Stellar Feedback Effects on the Mass Distribution of Clouds and Cloud Complexes

TL;DR

This study investigates how stellar feedback shapes the mass distribution of gas clouds in dwarf galaxies by running RAMSES-RT simulations with varying feedback modes (SNe, winds, and ionizing radiation) and identifying clouds via isocontours at $n>10$, $n>10^{1.5}$, and $n>10^{2}\ \mathrm{cm}^{-3}$. The key finding is that the mass function of dense clouds ($n>100\ \mathrm{cm}^{-3}$) is largely insensitive to feedback, suggesting gravity dominates their structure, while cloud complexes ($n>10\ \mathrm{cm}^{-3}$) become more top-heavy in the presence of radiation due to radiative heating and reduced fragmentation. Winds do not significantly alter the mass distributions, implying that pre-SN feedback effects on cloud masses are mainly driven by radiation-induced thermodynamics rather than mechanical feedback. The results imply that the observed changes in the star cluster mass function attributed to feedback arise from intra-cloud regulation of star formation and radiation-driven temperature changes, rather than direct shaping of the parent cloud mass spectrum.

Abstract

Galaxy evolution is sensitive to how stars inject feedback into their surroundings. In particular, stellar feedback from star clusters strongly affects gas motions and the baryonic cycle, with more massive clusters having stronger effects. Our previous results show that the star cluster mass distribution in dwarf galaxies depends on feedback, as strong pre-SN feedback, particularly ionizing radiation, results in fewer high-mass clusters. We investigate the mass distribution of gas clouds in dwarf galaxies. Since clusters form from collapsing gas clouds, we expect a similar feedback dependence in both distributions, so we hypothesize that pre-SN feedback yields fewer high-mass clouds. To test this, we use an isocontour analysis at cutoff densities of $10,\ 10^{1.5},\ 10^{2}$ cm$^{-3}$ to identify clouds in dwarf galaxy simulations run with the RAMSES adaptive mesh refinement code. We calculate mass distributions for models with different combinations of SNe, stellar winds, and ionizing radiation. We find that the mass distribution for clouds with $n>100$ cm$^{-3}$ is independent of feedback, but the distribution for complexes with $n>10$ cm$^{-3}$ is more top-heavy in the presence of radiation. Winds do not affect the distribution at any scale. This contradicts our hypothesis that cloud and cluster mass distributions respond similarly to feedback. Instead, the dense cloud mass function shows no feedback dependence, suggesting its shape is set by gravity. We conclude that the cluster mass function must be shaped by intra-cloud feedback regulating star formation and, in the case of radiation, effects on parent cloud temperature. (shortened)

Figures (9)

• Figure 1: Cluster initial mass function, normalized to the total number of clusters (defined as stellar groups with ages $<25$ Myr) found in each simulation after 300 Myr. The error bars indicate the 16th and 84th percentiles computed using bootstrapping. The dashed line shows the slope of scale-free formation models, often associated with the initial cluster mass function (from andersson24starclusters, where more details are provided).
• Figure 2: Root node contours in projected gas density plots for different simulations (SNe in the top row, SNe+radiation in the bottom row) and density thresholds ($n >$10 cm$^{-3}$ in the left column, and $n>10^2$ cm$^{-3}$ in the right column). As expected, astrodendro selects fewer and smaller structures at a higher density cutoff.
• Figure 3: Mass distributions for cloud complexes with $n>10$ cm$^{-3}$ (top), big clouds with $n>10^{1.5}$ cm$^{-3}$ (middle), and clouds with $n>10^2$ cm$^{-3}$ (bottom). The lines show the median values, while the shaded regions show the 16th--84th percentile range. All distributions shown contain structures from multiple snapshots, where consecutive snapshots are 25 Myr apart. Dashed line shows ${\rm d}N/{\rm d}\log(M) \propto M^{-1}$, i.e., ${\rm d}N/{\rm d}M \propto M^{-2}$.
• Figure 4: Cloud mass distributions in 175 Myr time bins. Feedback is encoded through color as in Figure \ref{['fig:massfunction']}, while the number of clouds in each distribution is given in the legend. The lines show the median values, while the shaded regions show the 16th–84th percentile range.
• Figure 5: Gas fraction $f_{\rm gas}$ (Eq. \ref{['eq:gas-frac']}) throughout time for all simulations as shown in the legend.