\,\rm{dex}$ in dynamical mass. Furthermore, relying on bright emission features will bias towards an increasingly steep relation.