Cu + organic molecule crystal

[pierre.an...@gmail.com]

Dear CP2K users,

I am having a hard time with a crystal containing six copper atoms, six carnosine molecules, and twelve water molecules. A carnosine molecule forms three bonds with a copper atom and one with another. The carnosine molecule of the other copper form a fourth bond with the first copper atom. Then a water molecule forms a fifth bond with the copper atom. There is no sixth bond. Each copper atom is in the second oxidation state (Cu2+). It comes that an isolated copper with its five ligands is a doublet. This is confirmed by a Gaussian calculation.

From the description above, it appears that the copper atoms come as a pair as they are bound by the same carnosine molecules. Gaussian calculations on the pair show it is in a singlet state (singlet 4 kcal/mol lower than the triplet). However, when running the crystal with CP2K, the system runs only if set into a triplet state. Furthermore, VASP fails at running it. I cannot find any issue with the crystal (not missing hydrogen or anything of the sort).

CP2K converges and runs an optimization without issue when the system is set in a triplet state. Yet, I have a hard conceiving that a crystal made of three pairs of Cu+carnosine, each pair being a singlet, could be anything but a singlet. What could be the reason for such a system to be a triplet instead? Is there any issue with the parameters I used for the calculation? Substituting copper with zinc leads to well-behaved singlet crystals.

I am looking forward to your insights. To that end, I attach the restart file of a 2x2x2 supercell calculation and a POSCAR (easier to visualize the crystal with VESTA).

Regards,

Pierre