I have simulated iron crystal (BCC) in Lammps using many different potential files in order to make an isosceles TB. I modeled the TB by rigorously calculated the offset distances of the two crystal phases according to what an isosceles TB should be like. However, when the relaxation started, the prefabricated isosceles TB immediately changed to ordinary mirror symmetrical TB. In contrast, from literature, for metal iron, it is easier to format isosceles TB instead of mirror symmetrical TB. Why is that?