Physical properties and H-ionizing-photon production rates of extreme nearby star-forming regions

Measurements of the galaxy UV luminosity function at z>6 suggest that young stars hosted in low-mass star-forming galaxies produced the bulk of hydrogen-ionizing photons necessary to reionize the intergalactic medium (IGM) by redshift z~6. Whether star-forming galaxies dominated cosmic reionization, however, also depends on their stellar populations and interstellar medium properties, which set, among other things, the production rate of H-ionizing photons, $\xi_\text{ion}^\star$, and the fraction of these escaping into the IGM. Given the difficulty of constraining with existing observatories the physical properties of z>6 galaxies, in this work we focus on a sample of ten nearby objects showing UV spectral features comparable to those observed at z>6. We use the new-generation Beagle tool to model the UV-to-optical photometry and UV/optical emission lines of these Local 'analogues' of high-redshift galaxies, finding that our relatively simple, yet fully self-consistent, physical model can successfully reproduce the different observables considered. Our galaxies span a broad range of metallicities and are characterised by high ionization parameters, low dust attenuation, and very young stellar populations. Through our analysis, we derive a novel diagnostic of the production rate of H-ionizing photons per unit UV luminosity, $\xi_\text{ion}^\star$, based on the equivalent width of the bright $[\text{OIII}] \lambda 4959,5007$ doublet, which does not require measurements of H-recombination lines. This new diagnostic can be used to estimate $\xi_\text{ion}^\star$ from future direct measurements of the $[\text{OIII}] \lambda 4959,5007$ line using JWST/NIRSpec (out to z~9.5), and by exploiting the contamination by $\text{H}\beta + [\text{OIII}] \lambda 4959,5007$ of photometric observations of distant galaxies, for instance from existing Spitzer/IRAC data and from future ones with JWST/NIRCam.
Comment: 10 pages, 4 figures, 5 tables, accepted for publication on MNRAS