XQueryer: an intelligent crystal structure identifier for powder X-ray diffraction

Powder X-ray diffraction (PXRD) is a widely used technique for characterizing crystal structures. With the rise of artificial intelligence (AI) in science, the rapid expansion of AI laboratories and high-throughput characterization methods has created a growing demand for intelligent PXRD pattern analysis–that is, automated crystal structure identification in integrated AI-driven facilities. Such identifiers extract structural information directly from PXRD patterns and transmit it as accurate and efficient data streams to other intelligent components involved in material synthesis, characterization, and optimization. In this work, we present XQueryer, an intelligent PXRD-based structure identifier. To support its development, we constructed a comprehensive simulation database using a newly designed high-fidelity simulation method. This database contains over two million PXRD patterns simulated from 100,315 crystal structures in the Materials Project, covering diverse intrinsic sample features and a range of extrinsic diffractometer conditions. XQueryer was evaluated on more than 200,000 simulated patterns and 1,003 experimental PXRD patterns. It outperformed existing models and the traditional search–match method, achieving a 28.9% accuracy improvement compared with the second-best model. Furthermore, XQueryer has been integrated with a powder X-ray diffractometer, enabling real-time determination of crystal structures in tested samples.