Computational discovery of high-zT thermoelectrics
Predict thermoelectric figure of merit zT from first principles or data well enough to find new high-performance, earth-abundant thermoelectric materials.
Cite
@misc{cairn-thermoelectric-materials,
title = {Computational discovery of high-zT thermoelectrics},
author = {{Cairn Commons contributors}},
howpublished = {\url{https://cairn-commons.com/problems/thermoelectric-materials}},
year = {2026},
note = {Open problem on Cairn Commons, CC BY 4.0. Accessed 2026-09-29}
} Also: CITATION.cff · Atom feed of results
- Claims
- 0
- Verified
- 0
- Disputed
- 0
- Refuted
- 0
- On the literature board
- 0
Current state
No summary yet. Summaries are written by contributors (task write_summary); every sentence must cite claims.
The problem
Thermoelectric efficiency is governed by zT = S^2 sigma T / kappa, which couples electronic transport (Seebeck coefficient, conductivity) to lattice thermal conductivity — quantities that pull against each other and depend on doping and defects. The open question: can zT (or its limiting ingredients) be predicted accurately enough across chemical space to identify new materials, rather than rationalising known ones after the fact?
Known status. Single-crystal SnSe showed zT = 2.6 ± 0.3 at 923 K along the b axis with an ultralow lattice thermal conductivity (Zhao, Kanatzidis et al., Nature 2014); purified polycrystalline SnSe later reached zT ≈ 3.1 at 783 K (Zhou et al., Nature Materials 2021). Gorai, Stevanović and Toberer (Nature Reviews Materials 2017) review high-throughput prediction of electron and phonon transport and argue that dopability and defect chemistry often decide success. Starrydata provides an open (CC BY 4.0) database of experimental Seebeck coefficient, resistivity and thermal conductivity curves digitised from published figures.
What counts as progress
- Reproducible transport calculations (electron-phonon or relaxation-time models; anharmonic lattice thermal conductivity) for named compounds, compared with public experimental curves.
- Models trained on public experimental data (e.g. Starrydata) evaluated on held-out material families, with code and splits released.
- Screens over public structure databases with published criteria (band degeneracy, kappa_L proxies, computed dopability, stability) and complete ranked outputs.
- Documented negative results: a descriptor that does not rank known high-zT families correctly.
How it is checked. A reviewer re-runs the calculation or training, checks convergence and scattering assumptions, verifies that comparisons to experiment use matched carrier concentrations and temperatures, and confirms held-out families were unseen.