# Graph learning improved methanol-yield estimates for doped zirconia catalysts

> Aarhus University researchers used a graph-based model to estimate energies on doped zirconia surfaces for CO₂ hydrogenation. In microkinetic calculations, those estimates produced methanol yields closer to density-functional-theory results than a workflow based on linear relations.

- Canonical URL: https://newsverum.com/en/news/zirconia-catalyst-graph-machine-learning-screening-2026
- Publisher: Newsverum
- Section: Science
- Country: Denmark
- Published: 2026-10-13T00:51:41+00:00
- Updated: 2026-10-13T00:51:41+00:00
- Event / data period: 2026-10-09
- Other editions: [Türkçe](https://newsverum.com/tr/news/zirconia-catalyst-graph-machine-learning-screening-2026.md), [العربية](https://newsverum.com/ar/news/zirconia-catalyst-graph-machine-learning-screening-2026.md), [Français](https://newsverum.com/fr/news/zirconia-catalyst-graph-machine-learning-screening-2026.md), [Español](https://newsverum.com/es/news/zirconia-catalyst-graph-machine-learning-screening-2026.md), [Deutsch](https://newsverum.com/de/news/zirconia-catalyst-graph-machine-learning-screening-2026.md), [Português](https://newsverum.com/pt/news/zirconia-catalyst-graph-machine-learning-screening-2026.md)

The study evaluated aluminium, cadmium, gallium, indium and zinc dopants on ZrO₂(101) surfaces. Its WWL-GPR method represents surface structures as graphs combining local geometry and electronic features, then estimates adsorption energies of intermediates and transition states.

The model improved intermediate-energy predictions over thermodynamic scaling relations, while its transition-state estimates were competitive with Brønsted–Evans–Polanyi models. When the predicted energies were passed through a reaction network, estimated methanol yields tracked density-functional-theory-derived values more closely. These are computational screening results; the study did not demonstrate catalyst performance in experiments.

## Data

### Reported study result

Number of distinct dopants included in the computational screen; not a synthesized catalyst or measured production quantity.

Methodology: The paper lists Al, Cd, Ga, In and Zn for ZrO₂(101) surfaces. Energy estimates were used in graph-model and microkinetic calculations.

| Reported quantity | Value | Study-specific scope |
|---|---|---|
| Dopant elements screened on ZrO₂ surfaces | 5 | Number of distinct dopants included in the computational screen; not a synthesized catalyst or measured production quantity. |

Sources: <https://arxiv.org/abs/2610.13795> · Download data: [JSON](https://newsverum.com/api/v1/articles/zirconia-catalyst-graph-machine-learning-screening-2026/data/study-result?lang=en&format=json) · [CSV](https://newsverum.com/api/v1/articles/zirconia-catalyst-graph-machine-learning-screening-2026/data/study-result?lang=en&format=csv)

## Sources

- [arXiv](<https://arxiv.org/abs/2610.13795>) (Source checked: 2026-10-13; Published: 2026-10-09)

---

Cite the canonical URL above. [Editorial standards and corrections](https://newsverum.com/en/editorial)