Canada’s approach to deep geological repositories for nuclear waste relies on a multi-barrier system, including copper-coated steel containers surrounded by a bentonite buffer. The performance of bentonite is largely governed by montmorillonite (MMT), a negatively charged swelling clay stabilized by interlayer cations such as \(Na^+\) and \(Ca^{2+}\). Copper corrosion products may interact with the clay and drive ion exchange, potentially altering the hydration, swelling, and mechanical response of the bentonite barrier.

To investigate these effects, we employ a multiscale computational approach. Density functional theory (DFT) simulations are first used to characterize \(Cu^{2+}\) interactions with MMT at the electronic and atomic scales. Molecular dynamics (MD) simulations then quantify the energetic and structural effects of \(Cu^{2+}\) exchange with native interlayer cations, including changes in layer spacing, swelling behaviour, hydration structure, and ion mobility. To extend the modelling framework beyond the atomistic scale, we also develop mesoscale clay platelet models, in which physically motivated interaction terms are combined with Gaussian process regression (GPR) corrections to predict larger-scale swelling and mechanical behaviour under different environmental conditions. This framework bridges atomistic interactions and mesoscale bentonite properties, providing a predictive approach for modelling the swelling and mechanical behaviour of clay-based engineered barriers.

Also discussed is the ongoing work on pyKMC, a Python framework for adaptive off-lattice kinetic Monte Carlo simulations, as a route to access longer-time-scale structural evolution in materials beyond the timescales accessible to conventional molecular dynamics.