Phase Boundary Mapping to Obtain n-type Mg₃Sb₂-Based Thermoelectrics

Zintl compounds make excellent thermoelectrics with many opportunities for chemically tuning their electronic and thermal transport properties. However, the majority of Zintl compounds are persistently p-type even though computation predicts superior properties when n-type. Surprisingly, n-type Mg₃Sb₂-based thermoelectrics have been recently found with exceptionally high figure of merit. Excess Mg is required to make the material n-type, prompting the suspicion that interstitial Mg is responsible. Here we explore the defect chemistry of Mg₃Sb₂ both theoretically and experimentally to explain why there are two distinct thermodynamic states for Mg₃Sb₂ (Mg-excess and Sb-excess) and why only one can become n-type. This work emphasizes the importance of exploring all of the multiple thermodynamic states in a nominally single-phase semiconductor. This understanding of the existence of multiple inherently distinct different thermodynamic states of the same nominal compound will vastly multiply the number of new complex semiconductors to be discovered for high zT thermoelectrics or other applications. While there has been good progress in finding high thermoelectric efficiency (zT) p-type Zintl compounds, high zT n-type Zintl compounds have eluded discovery for 10 years despite the theoretical predictions that these would make even better thermoelectric materials. Here we show that even in a “line compound” multiple thermodynamic states exist that profoundly affect the electronic properties by suppressing the formation of unwanted defects differently. To form the desired n-type Mg₃Sb₂-based compound it is most critical to make the Mg-excess thermodynamic state. This understanding suggests a synthesis strategy we call “phase boundary mapping” that could be counter-intuitive from the normal perspective that favors thermodynamic phase purity: to add excess constituents until the impurity phase is identified. This strategy can help discover many different versions of any compound, even ones considered to be a “line compound” with no measurable compositional variation. Despite theoretical predictions of high thermoelectric performance (zT) in n-type Zintl compounds, zT > 1.5 has only been found in Mg_3+xSb_1.5Bi_0.5-yTe_y with excess Mg recently. We show here that this has been due to killer Mg-vacancy defects that make Sb-excess material p-type. Recognizing that another thermodynamic state (Mg-excess) exists, we explain the recent discovery and demonstrate zT n-type Mg₃Sb₂ even without Bi and much less excess Mg. Such phase boundary mapping will enable discovery of new semiconducting states of nominally the same compound.