Shift of Fermi level by substitutional impurity-atom doping in diamond and cubic- and hexagonal-boron nitrides II. Generalized Gradient Approximation

TL;DR

This paper refines previous calculations of impurity-induced Fermi level shifts in diamond and boron nitrides by employing the generalized gradient approximation (GGA) instead of LDA, confirming the robustness of impurity concentration estimates.

Contribution

It introduces GGA into the KKR scheme for impurity calculations, demonstrating minimal impact on impurity concentration values compared to LDA.

Findings

Impurity concentrations at Fermi level positions are consistent with previous LDA results.

GGA-based calculations show similar impurity concentration sensitivities as LDA.

The impurity concentration values are not significantly affected by the choice of exchange-correlation approximation.

Abstract

In succession to the first paper (arXiv 1406.6204v5), the impurity-atom concentrations when the Fermi levels are either at the valence band maximum (VBM) or the conduction band minimum (CBM) were identified for diamond, cubic boron nitride (cBN), and hexagonal boron nitride (hBN) using the Korringa-Kohn-Rostoker (KKR) scheme using the local density approximation (LDA). In the present paper, the generalized gradient approximation (GGA) was used instead of the LDA for exchange-correlation. The impurity atoms were B and N for diamond, Be, Si, and C for cBN, and Be for hBN; these impurity atoms were known in the first paper to form degenerate states by increased impurity-atom concentrations. The impurity-atom concentrations when the Fermi level was located either at the VBM or the CBM were as follows: (i) the B concentration was 0.27 at.% in B-doped diamond, (ii) the N concentration was 0.25 at.% in N-doped diamond, (iii) the concentration of Be substituting B was 0.88 at.% in cBN, (iv) the concentration of Si substituting B was 0.06 at.% in cBN, (v) the concentration of C substituting B was 0.07 at.% in cBN, (vi) the concentration of C substituting N was 0.88 at.% in cBN, and (vii) the concentration of Be substituting B was 1.80 at.% in hBN. The values of (iv) and (v) were significantly smaller than the corresponding values in paper I, but it was attributed to the input parameters used in the present paper, hence it was concluded that the computed concentrations were not sensitive to the GGA used.