In my DFT calculations on nanoflakes of 2D quantum dots, I observe a few small negative frequencies in the range of 10–30 cm⁻¹, even after full convergence. I have used both effective core potentials (ECPs) and a large basis set (def2-TZVP). The IR spectrum in that region is nearly flat, and the associated vibrational modes appear to involve soft, floppy motion.
I also slightly displaced the geometry along the fluctuating directions (corresponding to those modes), but the re-optimized structure yields similar results with no significant changes.
Is it generally acceptable in this case to proceed with the current configuration, or is further refinement necessary?
You may want to play with the SCF or geometry optimization convergence criteria. If the potential landscape is rather flat along a vibrational mode, the optimization may "believe" that it is done despite being slightly off the actual minimum.
Jürgen Weippert Thanks for the suggestion. The convergence criteria is chosen to be 10^-8.
Saraf Mohaimen Chowdhury Nirjhor
For which of the two? SCF or geometry? These are two different parameters which may have to be tuned independently.
Jürgen Weippert It is the maximum force criteria of the geometry optimization. But the SCF convergence was set to its default mode: 10⁻⁶ Hartree.
I have also tried no symmetry approach for one configuration (MX4), but it's completely distorted the geometry.
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India