....how warm could the soil particles be I wonder? After all we've seen that frosts can form on the surface, which should be the warmest part of the soil. My understanding might be off here, but wouldn't a forming frost involve the deposition of monolayers of freely moving water molecules to start with? I hesitate to use the term 'liquid'.

Can frosts form, at the temperatures and pressures seen at the phoenix site, without these 'liquid' monolyers?

This does seem pretty mysterious to me. I'd be very interested to know what the exact mechanism keeping the soil from interacting with the vapour is. I think some time in a mars simulation chamber would be a good place to start (wish I had half a million quid and a year to build one). Maybe the soil has insulating properties that keep the lower layers warmer than the surface?

Or perhaps as bruce suggested there is a static charge that repels the water, I believe water has a positive charge on areas of its envelope, due to the H atoms having to share their only electron with the oxygen.