(247h) Molecular-Level Characterisation of Crystal/Solution Interfaces: The Case of Ibuprofen | AIChE

(247h) Molecular-Level Characterisation of Crystal/Solution Interfaces: The Case of Ibuprofen

Authors 

Marinova, V. - Presenter, University College London
Salvalaglio, M., University College London

In the pharmaceutical industry active pharmaceutical ingredients (APIs) are often manufactured as crystalline solids. The solid form of a drug provides several advantages in the isolation, handling and storage of the material, such as higher purity, improved flow, and higher chemical and physical stability. Probably the most critical decision in the process and product design of the API manufacture is the choice of the solvent for the crystallisation step. The solvent can affect the size and shape distribution of the resulting crystalline particles, which in turn can have a significant impact on its chemical and physical properties, as well as it bioperformace 1. Predicting crystal shapes is therefore desirable in order rationally identify favourable crystallization conditions and develop efficient API manufacturing processes.

The morphology of solution-grown crystals is affected by the solvent due to specific surface-solvent interactions. These interactions can alter the relative growth rate of morphologically dominant crystallographic faces, resulting in a modified crystal shape. Mechanistic models and molecular simulations have been used over the last decade in attempt to account for solvent effects and provide an accurate crystal shape prediction. Developing such models is challenging due to the diversity of factors that may affect growth rates, usually present within a single system. Factors such as solubility, formation of hydrogen bonds, strong non-bonded interactions, degree of conformational flexibility of both solute and solvent molecules can all play a role in determining the growth mechanism at the crystal surface or have an impact on the rate-determining step of the process. The level of complexity increases further since crystallographic faces within the same system expose different functional groups to the solution and a different rate-altering factor dominates. Solvent-specific crystal shape prediction has been achieved for several systems 2,3, however, the development of a universal and systematic crystal shape prediction model still presents outstanding challenges 4.

In our research we lay the groundwork of tackling these challenges and propose a systematic way of accounting for the effects of specific surface-solvent interactions on the morphology of solution-grown crystals. Our efforts are focused in two main directions. Firstly, we study the role of conformational flexibility of the solute molecule in the context of crystal growth 7. To this aim, we study the conformational rearrangement of ibuprofen with the aid of molecular simulations. We investigate the thermodynamics and dynamics of ibuprofen conformational transitions in the states involved in the process of incorporation of a molecule in the crystal bulk (in solution, at the surface/solution interface and in the crystal lattice). Secondly, we investigate the solvent behaviour as a function of distance from the morphologically dominant crystal faces of ibuprofen for several solvents. We obtain information on the thermodynamics and dynamics of solvent molecules at the crystal-solution interface. We account for the free energy difference between a molecule adsorbed on the crystal surface compared to bulk solution, as well as the exchange rate of an adsorbed solvent molecule with a molecule from the bulk solution or a subsequent solvent layer. In our study we consider the morphologically dominant {002}, {011}, {110}, {100} polar and {100} apolar ibuprofen crystal faces in 9 different solvents – water, 1-butanol, toluene, cyclohexanone, cyclohexane, acetonitrile, trichloromethane, methanol and ethyl acetate.

This work is carried out with the aid of molecular dynamics (MD) simulations. We investigate the conformational flexibility of ibuprofen by applying state of the art enhanced sampling molecular simulation techniques 5. State-to-state kinetics and conformational rearrangement mechanisms are then reconstructed within the theoretical framework of Markov State Models 6. Studying the structure of the solution in the presence of a crystal surface is achieved by extensive post-processing of MD trajectories with Plumed 2.38.

Our study reveals that the conformational free energy landscape of ibuprofen is strongly dependent on the environment and the relative position with respect to the crystal surface. While in solution we find a total of 6 conformers, all characterised by non-negligible populations, in the crystal bulk the distribution of conformers consists almost exclusively of one conformer. At the crystal/solution interface the distribution of conformers varies depending on the solvent and on the structure of the interface. The system relaxation time is found to also depend on both the solvent and the environment. Furthermore, when moving from solution to the crystal surface the mechanism of conformational rearrangement changes significantly. Last but not least, the structure of the solid/liquid interface is found to be strongly dependent on the solvent, affecting the conformational flexibility of molecules at the surface.

The investigation of the solution structure reveals that the arrangement of solvent molecules in the solution and the stability of the absorbed state varies greatly depending on their interaction with the morphologically dominant crystal faces. We find that while for some surface/solvent combinations an adsorbed state is practically non-existent, in other cases surface-solvent interactions promote a stable adsorbed state and can furthermore induce structuring effect in the solution. In particular long-range effects are observed in the case of surface-induced stacking of solvent molecules, which propagates beyond the solvent molecules in direct contact with the crystal surface. In such cases the rate of exchange of solvent molecules between adsorbed layers is remarkably slow, with characteristic exchange times up to three orders of magnitude slower than in cases in which long-range layering is not observed.

In the context of crystal growth our findings have important consequences. On the one hand, we find that incorporation of a solute molecule from bulk solution to an adsorbed state may be limited by conformational rearrangement due to the difference in conformational distribution between states. This process, however can also be significantly impaired by the structure of the solution. In fact, for cases where we observe solvent layering, the kinetics of a mass transport driven process such as rough growth can be significantly affected. Furthermore, under the assumption that a lifetime of an adsorbed state of a solvent molecule is correlated with the vacation of a kink site, the solvent kinetics at the surface/solution interface may carry a great significance in the growth rate of the particular face and provide a crude approximation of the solvent effect on the crystal shape. We find that the surface-solvent interactions also affect the degree of flexibility of a solute molecule in the crystal surface and the resulting freedom of conformational rearrangement can affect the rate of solute ordering at the crystal surface. Overall, we believe our work paves the way towards a formulation of solvent-dependent kinetic models for the growth and dissolution of flexible molecules.

References

  1. N. Variankaval, A. Cote, AIChe Journal, 2008, 54(7),1682
  2. M. Salvalaglio, T. Vetter, M. Mazzotti, M. Parinello, Angewandte Chemie Int Edition, 2013, 52(50), 13369
  3. J. Li, M. Doherty, Crystal Growth and Design,2017, 17(2), 659
  4. E. Elts, M. Greiner, H. Briesen, Crystals, 2017, 7, 288
  5. A. Barducci, G. Bussi, M. Parrinello, Phys, Rev. Letters, 2008, 100, 020603
  6. G. R. Bowman, V. S. Pande, An introduction to Markov State Models and Their Apllication to Long Timescale Molecular Simulations, Springer, USA
  7. V. Marinova, G.P. Wood, I. Marziano, M. Salvalaglio JCTC, 2018, 14(12), 6484
  8. G. Tribello, M. Bonomi, D. Branduardi, C. Camilloni, G. Bussi, Computer Physics Communications, 2013 185, 604