---
title: "Introducing the Thermoelectrics API"
description: "Fast screening of inorganic crystal structures for thermoelectric performance from a CIF file."
url: "https://ouro.foundation/posts/mmoderwell/introducing-the-thermoelectrics-api"
id: "019d063b-22d4-71d8-b46b-9b508d31d66b"
type: "post"
---

# Introducing the Thermoelectrics API

Post by @mmoderwell in #thermoelectrics (all) · 2026-03-19

- Asset id: `019d063b-22d4-71d8-b46b-9b508d31d66b`
- Page: https://ouro.foundation/posts/mmoderwell/introducing-the-thermoelectrics-api

# Introducing the Thermoelectrics API

If you're exploring new thermoelectric compounds, one of the biggest bottlenecks is deciding which candidates are actually worth deeper computation or synthesis effort. This API is built for that first pass.

Instead of running a full transport workflow on every structure, you can upload a CIF and get a fast screening readout of the properties that matter most for thermoelectric triage: structural stability signals, phonon-derived transport quantities, Seebeck response, band gap, and an overall ZT estimate.

> Embedded service: [Thermoelectrics API](https://ouro.foundation/services/mmoderwell/thermoelectrics-api)

## Why use it

- Screen more candidates before committing serious HPC or lab time.
- Compare materials with a shared, reproducible pipeline instead of ad hoc scripts.
- Catch obvious problems early, like imaginary phonon modes or weak electronic response.
- Get both the component properties and a combined ZT estimate in one workflow.

This is especially useful if you're mining databases, generating hypothetical compounds, or narrowing a large candidate set down to the few materials that deserve more expensive follow-up.

## What the API returns

The main route takes a crystal structure as a CIF file and returns a single JSON response with the core pieces needed for thermoelectric screening:

- structure relaxation metrics
- phonon stability summary
- Gruneisen parameter
- lattice thermal conductivity estimates, including `kappa_slack` and `kappa_min`
- p-type and n-type Seebeck coefficients
- band gap estimate
- ZT estimates across temperature

> Embedded route: [Estimate ZT and key thermoelectric properties](https://ouro.foundation/routes/mmoderwell/estimate-zt-and-key-thermoelectric-properties)

The service also exposes separate routes for relaxation, phonons, thermal conductivity, and Seebeck/band gap if you only want one stage of the workflow instead of the full pipeline.

## How it works

1. **Structure relaxation**: the input CIF is relaxed with NequIP-OAM-XL so downstream properties are computed from a more physically reasonable geometry.
2. **Phonons**: phonopy + NequIP forces are used to compute phonon frequencies, group velocities, Debye-scale information, and instability flags.
3. **Thermal conductivity**: the phonon outputs are used to estimate both a WTE-based minimum thermal conductivity and a Slack-model lattice thermal conductivity.
4. **Electronic properties**: ALIGNN is used to predict Seebeck coefficients, while the band gap comes from ABACUS DFT when available or an ML fallback when needed.
5. **ZT estimation**: those ingredients are assembled into a fast temperature-dependent ZT screening estimate.

The goal is not to replace careful first-principles transport calculations. The goal is to give you a strong, fast filter that helps answer: *is this material worth deeper attention?*

## Where it creates value

A lot of thermoelectric work has a funnel shape: you may start with hundreds or thousands of possibilities, but only a small number can justify high-cost calculations or experimental effort. This API helps at that funnel stage.

Good use cases include:

- screening generated or database-derived structures
- comparing composition families before synthesis
- deciding which candidates are worth deeper DFT or transport calculations
- getting quick feedback on whether low thermal conductivity and strong Seebeck response appear plausible together

## Important caveats

This is a screening tool, not a final word on performance.

- `kappa_min` is a lower bound, not the realized thermal conductivity of a material.
- ZT is an estimate assembled from model outputs, so it should be used for ranking and prioritization rather than as a publication-ready number.
- Structures with imaginary phonon modes or unusual chemistry should be treated with extra caution.

If you're working on thermoelectric discovery, feedback is very welcome on which outputs, routes, or follow-up tools would be most useful to add next.

## Assets in this post

- [Thermoelectrics API](https://ouro.foundation/services/mmoderwell/thermoelectrics-api) (service, id `7ad9122e-8149-4802-a0f7-0b856f62d66a`)
- [Estimate ZT and key thermoelectric properties](https://ouro.foundation/routes/mmoderwell/estimate-zt-and-key-thermoelectric-properties) (route, id `a8e05903-b077-4e5f-8de5-bdbeac459db0`)