In a groundbreaking advancement set to transform the understanding and manipulation of amorphous materials, researchers have unveiled a novel methodology enabling the atomic-scale reconstruction of their three-dimensional (3D) structures. Unlike crystalline solids that benefit from long-range order and periodicity enabling relatively straightforward structural determination, amorphous materials such as thin-film electronics, solar cells, and phase-change memory defy easy characterization due to their lack of periodic atomic arrangements. This absence of long-range order has historically rendered the precise atomic-scale mapping of these materials a formidable challenge, limiting insights crucial for optimizing their diverse technological applications.
The investigative team, spearheaded by Liao, Sha, O’Leary, and collaborators, has deployed Atomic Electron Tomography (AET)—a sophisticated imaging technique capable of mapping atomic positions in 3D—to tackle the intricacies inherent in non-crystalline structures. Their rigorous study, published in Nature in early 2026, details a robust, multi-step analytical framework that overcomes prior limitations by integrating advanced image preprocessing, denoising algorithms, and meticulous projection alignment and normalization. This comprehensive workflow ensures the extraction of accurate and reliable 3D atomic coordinates along with elemental identification, a crucial leap in characterizing amorphous solids beyond traditional means.
Central to their approach is the optimization of AET data handling. The researchers emphasize how pre-processing steps, including thorough noise reduction and normalization, significantly enhance the quality of the tomographic reconstructions. By employing advanced reconstruction algorithms tailored to mitigate artifacts typical in amorphous material imaging, the team achieves stringent positional precision, enabling visualization of atoms with sub-angstrom accuracy. This level of precision is pivotal in differentiating subtle atomic arrangements that define properties unique to amorphous phases.
The study makes a comparative evaluation against earlier methodologies, demonstrating superior performance across multiple parameters. Utilizing multislice simulations on amorphous nanoparticles composed of silicon, silicon-germanium-tin (SiGeSn), and cobalt-palladium-platinum (CoPdPt), their workflow exhibits enhanced positional precision and elemental classification accuracy at various noise thresholds. This comparative analysis not only validates the methodological rigor but also establishes quantitative benchmarks that future investigations can adopt to guarantee reliability in amorphous structure determination.
Intriguingly, for the ternary alloy CoPdPt, the method achieved remarkable elemental identification rates: detecting 95.1% of cobalt atoms, 99.0% of palladium, and a flawless 100% of platinum atoms. Even more impressive was the pinpoint positional accuracy, measured at 29 pm for cobalt, 12 pm for palladium, and an extraordinary 6 pm for platinum atoms. These values are unprecedented for amorphous materials, particularly given the realistic electron dose conditions under which the reconstructions were obtained. Such precision heralds new possibilities for linking atomic arrangement to macroscopic properties directly, providing insights that have eluded materials science for decades.
Beyond elemental mapping, the research highlights the broader implications of this breakthrough. The refined ability to resolve local atomic packing, short- to medium-range order, and heterogeneity in amorphous systems paves the way for optimizing material functionalities across several domains. This has immediate relevance for applications reliant on thin-film semiconductors, amorphous magnetic components, and emerging quantum devices, where atomic-level understanding could translate into enhanced performance, stability, and longevity.
The rigorous computational underpinning of the proposed framework merits attention. By combining physics-based simulation models with machine learning-driven classification techniques, the researchers circumvent common pitfalls such as atom misidentification or inaccurate position refinement. The use of multislice simulations to generate synthetic datasets ensures the method’s applicability across different compositions and experimental conditions, showcasing its versatility and robustness. Importantly, this hybrid approach harmonizes experimental data with theoretical constructs to push the boundaries of what is experimentally achievable.
Another remarkable aspect of the study is its holistic nature: it does not merely depend on single components of the imaging or processing pipeline but underscores the synergy among image denoising, alignment, tomographic reconstruction, atom tracing, elemental classification, and final atomic position refinement. Each step in the workflow incrementally enhances the fidelity of the final 3D atomic model, demonstrating that state-of-the-art results demand a comprehensive, well-integrated approach rather than piecemeal improvements.
The adoption of this refined AET methodology could revolutionize the way scientific communities approach the structural study of disordered systems. Historically, amorphous materials have often been characterized by indirect techniques such as X-ray diffraction pair distribution functions, neutron scattering, or reverse Monte Carlo simulations, which provide averaged or probabilistic atomic information. The ability to reconstruct individual atomic positions in three dimensions with elemental specificity constitutes a quantum leap from these averaged structural models.
Moreover, the implications extend beyond materials science. In biophysics, chemistry, and nanotechnology, where understanding non-periodic atomic structures is crucial, this imaging and analysis framework can inform the rational design of novel materials and molecular complexes. It has the potential to inspire advancements in phase-change memory devices, enhance the efficiency of photovoltaic materials, and contribute to the development of next-generation quantum sensors and detectors by delivering atomic-scale insights previously deemed unattainable.
While the presented advances are remarkable, the authors acknowledge ongoing challenges and future directions. Further refinement in dose efficiency, expansion to higher atomic number materials, and integration with in situ experimentation will be key to broadening the applicability of this technique. Nonetheless, the current findings lay a solid foundation, providing a meticulously validated protocol and performance metrics that can serve as a blueprint for subsequent studies aiming to elucidate the atomic landscape of amorphous solids.
In sum, this pioneering work establishes clear, practical guidelines to achieve accurate 3D atomic resolution in non-crystalline materials using Atomic Electron Tomography augmented by advanced computational processing. It convincingly demonstrates that the longstanding barriers to direct atomic mapping of amorphous materials can be overcome. This study is poised to influence material science profoundly, opening doors to innovations in material design, characterization, and application that hinge on detailed atomic-scale comprehension.
As the scientific community grapples with the complexity of amorphous matter’s structural mysteries, the integration of experimental prowess with computational ingenuity in this research provides a transformative toolkit. It challenges the notion that disorder equates to inscrutability at the atomic level and instead frames amorphous materials as accessible entities, whose secrets can be cracked with the right combination of technology and analytical sophistication.
Ultimately, this work not only illuminates the atomic structure of amorphous solids but also reshapes the methodological landscape of 3D structural analysis in materials science. It stands as a testament to the potential unlocked when experimental electron microscopy and computational modeling converge harmoniously, setting a new standard for atomic resolution imaging in the presence of disorder.
Subject of Research:
Determination of three-dimensional atomic structure of amorphous materials using Atomic Electron Tomography.
Article Title:
Accurate determination of the 3D atomic structure of amorphous materials
Article References:
Liao, Y., Sha, H., O’Leary, C.M. et al. Accurate determination of the 3D atomic structure of amorphous materials. Nature 649, 1123–1129 (2026). https://doi.org/10.1038/s41586-025-09857-4
Image Credits:
AI Generated
DOI:
10.1038/s41586-025-09857-4
Keywords:
Amorphous materials, Atomic Electron Tomography, 3D atomic reconstruction, amorphous Si, SiGeSn, CoPdPt nanoparticles, atomic-scale imaging, elemental classification, tomographic reconstruction, nanoscale structure, electron microscopy, positional precision
