(91g) Machine Learning Enhanced Modeling of Orientation-Dependent Van Der Waals Interactions in Nanostructured Cuboids with Surface Roughness Effects

Van der Waals (vdW) interactions play a critical role in governing self-assembly, colloidal stability, and interparticle forces. Given the prevalence of anisotropic particle geometries and surface roughness in real-world colloidal systems, we have developed a computational framework that combines Hamaker’s approach with numerical modeling of non-spherical particles featuring surface roughness. While this method significantly reduces computational time compared to traditional techniques, it still imposes a considerable computational burden, limiting our ability to fully explore how vdW potentials depend on morphology, configuration, surface roughness, and interparticle distance.

To overcome these limitations, we are integrating machine learning (ML) frameworks trained on simulation data generated from our previously established vdW models. This strategy not only accelerates prediction of interaction potentials but also broadens the applicability of our approach to include various particle shapes such as cubes, cuboids, and complex polygons. Our findings reveal that the power-law exponents governing vdW interactions are sensitive to surface roughness and particle configurations, offering deeper physical insight. Current research efforts are focused on extending this framework to capture emergent behaviors in many-particle ensembles, moving beyond simple pairwise interactions.