Related Experiment Video
Updated: Feb 18, 2026
07:45
Electrophoretic Crystallization of Ultrathin High-performance Metal-organic Framework Membranes
Published on: August 16, 2018
10.5K
Rational Design of High-Quality ZIF-8 Membranes: Machine Learning-Guided Optimization of ALD ZnO Conversion
Kevin Dedecker1, Martin Drobek1, Mikhael Bechelany1
1Institut Européen des Membranes (IEM); CNRS, ENSCM, Univ Montpellier; Place Eugène Bataillon, 34095 Montpellier, France.
ACS Applied Materials & Interfaces
|February 16, 2026
Summary
Machine learning accurately predicts the conversion of atomic layer deposited (ALD) ZnO films into high-quality zeolitic imidazolate framework-8 (ZIF-8) membranes. This approach optimizes synthesis conditions, reducing experimental trials for ZIF-8 membrane fabrication.
Area of Science:
- Materials Science
- Chemical Engineering
- Data Science
Background:
- Fabricating high-quality zeolitic imidazolate framework-8 (ZIF-8) membranes from atomic layer deposited (ALD) zinc oxide (ZnO) thin films is challenging due to complex synthesis conditions.
- Optimizing ZIF-8 membrane synthesis requires extensive experimentation to identify ideal solvent systems, temperatures, and reaction times.
Purpose of the Study:
- To develop a machine learning (ML) framework for predicting the outcome of converting ALD ZnO films into ZIF-8 membranes.
- To identify key experimental parameters influencing the quality of ZIF-8 membrane formation.
- To accelerate the optimization of ZIF-8 membrane synthesis protocols.
Main Methods:
- Evaluated seven classification algorithms, including k-nearest neighbors (k-NN), random forests, neural networks, and decision trees.
- Utilized stratified 10-fold cross-validation for systematic algorithm assessment.
- Employed feature importance analysis and decision tree analysis to understand parameter influence and identify critical thresholds.
- Applied Synthetic Minority Oversampling Technique (SMOTE) to enhance minority class detection.
Main Results:
- An optimized k-NN classifier (k=5) achieved 92.6% accuracy and a Kappa statistic of 0.791 in predicting membrane quality.
- The primary solvent was identified as the most significant predictor, followed by temperature and reaction duration.
- A critical temperature threshold of 80 °C was found for methanol-based systems, influencing required reaction times.
- The predictive framework demonstrated over 90% confidence in screening conversion conditions.
Conclusions:
- A data-driven ML approach effectively predicts ZIF-8 membrane quality from ALD ZnO films.
- The developed model significantly reduces the need for experimental trials in membrane synthesis.
- This methodology offers a blueprint for applying ML to optimize other metal-organic framework systems and materials synthesis.
