In nuclear reactor accident safety studies, the radiological source term is a metric that quantifies the release of radiological material from the reactor to the environment. The present work evaluates heat transfer between high-temperature vapor bubbles and the surrounding coolant and the effect these interactions have on the source term for postulated core disruptive accident scenarios associated with an oxide-fueled, liquid metal–cooled fast reactor class. It is shown that aerosol particle size can influence heat transfer, and it is suggested that the extent of the influence depends on the fineness of the particles in the aerosol. The results are consistent with legacy experiments conducted in the Fuel Aerosol Simulant Test (FAST) facility at Oak Ridge National Laboratory and offer a more comprehensive assessment of vapor condensation by treating the bubble constituents, in the context of radiation heat transfer, as participating media. The model, which couples classical scattering theory to the equation of radiative transfer and the energy equation, provides a means for estimating size-affected radiative cooling times. Solutions are obtained via the P-1 method of spherical harmonics with improved, higher-order boundary conditions. Outcomes include the development of an “extinction-time ratio” criterion for assessing whether ejection of aerosol from the bubble to the cover region is likely. Aerosol release from the coolant pool is evaluated using this criterion with the potential to extend this work to reactor-scale accidents. A baseline evaluation is provided that shows that omission of participatory effects could lead, in a relative sense, to cooling time offsets in excess of 14%. In addition to enhancing previous evaluations of FAST results, these modeling outcomes contribute to knowledge management efforts aimed at developing a more mechanistic assessment of the source term while suggesting potential enhancements to severe accident safety analysis through the use of more comprehensive radiative heat transfer models.