We examine the conditions under which ions impacting the Hermean surface can act as a regional source of enhanced atmospheric column through ion implantation and subsequent release at the surface. At most latitudes, energetic (a few keV) Na or K ions which impact the nighside surface, are released quickly (well before noon) upon warming of the surface. We show that the relative sunrise/sunset difference produced by ion implantation is {<n>}SR-<n>SS/<n>Ave=f, where f is the fraction of the photo-ions recycled, <n>SR is the average zenith column above the sunrise portion of the illuminated hemisphere, <n>SS is the average zenith column over the sunlit disk. Thus, to produce a large sunrise/sunset difference via ion implantation and subsequent release requires efficient (close to total) recycling. We show that the most extensive set of available data reduced to Na column abundance does not show any sunrise enhancement. We argue that the K data not permit an unambiguous interpretation in favor of sunrise/sunset differences.

We further find that if an efficient surface loss process for the alkali is not operating after sunrise, the initial Na and K distribution will relax into the bulk of the solid. Preserved abundance gradients of Na and K in lunar glasses suggest that the Arrhenius coefficients for impact glasses are likely more modest than those for laboratory glasses derived from rock by nonimpact processes, but the effect of the more modest diffusion rates is only to delay the efficient loss of Na by a few Earth days. We argue that implantation can lead to observable regional increases in the observed Na or K column densities only it occurs at very high latitudes, where diffusion is slow. It is typically lower energy ions which impact at high latitude and these are both more numerous than the high energy ions and possess smaller average penetration depths; thus there are additional reasons to favor a high-latitude locus for any possible prompt return related increases in zenith column. We find that we cannot rule out sputtering as a source process, as the low-energy ions are efficient sputterers. Finally, we argue that the observed Na/K ratio in the atmosphere may be the expression of their different asymptotic rates of loss from the interiors of the regolith grains. If so, this allows us to fix the importance of sputtering and photon stimulated desorption relative to impact vaporization.