Ionic Relaxation getting stuck in local minima

[heisen]

Dear All,
I have been performing ionic relaxations of Si (100) in order to see the 2x2 surface reconstruction. I have a supercell with 64 Si atoms (4x4x2 Si 8-atom 8-cells). The bottom layer of Si atoms is fixed and passivated with H atoms. The rest are free to relax. I use IBRION=2 and POTIM=0.5 and EDIFF = 1E-5 in my relaxation. The ionic relaxation keeps getting stuck in local minima close to my starting atomic positions, instead of the known global minima. For Si(100) relaxation I know how to move the surface atoms to be close to the known global minima, however I am seeing a similar phenomena of ionic relaxations getting stuck in local minima in other surface calculations (Au(111) substrate with thiols attaching to surface) where I do not know in advance the global minima.
How can I ensure my ionic relaxation reaches the global minima. Do I need to alter IBRION and POTIM?

[merzuk.kaltak]

Can you try using a negative value for EDDIFG, that is to stop the relaxation after all forces on the ions are below this threshold?