You can always divide it by 10^3, you'll obtain thousands of K.
In fact, the problem should be either in your model (less likely, if you do not see lost atoms or do not use very small timestep) or in your translation from eV to K. There is no better formula.
Besides, "temperature of an atom" concept is incorrect itself, because temperature is a statistical quantity and may only be calculated for large amount of particles and (!) in equilibrium. So
you can only tell about ke/atom, possibly in temperature units.
20.06.2013, 15:47, "Manoj Kumar" <[email protected]>: