Quantifying nanoscale forces using machine learning in dynamic atomic force microscopy

Area of Science:

• Materials Science
• Nanotechnology
• Data Science

Background:

• Dynamic atomic force microscopy (AFM) is crucial for material characterization.
• Quantifying nanoscale tip-sample forces typically requires complex physical models.
• A challenge exists in analyzing these forces directly from experimental data.

Purpose of the Study:

• To develop a machine learning approach for quantifying nanoscale tip-sample forces in dynamic AFM.
• To characterize forces purely from experimental data, bypassing complex modeling.
• To demonstrate the method's application on polymer blends.

Main Methods:

• Utilized machine learning and data science algorithms.
• Trained the machine learning model on standard AFM models.
• Applied the trained algorithm to experimental data from polystyrene (PS) and low-density polyethylene (LDPE) blends.

Main Results:

• Successfully characterized tip-sample forces with sub-microsecond resolution.
• Probed the complex physics of tip-sample contact in polymers.
• Estimated elasticity and provided insights into energy dissipation during contact.

Conclusions:

• Machine learning offers a novel route for dynamic AFM force characterization.
• This method enables real-time analysis of transient phenomena.
• Potential applications include studying phase transformations and biological processes.