Abstract:Understanding the processes of perovskite crystallization is essential for improving the properties of organic solar cells. In situ real-time grazing-incidence X-ray diffraction (GIXD) is a key technique for this task, but it produces large amounts of data, frequently exceeding the capabilities of traditional data processing methods. We propose an automated pipeline for the analysis of GIXD images, based on the Faster R-CNN deep learning architecture for object detection, modified to conform to the specifics of the scattering data. The model exhibits high accuracy in detecting diffraction features on noisy patterns with various experimental artifacts. We demonstrate our method on real-time tracking of organic-inorganic perovskite structure crystallization and test it on two applications: 1. the automated phase identification and unit-cell determination of two coexisting phases of Ruddlesden-Popper 2D perovskites, and 2. the fast tracking of MAPbI$_3$ perovskite formation. By design, our approach is equally suitable for other crystalline thin-film materials.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to use deep - learning - based feature detection methods to efficiently analyze two - dimensional X - ray scattering data during the perovskite crystallization process. Specifically, the researchers are concerned with how to automatically process and analyze a large amount of image data generated by in - situ real - time grazing - incidence X - ray diffraction (GIXD) through an improved Faster R - CNN deep - learning architecture. This is not only to improve the understanding of the crystallization process of perovskite materials but also to optimize the performance of organic solar cells. ### Background and Objectives of the Paper Perovskite materials are regarded as the future development direction in solar cell technology because of their low cost and high efficiency. Understanding the crystallization process of perovskite materials is crucial for optimizing their structure and photoelectric properties. GIXD is a key technology that can achieve real - time, non - destructive detection of perovskite crystal structures. However, with the progress of detector technology, the amount of raw data collected per measurement can reach millions of images, far exceeding the capacity of traditional data - processing methods. Therefore, important processes are often hidden or undiscovered, and the full potential of the technology has not been fully utilized. ### Solutions To this end, the paper proposes an automated pipeline based on the Faster R - CNN deep - learning architecture for the analysis of GIXD images. The model is modified to adapt to the characteristics of scattering data and can detect diffraction features on noise patterns with high precision. The paper demonstrates the application of this method in real - time tracking of the crystallization of organic - inorganic perovskite structures and tests two specific applications: 1. **Automatic phase identification and unit - cell determination**: For two co - existing Ruddlesden - Popper 2D perovskite phases. 2. **Rapid tracking of the formation of MAPbI₃ perovskite**: The change process from the intermediate phase to the final phase. ### Methods and Results - **Data - driven model modification**: The researchers made several adjustments to the Faster R - CNN framework, including polar coordinate conversion, handling of simple features and complex experimental artifacts, the "fractal" characteristics of detection objects, asymmetric object shapes at different scales, and high - dynamic - range processing. - **Performance evaluation**: The model was trained on 48,000 simulated images and validated on 10,000 test images. The results show that the model can detect peaks with extremely high precision (99.22%), with a very low error rate (0.78% missed - detection rate and 0.25% false - detection rate). - **Practical applications**: The paper demonstrates the effectiveness of this method in two specific application scenarios, including lattice - parameter refinement of 2D halide perovskites and real - time tracking of the formation of 3D MAPbI₃ perovskites. ### Conclusions Through this automated deep - learning method, the researchers not only improve the speed and accuracy of GIXD data analysis but also provide a new tool for scientific research in the fields of materials science and organic photovoltaics. The application prospects of this method are wide - ranging, including real - time adjustment of experimental conditions and high - throughput screening of possible perovskite combinations.

Tracking perovskite crystallization via deep learning-based feature detection on 2D X-ray scattering data

A Deep Learning Based Dislocation Detection Method for Cylindrical Crystal Growth Process

Harnessing interpretable and unsupervised machine learning to address big data from modern X-ray diffraction

Perovskite or Not Perovskite? A Deep Learning Approach to Automatically Identify New Hybrid Perovskites from X‐ray Diffraction Patterns

Machine learning for ultrafast X-ray diffraction patterns on large-scale GPU clusters

Rapid detection of rare events from in situ X-ray diffraction data using machine learning

Fast and interpretable classification of small X-ray diffraction datasets using data augmentation and deep neural networks

Free Field of View Infrared Digital Holography for Mineral Crystallization

Data-driven approach for synchrotron X-ray Laue microdiffraction scan analysis

Exploring Supervised Machine Learning for Multi-Phase Identification and Quantification from Powder X-Ray Diffraction Spectra

Enhancing deep-learning training for phase identification in powder X-ray diffractograms

Artifact Identification in X-ray Diffraction Data using Machine Learning Methods

Adaptively driven X-ray diffraction guided by machine learning for autonomous phase identification

Dynamic X-ray speckle-tracking imaging with high-accuracy phase retrieval based on deep learning

Automatic Detection of Grains in Partially Recrystallized Microstructures Using Deep Learning

Rapid detection of rare events from in situ X‐ray diffraction data using machine learning

Machine Learning Recognition of hybrid lead halide perovskites and perovskite-related structures out of X-ray diffraction patterns

Machine learning-based protein crystal detection for monitoring of crystallization processes enabled with large-scale synthetic data sets of photorealistic images

Multispectral Microscopic Multiplexed (3M) Imaging of Atomically‐Thin Crystals Using Deep Learning

Hyperspectral nanoscale mapping of hybrid perovskite photophysics at the single grain level