Grain boundary simulation

I am facing some difficulties in preparing a relaxed configuration which has multiple grain boundaries.

I am studying a sample which has 5 CSL grain boundaries arranged parallel to each other. The distance between the boundaries is about 30 A (3 nm).

I have studied two cases :
Case 1: boundary p p p
Case 2: boundary p p f (boundaries are stacked along z-direction)

I have done a NPT equilibration at T = 300 K to make the total Pressure = 0

Now I have plotted a field map for the local pressure for these configurations. It is seen that the loacl pressures at the boundaries are not similar. If we tag the boundaries as GB1, GB2, GB3, GB4, GB5, the local pressure at these boundaries are like the following : (+), (-), (+), (-) and (+). Here (+) represents a positive pressure and (-) represents a negative local pressure. The bulk region of the system (the regions outside the grain boundary domain) are at nearly zero pressure domain.

I don’t understand the reason of such alternate (+) and (-) pressures at the grain boundaries. It does not seem to be physical.

Can anyone suggest some techniques to equilibrate such a configuration like, the choice of boundary condition, or choice of ensemble, or anything else ?

Thanks and best regards,

Some people do this with “annealing” by

ramping the temperature up and down.