Hello Robert,
at first sight, Turbomole seems to work like most quantum chemistry programs do,
but the usage is not:
input file -> run one (always the same) program -> results are somewhere in the output file
while the input file is unchanged! The control file (and its referenced files like coord, mos, ...) contains the input and the results of the last job. After a geometry optimization, the coordinates are those of the minimum structure, the orbitals are those of the SCF step of the minimum structure, and so on. The usage of Turbomole is more application or data oriented (if you do a spell check in a word processing program, you will not get a new text in a new file
)
After the SCF step, your orbitals are already the converged ones. You can easily check that. Either open the orbital file (mos or alpha, beta) in an editor, or much quicker:
head mos or
head alpha for UHF will either show:
$scfmo expanded format(4d20.14)
expanded is for start orbitals which you get when you generate some by define.
$scfmo scfconv=7 format(4d20.14)
are converged orbitals with 10^-7 Hartree SCF convergence
If SCF did not converge within the allowed maximum number of iterations, they keyword will look like this:
$scfmo scfdump=30 format(4d20.14)
those are orbitals which have not been converged after 30 SCF iterations. If you now simply restart the energy calculation, dscf or ridft will read in the orbitals and continue at that point - so after another 3 or 10 or 30 iterations it will converge, otherwise the number behind scfdump will be e.g. 60 (two times a maximum of 30 iterations).
The
use command in define is to project already converged orbitals from a former step to either a new basis set (useful if you first do an SCF step with a small basis set and then use those orbitals for a larger basis set) or from a job with different symmetry (if you e.g. lower the symmetry and want to re-use the symmetric orbitals). But
use in define will will always switch from '$scfmo scfconv=<N>' to '$scfmo expanded'.
From your post it is not clear if you did change the basis set (then a new SCF calculation is needed anyway), or not (then you do not need to run define at all).
Regards,
Uwe