A new PhD project is inviting applicants to work on the development and
implementation of Monte Carlo techniques to simulate glass alteration
kinetics for better understanding of the elementary mechanisms at play.
This work is part of a collaboration between CEA Marcoule in France and
PNNL (Pacific Northwest National Laboratory) in the USA with the focus
on the development of a predictive model to calculate compositional
effects both of the glass itself and of the solution chemistry on glass
durability. Further details are in the attached file.
The candidate should have experience with computational modeling,
especially with atomistic modeling of materials. The project primarily
involves computational work, but with close connections to
experimentalists. A Master internship (6 months) is also possible on
this topic before starting the PhD project.
If you have suitable and interested candidates among your students, I
would greatly appreciate your letting them know about this opportunity
and letting them contact the thesis advisers directly at the addresses
given in the attached announcement.
Thank you very much in advance,
Dr. Andrey G. KALINICHEV, Directeur de Recherche
SUBATECH (UMR 6457 - IMT Atlantique, UniversitÃ© de Nantes, CNRS/IN2P3)
4 rue Alfred Kastler, La Chantrerie - CS 20722
44307 Nantes Cedex 3, FRANCE
TÃ©l: +33 (0)126.96.36.199.80
To join or leave the molecular-dynamics-news email list, go to:
Development of an advanced Monte Carlo method to simulate glass alteration
= = = = = = = = = = = = = = = = = = = = = = = = = =
• Beginning: October 2019
• Location: CEA Marcoule (with a stay at PNNL if possible)
• Contact/supervision (CEA Marcoule):
Jean-Marc Delaye, <[log in to unmask]>; Stéphane Gin <[log in to unmask]>
• Thesis directors: Andrey Kalinichev (Subatech/Nantes), <[log in to unmask]>;
Stéphane Gin (CEA Marcoule)
• Collaborator: Sebastien Kerisit (Pacific Northwest National Laboratory, USA)
Host laboratory: LCLT / CEA Marcoule
- - - - - - - - - - - - - - - - - - -
CEA Marcoule is a research center that includes around 2000 people. It is located near Avignon in South of France. The research conducted in Marcoule is mostly focused on nuclear fuel cycle issues (waste management and reprocessing), and dismantling. It represents broad scientific domains including physics, chemistry, and computing.
The LCLT (Long-Term behavior of glassy waste forms Lab) at CEA Marcoule includes around 25 people and is part of the DE2D (Waste Treatment Department). This laboratory is highly renowned in the field of glass durability and many studies have been performed in this domain in recent years coupling multi-scale experiments and simulations. Through its network of national and international collaborations, the LCLT have access to a large panel of tools to conduct scientific research at the best level.
- - - - - - - -
Glass corrosion is an important topic in materials science with implications in many fields such as Earth science, industry, health, and waste management. The microscopic processes responsible for glass alteration depend on the temperature, the pH, the solution composition, its renewal rate … These processes have been under investigation for many years.
In the nuclear industry, glasses are used as matrices to confine the radionuclides and are aimed to be stored in a deep, low-permeable geological formation. In that sense, glass alteration has to be characterized to assess the fate of radionuclides over geological time scales .
In the last few decades, many studies, based on experimental and computational tools, were undertaken to explain and model how glass behaves in various aqueous environments. The use of various analytical techniques such as solid-state NMR, ToF-SIMS, X-Ray and neutron diffraction, TEM, and Atom Probe Tomography, yielded atomic-scale insights into the basic mechanisms of alteration. In particular, in a PhD thesis defended in 2018 [2,3], new information were obtained about the nature of the atomic exchanges between the solution and the glass matrix.
The aim of this thesis is to incorporate these new findings into a Monte Carlo algorithm to simulate glass alteration kinetics, according to the elementary mechanisms at play. This model is a first step toward the development of a predictive model aimed at calculating compositional effects both of the glass itself and of the solution chemistry on glass durability. This work is part of a collaboration already existing between CEA Marcoule in France and PNNL (Pacific Northwest National Laboratory) in the USA.
A new tool will eventually be available to simulate glass or mineral alteration with an improved accuracy, that could be applied to various types of glasses and silicate minerals in various environments. Such a versatile tool is requested in many fields (geoscience, glass industry, waste management) and could contribute to address major environmental concerns such as CO2 sequestration, safety assessment of waste disposals, soils transformation and chemical erosion, impact of global warming on the chemical composition of the oceans…).
Description of the work
- - - - - - - - - - - -
Current state of the Monte Carlo approach
The first version of the Monte Carlo method was developed several years ago in the framework of a collaboration between Ecole Polytechnique and CEA Marcoule [4,5]. The aim of this technique is to simulate step by step the glass alteration by calculating various output data such as the release of tracer elements into the solution, the displacement of the glass–solution interface, and the structure of the gel layer. The input data are mainly the glass and solution compositions.
The global system (water + glass) is represented by an ordered network. One part of the system represents the glass. The glass elements, called the formers (Si, B, Al), are located at the nodes of the network. The O atoms, connecting the formers in a real glass structure, are not represented. The second part of the system represents the water molecules, also placed at the nodes of the ordered network. Hence an interface between the glass and the solution is formed. Figure 1 displays snapshots from the simulation of the alteration in deionized water of a simple borosilicate glass at different time points. The migration of the glass-water interface results in the formation of a porous gel layer whose structure also evolves with time.
The glass alteration is simulated using a set of parameters representing the probabilities for an atom to be dissolved. For instance, as the dissolution of a B atom is easier than the dissolution of a Si atom, the probability for the B dissolution will be larger than the probability for the Si dissolution. The local connectivity around the atoms is also considered because it is known that an atom dissolved more easily if it is less connected to the network. In the same way, the recondensation of the dissolved species at the glass surface is simulated using different probabilities. This method leads to the formation of a gel layer enriched in Si and Al, which can, in some circumstances, limit the mass transfer between the glass and the bulk solution. In this case, the glass dissolution slows down.
The Monte Carlo technique, although based on simple processes, successfully explained the non-linear effect of some key elements on glass durability [4,5,6]. However, recent studies on water dynamics in passivating gels have pointed out the need for improving the model to link the glass dissolution rate to the physical properties of the gel .
The objective of this Ph.D. work is to refine the basic elementary mechanisms implemented in the Monte Carlo algorithm to simulate more realistically the alteration of silicate glasses. Several changes will be considered.
First, in order to reproduce the residual alteration rate (this rate corresponds to the steady state achieved once the passivating gel layer forms and dissolves at the same rate), not accessible by the present Monte Carlo version, diffusion of both atoms dissolved at the glass surface and water molecules will be implemented. These elementary diffusion mechanisms will be first characterized by different molecular simulation methods (ab initio or classical molecular dynamics), and using isotopic tracing experiments prior their implementation in the Monte Carlo code.
Second, the simulation of the hydrolysis will be modified by refining the Monte Carlo network representing the glass to better mimic what really happens.
Third, to better represent the ripening of the alteration layer observed experimentally, aging mechanisms will be introduced to enable simulations of the evolving gel porosity and its influence on corrosion. This will essentially consist in improving the dynamics of the void population (migration, coalescence …) of the alteration layer.
All the Monte Carlo developments will be performed in collaboration with PNNL, as this laboratory has also significant expertise in this method [6,7,8]. A long-term residence at PNNL (typically 6 to 12 months) during the PhD could be considered for the candidate.
At each step of the work, comparison will be done with the experiments already available or that will be performed during the PhD. Typically, glass cations and water dynamics in the gels will be studied through isotopically-tagged glass specimen or solutions, followed by quantitative analyses (MC-ICP-MS for the liquid, ToF-SIMS and Nano-SIMS for the solid).
Profile of the candidate
- - - - - - - - - - - -
The candidate will have a pronounced passion for computer science and will already have experience in molecular modelling. This thesis primarily involves computational work with connections to experiments. A Master internship (6 months) is also proposed on this subject upstream from the PhD.
- - - - - -
 Gin S., Jollivet P., Fournier M., Angeli F., Frugier P., Charpentier T., Nature Communications, 6 (2015) 6360.
 Collin M., Gin S., Dazas B., Mahadevan T., Du J., Bourg I.C., The Journal of Physical Chemistry C, 122 (2018) 17764.
 Gin S., Collin M., Jollivet P., Fournier M., Minet Y., Dupuy L., Mahadevan T., Kerisit S., Du J., Nature Communications, 9 (2018) 2169.
 Cailleteau C., Devreux F., Spalla O., Angeli F., Gin S., The Journal of Physical Chemistry C, 115 (2011) 5846.
 Arab M., Cailleteau C., Angeli F., Devreux F., Girard L., Spalla O., Journal of Non-Crystalline Solids, 354 (2008) 155.
 Kerisit S., Ryan J.V., Pierce E.M., Journal of Non-Crystalline Solids, 378 (2013) 273-281.
 Kerisit S. and Pierce E.M., Geochimica et Cosmochimica Acta, 75 (2011) 5296-5309.
 Kerisit S., Pierce E.M., Ryan J.V., Journal of Non-Crystalline Solids, 408 (2015) 142-149.
To join or leave the molecular-dynamics-news email list, go to: