Ionised gas kinematics and dynamical masses of $z\gtrsim6$ galaxies from JADES/NIRSpec high-resolution spectroscopy

TL;DR

This study uses JWST/NIRSpec high-resolution spectroscopy of rest-frame optical lines in six z>5.5 galaxies to extract ionised-gas kinematics via forward-modelling that accounts for instrument PSF, MSA geometry, and pixellation. The analysis finds a mix of rotation- and dispersion-dominated systems with v(r_e) ~ 100–150 km/s and σ0 ~ 30–70 km/s, leading to dynamical masses of ~10^9–10^10 M⊙ that exceed stellar masses by up to ~40 and baryonic masses by ~3–4. The large dynamical-to-stellar mass discrepancy, coupled with substantial gas fractions (~10× M_*), suggests either dark matter dominance in the centers or reduced star-formation efficiency, while acknowledging significant systematic uncertainties in mass estimates and potential merger-driven kinematics. The work demonstrates the power of NIRSpec MOS in probing the dynamics of extremely low-mass, high-redshift galaxies and sets the stage for larger statistical studies of disc settling in the early Universe.

Abstract

We explore the kinematic gas properties of six $5.5<z<7.4$ galaxies in the JWST Advanced Deep Extragalactic Survey (JADES), using high-resolution JWST/NIRSpec multi-object spectroscopy of the rest-frame optical emission lines [OIII] and H$α$. The objects are small and of low stellar mass ($\sim 1\,$kpc; $M_*\sim10^{7-9}\,{\rm M_\odot}$), less massive than any galaxy studied kinematically at $z>1$ thus far. The cold gas masses implied by the observed star formation rates are $\sim 10\times$ larger than the stellar masses. We find that their ionised gas is spatially resolved by JWST, with evidence for broadened lines and spatial velocity gradients. Using a simple thin-disc model, we fit these data with a novel forward modelling software that accounts for the complex geometry, point spread function, and pixellation of the NIRSpec instrument. We find the sample to include both rotation- and dispersion-dominated structures, as we detect velocity gradients of $v(r_{\rm e})\approx100-150\,{\rm km\,s^{-1}}$, and find velocity dispersions of $σ_0\approx 30-70\,{\rm km\,s^{-1}}$ that are comparable to those at cosmic noon. The dynamical masses implied by these models ($M_{\rm dyn}\sim10^{9-10}\,{\rm M_\odot}$) are larger than the stellar masses by up to a factor 40, and larger than the total baryonic mass (gas + stars) by a factor of $\sim 3$. Qualitatively, this result is robust even if the observed velocity gradients reflect ongoing mergers rather than rotating discs. Unless the observed emission line kinematics is dominated by outflows, this implies that the centres of these galaxies are dark-matter dominated or that star formation is $3\times$ less efficient, leading to higher inferred gas masses.

Figures (10)

• Figure 1: Sample of six spatially-resolved high-redshift objects in JADES. Left panels show cutouts of the emission lines in the 2D rectified and combined spectra obtained with the high-resolution G395H grating. Negatives in the cutouts are the result of the background subtraction method used. Right panels show NIRCam image cutouts for each object (JADES, FRESCO), for the band that most closely resembles the emission line morphology (Section \ref{['sec:nircam']}). The 3-shutter slits and 3-point nodding pattern used result in an effective area of 5 shutters: the shutter encompassing the source is shown in orange, and the shutters used for background subtraction are shown in purple.
• Figure 2: Sample distribution in the parameter space of the stellar mass, SFR and emission line half-light radius, colour-coded by the redshift. We compare with a selection of ground-based near-infrared spectroscopic surveys that used the H$\alpha$ and [O iii] lines to measure galaxy kinematics at $z\sim1-4$Turner2017SINS2018Wisnioski2019Price2020; we use UV+IR SFRs from Whitaker2014 where available for the Turner2017 sample. We note that all objects at $z>1$ from KMOS3D are shown in this figure, but not all objects were used in the kinematic studies due to low SNR or their size being too small. The sample selected from JADES (hexagons) probes a very different regime: JADES targets are at higher redshift and lower stellar mass than ground-based facilities have probed thus far.
• Figure 3: Example of the fitting procedure for object JADES-NS-00016745 (Fig. \ref{['fig:sample_overview']}). Although the final combination of all exposures (left) was used for our initial visual inspection and sample selection, the pixels in this spectrum are highly correlated. Instead of using this combined spectrum, we simultaneously fit to all individual exposures obtained. In the case of JADES-NS-00016745 two exposures were taken per nodding position, resulting in six independent measurements for one 3-point nodding pattern with NIRSpec. To combat the undersampled PSF of NIRSpec, we perform our modelling in the detector plane, propagating parametric models to the exact same location on the detector as the observed data. The likelihood is then computed from the combination of all residual images. Pixels flagged by the reduction pipeline as affected by cosmic rays are masked and shown in grey.
• Figure 4: Corner plot for the 11-parameter thin-disc model for the object in Fig. \ref{['fig:method']}. Histograms show the posterior probability distributions, with orange lines indicating the prior probability distributions. We generally find good constraints on the kinematic parameters $v_{\rm a}$ and $\sigma_0$, although the turnover radius $r_{\rm t}$ is poorly constrained and degenerate with the rotational velocity. The top right panels show (in blue) the parameters that we derive from the model and are discussed in Section \ref{['sec:results']}.
• Figure 5: Velocity dispersion of the ionised gas as a function of redshift. The dashed line shows the fit from Uebler2019 for ionised gas at $0.6<z<2.6$ extrapolated to higher redshifts, while circles show results from a selection of ground-based IFU surveys in the near-infrared Turner2017SINS2018 and squares the results from ground-based near-infrared MOS data Price2020. Blue triangles show results from various studies with ALMA, which measure the kinematics of the cold gas for massive, dusty star-forming galaxies Neeleman2020Rizzo2020Fraternali2021Lelli2021Rizzo2021Herrera2022Parlanti2023. The high-redshift objects observed with JWST are dynamically approximately equally turbulent to the observations of more massive galaxies at cosmic noon, and do not follow the trend of increasing $\sigma_0$ with redshift observed at $z<4$.