Imagine a planet where the atmosphere hasn't *quite* got enough greenhouse effect to maintain its temperature and pressure. That atmosphere would collapse into surface liquids and/or solids, depending on the equilibrium temperature. Planets with eccentric orbits would experience this sort of effect every "year" as they approach and then recede from their primary (e.g. Pluto).

Now, consider the effect of a giant impact into an icy-surfaced planet - there is an immediate thermal effect, boiling off ices in the target zone, in the shocked zone around it, and in re-entry debris and secondary impact.

In a timescale of minutes to hours, the impact can deliver to the surface of the planet more energy than the sun gives in years or decades.

These ices become, temporarily, surface liquids and atmospheric gases. If enough gas is produced, the atmosphere can be flipped into a new metastable state. Until the atmosphere loses its energy, it will remain as a thicker, warmer, wetter system.

The lifetime of such an atmosphere might be measured in days, or in kiloyears. What would encourage it to remain? Well, sources of heat and gas are a good way. And one very probable source of these is the slow melting and escape of subsurface ice in and around the impact site and the impact debris layer, as heat conducts from glowing hot areas to icy cold ones. It takes a long time for an impact melt sheet to cool off - the valley of 10,000 smokes was degassing for decades after a mere volcanic eruption. Impact scientists estimate that major imapct melt pools and sheets would have taken thousands of years to cool, during which time they are delivering heat to the atmosphere.

Norman Sleep, in his studies of the Early Earth, emphasised the different impact-induced atmosphere that would have existed at a time when impacts were laarge enough and frequent enough for their thermal effects to overlap.

Scientific papers have been published on this for Mars by Tony Colaprete and his student, Theresa Segura. See this LPSC abstractfor an example. (or this paper. There are some issues here - they have looked at too recent an epoch on Mars and at impacts that are too small, but it is "a well-known fact" that the giant early impacts such as formed the Isidis basin or even Hellas, were so large that the planet was transformed across its whole surface. Somewhere between these extremes lies an epoch on Mars when Ephemeral atmospheres formed and collapsed and reformed. Each collapse led to torrential downpours of relatively warm water onto an icy planet, prior to the final snowstorm and atmospheric collapse.

See also my early ideas on this theme. Anda pre-canned Google search to delve deeper...