I was experimenting with mixed solvents and found that if you create a box around a molecule say 10A on a side, then de-select "remove solvent" and then create a second box, say 2A on a side, that second box forms around the first. You can then constrain the second box which will for the most part retain the first and the parent molecule during a simulation. This allows visualization of the live trajectory as well. I used MON around water and he process is faster to create than selecting the six outside surfaces of a solvent box. Paul