UT theoretical chemistry code forum

Hi Everyone, I am new in VASP calculation. I am running some Transition State calculation. Following is the DIMCAR file parameters from one of my dimer calculation. I am seeking suggestions about continuation of my current job.

Step Force Torque Energy Curvature Angle
49 0.47118 2.15055 -479.21594 -6.95128 3.17723
49 0.47118 2.79488 -479.21594 -7.00177 1.39095
49 0.47118 1.41243 -479.21594 -7.06950 1.65813
49 0.47118 1.71247 -479.21594 -7.07435 1.00805
50 0.30796 1.08057 -479.21816 -7.11068 1.08884
50 0.30796 1.06954 -479.21816 -7.13804 0.63877
50 0.30796 0.67130 -479.21816 -7.16042 0.56939
51 0.30258 1.74327 -479.21991 -7.64959 3.63823
51 0.30258 1.77362 -479.21991 -7.41852 0.91496
51 0.30258 0.95417 -479.21991 -7.46057 0.95041
52 0.33720 1.41692 -479.22124 -7.57534 1.25337
52 0.33720 1.41589 -479.22124 -7.53786 0.79535
52 0.33720 0.88980 -479.22124 -7.64845 0.72726
53 0.23311 1.84237 -479.22316 -7.55686 3.21589
53 0.23311 2.47447 -479.22316 -7.58759 1.10038
53 0.23311 0.83485 -479.22316 -7.63741 1.17236

Also, following is some of my INCAR parameters-

EDIFF = 1E-7 ; EDIFFG = -0.03
IBRION = 3 ; POTIM = 0.0
ICHAIN = 2
IOPT = 2


Do I need to change any parameters to get force converged?
I have also read in this forum about changing POTIM to ~ 0.2, could it help in this TS calculation? If does, how?
And do i need to change the IOPT to 3?

It looks like it is converging. The force is dropping and the curvature is negative.

Thanks Dr. Henkelman. But I have another concerns.

In the website, it is mentioned that force should be lower than the value of -ediffg in the INCAR file, and the curvature is negative.

For example, consider a scenario from DIMER calculation-

1. Force =< 0.10 ev/A which is slightly higher than -EDIFFG (i.e. ediffg= -0.03)
2. Curvature is consistently negative
3. Result is converged

My question is, could I consider this DIMER calculation as converged one? Does it indicate actual saddle point? Or do I need to give more runs?

It is not converged, but it is converging. The calculation needs to be continued.