Thermoelectrics convert a temperature difference directly into voltage with no moving parts; the physics is settled and modules are being built from cheap Heusler and Bi2Te3 alloys, but conversion figures of merit remain low enough that the technology stays confined to niches while a newer transverse, spin-based branch is still at the laboratory-demonstration stage.
Summary
A thermoelectric material generates a voltage when its two ends are held at different temperatures (the Seebeck effect) and pumps heat when a current is passed through it (the Peltier effect). Performance is captured by the dimensionless figure of merit zT, which rises with the Seebeck coefficient and electrical conductivity and falls with thermal conductivity. These three quantities are coupled in most solids, so materials work consists of decoupling them: engineering band structures with high degeneracy or unusual dispersion to raise the power factor, while scattering phonons to suppress the lattice thermal conductivity. Both levers appear repeatedly in the current literature. High-entropy engineering of Fe2VAl full Heuslers, combining heavy-element doping with controlled off-stoichiometry to create substitutional disorder on every lattice site, yields a lattice thermal conductivity of about 2.3 W m-1 K-1, among the lowest reported for that family, and an average zT of roughly 0.3 between 300 and 500 K. On the band-structure side, layered Zintl compounds XZnBi (X = Rb, Cs) are predicted to combine six degenerate hole pockets with a quasi-two-dimensional pudding-mould conduction band that mixes heavy and light electron masses.
A structurally different branch uses transverse geometries, where the electric field appears perpendicular to the heat flow. Spin caloritronics, reviewed as now transitioning from fundamental condensed-matter physics towards materials science and engineering, covers the spin Seebeck effect and the anomalous Nernst effect. Transverse devices avoid the series-connected pillar architecture of conventional modules, but have been limited by spin and magnon diffusion lengths to thin films; nanostructured bulk composites of Pt-coated yttrium iron garnet powders, made by dynamic powder sputtering and low-temperature sintering, have now shown isotropic bulk-scale spin Seebeck signals with power analysis pointing to volumetric rather than diffusion-limited scaling. Anomalous Nernst materials are being engineered by composition: Gd-Co-Pt ternary films reach a heat-flux sensitivity of about 0.24 micrometres per amp, with Pt driving a sign reversal of the Nernst coefficient, and the kagome magnet YCo3 shows a bipolar anomalous Nernst effect driven by a topological Lifshitz transition below its 225 K Curie temperature, relevant to thermopiles that need both signs from one material.
The deciding parameters are therefore: zT and the temperature window over which it is sustained; material cost, toxicity and scalability of synthesis; module-level losses at contacts and interfaces; and, for the transverse branch, whether large enough transverse coefficients can be found in materials that also survive bulk fabrication.
Evidence base
- 16 June 2026: high-entropy engineered Fe2VAl full Heuslers reach a lattice thermal conductivity of about 2.3 W m-1 K-1, among the lowest reported for the family, giving an average zT of roughly 0.3 from 300 to 500 K, with a full module built from hot-pressed scaled-up batches.
- 22 July 2026: nanostructured bulk composites of Pt-coated yttrium iron garnet powders show isotropic spin Seebeck signals at bulk scale, with power analysis indicating volumetric generation beyond the diffusion-limited thin-film geometries that have constrained the effect until now.
- 7 July 2026: Bi2Te3/Sb2Te3 hybrid inks in a PMMA matrix with a hexanedithiol binder give best performance at 80 per cent nanoparticle loading, with finite-element modelling used to optimise the p- and n-part cross-sectional area ratio.
- 16 June 2026: Gd-Co-Pt ternary films reach an anomalous Nernst heat-flux sensitivity of about 0.24 micrometres per amp, with Pt inducing a sign reversal of the Nernst coefficient and Gd reducing thermal conductivity.
- 27 July 2026: intrinsic nonlinear Seebeck, Nernst and mixed-directional thermoelectric responses are observed up to room temperature in Td-WTe2 and TaIrTe4 without magnetic fields or magnetic materials.
- 5 May 2026: a field review places spin caloritronics at a turning point, transitioning from fundamental condensed matter physics to materials science and engineering application.
Open questions
- Can high-entropy Heusler chemistry push average zT materially above the reported 0.3 over 300 to 500 K without reintroducing cost or toxicity, and what module conversion efficiency does that translate into?
- Do the bulk Pt-coated YIG composites actually deliver volumetric power scaling in a multi-centimetre device, and how does that power density compare with a longitudinal Bi2Te3 module of the same volume?
- How many of the DFT-predicted candidates in this literature, such as CrSi2N4, beta2SnGeX6 or XZnBi, are ever synthesised, and how far do measured power factors fall short of the calculated values?
- Can an anomalous Nernst material combine bipolar operation with a Curie temperature well above room temperature, given that the YCo3 demonstration sits below 225 K?
Assessment drafted 2026-08-31 from up to 18 KB sources using the technology-scorecard framework; scores are a draft read pending review.