Machine Learning Approaches to Analyzing Atomic Force Microscopy-Force Spectroscopy Measurements of Cross-Linked Polyethylene

Cross-linked polyethylene is being increasingly used for water transport and heating applications, and it is important to understand the ageing, degradation and failure mechanisms of cross-linked polyethylene (PEX-a) pipe to ensure its long-term reliability. We used atomic force microscopy-force spectroscopy (AFM-FS) to quantify changes in the morphology and mechanical properties of the inner surface of PEX-a pipe. High resolution maps of parameters such as stiffness, modulus, and adhesion were collected on pipes of different formulations and subjected to different processing and accelerated ageing conditions. To process this large amount of data, we used machine learning techniques such as regression analysis, classification analysis, and k-means clustering to reveal and correlate distinct changes in the morphology and mechanical properties with ageing. These techniques allow us to create models that predict the effective age of the pipes, to compare the relative performance of different formulations, to quantify the effects of different processing conditions, and to elucidate the mechanisms responsible for pipe degradation and failure. Our machine learning approaches to the analysis of this large amount of AFM-FS data reveals new information that goes beyond the results of traditional materials science analysis methods and complements our deep generative modeling of infrared images of the same pipes [1].

[1] M. Grossutti et al., ACS Appl. Mater. Interfaces 15, 22532 (2023).