Probing the heating of the neutral atomic interstellar medium in the Dwarf Galaxy Survey through infrared cooling lines

Context. Star formation in galaxies is regulated by dynamical and thermal processes. In the Milky Way and star-forming galaxies with similar metallicity, the photoelectric effect on small dust grains usually dominates the heating of the neutral atomic gas, which constitutes the main star-forming gas reservoir. In more metal-poor galaxies, the lower dust-to-gas mass ratio together with the higher occurrence and luminosity of X-ray sources suggest that other heating mechanisms may be at play. Aims. We aim to determine the contribution of the photoelectric effect, photoionization by UV and X-ray photons, and ionization by cosmic rays to the total heating of the neutral gas in a sample of 37 low-metallicity galaxies. In particular, we wish to assess whether X-ray sources can be a significant source of heating. We also attempt to recover the intrinsic X-ray fluxes and compare them with observations when available. Methods. We used the statistical code MULTIGRIS together with a photoionization grid of Cloudy models propagating radiation from stellar clusters and potential X-ray sources to the ionized and neutral gas. This grid includes physical parameters such as metallicity, gas density, ionization parameter, and radiative source properties. We describe a galaxy as a combination of many 1D components linked by a few physical hyperparameters. We used infrared cooling lines as constraints to evaluate the most likely combinations and parameters. Results. We constrained the heating fractions for the main mechanisms for the first time in a low-metallicity galaxy sample. We show that for the higher metallicity galaxies, the photoelectric effect dominates the neutral gas heating. At metallicities below 1 /8 the Milky Way value, cosmic rays and photoionization can become predominant. We computed an observational proxy for the photoelectric effect heating efficiency on polycyclic aromatic hydrocarbons (PAHs) using the total cooling traced by [CII] 158 mu m + [OI] 63 mu m. We show that this proxy can match theoretical expectations when accounting for the fraction of the heating due to the photoelectric effect according to our models. Finally, we show that it is possible to predict the X-ray fluxes reasonably well in the 0.3-8 keV band from the gas cooling lines for most of the galaxies observed in this band. With the current grid and assumptions, determining the exact heating fraction due to cosmic rays remains difficult, but we speculate that heating from X-ray sources is more important. Conclusions. As expected from the low abundance of dust and PAHs in metal-poor galaxies, heating mechanisms other than the photoelectric effect heating must be accounted for. Bright X-ray sources may deposit their energy on large scales in such transparent, dust-poor interstellar medium, and thus they represent promising avenues to understand the physical properties of the main star-forming gas reservoir in galaxies. The modeling strategy adopted here makes it possible to recover the global intrinsic radiation field properties when X-ray observations are unavailable, such as in early universe galaxies.