Thesis subject proposed by CEA

Simulation of reactive transport on the nanoscopic scale: application to the dissolution of R7T7-type glasses

Practical information • Beginning of the thesis: October 2018• Place: CEA Marcoule• Contact/supervision: Frédéric Bouyer, Jean-Marc Delaye, Stéphane Gin (frederic.bouyer@cea.fr; jean-marc.delaye@cea.fr; stephane.gin@cea.fr)• Collaborators: Ian Bourg (Princeton University), Jincheng Du (North Texas University), Sebastien Kerisit (Pacific Northwest National Laboratory)

ContextIn diphasic solid/liquid systems, it is presently established that the microporosity of a solid medium considerably modifies the properties of the liquid phase (Wang 2014). In the case of water, the dielectric constant, density or even the freezing point are different than those measured in an open environment. In the presence of ions, the number of water molecules in hydration shell decreases with the pore size of the solid phase. This also affects speciation, solubility of mineral phases, etc. Qualitatively this is explained easily by the fact that the surface effects (proportion of water molecules interacting at the surface of the pores) increase when the size of pores decreases. This becomes very significant when the size of the pores is less than 5 nm. On the other hand, the effects of these interactions are still not well understood. Size is however the issue and the scientific communities that work in this domain are vary widely (energy storage, CO2 storage in geological formations, oil extraction, nanotechnology, nuclear biology, ...).

R7T7 glasses, upon their alteration by water, develop a microporous silicated layer on their surface that, when water is not renewed, quickly becomes transport limiting. This phenomenon is at the origin of the decrease of 4 to 5 orders of magnitude of the alteration rate of glass relative to the initial speed. The expected durability of glasses in geological storage thus depends in large part on the properties of this layer. Recently it has been shown that in silica saturated conditions the layers formed on the glass present pore sizes less than nm (Gin et al., 2015). On this scale, there is no longer any free water and the ions extracted from the glass by ionic exchange have very limited mobility; this happens even if porosity predominantly remains open (the apparent diffusion coefficients of water in this layer are of the order of 10 -22 m2/s, which is 10 to 13 orders of magnitude less than the diffusivity of ions in bulk water). Up to now the behavior of the glass is modeled based on laws using macroscopic parameters (Frugier et al., 2008). Furthermore, understanding of the processes on the nanoscopic scale would first allow reinforcement of the robustness of models

predicting long-term behavior of the glasses but would also offer possibilities of concept generalization to optimize the storage environment and guide the choice of new glass formulations.

A first thesis on the subject (Collin, 2018), integrated with the American EFRC project funded by DOE (https://efrc.osu.edu/), gives the first response elements based on an experimental methodology and the first simulations by classical molecular dynamics (Collin et al., 2017). This work shows that the dynamics of water in the passivating gel layer is strongly influenced by the presence of alkaline ions in the gel porosity. Besides, it was established that the passivating layer sufficiently reduces the flux of water molecules coming to the surface of the glass to reduce its alteration rate by several orders of magnitude. Recent developments in atomistic modeling open up new perspectives to explore more deeply these effects and this is exactly the subject of this thesis.

Development of the thesisThe structure of several glasses of simplified composition will be studied by classical molecular dynamics, typically with 6 oxides (SiO2, B2O3, Na2O, Al2O3, CaO, ZrO2), representing more complex nuclear glasses, as well as the corresponding passivating gels formed by the release of the most soluble elements (Na, B and partially Ca) and restructuring of the network. The topology of the porous network will be studied: geometry, density and pore connectivity. These simulations will be confronted with analyses by Atom probe tomography where the sample, formed by alteration of a glass in conditions of saturation in silica remains frozen after it is taken out of the solution until analysis to keep the water molecules in the pores in spite of putting the sample in a vacuum (analytical development in progress at PNNL in the context of the EFRC project). This first part will allow validation of the simulation method of the structure of a nanoporous passivating layer and from there exploration of the effect of the composition of the glass on the topology of these layers.

The dynamics and interactions between water and the pore walls can then be simulated by force field methods (REAXFF or CLAYFF type) taking advantage of the dissociative potentials of the water molecules. This work will be carried out in collaboration with the team of Jincheng Du at North Texas University, feasibility having been established at the end of Marie Collin's thesis. The reaction paths, possible phase transition, bonding energies, migration barriers etc. will be extracted from these simulations. Concerning the visible diffusion coefficients of water in the passivating gels, the values derived from isotopic tracing experiments are too low to be confronted with the simulations by molecular dynamics. One prospective part of the thesis will thus consist of trying to establish links with the macroscopic scale. Several approaches are possible. Methods of accelerated dynamics will be tested on out systems (Mousseau et al., 1998) to force the migration of water molecules. Definition of a kinetic Monte Carlo code may also be considered such as that developed in collaboration with Sébastien Kerisit at PNNL (CEA thesis of Amreen Jan). Hydrolysis/condensation parameters and structural disorder of the glass will be constrained by preceding studies in molecular dynamics. Finally, macroscopic models permitting extrapolation of results emerging from classical molecular dynamics on a laboratory scale may be used as well. Initial work will consist of implementing a synthesis of these methods to identify those best adapted for the case of glass - liquid systems. This second part will allow deconvolution of chemical interactions and transport of water molecules in the nanoporosity of the passivating layer.

Exogenous ions, brought to the solution upon alteration of the glass, will be introduced into the pore water to understand the ionic exchange reactions and the role of these ions on reactivity and transport of water in the nanoporosity. These simulations will benefit recent developments of interaction potentials between atoms. These simulations will be validated by experiments involving isotopic markers to follow the mobility of water in the nanoporous layer (method developed in the laboratory and consistent for tracing the isotopic ratio 18O/16O by ToF-SIMS in the passivating layer after immersion of the sample in water spiked with H2

18O). This third part will allow understanding of reactivity and dynamics of water in realistic systems.

The integration of this knowledge will lead to proposing a revision of kinetic laws describing the alterations of glasses.

Figures illustrating a simulation of the structure of glass with 6 oxides - (a): the orange, magenta, cyan and pink polyhedra correspond to those of silicon, aluminum, zirconium and boron - (b): zoom on the unit polyhedra [BO3] and [BO4]; the atoms in red, purple, green represent the atoms of oxygen, sodium and calcium

References

Frugier et al., Journal of Nuclear Materials, 380 (2008) 8-21.Collin et al. Nature Partner Journal-Materials Degradation, Submitted.Collin M., Géochimie en milieu nanoporeux (Geochemistry in a nanoporous medium). PhD thesis of the University of Montpellier, defense scheduled for June 2018.Gin et al., Nature Communications, N°7360 (2015).Mousseau et al., Physical Review E, 57 (1998) 2419-2424 Wang Y., Chemical Geology, 378-379 (2014) 1-23