Shape-dependent local strain in gold nanorods: data-driven atomic-resolution electron microscopy analysis

TL;DR

This study enhances the precision of local strain measurement in gold nanorods using data-driven electron microscopy analysis, revealing shape-dependent strain variations and their implications for nanoscale engineering.

Contribution

It introduces Gaussian process regression to improve strain measurement accuracy in atomic-resolution electron microscopy of nanoparticles.

Findings

Strain measurement precision improved to 0.2%.

Detected characteristic lattice expansion of +0.6% in nanorods.

Strain variations explained by curvature-dependent surface stress.

Abstract

The local variation in inter-atomic distances, or local lattice strain often influences significantly material properties of nanoparticles. Strain measurement with ~1% precision is provided by recent atomic-resolution electron microscopy. However, the precision has been limited by noises in the experimental data. Here, we have applied one of the data-driven analyses, Gaussian process regression to predict true form of strain. The precision has been improved to be sub-percent of 0.2 % and more for detection of local strain. Rod-shaped nanoparticles have been revealed to contain characteristic lattice expansion ~+0.6 % around the subsurface cap tip area. The experimental results are reproduced by molecular dynamics simulations of the corresponding shaped atomic models. The strain peculiar to nanorods are explained in terms of curvature-dependent non-uniform surface stress due to shape anisotropy. The present results bring a hint to nanoscale engineering to optimize the strain in nanoparticles by shape control.