Atomistic simulation of green rust, its structure, composition and potential in nuclear waste storage and reprocessing.

The aim of this thesis is to use computer simulation techniques to model the bulk structure and low index Miller surfaces of green rust 1 and green rust 2, along with uranyl minerals which could form in the interlayers of green rust, and thus provide a potential mechanism to sequester uranium from polluted groundwater in the environment. The intention is that transferable potential models be developed that can be used in further studies or alternative works.

In Chapter 1 the historic use of nuclear material and its subsequent reprocessing and storage is introduced. The aqueous contamination of groundwater by uranium is discussed, changing the oxidation state of the uranyl ion is examined and possible methods of remediation are investigated. Laboratory synthesis of green rust is reviewed along with the techniques used in experimentally determining the structure of green rust. The natural occurrence of green rust is explained and potential uses of the material are described, with a discussion of the difficulties involved in working with green rust; this leads on to a review of the benefits of computer simulation at an atomistic level, and the drawbacks associated with such techniques. Finally the questions which drive the objectives of this work are stated.

Chapters 2 and 3 discuss the methodologies applied in this work and describe the theory behind them. Chapter 2 introduces computational modelling techniques and the theoretical methods which underpin them; the use of the potential model, which is key to this work, is described in detail. Energy minimisation and molecular dynamics are the theoretical methods used in this thesis and they are described in Chapter 3. Chapter 3 also introduces surfaces, which are critical in the aims of this work, and describes different methods of surface energy calculation.

Chapter 4 describes the work on testing and refining the interatomic potential models for Fe(OH)2 and goethite. While Fe(OH)2 has few experimentally reported structures, the ones which are documented were reproduced with excellent results using the potential model; the potential model outperformed density functional theory (DFT) in its ability to match the structural parameters of the cell. Five low index Miller surfaces of Fe(OH)2 were modelled and the (0 0 1) proved to be the most stable. The potentials were then used in similar modelling of goethite; again the potential model was able to better reproduce experimentally determined structures than was the DFT model, with cell dimensions within 1% of reported structures. This demonstrated the viability of the potential sets to be used interchangeably in modelling larger systems and that sophisticated fitting procedures were not necessary.

Chapter 5 documents the work on atomistic modelling of uranyl minerals. The need for a reliable set of interatomic potentials for a range of uranyl minerals is introduced. Existing potentials are fitted to known mineral structures from the International Crystal Structural Database (ICSD) and compared to the results of DFT calculations; the methods used to develop these potentials are described in detail. Results show that the set of potentials produced are capable of working successfully and that these potentials can be validated by reference to empirical data or DFT calculations.

In Chapter 6 the potentials developed in Chapters 4 and 5 are developed further and used to model one example of each of green rust 1 and green rust 2. It was shown that sulfate green rust 2 could successfully be modelled using interatomic potentials and that the sulfate group, with respect to the oxygen atoms, takes up a tridentate orientation toward the hydroxide layer. Modelling low index Miller surfaces of sulfate green rust 1 showed the (0 0 1) surface to be the most stable of those modelled. The bulk structure of chloride green rust 1 was modelled without water in the interlayers but when water was added there was some dissociation of the H atoms from the hydroxide layers; resolution of this dissociation and progression to surface modelling was prevented by the research period coming to an end.

Lastly, Chapter 7 summarises the results presented in this thesis and the conclusions that can be drawn, with a suggestion of further work that could be carried out built upon these results and the potentials developed therein.