Machine Learning Approach to Characterize the Adhesive and Mechanical Properties of Soft Polymers Using PeakForce Tapping AFM
Bahram Rajabifar, Gregory F. Meyers, Ryan Wagner, Arvind Raman
Abstract
We develop an algorithm based on the enhanced Attard’s model (EAM) to simulate PeakForce tapping (PFT) atomic force microscopy (AFM) on soft adhesive polymers. The simulations enhance our understanding of microcantilever–surface interactions, predict surface dynamics, and illustrate the role of viscoelasticity and adhesion on PFT AFM observables. Behaviors predicted by the developed algorithm cannot be fully reproduced with alternative contact mechanics models. In the second part of this study, we utilize the output of our PFT AFM simulations to train a data analytics approach that quantitatively estimates a surface’s viscoelastic and adhesive properties from experimentally acquired PFT AFM data. We demonstrate the performance of a machine learning (ML) algorithm to estimate the properties of three elastomer grades with different nominal stiffnesses. The properties extracted from the PFT AFM data using the ML algorithm agree well with the bulk properties of these polymers.