In materials science, a multi-scale approach is essential for a comprehensive understanding of diffusive and reactive behaviour through a combination of experiments, modeling, and simulations. Bridging the gap from the atomic scale to the meso- or macro-scales and from nanoseconds to seconds or beyond, poses a crucial challenge in this field. This workshop aims to address this challenge by focusing on complex interfaces, including rough interfaces, small and deformable molecules, and evolving surfaces over time (durability, chemical attacks,  pH effects, etc.). The primary objective of this CECAM workshop is to explore methods that can effectively bridge the atomic and mesoscale realms for such interfaces. Recently, this has driven a lot of research, in particular in the molecular dynamics community for studying polymers, construction materials, porous materials, biomolecules, etc. [1, 2, 3]. Another particularly important step is going from the nanosecond to seconds, minutes or days in order to observe how chemical reactions can affect the atomistic structure and durability.

Coarse-grained models have emerged as valuable tools to represent the mesoscale[4]. These models represent groups of atoms as single particles, reducing the computational complexity while still capturing the essential features of the system. Coarse-grained simulations enable the study of larger structures, such as supramolecular assemblies or biological membranes, and longer time scales, making them well-suited for exploring mesoscale phenomena[4]. In order to do molecular dynamics at the coarse-grain level, it is necessary to determine the interaction potential between particles, which can be done by using experimental information (for instance mechanical strength, indentation modulus, etc.)[5], by using atomistic simulations to reconstruct the potential of mean force[6] or by employing generalized force fields[7,8]. Recently, enhanced sampling techniques have been used to help explore the conformational space and overcome energy barriers, enabling the study of rare events and complex dynamics[9], including replica exchange, metadynamics, and umbrella sampling. Besides, machine learning approaches have also gained attention in atomic-to-mesoscale modeling[10] as these methods can aid in the development of accurate interatomic potentials or potential of mean force reconstructions[11,12]. Machine learning techniques offer the potential to capture complex non-linear relationships and speed up simulations.

Recent work using multi-scale approaches, based on potential of mean force approaches, have helped to understand the properties of many materials such as cement[13,14], clays, zeolites[15], polymers, etc. Therefore, these approaches are key in current progress in materials sciences. However, up to now, one major issue is dealing with complex materials in which surfaces are difficult to characterize[16] or where chemical environments changes could lead to chemical reactions that will modify the grain-to-grain interactions[17]. This is particularly important when considering material durability, which is a key challenge nowadays. Being able to bridge the atomic scale and the mesoscale for complex interfaces would open new avenues of research that include rough interfaces, deformable molecules, surfaces that evolve over time due to time, or chemical attacks or pH, and more. The aim of this workshop is to discuss methods or approaches that could be developed to go in that direction.