Oral Presentation The 35th Biennial Conference of the Society of Crystallographers in Australia and New Zealand 2024 (Crystal 35)

Predicting crystal growth and morphology from accuracy to speed (108999)

Julian D Gale 1 , Blake I Armstrong 1 , Alvin J Walisinghe 1 , Peter R Spackman 1 , Paolo Raiteri 1 , Michael W Anderson 2
  1. Curtin University, Perth, WA, Australia
  2. Chemistry, University of Manchester, Manchester, UK

While for many materials the bulk structure is well-defined and characterised, there can remain many challenges in understanding the surface crystallography, especially when the system is in contact with a liquid. This includes knowledge of the morphology at the macroscopic level down to individual surface sites at the atomic or molecular level, as well as the interfacial ordering of solvent. Not only are the structural properties important at the surface, but also the thermodynamics of distinct interfacial sites and the kinetics of solvent exchange, both of which influence the rate of crystal growth or dissolution, depending on the conditions. Here computer simulation can be a valuable complement to experiment, being well-suited to providing atomic detail. 

In this presentation we will examine how molecular dynamics simulation, based on carefully selected models, can provide valuable data for interfacial processes including comparison of solvent structure with experiment and accurate thermodynamics for individual surface site including kinks using calcite (CaCO3) as a model system. Having established accurate methods for characterising the free energy of surfaces as a function of structure, the remainder of the presentation will examine alternative approaches for fast prediction and screening of thermodynamics and solubility, using molecular crystalline systems as examples. It will be demonstrated that this approach provides an important and rapid starting point for predicting the variation of crystal morphology, including as a function of supersaturation, using Monte Carlo-based approaches, such as the CrystalGrower model.