Hi,
there is not need to define dummy atoms for SiNCS, since define uses internal coordinates, not Z-matrices.
By the way, dummy atoms are allowed in define, the are called 'q', but they are ignored when defining internal coordinates.
The probably simplest way to do what you want is to read in both molecules and let define generate automatically internal coordinates. If you generate non-redundant ones, just say iaut in the geometry menu and then *. Check with imet (after that the number of internal coordinates #k= must be the same as the number degrees of freedom #ideg=) if they are linearly dependent and look at them (disi). If you find a simple stretch from one fragment to the other one, fix all other internal coordinates by entering (i and then ic).
Example:
If you have, say, 20 internal coordinates, and coordinate number 11 is the one you would like to optimize while fixing all other ones, the command in the INTERNAL COORDINATE MENU would be
ic 1-10,12-20 f
After that, the first line in your current menu should look like that:
INTERNAL COORDINATE MENU ( #ideg=20 #k=1 #f=19 #d=0 #i=0 )
while the d and i numbers might be different and can be ignored.
Another idea would be not to perform a geometry optimization, but do single-point jobs with different intermolecular distances.
Regards,
Uwe