[gmx-users] nematic phase disappears with reducing the time step

[Argyrios Karatrantos]

hi, 

i am doing simulations NVT simulations of rigid charged rods (with constrained bonds and stiff bending potential)in vacuum using PME for the electrostatics with a time step of 0.0015 ps (which i think is a pretty reasonable value for a corse grain model)
A nematic phase of the rods appears at specific concentration (i have checked the Total-energy, Pressure, Temperature of the system and it is seems the simulation is in equilibrium)

however when i do simulation of the same system with smaller time steps such as 1.0 fs or 0.5 fs, the nematic phase disappears (there is only a small increase of the potential energy comparative to that of the nematic)

what is the reason that the time step can cause such an artifact phase?