Dear all, Thank you very much for your answers and advises! I have never had any experience in investigations of defects in crystals before.
Actually, I'm not interested in a very accurate thermodinamics of the defect. The most important things I want to know are structural and electronic band structure changes due to Ni defect in orthorhombic SbSI ferroelectric crystal. There are still some questions I would like to find out: 1) how large supercells should I use to reach a converged result? The lattice parameters of the host orthorhombic material are: a=8.48 A, b=10.12A, c=4.08 A, and there are 12 atoms in a unit cell (4 formula units). The Ni atom substitutes the Sb one, and from the experiment it follows that defect atom will behave like Ni+2 ion (Sb behaves in the host material as Sb+3 ion). I already performed calculations with 1_1_2 (23 host atoms and 1 defect) and 1_1_3 supercells (35 host atoms and 1 defect) with the constructed pseudopotential for neutral Ni atom (the parameters for the pseudopotential construction are the ones, which I mentiond in my previous letter). The distance between Ni defect atom and its periodic repetition in the next cell was in avarage close to 8-10 A in all directions. The structure of local environment of Ni defect, which I obtained after full relaxation of these 2 supercells almost coincide within 0.005 A (the forces on atoms after optimization are less then 0.005 eV/A). But I'm confused with the fact, that the symmetry of the supercell after full structural optimization changes from simple orthorhombic to simple monoclinic (deformation in ab plane). Experimental investigations, held with an amount of Ni of 2 mol.%, show that the symmetry of the crystal with defect is orthorhombic with slightely smaller lattice parameters. 2) Is this fact caused by a high concentration of defects or I'm doing something wrong? 3) It's not quite clear for me, why electron configuration 4s1 3d9 for Ni atom is more preferable, then 4s2 3d8 one? I have tested ''my'' pseudopotential for transferability and it was OK. But anyway, I'm going to try another one with parameters, kindely proposed by Andrei. I'm very greatefull for your reasonable remarks. I appreciate it! I will appreciate any help concerning these questions also! Best regards, K.R. -- K.Rushchanskii, Dr. Theoretische Physik, Universität Regensburg 93040 Regensburg Germany tel: +49 (0) 941-943-2466 fax: +49 (0) 941-943-4382
