The "bubbly" interstellar medium as origin for the inhomogeneous internal metallicity distributions in large disk galaxies

TL;DR

The paper addresses the origin of small-scale metallicity fluctuations in local disk galaxies, which are not readily explained by well-mixed-ISM models. It introduces a physically motivated superbubble scenario, wherein tightly clustered core-collapse SNe create metal-enriched bubbles that expand until breaking out at the disc scale height, and it formalizes this with a four-parameter geostatistical forward model fitted to PHANGS-MUSE maps. The main findings are that the characteristic fluctuation scale is φ ≈ $300$ pc and the local star-formation efficiency in superbubble regions is ε ≈ $0.1$–$0.2$, with a metallicity amplitude ΔZ ≈ ε y and y = 0.015, and that these small-scale parameters correlate with global galaxy properties (φ with $M_*$ and SFR; ε with $M_*$). The significance lies in linking SN feedback and disc structure to chemical enrichment, yielding testable predictions (e.g., direct $T_e$ metallicities, velocity signatures, dust holes) and offering a framework to infer disc scale heights and ISM burstiness from resolved metallicity maps.

Abstract

Resolved metallicity studies of local disk galaxies have revealed that their interstellar media (ISMs) are far from chemically homogeneous, displaying significant ($\sim 0.05$ dex) variations in the metallicity on characteristic scales of a few hundred parsecs. Such data is at odds with most analytical models, where the ISM is predicted to be more well-mixed. Here, we suggest that the observed small-scale features seen in galaxies may be superbubbles of metal-enriched gas created by a collection of core collapse supernovae with tight spatial (and temporal) correlation. In this scenario, the size of the metallicity fluctuations (superbubble radius, $φ$) is set by the disk scale height of the galaxy in question (after which point shock breakout favours preferential expansion along directions perpendicular to the dense disc), and the amount of additional metals contained within a fluctuation is proportional to the star formation efficiency in superbubble regions ($ε$). To test this theory, we analysed metallicity maps from the PHANGS-MUSE sample of galaxies using a geostatistical forward-modelling approach. We find $φ\simeq 300$ pc and $ε= 0.1-0.2$, in good agreement with our theoretical model. Further, these small-scale parameters are found to be related to the global galaxy properties, suggesting that the local structure of the interstellar medium of galaxies is not universal. Such a model of star formation paints a new picture of galaxy evolution in the modern universe: in large local galaxies, star formation appears steady and regular when averaged over large scales. However, on small scales, these large galaxies remain intrinsically bursty like their smaller, high-redshift counterparts.

Figures (7)

• Figure 1: An illustration of the ISM model we propose as the explanation for the metallicity fluctuations observed in Metha+21, inspired by the "chimney" model of Norman+Ikeuchi89. Left: A burst of star formation occurs in a dense pocket of gas, leading to $10-100$ core collapse supernovae occurring at a similar location and time. Centre: These supernovae create a spherical region of expanding hot metal-enriched gas, shown in red. This sphere of hot gas expands until it reaches the scale height of the galaxy (right), after which point the hot gas preferentially expands away from the high-density galaxy disc. This process would leaves a metal-enriched pocket of gas as an observable, with the lower metallicity ambient gas of the ISM blown away from the site of the supernovae.
• Figure 2: The metallicity distribution of NGC 1385, computed using the O$_3$N$_2$ metallicity diagnostic of Curti+17. In this Figure, both a large scale trend in decreasing metallicity (corresponding to a negative metallicity gradient) and small-scale (sub-kpc) fluctuations (which we argue correspond to face-on superbubbles) can be seen.
• Figure 3: Small scale parameters for the metallicity fluctuations identified in the PHANGS galaxy sample, using the O$_3$N$_2$ metallicity diagnostic. The top panel shows how the characteristic size of fluctuations in a galaxy, $\phi$, changes throughout the galaxy sample. Galaxies here are ordered from the smallest (left) to greatest (right) stellar mass. The dashed black line in the top panel is the median scale height of thin stellar discs in the galaxy sample presented in Appendix B of Comeron+18, with the dark- and light-shaded grey regions indicating the $1\sigma$ and $2\sigma$ spreads in this population. We see that values of $\phi$ are generally consistent with typical values of the scale heights of galaxies. In the lower panel, we plot the estimates of $\epsilon$, the star formation efficiency within the regions which become superbubbles. With this metallicity diagnostic, we find typical star formation efficiencies of $0.1-0.2$, indicating that highly efficient star formation occurred in order to produce the metallicity fluctuations observed.
• Figure 4: Moderate correlations can be seen between the small-scale properties of the metallicity fluctuations seen in galaxies and the global properties of the galaxies. In the top panel, we show the positive correlation between stellar mass and the star forming efficiency of gas in superbubbles, $\epsilon$$(\rho=0.42)$. In the middle and lower panels, we show the positive correlations of $\phi$, the characteristic spatial scale of metallicity fluctuations, compared to two global galaxy properties: the stellar mass of galaxies (middle panel; $\rho=0.53$), and their star formation rates (lower panel; $\rho=0.54$).
• Figure 5: Two-point correlation functions revealing the spatial structure of the star formation profile residuals (red) and the metallicity residuals (blue) of NGC 1385, computed using H$\alpha$ brightness maps and the O$_3$N$_2$ diagnostic, respectively. Stars indicate the locations at which each of these correlation curves first drop below a value of $0.15$. We see that the correlation function for SFR approaches zero more rapidly than the metallicity correlation function, indicating that SFR is correlated on smaller spatial scales than metallicity, in line with our theory.