Density-functional study ofS2−defects in alkali halides

Abstract

Density-functional methods, as implemented in the Amsterdam Density Functional program, are used to calculate the electron paramagnetic resonance ͑EPR͒ and electron nuclear double resonance ͑ENDOR͒ parameters of the S 2 Ϫ defect in a halide monovacancy in various alkali halides (M Z:M ϭNa, K, Rb and ZϭCl, Br, I͒ lattices. The calculations were performed on cluster in vacuo models for the defect and its lattice surroundings, involving up to 88 atoms in order to limit boundary effects. For all M Z lattices, the calculated g and 33 S hyperfine tensors of the S 2 Ϫ molecular ion are in very good agreement with the available EPR data, explicitly supporting the monovacancy model for the defect. In addition, computational results for the principal superhyperfine and quadrupole values and axes of the nearest shells of M ϩ and Z Ϫ ions are compared with experimental ENDOR data. The merits and shortcomings of the applied cluster in the vacuo method are critically evaluated.

Ab initio EPR study of S and Se defects in alkali halides

“…The ADF program was also selected to calculate the g [22] and Q tensors [23] for both defect models, again using the VWN functional in combination with basis set IV. Using this approach, very accurate g and Q values for diatomic chalcogen defects were obtained [11,12] but substantial discrepancies with experimental (super)-hyperfine data were observed. For this reason, a hybrid functional shall be used for calculating the (super)hyperfine values, as they produce more reliable results for these properties [24].…”

“…[11,12]. For the trivacancy model, a cluster composed of 140 atoms was used to optimize the structure of the X 3 Ϫ molecular ions.…”

“…Despite the accuracy of the experiments and the analyses of these data, our knowledge about the structure of these defects is still incomplete. Recently, cluster in vacuo density functional theory (DFT) calculations [11,12] for S 2 Ϫ , SSe Ϫ and Se 2 Ϫ molecular ions have shown that the ground state, the g and hyperfine tensor may be satisfactorily reproduced in nine alkali ha-lide (MZ, M ϭ Na, K, Rb and Z ϭ Cl, Br, I) host lattices. This opens favorable perspectives to solve at least some of the remaining problems concerning sulfur and selenium defects in these lattices, such as their precise structure or groundstate properties.…”

Molecular impurities in crystals: (ReO4)2− Jahn‐Teller ion in KCl

“…In several cases we found significant differ- ences from the equilibrium K-Cl distance (3.146 Å). For example, for the N c ϭ 81 cluster the optimized distances from the central Cl Ϫ ion to the K (1) , Cl (2) , and K (3) shells of atoms are 3.264, 4.895, and 6.166 Å, respectively, while the experimental ones are 3.146, 4.449, and 5.449 Å, respectively. For this reason, we introduced in all calculations the Madelung electrostatic potential due to the rest of the KCl lattice simulated by means of a set of point charges.…”

“…In order to find the optimum cluster size not only for modeling the ReO 4 2Ϫ molecular ion but also the distortions of the KCl lattice, we considered clusters with a different number of ions, N c , centered around the ReO 4 2Ϫ ion substituting a Cl Ϫ ion of the KCl lattice. We considered clusters with N c ϭ 31 (including three shells of atoms, K (1) , Cl (2) , and K (3) , around ReO 4 2Ϫ ), N c ϭ 61 (adding Cl (4) and K (5) , shells of atoms) and N c ϭ 85 (adding Cl (6) shell). A picture of the cluster with 85 atoms is given in Figure 1.…”