My Community

Hi,

I've been trying to perform NEB calculations for my system. I've gotten convergence using 1 image and am now trying to expand to 3. I'm at the point where the energies of the image files oscillate with a difference between high and low energies of 0.006 eV, so it doesn't seem like the calculations will ever converge. The only thing that I can think that might be a problem is that images 01 and 02 are less than 0.01 eV less stable that image 00. That is, the energy barrier from 00 is basically nonexistent. Also, there are numerous local minima with similar geometries, but I'm fairly certain that my two endpoints are true minima. So my question is...should this converge (is there any reason why it shouldn't)? If not, how do I handle this system to calculate a transition state energy?

Thanks,
Matt

The pertinent settings are

SPRING = -5.0
IBRION = 1
POTIM = .1
NFREE = 2
NELMIN = 4
EDIFF = 1e-5

Try IBRION=3. This routine does not rely on estimating second derivaties, so it has a larger radius of convergence and tends to be more stable (and less efficient). Also, there is no Hessian consistent with the NEB forces, and I don't think it's well understood how the quasi-Newton methods deal with non-conservative forces.