Strategic Feedback Control of Pharmaceutical Crystallization Processes

很大一部分材料是以晶体形式生产的。这些晶体中有许多是通过在溶液中成核和生长而获得的。这种类型的晶体生产通常被称为工业结晶。结晶是大多数制药、食品和精细化工工艺中关键的分离纯化单元，对整个工艺的效率和盈利能力有重大影响。超过90%的药品含有以晶体形式生产的活性成分，单批活性药物成分的典型原材料成本为100万至200万美元。不符合产品规格会产生巨大的成本。为了高效的下游操作（如过滤和干燥）和产品有效性（如生物利用度，片剂稳定性），晶体纯度，尺寸分布和形状的控制可能是至关重要的。晶体的大小和形状影响溶解速率，这是药用晶体的一个重要特性。在制药工业中，药物益处与不良副作用的相对影响可能取决于溶出速度。通过对晶体大小和形状的控制，可以优化溶出速度，使效益最大化，同时使副作用最小化。对晶体大小和形状控制不佳也会导致过滤或干燥时间过长，或在额外的加工步骤中，如重结晶或碾磨，并可能影响产品的纯度，这在消耗晶体的食品和制药工业中尤为重要。改进的结晶过程控制为更好的产品质量和改进的过程效率提供了可能性，例如，通过缩短上市时间（并延长专利到期前的时间长度），减少受损批次，从而显著提高生活质量，例如，通过更快地以更低的成本获得新药。然而，由于其高度非线性和对工艺条件的高度敏感性，控制结晶是一项具有挑战性的工作。该研究的目的是开发一个系统和全面的框架来控制药物晶体的形成，该框架结合了第一性原理仿真模型，有效的动态优化和基于模型的控制算法，以及新颖的数学分析技术。该方法将允许通过实时反复求解约束非线性优化问题来控制晶体的形状和尺寸分布的整体形式，该问题将调整操作条件以达到期望的目标，并保证该过程在可行的条件下运行。操作条件中的不确定性将被纳入控制器设计中，以减少产品质量与其期望值的可变性。现场过程分析技术提供的测量结果将实时用于反馈控制策略，以估计和预测不同操作条件下的产品质量。该技术将有助于处理几个工业上重要的结晶关键问题，例如控制所需多晶体的形成和/或实现一致的产品质量，尽管由于放大而存在不确定性。该项目的最终结果将是一种新的结晶控制方法，它将提供一个全面的框架（包括模型，算法，软件和设备），用于特定应用（例如药物输送和剂量，或蛋白质组学）所需的多晶型，晶体形状以及晶体尺寸分布形式的强大设计，为未来的系统晶体工程开辟道路。

A significant proportion of materials are produced in crystalline form. Many of these crystals are obtained by nucleation and growth from solution. This type of crystal production is often referred to as industrial crystallization. Crystallization is a key separation and purification unit in most of the pharmaceutical, food and fine chemical processes, with a significant impact on the efficiency and profitability of the overall process. Over 90% of all pharmaceutical products contain active ingredients produced in crystalline form and typical raw material cost for a single batch of active pharmaceutical ingredient is $1 to $2 million. Failure to meet product specifications incurs significant costs. For efficient downstream operation (such as filtration and drying) and product effectiveness (e.g. bioavailability, tablet stability) the control of crystal purity, size distribution and shape can be critically important. The crystal size and shape affect the dissolution rate, which is an important property of crystals for medicinal use. In the pharmaceutical industry, the relative impact of drug benefit versus adverse side effects can depend on the dissolution rate. Control of crystal size and shape enables the optimization of the dissolution rate to maximize the benefit while minimizing the side effects. Poor control of crystal size and shape can also result in unacceptably long filtration or drying times, or in extra processing steps, such as recrystallization or milling, and can influence the purity of the product which is especially important in the food and pharmaceutical industries, in which the crystals are consumed. Improved control of crystallization processes offer possibilities for better product quality and improved process efficiency, for example by reducing time to market (and extending the length of time before patent expiration), and the reduction of compromised batches, therefore providing significant increase in quality of life, for example by making new drugs available more quickly and at lower cost. However, controlling crystallization is challenging due its high nonlinearity and its high sensitivity to process conditions. The aim of the research is to develop a systematic and comprehensive framework for controlling pharmaceutical crystal formation that incorporates first-principles simulation models, efficient dynamic optimization and model based control algorithms, as well as novel mathematical analysis techniques. The approach will allow to control the shape of the crystal and the overall form of the size distribution by repeatedly solving a constrained nonlinear optimization problem in real-time that will adjust the operating conditions to achieve the desired targets, and guarantees that the process operates within feasible conditions. Uncertainties in the operating conditions will be incorporated in the controller design to reduce variability of the product quality from its desired value. Measurements provided by in situ process analytical technology will be used in real-time by the feedback control strategy to estimate and predict the product quality for different operating conditions. This technique will be useful in treating several industrially important key problems in crystallization, such as controlling the formation of desired polymorphs and/or achieving consistent product quality despite of uncertainties due to scale-up. The end result of the project will be a novel methodology for crystallization control, which will provide a comprehensive framework (including model, algorithm, software and equipment) for the robust design of desired polymorph, crystal shape as well as the form of the crystal size distribution for specific applications (e.g. drug delivery and dosage, or proteomics), opening the way toward systematic crystal engineering in the future.