Quantifying molecular interactions linking disordered and ordered phases to predict crystallisation

尽管结晶是化工和制药行业中最常用的单元工艺之一，但如果不进行实验，仍然不可能预测结晶的结果。目前结晶筛选的做法在时间和人力上为每个化合物花费了工业50K到12万欧元。在计算晶体结构预测的帮助下简化这一方法目前受到计算能力和预测结构与产生它们所需的实验条件之间缺失的链接的限制。目前研究溶液状态聚集及其对结晶结果的影响的尝试显示出不同的成功，是孤立的，不允许进行研究以产生普遍的见解。这个项目的目的是阐明结晶前状态的分子相互作用并将这些与结晶结果联系起来。使用光谱和总散射技术的组合，我将从单组分相(熔体或玻璃)中提取均相相互作用的强度和结构，从溶液中提取溶质-溶剂相互作用，以及二元溶剂微观非均质性和溶剂-溶剂界面对溶液中聚集的影响。这种方法将产生基本的理解，从而能够预测本研究中没有研究的化合物的结晶行为。此外，我将把这些基础知识转化为相互作用强度的标尺和易于使用的规则，可与利平斯基的五项规则或生物药物分类系统相媲美，从而允许将该项目的基本见解应用于结晶科学家的日常工作。这将导致社区如何实现结晶的一步改变，并允许一种量身定制和受控的方法来取代目前的蛮力筛选。

Even though crystallisation is one of the most used unit processes in the chemical and pharmaceutical industries, it is still impossible to predict the crystallisation outcome without conducting an experiment. The current practice of crystallisation screening costs industry between 50K and 120K Euros per compound in time and manpower. Streamlining this approach with the help of computational crystal structure prediction is currently curbed by computing power and the missing link between the predicted structures and the experimental conditions needed to produce them. Current attempts to investigate solution-state aggregation and its influence on the crystallisation outcome show varying success, are isolated, and do not allow for investigation to yield generalised insights.The aim of this project is to elucidate the molecular interactions in pre-crystallisation states and to link these with crystallisation outcomes. Using a combination of spectroscopic and total scattering techniques, I will extract strength and structure of homomeric interactions from the single component phase (melt or glass), solute-solvent interactions from solution, and the impact of binary solvent microheterogeneity and solvent-solvent interface on aggregation insolution. This approach will yield the fundamental understanding to enable the prediction of the crystallisation behaviour of compounds not investigated in this study. In addition, I will translate this fundamental knowledge into scale bars of interaction strength and easy-to-use rules comparable to Lipinski's 'Rule of Five' or the Biopharmaceutical Classification System allowing the application of the fundamental insights of this project in the everyday work of crystallisation scientists. This will lead to a step-change in how the community approaches crystallisation and enable a tailored and controlled approach to replace the current brute-force screening.