Crystal property prediction, governed by quantum mechanical principles, is computationally prohibitive to solve exactly for large many-body systems using traditional density functional theory. While machine learning models have emerged as efficient approximations for large-scale applications, their performance is strongly influenced by the choice of atomic representation.

We introduce PRDNet that leverages unique reciprocal-space diffraction besides graph representations. To enhance sensitivity to elemental and environmental variations, we employ a data-driven pseudo-particle to generate a synthetic diffraction pattern. PRDNet ensures full invariance to crystallographic symmetries.

Multiphase powder X-ray diffraction (PXRD) analysis remains a fundamental bottleneck in structure identification, as real-world synthesis often produces complex mixtures whose constituent phases (components) cannot be reliably disentangled. While recent advances in representation-based crystal retrieval and generation suggest the possibility of inferring structures directly from PXRD, existing approaches largely assume single-phase inputs and break down in multiphase settings. Here, we present XDecomposer, a prior-free framework for joint decomposition and identification of multiphase XRD patterns without requiring candidate phase lists, structural templates, or prior knowledge of phase number. We formulate multiphase diffraction analysis as a set prediction problem, where the model infers an unordered set of phase-resolved components, their mixture proportions, and corresponding structural representations within a unified architecture. A phase-query-driven decomposition mechanism, together with diffraction-consistent physical reconstruction, enables accurate source separation while preserving crystallographic fidelity. Extensive experiments on both simulated and experimental datasets show that XDecomposer substantially improves reconstruction accuracy and phase identification across diverse chemical systems, while maintaining strong generalization to unseen mixtures. These results provide a practical route toward data-driven, source-resolved multiphase XRD analysis and reduce long-standing dependence on prior-guided iteratively phase matching.

We present WPEM, a physics-constrained workflow for X-ray diffraction (XRD) analysis that integrates Bragg’s law into a batch expectation–maximization framework. WPEM models full diffraction profiles as probabilistic mixtures, iteratively inferring component-resolved intensities while ensuring Bragg-consistent peak positions, producing stable, physically valid representations even under severe overlap or mixed phases. It outperforms standard packages on reference patterns and generalizes to complex experimental scenarios, including multiphase thin films, disordered oxides, semicrystalline polymers, operando cathodes, and historical samples. WPEM bridges AI-generated phase hypotheses with diffraction-ready structure refinement.

We developed XQueryer, an intelligent agent for simulating, recognizing, and analyzing powder X-ray diffraction (PXRD) patterns. Trained on over two million high-fidelity simulated spectra, XQueryer achieves significantly higher accuracy—28.9% better than existing AI models and traditional methods. Integrated with a powder diffractometer, it enables real-time structural analysis of crystal samples.

A freely accessible online platform is available at https://xqueryer.caobin.asia/

Powder X-ray diffraction (XRD) patterns are highly effective for crystal identification and play a pivotal role in materials discovery. While machine learning (ML) has advanced the analysis of powder XRD patterns, progress has been constrained by the limited availability of training data and established benchmarks. To address this, we introduce SimXRD, the largest open-source simulated XRD pattern dataset to date, aimed at accelerating the development of crystallographic informatics.

We developed a novel XRD simulation method that incorporates comprehensive physical interactions, resulting in a high-fidelity database. SimXRD comprises 4,065,346 simulated powder XRD patterns, representing 119,569 unique crystal structures under 33 simulated conditions that reflect real-world variations. We benchmark 21 sequence models in both in-library and out-of-library scenarios and analyze the impact of class imbalance in long-tailed crystal label distributions.

We present a spatially adaptive active-learning framework with closed-loop experimentation for targeted catalyst optimization. Bayesian optimization and a conditional variational autoencoder first identify a low-overpotential stability subspace, followed by active learning to pinpoint the most stable candidate.

This strategy leads to the discovery of a Cu–RuO₂ catalyst with outstanding durability (625 h) and a low overpotential of 177 mV at 10 mA cm⁻². Our results highlight an efficient AI-driven pathway for accelerating the design of stable acidic OER catalysts.

Materials discovery is a fundamental driver of technological advancement with direct impact on real-world challenges. From energy systems and electronics to biomedical devices and sustainable manufacturing, novel materials enable new functionalities and improved performance.

Our survey offers a detailed and comprehensive overview of material representations, particularly focusing on crystal structures and their precise mathematical definitions. We systematically categorize and compare a wide range of existing techniques, supported by a clear development timeline. To facilitate research and practical application, we provide abundant resources, including links to open-source code and datasets. Additionally, we highlight current challenges and propose future research directions to inspire continued innovation in the field.

In 2025, I led the development of the user interface software for Bgolearn, called BgoFace, published in MGR Advances. BgoFace is a user-friendly platform designed to accelerate materials innovation through active learning. It streamlines Bayesian global optimization by tackling key challenges such as experimental–computational interoperability and algorithm accessibility. With its intuitive interface and built-in support for experimental constraints, BgoFace enables efficient materials discovery without requiring deep machine learning expertise.

Using default settings and simple button clicks, BgoFace guided six iterations of optimization, recommending four infill points per iteration via four different utility functions—for a total of 24 suggestions. The optimal sample, with a highest QY of 37.25%, was discovered in the fifth iteration—nearly doubling the initial value. The software and source code are openly available at https://github.com/Bgolearn/BgoFace.

A notable difficulty in applying machine learning to this domain is the lack of sufficiently sized experimental datasets, which has constrained researchers to train primarily on simulated data. However, models trained on simulated pXRD patterns showed limited generalization to experimental patterns, particularly for low-quality experimental patterns with high noise levels and elevated backgrounds.

With the Open Experimental Powder X-Ray Diffraction Database (opXRD), we provide an openly available and easily accessible dataset of labeled and unlabeled experimental powder diffractograms. Labeled opXRD data can be used to evaluate the performance of models on experimental data and unlabeled opXRD data can help improve the performance of models on experimental data, e.g. through transfer learning methods. We collected 92552 diffractograms, 2179 of them labeled, from a wide spectrum of materials classes.

In 2022, I open-sourced the Bgolearn framework and have since actively maintained and promoted it in collaboration with experimental researchers. My goal is to foster materials innovation through attachable and accessible machine learning tools.

Bgolearn is a lightweight and extensible Python package for Bayesian global optimization, tailored for accelerating materials discovery and design. It supports both regression and classification tasks, includes a variety of acquisition strategies, and provides a seamless pipeline for virtual screening, active learning, and multi-objective optimization.

• 📦 Official PyPI: pip install Bgolearn
• 🎥 Code tutorial (BiliBili): Watch here
• 🚀 Colab Demo: Run it online

I’m truly glad to see several groundbreaking innovations in materials science enabled by the Bgolearn framework:

2025

1. Nano Letters: Self-Driving Laboratory under UHV — Link
2. Small: ML-Engineered Nanozyme System for Anti-Tumor Therapy — Link
3. Computational Materials Science: Mg-Ca-Zn Alloy Optimization — Link
4. Measurement: Foaming Agent Optimization in EPB Shield Construction — Link
5. Intelligent Computing: Metasurface Design via Bayesian Learning — Link

2024

6. Materials & Design: Lead-Free Solder Alloys via Active Learning — Link
7. npj Computational Materials: MLMD Platform with Bgolearn Backend — Link

In this work, we present a crystal generative framework based on Wyckoff generative adversarial network (CGWGAN) to efficiently discover novel crystals.

The CGWGAN includes three modules: a generator of crystal templates, an atom-infill module, and a crystal screening module. The generator uses a generative adversarial network (GAN) to produce crystal templates embedded with asymmetry units (ASUs), space groups, lattice vectors, and the total number of atoms within the lattice cell, ensuring that the generated templates precisely match all requirements of crystals. These templates become crystal candidates after filling in atoms of different chemical elements. These candidates are screened by M3GNet and the passed ones are subjected to density functional theory (DFT)-based calculations to finally verify their stability. As a showcase, the CGWGAN successfully discovers seven novel crystals within the Ba-Ru-O system, demonstrating its effectiveness. This work provides a knowledge-guided Artificial Intelligence generative framework for accelerating crystal discovery.