The binary antimonide Mo₃Sb₇ is metallic with poor thermoelectric properties. Since band structure calculations indicated a possible change to semiconducting behavior by adding two more valence-electrons, the substitutional series Mo₃Sb_7 − δTe_δ has been investigated. Te atoms can replace up to 2.3 Sb atoms in Mo₃Sb₇ without noticeable structural changes. This substitution leads to a dramatic enhancement in the thermopower, indicating that in fact semiconducting properties have been achieved with the synthesis of Mo₃Sb₅Te₂.

A large variety of different interactions between the chalcogen atoms, Q, occur in the solid state structures of polyselenides and polytellurides, including both molecular and infinite units. The simplest motifs are classical Q₂^2– dumbbells and nonlinear Q_n^2– chains (n = 3, 4, 5, ..), e.g. found in alkali metal polychalcogenides. In addition, nonclassical so-called hypervalent motifs exist in the form of linear Q₃^4– units or within larger units such as Q₄^4– and Q₅^4–. Infinitely extended Q units include zigzag, cis/trans and linear chains, as well as planar and slightly puckered layers. Several of those are susceptible to Peierls distortions, leading to the formation of both commensurate and incommensurate superstructures and anomalies in transport properties, including metal-nonmetal transitions.

Due to increasing global energy concerns, alternative sustainable methods to create energy such as thermoelectric energy conversion have become increasingly important. Originally, research into thermoelectric materials was focused on tellurides of bismuth and lead because of the exemplary thermoelectric properties of Bi₂Te₃ and PbTe. These materials, however, contain toxic lead and tellurium, which is also scarce and thus expensive. A viable alternative material may exist in Mg₂Si, which needs to be doped and alloyed in order to achieve reasonable thermoelectric efficiency. Doping is a major problem, as p-type doping has thus far not produced competitive efficiencies, and n-type doping is problematic because of the low solubility of the typical dopants Sb and Bi. This investigation shows experimentally that these dopants can indeed replace Si in the crystal lattice, and excess Sb and Bi atoms are present in the grain boundaries in the form of Mg₃Sb₂ and Mg₃Bi₂. As a consequence, the carrier concentration is lower than the formal Sb/Bi concentration suggests, and the thermal conductivity is significantly reduced. DFT calculations are in good agreement with the experimental data, including the band gap and the Seebeck coefficient. Overall, this results in competitive efficiencies despite the low carrier concentration. While ball-milling was previously shown to enhance the solubility of the dopants and thus the carrier concentration, this did not lead to enhanced thermoelectric properties.