Visualizing nanoparticle surface dynamics and instabilities enabled by deep denoising

Area of Science:

• Materials Science
• Nanotechnology
• Surface Science

Background:

• Materials functionality is linked to atomic-level dynamics on millisecond timescales.
• Electron microscopy faces signal-to-noise limitations for high spatiotemporal resolution imaging.

Purpose of the Study:

• To overcome signal-to-noise limitations in electron microscopy for observing fast atomic dynamics.
• To investigate the dynamic behavior of metal nanoparticle surfaces under gas environments.

Main Methods:

• Developed an unsupervised deep denoising framework.
• Applied in situ electron microscopy to platinum nanoparticles on cerium oxide.
• Achieved time resolutions down to 10 milliseconds at moderate electron doses.

Main Results:

• Observed continuous transitions between ordered and disordered configurations on nanoparticle surfaces.
• Identified stress fields penetrating surfaces, causing defect formation and destabilization.
• Demonstrated nanoparticle fluxional behavior on millisecond timescales.

Conclusions:

• The unsupervised denoiser significantly enhances spatiotemporal characterization in electron microscopy.
• This approach opens new avenues for exploring atomic-level structural dynamics in materials.
• Understanding dynamic surface transitions is crucial for materials functionality.