Taming Two-Dimensional Polymerization by a Machine-Learning Discovered Crystallization Model.

Area of Science:

• Materials Science
• Chemical Engineering
• Computational Chemistry

Background:

• Rapid synthesis of high-quality 2D covalent organic frameworks (2D COFs) is essential for their applications.
• Current synthesis strategies rely on empirical understanding, lacking quantitative guidance for optimization.

Purpose of the Study:

• To develop a quantitative model for understanding and optimizing 2D COF crystallization.
• To overcome limitations of traditional models in describing non-classical crystallization processes.

Main Methods:

• Utilized a machine-learning approach to derive a bottom-up model (NEgen1) for 2D COF formation.
• Established correlations between synthesis conditions and kinetic parameters (nucleation, growth, bond formation).

Main Results:

• The NEgen1 model accurately describes the nucleation-elongation mechanism in 2D COF synthesis.
• Revealed the dynamics between nucleation and growth, challenging previous empirical models.
• Identified an optimal strategy of gradually increasing monomer addition speed for rapid crystal growth.

Conclusions:

• The NEgen1 model offers precise guidance for optimizing 2D COF synthesis conditions.
• Demonstrated rapid synthesis of large COF-5 colloids, improving crystal quality.
• Highlights the potential for systematic improvement of 2D COF synthesis for diverse applications.