Data for PhD thesis: Taming Crystallization with Light

Crystallization is one of the most widely used purification and separation processes applied in a multitude of industries such as pharmaceuticals, food &amp; beverages, agriculture, and fine chemicals. However, the initial step of the crystallization process, nucleation, is still poorly understood and highly stochastic. As a result, most crystallization processes lack proper control over the properties of the crystals produced. Among many techniques for achieving better control over the nucleation process, the application of non-photochemical laser induced nucleation (NPLIN) has gathered significant interest. This is because of its potential to improve product quality in crystallization processes by directly controlling the nucleation rate, both spatially and temporally. Additionally, NPLIN can induce crystallization in solutions that would otherwise take a long time to nucleate, offering a unique advantage over traditional methods. However, despite its promising capabilities, NPLIN is not widely used in practice yet. The fundamental mechanism behind NPLIN is not fully understood, making it unclear how it should be applied effectively in practice and for which systems NPLIN could be beneficial.<br>This Ph.D. project aims to delve into the fundamental mechanisms of NPLIN, by examining how specific laser and solution parameters influence nucleation kinetics, leveraging innovative experimental setups. Laser parameters being studied include laser-exposed volume, laser irradiation position, laser intensity, and laser wavelength, and solution parameters include supersaturation levels, solution filtration, and the presence of impurities or dopants, particularly nanoparticles.The thesis begins with a comprehensive review of the experimental and computational literature on NPLIN. It then presents a detailed study on the effect of the laser-exposed volume and laser irradiation position on the nucleation probability within partly illuminated supersaturated aqueous potassium chloride solutions. An increase in the laser-exposed volume resulted in a higher nucleation probability and a higher number of crystals per nucleated sample. Furthermore, laser irradiation, particularly through the air/solution interface, not only enhances nucleation probability but also influences the formation of different crystal morphologies. These observations are partly explained by the Nanoparticle Heating mechanism and the Dielectric Polarization model (Chapter 2).<br>The research then transitions to a microfluidic platform, which allows for high-throughput and crystallization detection using the deep learning method. This innovative approach addresses the need for large data sets in NPLIN research, which has been a significant challenge due to the manual nature of traditional experiments. The study examines the effects of laser intensity, wavelength, supersaturation, solution filtration, and intentional doping on nucleation probability in supersaturated potassium chloride solutions. Higher laser intensities and increased supersaturation significantly enhance nucleation probabilities. The laser wavelength effect was only observed for 355 nm at higher laser intensities. Solution filtration suppresses the NPLIN effect, whereas the addition of nanoparticles as dopants into the solution not only increases the NPLIN probabilities but also affects the crystal morphology. The results highlight the importance of impurities in the solution and support the hypothesis that nanoparticle or impurity heating could be the key mechanism in understanding NPLIN (Chapter 3).<br>The study finally investigated the effects of solution filtration, laser intensity, and nanoparticle properties including nanoparticle concentration and material on NPLIN probability in supersaturated aqueous urea solutions. The study highlights the significant role of impurities in NPLIN, demonstrating that doping with different nanoparticle materials leads to varied nucleation probabilities. In particular, gold nanoparticles were found to enhance nucleation more effectively than silica nanoparticles. Additionally, it was observed that NPLIN probabilities followed a Poisson distribution to changes in nanoparticle concentration and laser intensity respectively. The findings in this chapter enhance our understanding of the critical role of impurities in comprehending the NPLIN mechanism (Chapter 4).

结晶是应用于制药、食品饮料、农业及精细化工等众多行业的最广泛使用的纯化与分离工艺之一。然而，结晶过程的初始步骤——成核，目前仍未被充分理解且具有高度随机性。因此，绝大多数结晶过程难以对所制备晶体的性能实现精准调控。在众多可实现成核过程精准调控的技术中，非光化学激光诱导成核（non-photochemical laser induced nucleation, NPLIN）已受到广泛关注。这是因为该技术可通过时空维度直接调控成核速率，有望提升结晶过程的产品质量。此外，NPLIN技术可在原本需要漫长时间才能发生成核的溶液中诱导结晶，相较传统方法具备独特优势。尽管NPLIN技术颇具应用前景，但目前尚未在工业界得到广泛应用。其背后的核心机制尚未完全阐明，导致无法明确其实际应用的有效方式，以及哪些体系可从NPLIN技术中获益。
本博士学位项目旨在借助创新实验装置，通过探究特定激光与溶液参数对成核动力学的影响，深入剖析NPLIN技术的核心机制。所研究的激光参数包括激光辐照体积、激光辐照位置、激光强度与激光波长；溶液参数则涵盖过饱和度水平、溶液过滤处理以及杂质或掺杂剂（尤其是纳米颗粒）的存在情况。本论文首先对NPLIN相关的实验与计算文献进行了全面综述。随后，针对部分辐照的过饱和氯化钾水溶液体系，详细研究了激光辐照体积与激光辐照位置对成核概率的影响。研究发现，激光辐照体积的增大会提升成核概率与每成核样品的晶体数量。此外，激光辐照（尤其是通过空气/溶液界面进行辐照）不仅可提升成核概率，还会影响不同晶体形貌的形成。上述观测结果可部分通过纳米颗粒加热机制与介电极化模型进行解释（第2章）。
随后，本研究转向微流控平台，该平台可借助深度学习方法实现高通量结晶检测。这一创新方案解决了NPLIN研究中对大规模数据集的需求难题——传统实验因人工操作属性，一直面临数据集规模不足的重大挑战。本研究探究了激光强度、波长、过饱和度、溶液过滤以及有意掺杂等因素对过饱和氯化钾溶液中成核概率的影响。结果表明，更高的激光强度与更高的过饱和度可显著提升成核概率。仅在较高激光强度下的355 nm波长处观测到了激光波长效应。溶液过滤会抑制NPLIN效应，而向溶液中添加纳米颗粒作为掺杂剂，不仅可提升NPLIN成核概率，还会影响晶体形貌。研究结果凸显了溶液中杂质的重要性，并支持了“纳米颗粒或杂质加热可能是阐明NPLIN核心机制的关键”这一假说（第3章）。
本研究最后针对过饱和尿素水溶液体系，探究了溶液过滤、激光强度以及纳米颗粒浓度、材料等纳米颗粒属性对NPLIN成核概率的影响。研究再次强调了杂质在NPLIN过程中的关键作用，证实使用不同纳米颗粒材料进行掺杂会得到各异的成核概率。尤其值得注意的是，相较于二氧化硅纳米颗粒，金纳米颗粒可更有效地提升成核概率。此外，研究观测到NPLIN成核概率分别随纳米颗粒浓度与激光强度的变化服从泊松分布。本章的研究结果进一步加深了我们对杂质在阐明NPLIN机制中关键作用的理解（第4章）。