Powder X-ray diffraction (PXRD) is the routine probe of crystalline matter, yet its analysis is the rate-limiting step as laboratories automate acquisition. Deep-learning analyzers excel on simulated patterns and degrade on measured ones. This simulation-to-real gap is structural, not additive: synthetic denoising gives no measurable lift on real spectra, whereas correcting a small peak-position drift more than doubles median retrieval correlation. Real-spectrum fine-tuning, peak-aligned reranking, and recalibration narrow what remains and restore the coverage synthetic anchors lose. Xtalyst integrates these in an agent-orchestrated system spanning phase identification, refinement, and calibrated property prediction. On a frozen held-out partition (n=534) each module measured on both splits reproduces its development finding -- including the synthetic-anchor under-coverage, whose magnitude differs between the two pools -- while held-out refinement converges and preserves symmetry without reaching profile-quality fits, and on a diffractometer its wet-dry recommend-rescan-reanalyze loop flips a blinded silicon standard to a gated PASS and changes which minor phase is resolved on a multi-metal alloy.
Crystal structure prediction (CSP) from powder diffraction data is a central challenge in materials chemistry. Machine learning (ML) models show promise, but most are trained on idealized simulated data, limiting reliability on real experiments. Here, we assess real-world behaviour using the previously published deCIFer model as an example of PXRD-conditioned generative CSP. deCIFer is an autoregressive transformer that conditions each step of structure generation on encoded PXRD data, guiding token-wise predictions of space group, lattice parameters, and atomic positions. Using controlled robustness tests, we quantify performance under realistic artefacts (noise, background, peak asymmetry, and Scherrer broadening) and introduce metrics for accuracy and predictive uncertainty. deCIFer adapts smoothly to signal distortions and improves over unconditioned baselines when diffraction features remain informative, while expressing appropriate uncertainty as the PXRD pattern becomes underdetermined. Experimental PXRD tests recover the known structures of Si and CeO2 and expose the expected limitations for lower-symmetry Fe2O3 and nanocrystalline CeO2. Overall, ML-based CSP is fundamentally limited by the information content of PXRD, but can accelerate expert workflows by rapidly generating chemically plausible candidates and quantifying uncertainty, making such models valuable human-in-the-loop tools for real-world structure determination.
Frederik L. Johansen, A. Sapnik, Erik B. Dam et al.· Digital Discovery· 0 citations
High-throughput experimentation and self-driving laboratories are drastically accelerating materials discovery, yet automated interpretation of X-ray powder diffraction (XRPD) data remains a critical rate-limiting step. Conventional search-match workflows rely heavily on expert manual intervention, while pure data-driven machine learning approaches suffer from limited generalizability across chemical systems and lack rigorous crystallographic interpretability. Here we present MatDiffract, a material-informed automated analysis platform for high-throughput XRPD characterization. Built on a first-principles density functional theory (DFT)-derived inorganic crystal structure database, Atomly, MatDiffract constructs a perturbation-augmented simulated diffraction database, embeds multi-scale diffraction features into indexable vectors, and integrates hierarchical vector retrieval with full-pattern fitting Rietveld refinement and quantitative phase fitting. Benchmarked on 875 single-phase experimental patterns, the platform achieves 91.3% Top-1 and 97.2% Top-10 identification accuracy after automated refinement. For binary and ternary multiphase mixtures, it delivers 85.0% and 70.0% Top-1 accuracy with mass fraction mean absolute errors as low as 1.2% and 1.8%, respectively. Beyond mere phase labeling, MatDiffract outputs full crystallographic results including refined structural models, fitted profiles, and quantitative compositions within tens of seconds per sample. Its modular vector-based architecture supports seamless incremental expansion to new material systems, providing an end-to-end solution to close the characterization throughput gap for autonomous materials discovery and high-throughput materials development.
Hongqing V. Wang, Ming-Wei Chen, Hong Luo et al.· 0 citations
High-Energy Diffraction Microscopy (HEDM) recovers crystallographic orientation, strain, and grain position from rotating-crystal X-ray diffraction patterns. Existing forward models in far-field (FF), near-field (NF), and point-focused (pf) HEDM are not differentiable, which forecloses gradient-based joint parameter refinement, physics-informed regularisation, and Bayesian uncertainty quantification. We present the first end-to-end differentiable HEDM forward model covering all three geometries, implemented in PyTorch with pixel-exact agreement against the established MIDAS reference simulators (162/162 FF, 2304/2304 NF including a non-zero detector-tilt sweep, and 1088/1096 pf-HEDM spots matched). Three demonstrations validate the framework: joint orientation-strain-position recovery in NF-HEDM at ~6 nm precision; round-trip refinement on a real 214-grain alpha-Ti FF-HEDM dataset reaching 100% grain recovery from a 1.5 degree initial perturbation with residuals matching the production fit to 0.3%; and joint refinement of all per-detector geometry parameters and per-grain state on a synthetic four-panel FF-HEDM setup, recovering panel rotations about the beam axis to ~10 mu-rad and a global rotation-axis wedge to ~26 mu-rad. The framework is released as the open-source midas-diffract package (pip install midas-diffract).
Hemant Sharma, Nina Andrejevic, Simon Zhang et al.· 1 citation
A novel approach to post mortem characterisation of electrochemical and photovoltaic devices is spatially-resolved diffraction using a hyper-focused, micron-width x-ray beam to examine the distribution of degradation products and strain, a technique called $\mu$-XRD. Aside from the experimental difficulties associated with beam focusing and sample preparation, which are themselves non-trivial, the analysis of resulting data is complex and challenging, with full Rietveld analysis rarely attempted in literature. The difficulty lies in the size of the data, which may consist of hundreds or even thousands of diffraction patterns with very different crystallographic phase compositions depending on position within the device, and the difficulty in fitting the data due to the presence of many phases at the same position, including possible degradation products which may be difficult to index and assign to known phases. In this paper, we present a fully-automated open access Python routine for performing phase identification and Rietveld analysis on 2D datasets of diffraction pattern taken at micron-scale positions, measured over the cross-section of a chemically inhomogeneous device with polycrystalline phases. Solid oxide electrolyser cells are a promising technology for green hydrogen production which can utilise waste heat to split water at higher efficiencies than low-temperature electrolysis techniques such as polymer electrolyte membranes, but exhibit many degradation modes due to the high operating temperatures. We present a case study using our analysis protocol on an SOEC fragment encompassing the air electrode, cation diffusion barrier, electrolyte, and fuel electrode. With modification, this protocol could be applied to other devices such as all-solid-state batteries, wet-electrolyte battery electrodes, solid oxide fuel cells, photovoltaic devices, and metal-oxide pseudocapacitors.
L. A. Nagle-Cocco, C. Crain, Michael J. Dzara et al.· 0 citations
X-ray diffraction (XRD) is the primary tool for identifying crystalline phases following synthesis, but automated phase identification remains challenging, particularly for multiphase samples with overlapping peaks and experimental artifacts. While deep-learning methods have been proposed to improve upon classical search-match algorithms, most formulate phase identification as a single closed-set classification problem, requiring one shared model to discriminate among all candidate phases. Here we introduce GALAXI, which instead decouples the identification task into independent one-versus-all binary classifiers that each specialize in recognizing a single phase. These pre-trained classifiers first narrow the search space to a small set of plausible phases, which are then evaluated through Rietveld refinement to identify the combination of phases that best explains the full diffraction pattern. On a curated set of experimental patterns, GALAXI identifies the correct phases with a micro-F1 score of 0.935, outperforming classical search-match and prior deep-learning models. The method remains robust to common experimental artifacts, including low impurity phase fractions, small crystallite size, peak shifts, sample displacement, and texture, and performs well when applied to time-resolved in-situ XRD data from solid-state reactions. Moreover, because the phase-specific models are independent, GALAXI can expand to large reference libraries without retraining existing models. This modular architecture enables us to train classifiers for 64,594 structures from the Crystallography Open Database and deploy them through a public web interface at https://galaxi-xrd.com.
Xin-Yang Tong, Ethan Jin, Jia-Hang Xu et al.· 0 citations
We introduce a data-efficient neural-vision assisted method to automate crystallographic alignment during molecular beam epitaxy (MBE) growth. Trained on reflection high-energy electron diffraction (RHEED) patterns from only 15 CdTe structures, our model - enabled by physics-aware postprocessing - reliably infers crystallographic directions, replacing manual frame-by-frame inspection. To this end, we design, test, and critically compare neural-network architectures based on 2D and 3D ResNet configurations, both with and without postprocessing that leverages the physical constraints of RHEED image acquisition. Our work delivers (i) a fully trained neural system ready for closed-loop deployment in future CdTe growth experiments and (ii) a generalizable pipeline for new materials where access to diverse RHEED datasets is limited. More broadly, this study represents a step toward AI-driven MBE growth and demonstrates the potential of machine-learning-assisted automation in thin-film synthesis.
Bartłomiej Turowski, Jakub Meixner, Róża Dziewiątkowska et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
