[lammps-users] Dislocation Loops... advice?

Here are a couple... sorry about that. Maybe a bit more about the process would be useful too:
http://www.prism.gatech.edu/~egillilan3/Screenshot.png
http://www.prism.gatech.edu/~egillilan3/Screenshot-1.png
The first is when I'm putting in the loop... the z direction is <1 1 1>, in this coordinate system we get a ABCABC stacking of planes... basically I'm sticking a circular section of A in between B&C. Then I minimize it... and get the second picture. It seems really disorganized!

As for my process, I have frozen (with setforce) the atoms in my loop, and minimized around it. If I don't do this, the loop moves to the top of the cell, leaving perfect crystal below. Boundary conditions are currently p p m. My cell is about 22x35x25 Angstroms. It's possible this cell is just too small to handle the system correctly, but I have some limitations on the number of atoms I can use, so I hope not to go too big. If anyone has an advice, thanks a lot.

Erin