Simplification of QPLDA: A practical method to the correction for the LDA band gap problem

Abstract

It is necessary to employ quasi-particle calculations to correct band gap problems in LDA. As an expedient way for the reliable but massive GWA, Quasi-Particle-LDA (QPLDA) is proposed by several authors, where the total computational cost scales with N. Historically, Sham and Kohn introduced the idea of the local mass operator based on local wavenumber similar to WKB, but did not execute actual numerical calculations. They took into account exchange interaction only, from which we could not expect precise treatments anyway. Later, Pickett and Wang had proposed a more qualitative method and had shown its potentiality in examples of semiconductors, such as silicon, diamond and GaP. They used a model analytic formula for the dialectic function and adopted a model energy dispersion, which is free-electron like one, except that, being accompanied with an artificial band-gap discontinuity. The latter method has two shortcomings. First, to execute the calculation, several parameters are needed, such as the macroscopic dielectric constant and the band gap, which should be evaluated by the calculation itself, rather than being prepared as parameters. Second, in their computation of the mass operator, two or three dimensional numerical integrations are needed on each point of the direct space mesh, which demand large computational costs in iterative procedures in determining local wavenumbers and quasi-particle energies. This may be the reason why QPLDA has not been widely used until now. The present work proposes a qualitative way of QPLDA which does not need any experimental parameters, being much speedier than the formalism of Picket and Wang, and more precise than that of Sham and Kohn. The computational method adopted by the present work is a modification of the original proposal by Sham and Kohn, rather than that of Picket and Wang.