image: (a) User interface of the ZhouTomo software, which integrates microscope control, autofocus, target tracking and tilt-series acquisition. (b) Volume rendering of the reconstructed Pt nanoparticle aggregate acquired fully automatically in about 1.5 hours. (c) A 1.4-Å-thick central slice through the reconstructed volume, showing clear lattice fringes and internal voids. (d) Surface rendering of the Pt nanoparticles, with {100} facets in yellow and {111} facets in blue. Credit: Zhang et al. / Peking University.
Credit: ©Science China Press
Solid–solid interfaces and surfaces largely determine how materials behave, yet their three-dimensional (3D) atomic structures are notoriously difficult to resolve because they are buried, weakly responding and intrinsically non-periodic. Atomic-resolution electron tomography (AET) offers a powerful route to determine 3D atomic coordinates without relying on long-range crystallinity, making it especially suited to local, non-periodic and buried interfacial structures. However, current AET studies are limited by low acquisition throughput and a lack of dynamic or in situ measurement capability.
A research team led by Prof. Jihan Zhou at Peking University now reports an integrated software–hardware instrumentation framework that advances quantitative and in situ AET for 3D interface characterization. The work is published in Advanced Scientific Instruments.
First, the team establishes a standardized strategy for evaluating AET reconstruction quality by combining Fourier shell correlation, F1 score, root-mean-square deviation, projection-space residuals and Zernike-based spherical symmetry. This provides a multi-level assessment of resolution, reproducibility, positional precision, data consistency and alignment accuracy, addressing the current lack of standardized metrics across studies.
Second, the team develops ZhouTomo, a visually guided automated acquisition platform that integrates microscope control, autofocus, target tracking and tilt-series acquisition. In a demonstration on an aggregated Pt nanoparticle assembly, ZhouTomo acquired a near-atomic-resolution HAADF-STEM tilt series fully automatically in about 1.5 hours, compared with the 4–6 h typically required for manual operation. Reconstruction revealed nine single-crystalline Pt nanoparticles with Wulff-like polyhedral morphologies and {100}/{111} surface facets.
Third, the team designs and validates a high-tilt in situ heating holder that enables large-angle tilting (up to ±75°) and controlled heating within the microscope. Using this holder, the researchers tracked the thermally driven structural evolution of Pd@Pt core–shell nanoparticles from room temperature to 900°C, observing shell shrinkage, interfacial blurring and progressive Pd–Pt alloying. They also evaluated a commercial cryogenic holder for high-tilt atomic-resolution imaging, demonstrating its potential compatibility with AET data acquisition under cryogenic conditions.
Together, these advances outline a pathway toward automated, faster, more reliable and more dynamic 3D atomic characterization of solid–solid interfaces. The authors note that integrating automated acquisition with physics-aware neural-network reconstruction and advanced in situ holders could extend AET to external-stimulus-driven experiments—thermal, electrical, electrochemical or gas-environment—and enable high-throughput statistical tomography of interfaces, defects and dynamic structural transformations. Beyond metallic nanoparticles, extending AET to thin-film interfaces, battery materials and semiconductor heterojunctions will require further advances in specimen preparation, dose tolerance, light-element imaging and experimental control.
Method of Research
Experimental study