Lessons Learned from Studying H$α$ Galaxy Kinematics with Mock JWST/NIRSpec IFU Observations at $z > 6$

TL;DR

This study evaluates how reliably JWST/NIRSpec IFU observations can recover disk kinematics and dynamical masses for $z>6$ galaxies. Using two SERRA zoom-in galaxies (Opuntia and Narcissus), the authors generate idealised and realistic mock NIRSpec data and analyze them with 3DBarolo to assess biases in $v_ ext{rot}$, $σ$, and $v/σ$ caused by non-circular motions and beam smearing. They find that non-circular inflows/outflows can mimic disk-like velocity gradients and inflate $σ$ by factors of $\sim2$–$3$, biasing $v/σ$ low, though dynamical masses remain relatively robust for axisymmetric systems when asymmetry drift is accounted for; asymmetries and outflows can still lead to substantial mass-recovery biases. The results emphasize the need to exploit full 3D IFU data and combine warm gas tracers (H$ ext{ }$) with cold gas tracers ([C II]) to accurately interpret high-redshift disk kinematics and mass budgets, and they caution against over-reliance on integrated spectra or single-tracer inferences.

Abstract

Galaxies with a disk morphology have been established at $z > 9$ with the James Webb Space Telescope (JWST). However, confirming their disky nature requires studying their gas kinematics, which can be challenging when relying solely on the warm gas observed by JWST. Unlike the cold gas traced by the Atacama Large Millimetre/Submillimetre Array (ALMA), warm gas is sensitive to outflows, complicating the interpretation of the disk dynamics. This elicits the question of how to compare information obtained from varied tracers, as well as how to physically interpret the low angular and spectral resolution observations generally available at high redshift. We address these challenges through comparative kinematic analysis of idealised and realistic NIRSpec/IFU mock observations derived from two galaxies in the SERRA suite of cosmological zoom-in simulations. With these synthetic data, we determine the robustness of dynamical information recovered from typical IFU observations, and test widely-used criteria for identifying disks and gaseous outflows at high redshift. We find that at the typical NIRSpec/IFU spectral and angular resolution ($\sim$ 0.05"/pixel), non-circular motions due to inflows or outflows can mimic the smooth velocity gradient indicative of a disk, and bias measured velocity dispersion upwards by a factor of $2-3\times$. As a result, the level of rotational support may be underestimated in the NIRSpec/IFU observations. However, the recovered dynamical mass appears to be relatively robust despite biases in $v_\text{rot}$ and $σ$.

Figures (14)

• Figure 1: Maps of stellar distribution, with a colorbar showing the stellar mass density per unit area. The field of view is 8kpc $\times$ 8kpc, corresponding to $\sim$ 1.4" $\times$ 1.4" at $z = 6.07$ (Opuntia, $M_\star = 1.2 \times 10^{10} \mathrm{M}_{\odot}$) and $\sim$ 1.5" $\times$ 1.5" at $z = 6.82$ (Narcissus, $M_\star = 1.0 \times 10^{10} \mathrm{M}_{\odot}$).
• Figure 2: Moment-0, -1, and -2 maps, respectively depicting integrated spectral intensity, line-of-sight velocities, and velocity dispersion for Opuntia (top three rows) and Narcissus (bottom three rows) inclined at 60$\degree$. The [C ii] and H$\alpha$ emission maps in the first and second columns are obtained directly from the simulations. See Table \ref{['tab:target_properties']} for the intrinsic kinematic measurements. The third and fourth columns show the emission line cubes that have been processed to represent an idealised warm gas observation and a realistic mock NIRSpec/IFU observation as described in Section \ref{['Creating_Obs']}, for which the moment-1 and -2 maps are masked at 25$\sigma$ (in the idealised case) and 5$\sigma$ (in the mock NIRSpec case). The $v=0$ isophote is plotted on the mock NIRSpec moment-1 maps and dotted lines indicate the region representing the 'outer extent of the galaxy' in the mock NIRSpec moment-0 maps. PSFs are shown in the moment-0 maps for the derived data.
• Figure 3: Kinematic axes obtained from 3DBarolo fitting (section \ref{['barolo']}), measured from the realistic mock NIRSpec observations, along with idealised H$\alpha$ and [C ii] observations, overlaid on the intrinsic H$\alpha$ moment-1 maps.
• Figure 4: Position-Velocity diagrams for the idealised observations (left) and mock NIRSpec observations (right), where the major axis diagrams are extracted along the axes shown in Figure \ref{['fig:m0pas']}. Contour levels are at $3^n\sigma$, where n=[1, 2, 3, 4, 5] and the $\sigma$ value used to define the contours is the RMS value of noise-dominated regions in the diagrams. Black contours trace the data, and red contours represent the disk model. Arrows indicate regions of the PV diagrams that deviate from the expected profile for a disk.
• Figure 5: Narcissus and Opuntia are here plotted in the PVSplit parameter space alongside simulated disks and non-disks Rizzo22. Narcissus occupies the non-disk region of parameter space, while Opuntia is close to the plan of division between the disk and non-disk populations R-O23. Uncertainties in the PVSplit fitting, propagated from uncertainties in e.g. the fitted centre and position angle from 3DBarolo, tend to act towards the non-disk region.