From the snapshot it looks that water is attached to the complex. what is the distance of Re-OH2?

Mrs/Miss/Mrs LS,

Could you upload the output error section?

I agree with Debasish Mandal and Mutasem Z. Bani-Fwaz. You need to check the distance in between the metal and water.

Sorry I meant the phenanthroline molecule is no longer attached to the Re complex after optimisation. Does anyone know what the problem could be? Thank you everyone!

I guess there is no problem, the change in structure is due to the improper input (initial structure). Moreover, the bonds are weakened not disappeared. You need to assign the right bond lengths and angles in your input structure.

I am also new in this field and believe me, I had also faced this question.

I don't mean to be dismissive, but what are you attempting to accomplish by optimizing such a structure? You have a goodly number of heavy atoms and a metal that makes things worse. There are many degrees of freedom and it is quite likely that the minimization process will find unnatural ways to organize the atoms to reduce the overall energy. As the optimization is taking place in vacuum, there will be self interactions of components of the molecule with itself that would not arise in solution, where those components would interact with the solvent. In that case, what you really require is the ensemble-averaged structure in solution, not the vacuum-optimized structure. So, what are your objectives?

If you are looking to create parameters for a molecular mechanics simulation, these are always done with small subsets of the whole. If you are looking to perform a QM/MM or even QM only simulation of some reaction mechanism, you have omitted all of the solvent in your vacuum optimization. Embedding the molecule in solvent will alter the structure but you won't necessarily care what the optimized structure is, only the energy differences along some pathway.

As a beginner, resist the urge to just do calculations. It can be comforting to become confident that you can execute them and thrilling to discover that you've burned 10,000 CPU hours in the interim but it is quite easy to do calculations that are not useful. It is, unfortunately, quite difficult to perform calculations that can be compared with experiment and provide real utility.

As a technical note, you would probably obtain better results with the LANL basis sets for Re and B3LYP might not be the best choice but it is a common one. Coupled cluster calculations might improve your results as well but, again, what are you trying to accomplish? Once that is defined, you can fiddle with basis sets and methods and struggle to get things to complete without error within the computational resources at your disposal.

Dear Mark, I am attempting to look at the excited state properties of this complex by using TD-DFT. You have raised very good points that I haven't considered, I will try to do the optimisation using LANL basis and water as the solvent.

May I ask why B3LYP is not a good choice?

Thank you.

B3LYP is a safe choice. I have used B3LYP almost exclusively but I have been most interested in non-metallic, enzymatic processes. It will provide reasonably good results and is widely used. There have been numerous methods developed since the introduction of B3LYP that can buy you a factor of two or so in accuracy but not orders of magnitude. Authors of such methods would undoubtedly question my blase attitude toward their efforts. You'd have to investigate if other methods would improve your results but you should be aware that other choices exist.

Most methods are tested by folks like Larry Curtiss against a relatively small (hundreds) number of small molecules. (See G3 or G4 test set.) They compile lists of enthalpies of formation, etc., for the various hybrid methods and their latest ab initio methods. Some methods work better for some systems than others. You might find one better than B3LYP for your system.

I don't have much experience with excited states but understand that TD-DFT provides uncommonly good results for excited states. So, I wouldn't worry initially about B3LYP or not B3LYP but it will be important to include solvent. Bon voyage!

I agree with Mark in that you might have better chance with lanl2dz however, I am not sure if including solvent model will improve the optimization process. I think the problem has to do with Re. Because of the large size of Re relativistic effects become important. Effective core potentials do take into account relativistic effects but they might not have been benchmarked against this particular molecular structure. Higher level calculations might help. In TD-DFT though solvent model does affect the result.