Hello
The definition of the dipole
moment of a charged species is well defined. The only problem with it (and with
all higher multipoles) is, that its value depends on the origin of the
coordinate system. Thus for comparison purposes the origin of the coordinate
system of different molecules should be comparable. A very common choice is the
centre of mass; another possibility is the center of nuclear charge, which is
the standard origin in Gaussian09, if one does not specify the 'Nosymm'
keyword. You can also turn the problem on its head and define the origin as that
point where the dipole is zero...
I'm assuming that you want to compare the absolute value of the
dipole between different molecules. In case you also want to compare the
components, you also should use comparable orientations of the coordinate
systems (e.g the coordinate system where e.g. the inertia (or polarizability or
...) tensor is diagonal). This, however, is independent of being charged or
uncharged.