Aims. Understanding the internal structure of HII regions is fundamental to constraining star-formation processes in galaxies. We investigate how the filling factor (FF) relates to luminosity, size, electron density, and Hα equivalent width in HII regions. We also explore the association between star-forming regions and PAH-to-dust emission, to assess their possible connection to the evolutionary stages of HII regions.
Methods. We investigate the FF in 622 HII regions in NGC 628, combining 475 from SIGNALS and 147 from PHANGS-MUSE, using a dedicated analysis pipeline. We derived luminosity, emission-line fulxes, radii, electron densities, and FF. We used PHANGS-JWST survey images to study the spatial relationship between the FF and PAH-to-dust emission, quantified by the PAH-to-dust ratio RPAH using JWST/MIRI emission bands at 7.7, 11.3, and 21 μm.
Results. Higher Log(FF) and high EW(Hα) values are found in luminous regions, whereas more extended regions with lower EW(Hα) display lower Log(FF) values. We show that the HII radius definition significantly affects the LHα–R relation. Low-luminosity, compact HII regions appear to reflect a transition from cluster-powered regions to nebulae ionized by single massive stars with the transition occurring around log(LHα) ∼ 37 erg s-1. The PAH-to-dust ratio RPAH correlates with the volumetric Hα luminosity density, LHα/R3, with a transitional range centered around log(LHα/R3) ∼ 32 erg s-1, corresponding to log(FF) ≈ -4.4. Regions with lower Log(FF) values exhibit higher RPAH values, suggesting less efficient PAH processing in more porous structures.
Conclusions. These results could be consistent with an evolutionary scenario in which the FF decreases with stellar cluster age in giant HII regions as they evolve toward fainter and more spatially extended states. The volumetric Hα luminosity density, LHα/R3, provides a useful representation that reduces the covariance between LHα and R induced by different region definition methods, enabling more consistent cross-catalog comparisons across segmentation approaches.