AI Science Tracker
EN

University of St Andrews

1 discovery from 2025

Discoveries

Learned photocatalyst recommendations identify a productive ruthenium alternative for an ATRA reaction

Source check: Source checked. Not independent scientific confirmation.

Finding: A reaction-language model trained on literature examples recommended photocatalysts for a previously reported atom-transfer radical addition reaction. A ruthenium catalyst omitted from the original catalyst screen gave the highest yield among the catalysts tested in the new study, with productive results also observed for two related substrates. This is a specific measured catalyst–reaction result within known chemistry, rather than discovery of a new transformation or proof that model confidence predicts yield.

Technology: BERT-based reaction-language model for photocatalyst recommendation

Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.

AI contribution
The learned language model maps substrate and reaction-core representations to rankings over 31 photocatalysts, selecting a previously unscreened catalyst alternative for subsequent laboratory comparison.
Test and result
Top-five model recommendations were tested in the published ATRA system, with quantitative NMR yields from two independent experiment repeats. Source-labelled Ru1 was absent from the original screen and gave the highest measured yield among the newly compared catalysts; two related substrates also gave productive outcomes.
Limitation / still unknown
The journal also evaluates other known reactions, but the represented finding is this specific catalyst alternative. Yield depends on photoreactor/light setup; the authors observe a discrepancy with an earlier eosin-Y result and explicitly state model confidence does not track yield. Co-catalysts, additives and reaction conditions are not model inputs. An earlier July 2025 preprint differs in the number of evaluated reactions; this record uses the fully read December 2025 journal version without claiming earliest public disclosure.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://doi.org/10.1002/anie.202514544 · https://pmc.ncbi.nlm.nih.gov/articles/PMC12828469/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 21 Nov 2026. Review scope: Full primary journal/PMC data curation, BERT-model Methods, ATRA experiments/Fig3–4 captions and discussion; exact front identity/date and fresh primary notice metadata. Earlier preprint-version result equivalence not asserted..

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open the interactive profile for the timeline, or browse all organizations and model families.