2012ApJ...757..108Y

The ionization parameter U is potentially useful as a tool to measure radiation pressure feedback from massive star clusters, as it directly reflects the ratio of radiation to gas pressure and is readily derived from mid-infrared line ratios. We consider a number of physical effects which combine to determine the apparent value of U in observations encompassing one or many H II regions. An upper limit is set by the compression of gas by radiation pressure, when this is important. The pressure of shocked stellar winds and the presence of neutral clumps both tend to reduce U for a given intensity of irradiation. The most intensely irradiated regions are selectively dimmed by internal dust absorption of ionizing photons, leading to a bias for observations on galactic scales. We explore these effects in analytical and numerical models for dusty H II regions and use them to interpret previous observational results. We find that radiation pressure confinement sets the upper limit log₁₀(U) ~= -1 seen in individual regions. Unresolved starbursts are known to display a maximum value of ≃ - 2.3. While lower, this is also consistent with a large portion of their H II regions being radiation pressure dominated, given the different technique used to interpret unresolved regions, and given the bias caused by dust absorption. We infer that many individual, strongly illuminated regions cannot be significantly overpressured by stellar winds, and that even when averaged on galactic scales, the shocked wind pressure cannot be large compared to radiation pressure. Therefore, most H II regions cannot be adiabatic wind bubbles. Our models imply a metallicity dependence in the physical structure and dust attenuation of radiation-dominated regions, both of which should vary strongly across a critical metallicity of about one-twentieth solar.