:orphan: .. _case_CmF: =========================== Electronic structure of CmF =========================== We want to investigate the structure of the diatomic molecule CmF (Curium Fluoride). Following :cite:`Mochizuki_JCP2003` we assume that curium has valence configuration in :math:`5f^77s^2` in the molecule, in particular since curium is a late actinide and thus quite similar to lanthanides. The molecule will therefore have an atomic-like open :math:`5f^7` shell. This may cause convergence problems since DIRAC by default assumes open-shell orbitals to have energies above close-shell ones, which is likely to not be the case here. In order to guide DIRAC we shall therefore employ the following strategy: 1. We calculate the curium atom. 2. We import the :math:`5f^7` orbitals into the molecular calculation, place them in the open shell and keep them frozen in a first SCF calculation. 3. We then freeze all closed-shell orbitals, allowing the :math:`5f^7` orbitals to relax in the molecule. The curium atom ~~~~~~~~~~~~~~~ The ground state configuration of curium (Cm, Z=96) is :math:`[Rn]5f^76d^17s^2` (see `here `_ for general information and `also here `_), so with two open shells. We follow a two-step procedure to assure convergence: 1. We calculate the trivalent cation :math:`Cm^{3+}` with electron configuration :math:`[Rn]5f^7`. 2. Once we have identified the :math:`6d` and :math:`7s` orbitals amongst the virtuals we can easily calculate the neutral atom using reordering of orbitals and overlap selection. For the molecular calcualtion we do not neccessarily need the fully converged :math:`5f` orbitals of the neutral atom, but we follow this strategy in part for educational purposes. For ease of analysis we shall impose linear symmetry also in the atomic calculation. We achieve this through the introduction of a ghost center .. literalinclude:: Cm.xyz The menu file for the trivalent cation then reads .. literalinclude:: CmIII.inp Note that 1. we perform Mulliken population on all positive-energy orbitals. 2. we use the keyword SHELL to only indicate atomic shell type since we do not need more detailed information 3. we specify that the ghost center has no basis. We run the calculation using:: pam --inp=CmIII --mol=Cm.xyz --get "DFCOEF=cf.CmIII" (keywords for parallel run omitted). Here is a snippet of the Mulliken population analysis .. literalinclude:: CmIII.pop One clearly sees that the fractionally occupied :math:`5f` orbitals are followed by first virtual :math:`7s`, then virtual :math:`6d` orbitals. We want to reorder them as :math:`\left(5f,7s,6d\right)\rightarrow\left(5f,6d,7s\right)` such that both open shells follow the closed :math:`7s` shell in the neutral atom. We therefore set up the input file .. literalinclude:: Cm.inp This assures that the orbitals are correctly ordered when we start the SCF iterations. In order to keep them in the desired order, we invoke overlap selection. We impose non-dynamic overlap selection, which means that orbitals in a given iteration are selected based on their overlap with the *initial* orbitals. We run the calculation using:: pam --inp=Cm --mol=Cm.xyz --get "DFCOEF=cf.Cm" --put "cf.CmIII=DFCOEF" This calculation converges smoothly in 13 iterations, and we are now ready to attack the molecule. The molecular calculation: ~~~~~~~~~~~~~~~~~~~~~~~~~~ 1. Importing curium :math:`5f` orbitals --------------------------------------- The menu file for the first step of the molecular calculation reads .. literalinclude:: CmF_5f_frz.inp In the SCF section it may be noted that we specify 49 closed-shell orbitals, followed by seven open-shell ones, with seven active electrons. Next we specify that we will take orbitals 45 to 51 from the coefficient file of the cerium atom, that we name AFCMXX, put them in position 50 to 56 and keep them frozen. This also implies that we are effectively doing a closed-shell calculation, which helps convergence as well. We run the calculation using the command:: pam --mw=200 --inp=CmF_5f_frz --mol=CmF.xyz --put "cf.Cm=AFCMXX" --get "DFCOEF=cf.CmF_5f_frz" The calculation converges in 18 iterations. Here is a snippet from the population analysis: .. literalinclude:: CmF_5f_frz.pop It shows that the upper closed-shell orbital (49) is dominated by cerium :math:`(7)s`. The open-shell cerium :math:`5f` orbitals are frozen and keep their energies from the atomic calculation, with :math:`5f_{5/2}` and :math:`5f_{7/2}` orbitals having energies -0.5359 and -0.4840 :math:`E_h`, respectively. 2. Relaxing the curium :math:`5f` orbitals ------------------------------------------ We now want to relax the curium :math:`5f` orbitals in the molecule whilst keeping the other orbitals frozen. This step is not necessarily requred, but allows us to see how the :math:`5f` orbital energies evolve. We use the following input file .. literalinclude:: CmF_5f_relax.inp It will be seen that we now import the coefficient file of the previous run and freeze all closed-shell orbitals. We use the command:: pam --mw=200 --inp=CmF_5f_frz --mol=CmF.xyz --put "cf.CmF_5f_frz=DFCOEF" --get "DFCOEF=cf.CmF_5f_relax" The calculation converges in 14 iterations. Again a snippet from the Mulliken population analysis: .. literalinclude:: CmF_5f_relax.pop We note that the curium :math:`5f` orbital energies are split by the molecular field, also that they are somewhat lowered. We also note that orbitals 46 - 48, dominated by fluorine :math:`2p` orbitals, are almost degenerate with the curium :math:`5f` orbitals, whereas orbital 50 is an almost pure curium :math:`7s` orbital.