grimme-lab
Zr - F failed interactions
Describe the bug When running xtb optimization of defective MOF-808 (with H2O and OH) with GB (sarin), F - P bond breaks and F - Zr forms. In addition, an OH from the defective MOF site bonds to the Phosphorus. Thus a chemical reaction takes place during an optimization calculation.
To Reproduce Steps to reproduce the behaviour:
- Simply run the xtb optimizer using the same system. (all files attached)
- input structure for both calculations are the same.
Expected behaviour The optimization was expected to move GB closer to the defective cluster.
Additional context Using the same setup as the xtb optimization calculation, I ran an orca optimizer which yielded the expected behaviour. all files are attached.
Hi, and thank you for your interest in the xTB program. I can confirm this is happening for your attached system. I tested with the bleeding edge version a57fabb and the "reaction" takes place. I also tested this with the Native GFN2-xTB implementation in ORCA 6.1.0, and there it also happens, meaning that this is a potential energy surface problem and not a geometry optimization algorithm problem. Note that you used in your ORCA calculation PBE-D3(BJ)/def2-svp as a level of theory, which is DF,T and GFN2-xTB is "only" tight-binding. This indicates that the GFN2-xTB PES is just too inaccurate. However I also tested the GFN-FF method (a force field), and there it seems to work. This is likely due to the fact that GFN-FF cannot form bonds.
