Determining the three-dimensional atomic structure of an amorphous solid

Yao Yang, Jihan Zhou, Fan Zhu, Dillan J. Chang, Dennis Kim, Yakun Yuan, Minh Tuan Pham, Arjun Rana, Xuezeng Tian, Yonggang Yao, Stanley Osher, Liangbing Hu, Peter Ercius, Jianwei MiaoNature 2021 · DOI · arXiv

Introduction

The determination of atomic structures has been a cornerstone of physical science since the inception of X-ray crystallography over a century ago. While the periodic nature of crystals allows for precise structural mapping through diffraction, amorphous solids—such as window glass or metallic glasses—lack long-range order, making their internal organization notoriously difficult to resolve. Historically, our understanding of these materials has been limited to statistical averages and computational models. This study, published in Nature, marks a significant breakthrough by directly mapping the three-dimensional coordinates and chemical species of individual atoms within an amorphous solid with unprecedented precision.

The Challenge of Disordered Systems

Amorphous materials are characterized by a lack of translational symmetry. While researchers have long hypothesized the existence of "short-range order" (local atomic clusters) and "medium-range order" (the arrangement of these clusters), observing these structures directly in 3D has been considered a "holy grail" of materials science. Without a clear map of these structures, linking the atomic arrangement of glasses to their macroscopic physical properties—such as strength, ductility, and thermal stability—has remained a challenge.

Experimental Methodology: Atomic Electron Tomography

The research team utilized Atomic Electron Tomography (AET) to bridge this gap. The process began with the synthesis of high-entropy alloy (HEA) nanoparticles containing eight distinct metallic elements (Co, Ni, Ru, Rh, Pd, Ag, Ir, and Pt). To ensure a disordered state, the team employed a "carbothermal shock" technique, achieving an ultra-fast cooling rate of 69,000 K/s to "freeze" the atoms before they could form a regular lattice.

Data acquisition involved capturing a "tilt series" of 51 high-resolution images using a Scanning Transmission Electron Microscope (STEM) as the nanoparticle was rotated. The critical innovation, however, lay in the reconstruction process. The team developed a new algorithm called RESIRE (REal Space Iterative REconstruction). Unlike traditional Fourier-based methods, RESIRE operates entirely in real space, iteratively minimizing the discrepancy between the experimental 2D projections and a 3D atomic model to pinpoint the location of every atom.

Unveiling Order Within Chaos

The analysis revealed that the amorphous solid is not a purely random distribution of atoms. By applying Voronoi tessellation and breadth-first search algorithms, the researchers identified hidden "Medium-Range Order" (MRO) networks.

Key Structural Insights:

  • Hidden Symmetries: The study identified four distinct types of crystal-like MROs coexisting within the sample: face-centered cubic (fcc), hexagonal close-packed (hcp), body-centered cubic (bcc), and simple cubic (sc).
  • Distortion and Anisotropy: These MRO networks are highly distorted and do not possess orientational order, which explains why they remain invisible to traditional diffraction experiments.
  • Chemical Drivers: The researchers observed that chemical interactions influence the structure. Specifically, bonds between certain types of atoms were found to be 2.4% shorter than expected, suggesting that chemical preferences drive the formation of these local clusters.

Quantitative Results and Validation

The precision of the reconstruction is remarkable, with atomic positions determined to within 21 picometers—approximately one-tenth the diameter of an atom. Validation through multislice simulations confirmed that 96.5% of the atoms were correctly identified and classified.

Furthermore, the study provided a rare glimpse into the early stages of crystallization. The team identified 19 small crystal nuclei within the disordered matrix, 18 of which were located near the surface. The identified MROs measured between 2.06 and 2.40 nm, a scale that corresponds significantly with "shear transformation zones," the regions responsible for deformation in metallic glasses.

Limitations and Future Directions

While this work is a landmark achievement, it faces certain limitations. Current Z-contrast imaging can distinguish between groups of elements based on atomic number but cannot yet differentiate between elements with very similar numbers, such as Cobalt and Nickel. Additionally, the study was performed on a 9nm nanoparticle; scaling this technique to bulk materials presents significant computational and experimental hurdles.

Looking forward, the authors propose "4D AET" as the next frontier. By adding time as a dimension—imaging samples as they undergo thermal treatment—researchers could potentially observe the glass transition and the nucleation of crystals atom-by-atom in real-time. This capability promises a new era where we can design glasses with specific properties by engineering their hidden atomic networks.

Summary updated

BibTeX
@inproceedings{yang2021determining,
  author = {Yao Yang and Jihan Zhou and Fan Zhu and Dillan J. Chang and Dennis Kim and Yakun Yuan and Minh Tuan Pham and Arjun Rana and Xuezeng Tian and Yonggang Yao and Stanley Osher and Liangbing Hu and Peter Ercius and Jianwei Miao and Jianwei Miao},
  title = {{Determining the three-dimensional atomic structure of an amorphous solid}},
  booktitle = {Nature},
  year = {2021},
  doi = {10.1038/s41586-021-03354-0},
}

Loading related content...

Figure