Development of a Flexible Flow System for the Synthesis of Macro and Nanocrystals

晶体性质如尺寸、纯度、结构和习性对晶体的物理化学性质和随后在特定应用中的性能具有相当大的影响，利用快速合成定制晶体的能力在几个化学工业中是极其重要的。连续冷却结晶作为一种环保的替代传统的批量结晶能够促进快速原位过程分析技术（PAT）。连续结晶有局限性本项目将通过详细的工艺和计算优化来解决这些问题，旨在找到可持续性和晶体性能性能目标之间冲突的非支配解决方案。本项目将对可持续发展目标产生重大影响，9（工业、创新和基础设施），12（负责任的消费和生产），13（气候行动）通过降低成本来改善现有结晶器的可持续性，材料利用率和能源使用率，并提高废物最小化。这项研究有助于几个英国净零研究和创新的挑战-'开发数字解决方案和解锁资源和能源效率'通过使用多变量，多目标贝叶斯算法在流动结晶。这将完全消除人为错误，提高安全性和筛选通量，但也将看到结晶成核和生长的机械理解的巨大飞跃。香草醛（宏观尺度）和铯铅卤化物钙钛矿（纳米级）已被选为冷却结晶的广泛参考模型实例，其将通过三段流和动力学调节自动输入结晶器（KRAIC）的迭代来输送。工艺优化将开始，首先评价定义三段流的试剂，以确定可持续性、产品收率和分段规律性之间的折衷。将对公用设施和工艺流的物理布置进行评价，以调查对产品收率的影响，同时寻找回收和去除流的机会，以帮助最大限度地减少工艺废物。核心硬件项目将使用增材制造（AM）等快速技术进行系统优化，根据材料选择和工艺效率改善可持续性方面，同时提高产品产量。将使用探针沿着PAT监测过程变量，如紫外-可见光谱法，其将在宏观尺度上实施，以提供快速的成分信息，并与光致发光光谱法结合，提供纳米级颗粒浓度、形状和尺寸的信息。补充成分信息和详细的结构信息将使用单晶（SC-XRD）和粉末（PXRD）X射线衍射技术揭示。使用诺丁汉大学的非原位设备和利兹大学的Flow-Xl等国家设施的原位设备以及Diamond Light Source进行。计算优化首先从原位PAT中提取数据并将其转换为适合基于Python的机器学习算法操作的格式。多变量算法将负责接收用户定义的输入和现场过程数据，并制定一个知情的输出，该输出将操纵过程输入以达到最初的单一目标（形态学），然后在多个冲突目标之间找到一组非支配解（帕累托前沿）。这两个平台都是为了在我选择的模型之外应用于流动结晶而创建的，评估我的宏观和纳米级系统是否适用于类似规模的其他晶体的合成将证明该系统是灵活的。

Crystal properties such as size, purity, structure, and habit have a considerable impact on a crystals physiochemical properties and subsequent performance in specific applications harnessing the ability to rapidly synthesise bespoke crystals is extremely important in several chemical industries. Continuous cooling crystallisation emerge as an environmentally conscious alternative to traditional batch crystallisation capable of facilitating rapid in-situ process analytical techniques (PATs). Continuous crystallisation has limitations this project will look to address these areas through detailed process & computational optimisation, with the intention of finding the non-dominated solution between conflicting sustainability and crystal property performance objectives.This project will have a significant impact on the Sustainable Development Goals, 9 (Industry, Innovation and Infrastructure),12(Responsible Consumption and Production),13(Climate Action) by improving on the sustainability aspect of pre-existing crystallisers through reduced costs, material utilization and energy usage with enhanced waste minimisation. This research contributes to several UK Net Zero Research & Innovation Challenges-'Developing digital solutions and unlocking resource and energy efficiency' through the use of multi-variable, multi-objective Bayesian algorithms in flow crystallisation. This will completely negate human induced errors, increase safety and screening throughput but would also see a huge leap in mechanistic understanding of the nucleation and growth in crystallisation. Vanillin (macroscale) and Caesium Lead Halide Perovskites (nanoscale) have been selected as widely referenced model examples of cooling crystallisation, which will be delivered through tri-segmented flow and an iteration of the kinetically regulated automated input crystalliser (KRAIC). Process Optimisation will begin with evaluating the reagents that define tri-segmented flow to determine the compromise between sustainability, product yield and segmentation regularity. The physical arrangement of utility and process streams will be evaluated to investigate the impact on product yield while also looking for opportunities to recycle and remove streams to aid with process waste minimisation. Core hardware items will be optimised systematically using rapid techniques such as Additive Manufacturing (AM), improving on sustainability aspects based on material choice and process efficiency while increasing product yield. Probes will be implemented to monitor process variables In-situ along with PATs such as UV-Vis spectroscopy which will be implemented at macroscale to provide rapid compositional information, & coupled with Photoluminescence Spectroscopy provides information on particle concentration, shape, and size at nanoscale. Supplementary compositional information and detailed structural information will be uncovered using single crystal (SC-XRD) & powder (PXRD) x-ray diffraction techniques. Performed using ex-situ equipment at the University of Nottingham & in-situ equipment at national facilities such as Flow-Xl at the University of Leeds, and Diamond Light Source. Computational optimisation begins with extracting data from in-situ PATs & converting it into a format that is suitable for manipulation by python-based machine learning algorithms. The multi-variable algorithms will be responsible for receiving user defined inputs and in-situ process data, & formulating an informed output that will manipulate process inputs to reach what will initially be a single objective (morphology), before progressing to finding the group of non-dominated solutions (pareto front) between multiple conflicting objectives. Both platforms have been created with the intention of being applied to flow crystallisation outside of my chosen models assessing whether my macro & nanoscale systems are applicable to the synthesis of other crystals of similar scale will justify the system as flexible