Artifact identification in X-ray diffraction data using machine learning methods

Abstract

In situ synchrotron high-energy X-ray powder diffraction (XRD) is highly utilized by researchers to analyze the crystallographic structures of materials in functional devices (e.g. battery materials) or in complex sample environments (e.g. diamond anvil cells or syntheses reactors). An atomic structure of a material can be identified by its diffraction pattern along with a detailed analysis of the Rietveld refinement which yields rich information on the structure and the material, such as crystallite size, microstrain and defects. For in situ experiments, a series of XRD images is usually collected on the same sample under different conditions (e.g. adiabatic conditions) yielding different states of matter, or is simply collected continuously as a function of time to track the change of a sample during a chemical or physical process. In situ experiments are usually performed with area detectors and collect images composed of diffraction patterns. For an ideal powder, the diffraction pattern should be a series of concentric Debye–Scherrer rings with evenly distributed intensities in each ring. For a realistic sample, one may observe different characteristics other than the typical ring pattern, such as textures or preferred orientations and single-crystal diffraction spots. Textures or preferred orientations usually have several parts of a ring that are more intense than the rest, whereas single-crystal diffraction spots are localized intense spots owing to diffraction of large crystals, typically >10 µm. In this work, an investigation of machine learning methods is presented for fast and reliable identification and separation of the single-crystal diffraction spots in XRD images. The exclusion of artifacts during an XRD image integration process allows a precise analysis of the powder diffraction rings of interest. When it is trained with small subsets of highly diverse datasets, the gradient boosting method can consistently produce high-accuracy results. The method dramatically decreases the amount of time spent identifying and separating single-crystal diffraction spots in comparison with the conventional method.

Advancing Efficiency in Mineral Construction Materials Recycling: A Comprehensive Approach Integrating Machine Learning and X-ray Diffraction Analysis

“…Bunn et al [13] even utilized a threshold of 5% of the maximum intensity value for feature extraction. In Yanxon et al [14], the weight proportions of the minor components were between 2% and 20% by mass for two-phase mixtures. Classical methods such as RF have an advantage, as they offer traceability and explainability in the decisions.…”

“…As already mentioned, Lee et al [7] showed an additional approach that uses, among various other methods, an RF. Yanxon et al [14] used kNN, extra tree, gradient boosting, and RF for the detection of single-crystal diffraction spots in XRD images so as to enable precise analyses of 1D powder diffraction patterns. Again, RF proved to be the most suitable method.…”

“…A train-test split of 0.3 was used, resulting in sets of 1176 test and 504 training data pieces. Based on the results from [6,7,14], RF was chosen as the prediction method. This decision was based on the insights from the literature described above, which identified RF, alongside NN, as the most suitable classifier.…”

Machine Learning Automated Analysis of Enormous Synchrotron X-ray Diffraction Datasets

“…Combining μ-XRD mapping with solid–liquid mixed samples in high-temperature and high-pressure environments presents the most challenging scenario for obtaining high-quality XRD data interpretation. μ-XRD is typically used to analyze micrograins in samples, but when dealing with mixed solid–liquid samples, common issues such as overexposure, imperfect diffraction, and preferred orientation can lead to artifacts in the XRD pattern . If the samples are entirely solid, these issues can still be manageable, and data of reasonable quality can be retrieved. − However, the presence of the liquid phase and extreme environments significantly amplifies these adverse effects, leading to distorted XRD data.…”

Closing the loop: autonomous experiments enabled by machine-learning-based online data analysis in synchrotron beamline environments

“…X-ray user facilities are amongst the largest scientific data producers in the world (Yanxon et al, 2023). While experiments performed at these facilities cover an extensive range of multi-disciplinary sciences, they typically share a common data generation pattern, namely precisely positioning a specimen in the path of the X-ray beam and recording data (e.g.…”