CrystalTracer: Automated Image Analysis of In Situ AFM Data Uncovers the Kinetics of Crystal Growth

Abstract

In situ atomic force microscopy (AFM) has emerged as a powerful experimental tool to investigate studies of materials growth, offering the potential ability to quantitatively characterize the time evolution of growth processes at increasingly small length scales. However, converting in situ AFM observations into robust, automated, and statistically rigorous quantitative data remains a substantial challenge and is often assessed manually. This work presents a novel, comprehensive image analysis pipeline named CrystalTracer that systematically transforms raw AFM images into high-fidelity topographical maps, enabling the precise and automated identification of individual crystals and their complex dynamic evolution over time. The framework consists of four key components: image preprocessing, crystal segmentation using YOLOv8, feature tracking using LapTrack, and quantitative analysis. This automated feature extraction enables CrystalTracer to be well-positioned to enable an automated investigation of fundamental growth phenomena including nucleation and growth kinetics through Avrami analysis, extraction of critical cluster sizes from nucleation trajectories, crystal growth rate evolution, coarsening behavior associated with Ostwald ripening, and the analysis of fractal dimension evolution in dendritic structures. These aspects are illustrated through three use cases featuring crystal growth in metal halide perovskites, metal–organic frameworks, and peptides. CrystalTracer provides a comprehensive tool set for analyzing crystal growth dynamics from in situ AFM data, significantly reducing the manual processing time while maintaining high accuracy and reproducibility, exemplified here in an analysis of a variety of different systems and growth habits.