Computational Chemistry and Molecular Modeling
Atomic, Molecular and Optical Physics
Wiley InterScience Backfile Collection 1832-2000
Chemistry and Pharmacology
Ab initio Hartree-Fock band structure calculations are presented for the first time for lithium phosphide (Li3P) and lithium arsenide (Li3As) in the hexagonal P6/mmm crystal structure. Results are compared to those for lithium nitride (Li3N). The new calculations for Li3N agree with previous Hartree-Fock calculations, except for the valence band structure where results of previous pseudopotential calculations are confirmed. Geometry optimization for Li3P yields a lattice parameter a of 4.45 Å and a c value of 4.80 Å. These values differ markedly from experimental results reported to be 4.271 and 7.590 Å, respectively. A similar discrepancy is found for lithium arsenide: a = 4.60 Å and c = 4.96 Å have to be compared to the reported experimental values of 4.397 Å for a and 7.824 Å for c. Force constants are derived for in-plane and interplane vibrations. The band structures for Li3P and Li3As are found to be very similar to the one calculated for Li3N. Using Li3P as an example, it is shown how the band structure of the insulator can be derived from the band structures of the two metallic constitutent Li2P and Li monolayers. The metal-insulator transition occurs if the inter-plane distance falls below 4.25 Å. Contrary to expectations raised earlier, it is found that the 3d electrons in arsenic are strongly localized, evidenced by a very narrow d band width of 0.1 eV. They cannot be used to explain the difference in conductivity between the phosphide and the arsenide. A Mulliken population analysis gives charge distributions close to the ideal ionic structure (Li+)3X3-, X = N, P, and As. Overall it is found that hexagonal lithium phosphide and lithium arsenide arsenide are more similar to lithium nitride and less anisotropic than suggested previously. This discrepancy could be due to the use of polycrystalline samples in earlier experiments.
Type of Medium: