Real Space and High-throughput Analysis of Crystallisation over Multiple Length Scales

Studying crystallisation, especially nucleation, is challenging due to the stochastic nature of the nucleation process. Advancement in microscopy in the past decades has made capturing nucleation and growth of a nucleus during crystallisation possible. However, there are various opportunities and challenges when it comes to analysing and interpreting experimental data from imaging the crystallisation
processes. This thesis aims to address the opportunities and challenges by introducing a novel way of analysing and interpreting the experimental data that can both help to reveal novel crystallisation mechanism and speed up the analysis process. In the first instance, this thesis uses the Topological Cluster Classification (TCC) and various other fluid analysis tools to sample fluid changes during crystallisation, especially those fluids surrounding the nucleus/crystals. The results reveal a novel two step crystallisation process in a specific colloidal system, the dipolar colloids. The fluids analysis results help to offer an explanation for the mechanism of crystallisation in this system. In the second instance, the nucleation kinetics of calcium sulfate under dif�ferent nucleating agents are being investigated. The experiment employs state of the arts microfluidic device to study thousands of aqueous droplets containing dissolved calcium sulfate in the presence of different nucleating agents. An image analysis algorithm that contains a specialised type of Convolutional Neural Networks, the U-Net, is developed specifically for the experimental set up and the calcium sulfate crystallisation. The final algorithm not only can reduce the amount of time it takes to get nucleation kinetics out of the microfluidic experi�ments, but also output statistics on the final calcium sulfate crystal morphology.The entire methodology allows for quick ranking and comparison for effective�ness of different nucleating agents at inducing calcium sulfate nucleation and canbenefit both academics and industries.