If all calculations were correct (the last SCF converged), all data should be in the gbw file.Try opening the converted molden in some other program, for example Gabedit. What is the size of a molden file?

Thanks for your reply Andrey. I have tried manually converting the .gbw file using orca_2mkl, but the file it produces has 0 kb. The .gbw files produced by the QM/XTB calculations are all around 850 kb, and the calculations have converged when I look through the .out file made.

850 kb is very little for 15 atoms with triple-zeta basis.

But I think I understood what the problem is. When using the ONIOM model and adding the keyword "%xtb ... end" ORCA considers that QM = XTB and you get that both layers are calculated by the XTB method

Ahh I see... I was following the ORCA tutorial, and added the %xtb as I expect that it'll need more than the default 250 cycles to optimise the QM/XTB. Would something like this overcome this issue?

!QM/QM2 LC-PBE def2-TZVPPD

%QMMM QM2CUSTOMMETHOD "xtb"

%xtb

XTBINPUTSTRING "--iterations 2000"

end

end

QMATOMS {0:14}

I doubt that this will help.

It would be better to try to obtain a geometry in advance, the self-consistency for which will converge in less than 250 iterations