Possibility of n-type doping in CaAl$_2$Si$_2$-type Zintl phase compound CaZn$_2X_2$ ($X$ = As, P)

TL;DR

This study uses first principles calculations to evaluate the feasibility of n-type doping in CaZn$_2X_2$ ($X$ = As, P), identifying La substitution at Ca sites as a promising method for electron doping.

Contribution

It provides a detailed theoretical analysis of defect formation energies, highlighting the potential of La substitution for n-type doping in CaZn$_2X_2$ compounds.

Findings

La substitution at Ca sites has the lowest formation energy and is likely to spontaneously provide electrons.

N-type doping is more feasible in CaZn$_2$As$_2$ than in CaZn$_2$P$_2$ due to lower defect formation energies.

Interstitial doping with alkaline earth metals and group 3 elements shows relative favorability among considered options.

Abstract

Motivated by a recent theoretical suggestion that doping electrons into various CaAl$_2$Si$_2$-type Zintl phase compounds may give rise to high thermoelectric performance, we explore the possibility of n-type (electron carrier) doping of CaZn$_2X_2$ ($X$ = As, P) using first principles calculation. We consider n-type doping of CaZn$_2X_2$ with the following two situations: interstitial-site doping of alkaline earth metals $AE$ (= Mg, Ca, Sr, Ba) and group 3 elements $G3$ (= Sc, Y, La), and $G3$ substitutional doping for the Ca site. The evaluation of the formation energy of these defects in various charged states reveals that the interstitial-site doping of $AE$ = Ca/Mg or $G3$ = Sc/Y, and $G3$ = La/Y substitutional doping for the Ca site are relatively favorable among the possibilities considered. In particular, the formation energy of the La substitutional doping for the Ca site is the lowest among the considered cases both for CaZn$_2X_2$ ($X$ = As, P) and is negative, which implies that La is expected to be substituted for the Ca site and provide electron carriers spontaneously. We also find that for each doping case considered, the formation energy of the defects is smaller for $X$ = As than for $X$ = P, which suggests that former is relatively easier to realize n-type doping than the latter.