Theoretical and Experimental Investigation of Chiral Separation by Crystallization

手性，一些分子以两个镜像(对映体)的形式存在的能力，具有相同的物理性质，是生命的基本成分。在进化过程中，生物体选择只使用其中一个镜像来构建它们的手性分子；例如，所有活着的有机体的氨基酸都具有相同的手性构型。这解释了为什么引入生物体的药物的药理或毒理特性会因所使用的对映体而有很大的不同。不幸的是，这并不总是被赏识。沙利度胺悲剧中的出生缺陷仍然是对这一事实的悲哀提醒。难怪监管部门现在制定了严格的指导方针，以防止这种情况再次发生，其他行业也变得越来越谨慎。例如，对映体纯杀虫剂比外消旋杀虫剂(对映体的混合物)更受欢迎，因为它们更有效，对环境的影响更小。我们对手性分子生物学活性的更好理解导致了对映体纯化合物市场前所未有的增长。然而，尽管在合成化学方面取得了进展，我们仍然远未达到自然界的完美，使用酶来产生手性产品，这些酶的作用被高度优化以区分对映体。通常，化学合成的最终产物是外消旋混合物，需要通过适当的分离方法，如结晶化，将其拆分成手性组分。不幸的是，外消旋熔体或溶液的结晶很少导致自发拆分(同手性晶体的机械混合物)。当不发生这种情况时，可以利用这两个对映体与对映体纯拆分剂相互作用不同的事实来实现分离，形成一对具有不同物理性质的盐或分子络合物(非对映异构体)。通过明智地选择拆分剂，这两个非对映异构体的溶解度将有很大的不同，通过结晶出较难溶解的一个，它们将有可能分离。尽管这种方法得到了广泛的应用，但拆分剂和工艺条件的选择仍然是基于反复试验。我们的研究旨在利用计算化学的进展和计算资源的增加来解决预测如何通过结晶分离对映体的挑战。我们现在有了非常精确的方法来在原子水平上模拟分子之间的相互作用。这将使我们能够从第一性原理预测非对映异构体的晶体结构、热力学稳定性和性质，并进一步提高这些算法的准确性和可靠性。这些进展将导致对自发拆分的理解和预测，当不发生这种情况时，将导致对给定的外消旋混合物和拆分剂的拆分效率的理解和预测。然后，我们将能够通过改变它们的分子结构来设计拆分剂，以实现尽可能最佳的分离，并随后优化工艺条件(例如温度)。不幸的是，开发和验证预测模型所需的全部实验数据，从晶体结构到分辨率效率，只适用于少数几个系统。我们将利用化学、化学工程和分子建模方面的多学科能力来生成所需的实验数据以及计算模型。这项研究既及时，又具有重大的工业意义，因为结果将有助于减少耗费资源的反复试验，并满足专业化学产品开发的激进时间表。该项目的成功将降低生产成本，增加药品、食品和农用化学品的可获得性。

Chirality, the ability of some molecules to exist as two mirror images (enantiomers) with identical physical properties, is a basic ingredient of life. In the course of evolution, organisms opted to use just one of the mirror images for chiral molecules they are built from; for example amino acids of all living organisms have the same chiral configuration. This explains why the pharmacological or toxicological properties of drugs introduced to an organism can be very different depending on the enantiomer used. Unfortunately, this has not always been appreciated. Birth defects in the thalidomide tragedy still serve as a sad reminder of this fact. It is no wonder that regulatory authorities have now set strict guidelines to prevent this from happening again and other industries are becoming increasingly cautious. For example, enantiomerically pure insecticides are preferred over racemic ones (mixture of enantiomers) because they are more effective and have a smaller environmental impact.Our better understanding of the biological activity of chiral molecules has resulted in an unprecedented growth of the market for enantiomerically pure compounds. However, despite progress in synthetic chemistry, we are still far from the perfection of nature which produces chiral products by using enzymes whose action is highly optimised to discriminate between enantiomers. Frequently, the end product of a chemical synthesis is a racemic mixture which needs to be resolved into its chiral components via a suitable separation method, such as crystallisation. Unfortunately, crystallisation from a racemic melt or solution rarely leads to spontaneous resolution (mechanical mixture of homochiral crystals). When this does not occur, separation can be achieved by exploiting the fact that the two enantiomers interact differently with enantiomerically pure resolving agents, forming a pair of salts or molecular complexes (diastereomers) with different physical properties. By judiciously choosing the resolving agent, the solubility of the two diastereomers will be substantially different and their separation will be possible by crystallising out the less soluble one. Despite the widespread use of this method, the choice of the resolving agent and process conditions is still based on trial-and-error experimentation. Our research aims to exploit the progress of computational chemistry and increased availability of computing resources to address the challenge of predicting how to separate enantiomers by crystallisation. We now have highly accurate methods to model the interactions of molecules at the atomic level. This will allow us to predict the crystal structure, thermodynamic stability and properties of the diastereomers from first principles, and further advance the accuracy and reliability of these algorithms. These developments will lead to the understanding and prediction of spontaneous resolution and, when this does not occur, the resolution efficiency for a given racemic mixture and resolving agent. We will then be able to design resolving agents by altering their molecular structure to achieve the best possible separation and subsequently optimise the process conditions (e.g. temperature). Unfortunately, the full range of experimental data needed to develop and validate the predictive models, from crystal structures through to resolution efficiencies, are only available for a few systems. We will use multidisciplinary capabilities in chemistry, chemical engineering and molecular modelling to generate the required experimental data alongside computational modelling. The research is both timely and of significant industrial importance, as outcomes will help reduce resource-consuming trial-and-error experimentation, and meet the aggressive timescales for the development of specialised chemical products. Success in this project will result in reduced production costs and greater availability of drugs, food products and agrochemicals.