An AI framework for real-time crystal-structure identification from powder X-ray diffraction (PXRD) patterns.
Explore the article website → bin-cao.github.io/XQueryer
The multilingual project page introduces the paper, simulation pipeline, model architecture, benchmark results, and real-time diffractometer integration.
Tip
Need a smaller model? Try XQueryer Lightweight → A lightweight implementation is available in its dedicated repository.
XQueryer combines high-fidelity, physics-guided PXRD simulation with a neural structure-identification model. It is designed for AI-driven laboratories where diffraction data should become useful crystal information quickly and automatically.
| Dataset | Model | Validation | Deployment |
|---|---|---|---|
| 2.3M+ simulated PXRD patterns from 100,315 Materials Project structures | FFT filtering, CNN features and cross-attention | +28.9% accuracy over the next-best model; 70.3% accuracy on 1,003 RRUFF patterns | Integrated with a PANalytical Aeris benchtop diffractometer for real-time analysis |
- Physics-guided data synthesis — models intrinsic sample factors and extrinsic diffractometer effects to create diverse PXRD patterns.
- Robust pattern understanding — frequency-domain filtering helps reduce noise and peak overlap before classification.
- Real-time workflow — automatically parses fresh diffractometer output and returns identified structures with Materials Project information.
- Open research resources — source code, simulation tools, tutorials, matching utilities, and benchmark code are linked below.
 Code & data |
 Learning |
 Related |
|---|---|---|
| Source code · Simulation · RRUFF–MP matching · Dataset | Model tutorial · Simulation tutorial · High-throughput simulation | XqueryerBench · Lightweight version · Video demo |
Bin Cao, Zinan Zheng, Yang Liu, Longhan Zhang, Lawrence W-Y Wong, Lu-Tao Weng, Jia Li, Haoxiang Li and Tong-Yi Zhang. XQueryer: an intelligent crystal structure identifier for powder X-ray diffraction. National Science Review 12, nwaf421 (2025). Read the article
Maintained by Bin Cao. For questions, suggestions, or issues, please open an issue or contact bcao686@connect.hkust-gz.edu.cn.