Prevalent elongated galaxies in the early Universe evidenced by stellar kinematics

Abstract

The Universe is now extensively populated by discy galaxies with coherent galaxy-wise stellar rotation. This disc prevalence has been deemed a late-time phenomenon because the penetrating cold gaseous streams in the early Universe ($z\gtrsim 2$) fuel the star formation in galaxies too intensively to allow for thin disc formation. However, recent images taken by the James Webb Space Telescope (JWST) unveiled a prominent population of low-mass galaxies at high redshifts with flattened shapes, widely interpreted as early significance of discs given the well-established connection between flattening and discy morphology seen in the local Universe. It is noticed, on the other hand, that these galaxies show far more flattened systems than can be accounted for by randomly oriented oblate discs, and the axial ratio distributions are better explained by elongated prolate ellipsoids, an extremely rare spindle-like configuration at low redshifts. The true morphological nature of these early low-mass galaxies is fundamental to understanding the structure evolution of their discy descendants we see today, including our Milky Way. In this work, we discriminate the oblate disc and prolate spindle scenario by a decisive experiment with stellar kinematics at its core. The result clearly supports the prolate spindle scenario, and evidences an early Universe widely inhabited by linear stellar systems contrasting the current era dominated by planar discy galaxies, which suggests a dimensional transition in galactic structure over cosmic time.

Figures (4)

• Figure 1: A schematic plot for the astrophysical experiment based on stellar kinematics that decisively differentiates the two possible configurations for the intrinsic shapes of the prevalent flattened galaxies at high redshifts revealed by JWST. The reddish cutout on the left shows a typical flattened galaxy seen in JWST images. While it could be an oblate disc galaxy viewed relatively edge on, a common case at low redshifts, this galaxy is also likely to be a prolate spindle galaxy with its side toward the observer as illustrated by the 3D figures. The light red and blue lines to the right of the 3D figures mark their major and minor axes respectively in three-dimensional space. The dramatically different geometries of an oblate disc and a prolate spindle galaxy are determined by the distinct kinematics of their stars which is summarized in the orange boxes. A critical prediction of such distinct stellar kinematics and the associated intrinsic shapes is that oblate disc and prolate spindle galaxies will manifest contrary apparent shapes versus observed velocity dispersion relation as illustrated in the part III. The observed relation is shown in \ref{['fig:2']}.
• Figure 2: The observed apparent shape vs. stellar velocity dispersion relation for the major galaxy population in the JWST high-$z$ sample with stellar masses $\mathrm{lg}(\mathcal{M}_{\star}/\mathcal{M}_{\odot})\sim9$. The experiment is conducted specifically with 122 galaxies in redshift range $z\in (1.5,4)$ and in a narrow stellar mass bin $9<\mathrm{lg}(\mathcal{M}_{\star}/\mathcal{M}_{\odot})<9.5$, which are further divided into apparently flattened (upper panel) and round (lower panel) subpopulation by their median minor-to-major axial ratio. In each panel, the black solid line and its associated gray band represent the stacked spectrum and its one-sigma uncertainty estimated via 100 times bootstrap resampling. The positions of prominent gaseous emission and stellar absorption lines are marked by vertical dotted lines, where in many cases the line emission and absorption are superimposed. The bracketed He I and [Ne III] denote their possible existence underlying the observed emission lines. The orange line shows the best fit stellar component in the spectrum, with the emission lines masked, and the measured stellar velocity dispersion $\sigma^{\prime}_\mathrm{star}$ and its 1000 times bootstrapping estimated 68% confidence interval are recorded in the small inset panel. We note that, as compared to the intrinsic values $\sigma_\mathrm{star}$, the measurements $\sigma^{\prime}_\mathrm{star}$ are systematically enlarged by the line broadening due to finite spectral resolution and statistical uncertainties in the measured redshifts for individual galaxies, both of which do not explicitly depend on the apparent shapes of galaxies (more discussion in the main text).
• Figure 3: The observations and theoretical predictions for the apparent shape vs. stellar velocity dispersion relations of oblate discs and prolate spindles. The red and blue tracks show how the effective stellar velocity dispersion $\sigma_\mathrm{e}$ varies with apparent ellipticity (2000 randomly sampled inclination angles) predicted by Jeans Anisotropic Models respectively for prolate and oblate ellipsoids of exponential light profiles, normalized to the $\sigma_\mathrm{e}$ at their median ellipticity. The shapes of these two tracks are not sensitive to the anisotropy level of dynamical models, and the clear discontinuity of the red track at low ellipticities reflect the fact that randomly oriented prolate ellipsoids are highly skewed toward flattened shapes. The blue squares and the error bars mark the median normalized $\sigma_\mathrm{e}$ and ellipticities with their 1000 times bootstrapping estimated 68% confidence intervals in the four quarters of the ellipticity distribution for the Milky Way analogs in MaNGA survey. The orange squares stand for the apparently round (the left two) and flattened (the right two) subsample of the analyzed JWST galaxies, with emission lines included in the fit (marked by central magenta dot) or masked. The measured velocity dispersion $\sigma^{\prime}_\mathrm{star}$ and uncertainties are normalized to the linearly interpolated value at the median ellipticity 0.5 of the whole sample, for the purpose of qualitatively illustrating the trend.
• Figure 4: Monte Carlo simulation assessment of the errors in the measured stellar velocity dispersion introduced by infilling emission lines. This figure shows the small fractional errors of the measured stellar velocity dispersion $\sigma^{\prime}_\mathrm{star}$ on the y-axis as a function of the intrinsic dispersion ratio between ionized gas and stars on the x-axis for 10k mock spectra, with red dots showing the individual results and the black solid line and the two shaded bands illustrating the median, the 68% and 95% interval of the distribution respectively.