Research driving innovation for industry: Predicting properties of matter
A case study by Professor of Chemical Engineering, Claire Adjiman.
"A given molecule can form a solid by adopting different patterns with completely different properties, including colour, shape, strength and speed of dissolution in a liquid. We have been using high performance computers to develop automated computer-based methods of predicting differing patterns, called polymorphs, of molecules.
"Knowledge of the polymorphs of molecules is critical in ensuring the quality of the many products that come in crystalline form. These include drugs, pesticides and foods such as chocolate and ice cream. We have developed a systematic method that has helped to identify the polymorphs of several molecules, especially in the context of the pharmaceutical industry. The predictions have been confirmed experimentally, giving confidence that this is a useful tool. This technique is invaluable for crystalline product development because the use of a systematic and predictive approach to predict the properties of matter can reduce the risks involved in developing and manufacturing new products, increasing the reliability of protocols and ultimately driving lower production costs, speedier production processes and enhanced quality.
"The application of automated computer-based methods is so vital to research because experimental studies to identify polymorphs and learn to control their production are very expensive and time-consuming and cannot guarantee that all polymorphs have been found. Computer predictions of the polymorphs of molecules can help to guide experimental studies and decrease the risk of polymorph-related problems occurring during manufacturing and use. For example, the anti-retroviral drug, Ritonavir, was withdrawn temporarily from the market in the late 1990s due to the appearance of a previously unknown polymorph. It is our goal to avoid future problems of this kind by designing computer-based methods that can predict all polymorphs.
"The method we have developed at Imperial College London is being used by research groups around the world to advance crystal structure prediction and our understanding of molecular solids. It is also being used in collaboration with industry partners in their search for new products. In developing this method, we have trained several highly-skilled researchers, who combine a deep understanding of the behaviour of matter with the ability to develop new mathematical techniques and advanced programming skills. These researchers go on to deploy their skills across a range of sectors. Finally, this exciting research offers a great platform to communicate to future generations of scientists and engineers and to show how work that combines a sound fundamental basis with high performance computing can make an impact on the world around us."