Modelling, Optimization And Experimental Study Of Protein Crystallisation Processes

Large molecule protein crystals have shown significant benefits in the delivery of biopharmaceuticals to achieve high stability, high concentration of active pharmaceutical ingredients (API), and controlled release of API. However, among the about 150 biopharmaceuticals on the market by 2004, only insulin has been marketed in crystalline form. The main technical reason led to this situation is due to the fact that protein crystallisation has a more complicated crystallisation environment and is affected by many factors. There is currently a lack of knowledge on commercial scale production of protein crystals. Therefore, in contrast to the majority of previous work on experimental and computational studies of protein crystallisation that has been centered on single crystal scale, the focus of the research reported in this thesis is placed on the computational and experimental study of protein crystallisation at process scale, investigating the growth behavior of the population of crystals grown from a crystallizer.The morphology and size distribution of protein crystals are key measures in quality control. However, research on population balance modelling of protein crystallisation processes is still very limited and restricted to simulation of crystal size distribution, ignoring crystal shape information because the size of a crystal is simply defined as the diameter of a sphere having the same volume as the crystal. A new morphological population balance model is proposed in this work for modelling protein crystallisation processes. In the model, the shape of a crystal is quantitatively defined as the distances of all crystal faces to the geometric centre of the crystal. For all the crystals in the crystalliser, a concept of multi-dimensional size distributions, which we call shape distribution for the crystal population, is proposed.The case study protein is Hen-Egg-White (HEW) lysozyme for which the thermodynamics and kinetics of crystallisation were examined. Faced crystal growth kinetic models were developed using non-linear regression. Process models were built for the crystallisation process of HEW lysozyme by combining the newly proposed morphological population balance model with mass and energy balance equations. The models were applied to simulate the evolving behaviour of shape distribution as well as size distribution of crystals of tetragonal HEW lysozyme.The morphological population balance model was applied to crystallization of HEW lysozyme for study of the effect of cooling rate, seed loading and seed size on size and shape distributions of product crystals as well as supersaturation. It was found that for growth only crystallization, low seed loading (0.001) and high cooling rate (2.0℃/day) lead to large crystals of low aspect ratio, but care has to be taken to avoid nucleation and major shape change such as width becoming larger than the length. Low cooling rate (0.7℃/day) results in slow growth. In addition, a two dimensional model for estimating seed loading was developed.In principle, the proposed morphological population balance model for protein crystallisation can be applied to investigate and control the growth of individual faces with the aim of obtaining desired crystal shape and size. Therefore process optimisation techniques were introduced to the model for optimising the product crystal shape distribution and size distribution. Optimal temperature and supersaturation profiles leading to the desired crystal shape and size distributions were derived. Genetic algorithm was investigated and found to be an effective optimisation technique for the current application. Since tracking an optimum temperature or supersaturation trajectory can be easily implemented by manipulating the coolant flowrate in the reactor jacket, the proposed methodology provides a feasible closed-loop mechanism for protein crystal shape tailoring and control.The modelling and optimisation study in this work is based on cooling protein crystallisation. Cooling rate control can be easily achieved and has good repeatability, and as a result has been widely used in industrial crystallisation of small molecules. However, there is a noticeable lack of systematic study on cooling crystallisation of proteins. In this study Linbro parallel crystallization experiments (24-well sitting-drop plate) was conducted to find the suitable conditions for carrying out cooling crystallisation investigation, including the initial concentration of HE W lysozyme solutions, precipitate concentration and pH value of solution. Real-time in-process imaging was used to record the morphological evolution of HEW lysozyme crystals during cooling crystallization using a hot-stage. Using a hot-stage, the effect of cooling rates on HEW lysozyme crystallization was investigated. The results provide experimental proof of the feasibility and effectiveness of cooling as a means for protein crystallisation.