Speeding up the Topography Imaging of Atomic Force Microscopy by Convolutional Neural Network

Area of Science:

• Surface science
• Nanotechnology
• Microscopy techniques

Background:

• Atomic Force Microscopy (AFM) offers high-resolution surface topography analysis.
• Slow scanning rates in AFM limit imaging speed and increase sample drift, reducing image fidelity.

Purpose of the Study:

• To develop a fast AFM imaging method that maintains high image quality and minimizes sample drift.
• To improve the efficiency of AFM surface topography research.

Main Methods:

• A hybrid scanning approach combining fast Raster scanning with slow μ-path subsampling.
• Utilizing a supervised Convolutional Neural Network (CNN) model trained on AFM data.
• Processing fast-scan AFM images through the trained CNN to reconstruct high-quality images.

Main Results:

• Achieved imaging speed improvements of up to 10.3 times compared to conventional methods.
• Reduced sample drift by 8.9 times during multiframe imaging of the same area.
• Demonstrated high-fidelity imaging quality comparable to slower methods.

Conclusions:

• The proposed method significantly accelerates AFM imaging while preserving high resolution and reducing drift.
• This approach is applicable to various scanning probe microscopy techniques, including scanning electrochemical microscopy.
• Enables rapid, high-quality surface topography analysis in nanotechnology and materials science.