Structure of stationary photodissociation fronts

The structure of stationary photodissociation fronts is revisited. H-2 self-shielding is discussed, including the effects of line overlap. We find that line overlap is important for N(H-2) greater than or similar to 10(20) cm(-2), with a factor-of-2 suppression of pumping rates at column densities N(H-2) approximate to 3 x 10(20) cm(-2). We compute multiline UV pumping models and compare these with simple analytic approximations for the effects of self-shielding. The overall fluorescent efficiency of the photodissociation front is obtained for different ratios of chi/n(H) (where chi characterizes the intensity of the illuminating ultraviolet radiation) and different dust extinction laws. The dust optical depth tau(pdr) to the point where 50% of the H is molecular is found to be a simple function of a dimensionless quantity phi(0) depending on chi/n(H), the rate coefficient R(T) for H-2 formation on grains, and the UV dust opacity. The fluorescent efficiency of the photodissocation region (PDR) also depends primarily on phi(0) for chi less than or similar to 3000 and n(H) less than or similar to 10(4) cm(-3); for stronger radiation fields and higher densities, radiative and collisional depopulation of vibrationally excited levels interferes with the radiative cascade. We show that the emission spectrum from the PDR is essentially independent of the color temperature T-color of the illuminating radiation for 10(4) K less than or similar to T-color but shows some sensitivity to the rotation-vibration distribution of newly formed H-2. The 1-0 S(1)/2-1 S(1) and 2-1 S(1)/6-4 Q(1) intensity ratios, the ortho/para ratio, and the rotational temperature in the nu = 1 and nu = 2 levels are computed as functions of the temperature and density for different values of chi/n(H). We apply our models to the reflection nebula NGC 2023. Apparent inconsistencies between published K-band and far-red spectroscopy of this object are discussed; we adjust the two sets of observations for consistency. We are able to reproduce approximately the (adjusted) observations with models having chi = 5000, n(H) = 10(5) cm(-3), and a reasonable viewing angle. Further observations of NGC 2023 will be valuable to clarify the uncertain spatial structure of the emission.