Computer modelling of pure and doped stoichiometric and congruent lithium niobate

Abstract

This investigation has modelled lithium niobate in its congruent form, stoichiometric form and paraelectric phase to determine whether the preference of doping schemes changes, specifically when the concentration of a dopant is varied. Mott-Littleton calculations and the supercell method were used to model defect structures of lithium niobate and assess the viability of four doping schemes, with a particular emphasis on the dopant Zn2+, however other dopants were used.
The Mott-Littleton calculations involved the doping of multiple divalent and trivalent dopants. The results obtained showed scheme 2 to be the scheme with the lowest solution energy across all the dopants. The supercell method was carried out using a stoichiometric supercell of 48 Li and 48 Nb sites, with another larger stoichiometric supercell being used that contained 162 Li and 162 Nb sites for comparison with the same sized congruent supercell. The congruent supercell, however, contained a Li/Nb ratio of 0.963, in order to make the cell congruent. The dopants used in both supercells were Zn2+, Ce3+, In3+ and Eu3+. The supercells showed the effect of dopant concentration on the scheme preference.
The paraelectric phase was only modelled using the Mott-Littleton method. The same dopants used in the stoichiometric Mott-Littleton modelling were employed. The results from the paraelectric phase were then compared to the stoichiometric form’s results, which showed no change in the preference of the doping schemes, with scheme 2 having the lowest solution energy across all divalent and trivalent dopants.
Overall, the results showed that the doping scheme preferred is Scheme 2, both for divalent and trivalent dopants, which involved doping at the Li site and Nb site, and that the effect of increasing the concentration of the dopant had little effect.