Investigating the H i mass-size relation using the Simba cosmological simulations

TL;DR

This study tests the robustness of the H I mass–size relation (H I MSR) against diverse feedback mechanisms using the Simba cosmological simulations. It employs five feedback variants and a Martini-based pipeline to create observationally consistent HI measurements, then fits the relation with Orthogonal Distance Regression to obtain the form $ \log_{10}(D_{\mathrm{H\,I}}/\mathrm{kpc}) = \alpha \log_{10}(M_{\mathrm{H\,I}}/\mathrm{M}_\odot) - \beta$, finding $\alpha \approx 0.50$ and $\beta \approx -3.3$ to $-3.5$, in agreement with Wang 2016. The outer HI radial profiles, when scaled by the HI radius, collapse to a universal exponential across HI masses and feedback scenarios, indicating a self-similar disc structure that underpins the invariance of the H I MSR. This suggests that the HI MSR is a fundamental consequence of outer disc self-similarity and provides a robust diagnostic for disc structure in galaxy formation models, with data made available upon request.

Abstract

Observational studies have established a remarkably tight power-law relationship between the H I masses and sizes of late-type galaxies, known as the H I mass-size relation. This relation has been shown to persist across various models of a galaxy's H I surface density profile. Using the Simba cosmological simulations, we investigate the robustness of this relation under different feedback prescriptions, including cases where specific feedback mechanisms are absent. While the global properties of galaxies are significantly affected by changes in feedback, the H I mass-size relation remains intact. Moreover, its parameters consistently align with the best available empirical measurements. We analyze the H I mass distributions of galaxies and demonstrate that, regardless of the feedback scenario, galaxies within a given H I mass bin exhibit outer H I radial profiles well approximated by an exponential function. Furthermore, the exponential decline rate remains remarkably similar across different physical prescriptions. We attribute the persistence of the H I mass-size relation to this inherent self-similarity in the H I mass distributions.

Figures (6)

• Figure 1: Twenty selected H i total intensity maps of galaxies from the full-physics (s50) run. The H i mass increases from left to right and top to bottom. The yellow contour in each panel marks the 1 M$_{\odot}$ pc$^{-2}$ surface density level, while the fitted ellipse is shown in red. The text in each panel provides the physical size of the fitted ellipse, the galaxy’s total H i mass, and the relative error percentage , $\Delta$ D$_{\text{H\,{\sc i}}}$/D$_{\text{H\,{\sc i}}}$ , between the fitted ellipse and the 1 M$_{\odot}$ pc$^{-2}$ density contour.
• Figure 2: Distributions of key galaxy properties—logarithmic H i mass, stellar mass, H i fraction, and stellar half-mass radius—are shown for five simulation variants: s50, s50noAGN, s50noX, s50nojet, and s50nofb. In each panel, histograms compare the full galaxy sample (blue) with the selected study sample (orange). The variant name is indicated to the right of each row. The fourth panel also includes the number of galaxies in each sample. Vertical red dashed lines denote resolution limits: M$_{\text{H\,{\sc i}}} >$ 10$^{9.1}$$M_{\odot}$ for H i mass and M$_{*} >$ 10$^{9.5}$$M_{\odot}$, the threshold in Simba where black holes are seeded once a galaxy reaches this stellar mass limit, the latter corresponding to the stellar mass threshold in Simba above which black holes are seeded. Bin sizes are 0.5 dex for all properties except R$_{50,*}$ which uses 1 dex bins.
• Figure 3: This plot represents 5 different H i mass-size relations (H i MSR) from the Simba-50 feedback variants. Each subplot shows the kernel density estimate (KDE) of the data with a colour map, along with the linear best-fit line (in red) derived from orthogonal distance regression (ODR) algorithm. The blue line represents the relation from wang2016h, with the grey shaded region indicating the 1$\sigma$ uncertainty of their fit. A vertical dashed line at log$_{10}$(M$_{\text{H\,{\sc i}}}$[$M_{\odot}$]) = 9.1 marks the threshold at which the best-fit line is fitted.
• Figure 4: The scaled H i radial profiles from the Simba-50 z = 0 are shown. These profiles are scaled to $R_{\text{H\,{\sc i}}}$, and by definition, they overlap at a surface density of $\Sigma_{\text{H\,{\sc i}}} =$ 1 M$_{\odot}$.pc$^{-2}$ . Each panel corresponds to a different mass bins, with a width of 0.25 in log-scale, covering the range log$_{10}$($M_{\text{H\,{\sc i}}}$) = [8,10.75]. The number of galaxies per mass bin is denoted as N$_{gal}$. The grey lines represent the individual galaxy profiles, the red line shows the mean H i radial profile. The green dashed line indicates $R_{\mathrm{turn}}$ , the turning point that separates the inner and outer regions of the H i profiles. The blue line represents the fit to the mean H i radial profile, obtained using the piecewise linear fitting (pwlf) method in Python. $\alpha_{in}$ and $\alpha_{out}$ represents the inner and outer slopes.
• Figure 5: The scaled mean H i radial profiles from the Simba-50 simulation at z=0 for different feedback variants. When normalised by each galaxy’s characteristic H i disc size, the profiles reveal a universal exponential shape across all H i mass bins and feedback scenarios, highlighting the self-similarity of the outer H i distributions regardless of feedback strength.