If you are using Gaussian, you can use CASSCF=SpinOrbit keyword to calculate spin-orbit coupling between two spin states. For details: http://www.gaussian.com/g_tech/g_ur/k_casscf.htm
plenty of programs do SO corrections. i used Orca's MRCI program and it works fine and fairly easy to use, see the example input below.
1. you need to specific a "block" or multiple blocks with certain multiplicity or multiplicities, for ground state Br atom it is 2 (hence: NewBlock 2 *, also in xyz 0 2, which is similar to G09 input).
2. you then need to define the active space as with all other CAS-type calculations, basically counting the "important" electrons and orbitals (e.g., CAS(5,3) for the 5 p-electrons and 3 p-orbitals of Br).
3. lastly you should specify how many states you want for obtaining the SO term, i use 5 (Nroots 5) here for Br, which is probably already overkill but again it depends on your specific system.
You can get trial version of ADF, Francis. As far as I remember, it's free for 1 month, or even two. To me ADF (with ZORA program in it) is the best for calculating relativistic effects.
Dear Francis, I have your problem too. I want to know How we can calculate the spin-orbit coupling?is the Gaussian package suitable and does it give reliable outputs?
Francis any information which you have found on this topic would be regarded.
I don't remember that I had problems with units, as far as I understand, you can convert them by yourself. What I've got from my calculations (about 10 years ago, I think) were NMR Shieldings in ppm. I totally forgot details on that project, but I am sure it is possible to figure out. Usually I take some nicely done paper on similar study and try to repeat by myself one calculation. Google my publications with Dybowski, C. and see what was done.
Thanks a lot. I will follow your paper. I just need to see the procedure because this is the first time to calculate SOC. Therefore feel problem to use the concerned data from the out file.