2020ApJ...889L..22C

An analysis of the dynamics of a star formation event is performed. It is shown that galaxies able to drive leftover gas to sufficient altitudes in a few million years are characterized by two basic properties: small sizes (<=1 kpc) and high star formation rate (SFR) surface densities (Σ_SFR≥ 10M_☉yr^–1kpc^–2). For the parameter space of relevance, the outflow is primarily driven by supernovae with radiation pressure being significant but subdominant. Our analysis provides the unifying physical origin for a diverse set of observed Lyman continuum photons (LyC) leakers, including the green-pea galaxies, [S II]-weak galaxies, and Lyα emitters, with these two characteristics as the common denominator. Among verifiable physical properties of LyC leakers, we predict that (1) the newly formed stellar masses are typically in the range of 10⁸-10¹⁰M_☉, except perhaps ultra-luminous infrared galaxies (ULIRGs), (2) the outflow velocities are typically in the range typically of 100-600 kms^–1, but may exceed 10³kms^–1 in ULIRGs, with a strong positive correlation between the stellar masses formed and the outflow velocities, (3) the overall escape fraction of galaxies is expected to increase with increasing redshift, given the cosmological trend that galaxies become denser and more compact with increasing redshift. In addition, two interesting by-product predictions are also borne out. First, ULIRGs appear to be in a parameter region where they should be prodigious LyC leakers, unless there is a large ram pressure due to infalling gas with a rate exceeding about 30 times the SFR. Then, toward the tail end of a ULIRG event when the ram pressure relents, advanced ULIRGs are expected to leak more LyC photons than earlier ULIRGs. Second, Lyman-break galaxies (LBGs) are not supposed to be prodigious LyC leakers in our model, given their claimed effective radii exceeding 1 kpc. Thus, if LBGs are observed to have LyC leakers, it may be that the effective radii of their star-forming regions have been overestimated by a factor of 2-4.