Thermoelectric generators can be used to generate electricity directly from waste heat and are thus very useful for reducing energy loss and waste. However, the most popular thermoelectric materials in current use are toxic to the environment and require the rare and expensive element tellurium, prompting a large effort to find materials which are just as good but are less toxic and made of more common elements.

Thermoelectrics rely on an intrinsic property of all materials called the Seebeck effect, where an electric voltage can be spontaneously generated between two sides of a material if they are at different temperatures. In most materials this effect is tiny, but in certain semiconductors it can be significant. To be useful in devices, the material must be a good electrical conductor so that a current can flow between the "hot" and "cold" ends, and it must also be a bad thermal conductor so that the hot end stays hot, and the cold end stays cold. It is this last property that the researchers investigated in a recent study, published in Nature Communications.

Usually, heat can be transported through a material either by the motions of its atoms or its electrons. For metals, where the electrons can flow freely, they conduct both heat and electricity very well, so they would not be good thermoelectrics. In semiconductors, by contrast, most of the electrons cannot flow freely and thus both the electrical and thermal conductivity might be balanced to meet the thermoelectric requirements. Whilst these materials are solids where the atoms cannot flow, heat can still move via the vibration of their atoms around their equilibrium position in the crystalline lattice. These collective vibrations maybe described mathematically as quasiparticles called phonons and it is these phonons that transport heat around the material.

The amount of heat that can be transported depends on the 'speed' of the phonons and how 'obstructed' they are. Materials with heavier atoms such as bismuth, lead and tellurium (Bi, Pb and Te) have low phonon velocities and therefore low thermal conductivities. The phonon velocity is also affected by how strongly the atoms are bonded to each other, as weaker bonding results in lower phonon velocities. In addition, the phonons can be impeded (or 'scattered') by gaps in the crystal structure called vacancies. This process is called defect scattering.

In this study, the researchers from China, Japan, Australia and elsewhere used the MARI spectrometer at ISIS to measure the phonon modes in several candidate thermoelectric materials formed as Zintl phases, which are characterised by strong ionic bonding, in particular the compounds: SrCuSb, SrAgSb and Sr₂ZnSb₂.

They found that SrCuSb and SrAgSb show similar spectra, except that the modes in the silver compound are lower in energy (have a slower phonon velocity) than the copper compound, due to silver being heavier. When Cu or Ag is replaced by Zn, only half of the atoms are replaced in order to ensure charge neutrality, leaving a large number of vacancies, which reduces the thermal conductivity through defect scattering. However, the phonon spectrum of Sr₂ZnSb₂ is significantly different to that of SrCuSb or SrAgSb, more so than can simply be explained by increased defect scattering from the vacancies.

In fact, using theoretical calculations and X-ray photoelectron spectroscopy measurements the group showed that the substitution of Cu or Ag by Zn actually weakens the ionic bonds between the Sr and Sb and thus the bonding between the Sr layers and the Zn-Sb layers in the material's structure. This change to the bonding explains the strong shifts and broadening in the phonon modes and the large change to the neutron spectrum. It also explains the large six-fold reduction in the thermal conductivity from SrCuSb to Sr₂ZnSb₂.

This is not only important for Sr₂ZnSb₂ as a potential thermoelectric material, but their finding that substituting an ion in a different oxidation state into an ionic compound can strongly reduce the inter-atomic bonds, and thus the thermal conductivity, could be used to engineer other thermoelectric materials and may be a more important consideration in the long term. 

The full paper can be found at DOI: 10.1038/s41467-024-46895-4​