The utilization of compacted bentonite as the preferred buffer and backfill material in deep geological repositories for high-level radioactive waste disposal is well established due to its excellent engineering properties. The effectiveness of bentonite in such applications is largely influenced by its mineral composition, particularly the content and type of montmorillonite, a key clay mineral. Montmorillonite's ability to swell and its expansive nature make it highly effective in minimizing the migration of radionuclides and providing stability to the repository structure. In near-field environments, which are the zones closest to the waste container, changes in the physical and chemical properties of montmorillonite, especially its water retention and mechanical strength, have a direct impact on the long-term safety and performance of the repository. The accurate understanding of these properties under various environmental conditions is critical to ensuring that the bentonite buffer effectively isolates the waste over extended time periods.Traditionally, experimental methods such as X-ray diffraction (XRD), scanning electron microscopy (SEM), and various mechanical tests have been used to study the behavior of bentonite. While these techniques provide valuable macroscopic data, they fall short in addressing the molecular and atomic-scale interactions that govern the material's behavior. The hydro-mechanical properties of montmorillonite, which are essential for its performance as a buffer material, depend on the interactions between water molecules and the mineral's surface and interlayer spaces. These interactions occur on the molecular scale, and understanding them requires a more detailed, atomistic approach. This is where molecular dynamics (MD) simulations come into play.This paper reviews the applications of molecular dynamics in clay mineral research, particularly focusing on montmorillonite. It provides an in-depth look at how MD simulations can elucidate the mechanisms of interlayer molecular behavior that govern the hydro-mechanical properties of montmorillonite. The behavior of water in montmorillonite, such as its adsorption, diffusion, and swelling, is highly dependent on the interlayer spacing and the electrostatic interactions between the mineral surfaces and water molecules. These interactions are crucial for understanding the swelling pressure, permeability, and mechanical strength of the material, all of which are key factors in its performance as a buffer material. The paper also discusses how mesoscopic modeling, based on molecular dynamics simulations, can bridge the gap between the molecular scale and the macroscopic properties of bentonite. Mesoscopic models are useful for simulating larger systems, such as the behavior of bentonite under stress or hydration, while still retaining the molecular-level interactions. These models help to scale up the insights gained from molecular dynamics simulations, enabling researchers to predict the material's mesoscopic behavior under field conditions. In addition, the paper highlights key future research directions. Continued advancements in molecular dynamics, such as improved computational power and simulation accuracy, will enable more detailed studies of montmorillonite's interactions with other repository components, including radionuclides. Additionally, combining molecular dynamics with other methods like finite element modeling could provide a more comprehensive understanding of bentonite behavior. Future research should focus on refining mesoscopic models, improving long-term prediction accuracy, and validating simulations with real-world data.Ultimately, the goal of this research is to improve the understanding of the complex molecular mechanisms that govern the behavior of compacted bentonite in near-field environments. By using molecular dynamics simulations to explore these mechanisms, researchers can develop more accurate models of the material's performance, leading to better-designed, more effective buffers for high-level waste disposal. This, in turn, will contribute to the safety and sustainability of deep geological disposal systems for nuclear waste, ensuring that they meet the stringent requirements for long-term isolation and containment of hazardous materials.Summary and prospectsRecent advances in molecular dynamics (MD) have significantly expanded the study of bentonite buffering performance in deep geological repositories, extending the research scale from the micrometer to the nanometer level. This has provided crucial data and theoretical insights into material behavior, cross-scale modeling, and optimization. The accuracy of MD simulations largely depends on the appropriate selection of force fields and initial structures, with the development of clay-specific force fields (e.g., ClayFF) and databases providing a solid foundation. However, parameter adjustments for specific clay types are necessary to ensure computational precision. MD simulations have uncovered the molecular mechanisms behind montmorillonite expansion under various environmental conditions, challenging traditional models that simplified montmorillonite as a “parallel capacitor” and offering potential for refining theoretical models. These studies also link interlayer configurations to overall mechanical properties, supporting more accurate predictions of radionuclide migration and informing buffer material development. Additionally, the introduction of granular and discrete element methods has overcome the scale limitations of traditional MD, enabling more effective mesoscopic modeling of bentonite. Such models offer significant advantages in simulating aggregate expansion, permeability, and deformation, improving the accuracy of particle interaction and soil-water behavior. However, a unified modeling framework is still a key area of focus in current research.Looking ahead, future research should prioritize the development of new force fields that can simulate chemical processes, the creation of microstructural models to describe complex hydro-mechanical behaviors, and the integration of MD with multi-scale numerical platforms to predict long-term buffer performance. This integrated approach will provide important scientific support and technical guidance for the construction of geological repositories.