Adaptive Machine Learning Discovers 9 Unknown Zeolite-Type Frameworks

AI editorPublished 2026.11.04Posted 2026.11.05 12:01 KST

A JACS paper reported that a high-throughput search based on adaptive machine learning examined mixed-ligand ZIFs and found 9 new structures. Six of them showed a bcv topology that had not previously been observed. The authors proposed ligand width and extension ratio as steric descriptors for topology selection. [1][2]

1/ AI application method: high-throughput search based on adaptive machine learning
2/ Applied task: ZIF structure exploration and prioritization of candidates for new mixed-linker ZIF synthesis
3/ Reported result: 9 new mixed-linker ZIFs were discovered, and bcv topology was confirmed in 6 of them
4/ The JACS paper published on 2026-11-04 explored the structure space of zeolitic imidazolate frameworks (ZIF) through high-throughput search combined with adaptive machine learning.
5/ The researchers discovered 9 new mixed-linker ZIFs through mixed-ligand ZIF synthesis, and reported that 6 of these adopted a previously unrealized bcv topology.
6/ They also proposed ligand width and extension ratio as steric descriptors for explaining topology selection.
7/ From a materials perspective, this result shows that ML-based candidate screening and synthesis prioritization can lead to actual structural discovery in porous metal-organic frameworks with large search spaces.
8/ However, the reported achievement is limited to a specific class of ZIF and the corresponding synthesis conditions, and it cannot be said to generalize directly to other material systems or a broader chemical space. [1][2]

How to apply this

Use ML to prioritize promising synthesis combinations in mixed-ligand porous frameworks and discover new structures.

Organize candidate ZIF combinations and measurable structural descriptors into a training dataset.
Use an adaptive ML model to score which combinations should be synthesized next and select the top candidates.
Synthesize the top candidates in practice, then determine the phase by XRD and additional structural analysis if needed.
When a new structure appears, add the result back into the training data and update the recommendations for the next round.

Data needed: Candidate compositions, ligand geometry information, synthesis condition records, and the experimentally synthesized samples and structural analysis data are needed.
Checking the result: Confirm the structure of samples synthesized according to the predicted priority by XRD and other necessary structural analyses, and verify whether the discovered structure is a new phase absent from the training data.
What can go wrong: If the model is fit only to the known composition distribution, it may repeatedly recommend familiar structures instead of new topologies.

이 가이드는 원문의 방법을 바탕으로 AI가 제안한 적용 방향입니다. 원문에 적힌 내용이 아니며 같은 결과가 재현된다는 보장도 아닙니다.

웹 자료를 바탕으로 AI가 자동 요약한 글입니다. 연구의 세부 조건과 결과는 연결된 원문을 확인해주세요.

Adaptive Machine Learning Discovers 9 Unknown Zeolite-Type Frameworks

How to apply this

Related concepts and practice

Sources