Atomistic simulations of co-crystal formation via mechanochemistry

Most chemical synthesis is performed in solution because in this phase it is easy to ensure that there are a large number of reactive collisions between reactant molecules. In addition, solution chemistry is well understood and we thus have a high degree of control over the reactions that can be performed and the products that can be synthesised. The problem with this approach is twofold. Firstly, the solvents many solvents are environmentally unfriendly and secondly separating the product from the solution at the end of the reaction often requires distillation, which requires a large input of energy and which introduces an extra step to the whole process. It would thus be enormously beneficial if this step could be avoided and if the solvent could be eliminated. Mechanochemical reactions allow for just this possibility. In these processes the reactants are powdered crystals. These powders are mixed together and mechanical work is done on the mixture in, for example, a mortar and pestle, a ball mill or an extruder. Experiments have demonstrated that it is possible to do a wide range of reactions in this way i.e., "mechanochemically". Furthermore, these mechanochemical processes are seen in some quarters to be the best way to synthesise systems known as co-crystals in which one or more chemical components are packed together into an ordered, crystalline structure. However, wider use of these processes and commercialization of these technologies is prevented because of the relative lack of understanding of the fundamental mechanisms that are in play in these reactions.

The aim of this project is to examine what happens in a mechanochemical reaction by performing molecular dynamics simulations using a computer. Such simulations are useful because it is possible to keep track of the positions of all the atoms at all times. This, however, is also the difficulty as specialized tools are required to make sense of large volume of high dimensional data that emerges from such simulations. One of our intentions is, therefore, to develop computational tools for studying these highly complex processes.

Throughout the work a reaction between two pharmaceutically active molecules, aspirin and meloxicam, will be studied. We will construct models for nanoparticles composed of each of these molecule types and will use non-equilibrium molecular dynamics simulations to force collisions between these particles to occur. Collisions will be performed for a range of collision velocities and for a number of different collision geometries. We will investigate head on collisions between the particles and glancing collisions as well as collisions in which we will change the relative orientations of the two crystal structures. For all these various kinds of collisions we will investigate the degree to which the two chemical components mix and the degree to which the crystallinity of the structure is disrupted by the collision.

This work will give us one of the first visualizations of the zone of reaction in a mechanochemical process. More importantly, however, it will provide us with a way of rationalising what is being observed in the reactive zone. This work will thus provide new fundamental insights into how and why these reactions proceed and will serve as a basis for future work on the comercial exploitation of these reactions.

The main deliverables for this project will be a set of new analysis tools that can be used to interpret the results from molecular dynamics simulations. Any new analysis techniques that are developed on the computational side within this project will be incorporated into the molecular dynamics code plugin PLUMED on which the PI is a lead developer. This code is released under an open source license and can be used either in tandem with any molecular dynamics engine or as a post processing tool. PLUMED has already been used successfully with gromacs, namd, amber, openMM, LAMMPS and the UK flagship MD code DL_POLY. This code is used widely within Europe with the original article on the paper already having accrued 91 citations in only two years. PLUMED also has a biannual user meeting and tutorial and all the new techniques developed as part of this work will be discussed within these forums.

In addition to these methodological advances we will also perform the first simulations of the zone of reaction in a mechanochemical reaction and thus examine the mechanism via which mechanochemical processes take place at the most fundamental level. This work addresses a real need within the experimental community using these processes for synthesis as the biggest hurdle preventing wider adoption and commercialization of these synthetic techniques is the lack of understanding of the mechanistic aspects of mechanochemical processes. If this barrier can be surmounted mechanochemistry offers enormous promise as a green and more-energy efficient way of doing chemistry as it effectively eliminates use of solvents. This research would thus be of interest to pharmaceutical and fine chemical companies such as DOW, BASF, Novartis, Pfizer, BP, Tosoh and Johnston Matthey More importantly, however, a spin off company (MOF Technologies) started by researchers in the school of chemistry and chemical engineering at Queen's University Belfast, exploits these techniques to synthesise metal organic frameworks (MOFs) a family of porous solids with applications in catalysis and storage of gasses. Throughout the course of the project we will work closely with these researchers.

One attractive aspect of these mechanochemical synthesis is that safe demonstrations of reactions can be performed in front of an audience. These demonstrations are routinely delivered to school children and sixth formers in Northern Ireland as part of Queen's outreach and engagement program. These demonstrations are popular because young people are able to interact with leading edge research. As part of this work we will engage with the organisers of these activities and provide movies from our simulations that can be used to accompany the experimental demonstrations. This will allow us to demonstrate to the young scientists just how simulations and computers can compliment experimental efforts. Furthermore, for older, A-level and undergraduate students we can create tutorials showing how much of the analysis that we do is based on calculations performed with vectors, which is material that should be familiar to A-level students.