Convergence in energies

The defect energy converges till region 1 radius of 20angstroms but suddenly shoots up when it is increased to 40angstroms or so.Where lies the problem or it is usual to observe such type of trend.Can GULP handle such large region radii and atoms or it cannot?Kindly provide me some suggestions.

Divergence at larger radii is certainly not normal for defect calculations. Whenever there is strange behaviour with defect calculations it’s important to check that the bulk structure is fully relaxed and has no imaginary phonons anywhere in the Brillouin zone. If this is not the case then the results can behave badly and diverge with increasing region size. The maximum size GULP can handle depends on how much memory your computer has. At present defect calculations aren’t parallelised and so that places some limits on how far you can go. There is also the possibility to compute defect energies from periodic supercells as well, in which case you can use parallelisation.