Neutral gas heating by X-rays in primitive galaxies: Infrared observations of IZw18 with Herschel
(abridged) The dominant thermal mechanisms in the neutral interstellar medium, which acts as a star-forming gas reservoir, are uncertain in extremely metal-poor galaxies. Our objective is to identify the heating mechanisms in one such galaxy, IZw18, and assess the diagnostic value of fine-structure cooling lines. We also seek to constrain the mass of H$_2$, which, despite being an important catalyst and tracer of star formation, remains elusive in this object. Building on a previous photoionization model within a multi-sector topology, we provide additional constraints from the [CII] and [OI] lines and the dust mass recently measured with Herschel. The heating of the HI region appears to be mainly due to photoionization by radiation from a bright X-ray binary source, while photoelectric effect (PE) is negligible. The [CII] and [OI] lines imply an average X-ray luminosity of $4\times10^{40}$ erg s$^{-1}$, while the [NeV] upper limits bring strong constraints to the soft X-ray flux arising from the binary. A negligible amount of H$_2$ is predicted, but $\lesssim10^7$ M$_\odot$ of H$_2$ may be hidden in sufficiently dense clouds of order $\lesssim10$ pc in size. Regardless of the presence of significant amounts of H$_2$, [CII] and [OI] do not trace the so-called CO-dark gas, but the almost purely atomic medium. Although the [CII]+[OI]/TIR ratio is close to values found in more metal-rich sources, it cannot be safely used as a PE heating efficiency proxy. This ratio seems to be kept stable due to a correlation between the X-ray luminosity and the star-formation rate. We propose that X-ray heating is an important process in extremely metal-poor sources. The weak PE heating due to the low dust-to-gas ratio could be compensated for by the larger occurrence and power of X-ray binaries in low-metallicity galaxies. We speculate that X-ray heating may quench star formation.