Antisolvent crystallization using fluidic devices without moving parts : experiments and modelling

Crystallization is an essential unit operation for separating and purifying Active Pharmaceutical Ingredients (APIs) in pharmaceutical industry. Conventionally, crystallizers with moving parts such as stirred vessels are always used for crystallization operation. However, their applications are limited by relatively low heat and mass transfer efficiency as well as difficulty of scale-up. In recent years, there has been a growing interest in using fluidic devices without moving parts for crystallization since they can offer better mixing performance, scale-up capability, and small equipment footprint. Meanwhile, continuous operation is preferred for its unique features such as better control of crystal attributes, higher productivity, and lower operating costs. However, one of the main challenges for the fluidic devices applied in continuous crystallization domains is the limited residence time. The residence time cannot be augmented due to adverse influence of mixing intensity. In the experimental work of present research, a recently optimized design of fluidic oscillator along with two other fluidic devices namely helical coil and coiled flow inverter was used for continuous antisolvent crystallization of paracetamol from methanol-water solution. A loop configuration was designed to realize a higher residence time up to ~13.5 min without jeopardizing mixing in the fluidic devices. An offline imaging method using microscope, digital camera, and ImageJ software was employed to measure particle size distribution (PSD) of manufactured crystals. The lognormal distribution was then used to characterize the experimentally observed PSD. The impact of key operating conditions such as operational mode, configuration of the fluidic devices and residence time on PSD of produced paracetamol crystals was evaluated. Population balance equation (PBE) based model as an effective tool is always used for better insight of crystallization process. In the modelling work of present research, a population balance model coupling with conservation equations and crystallization kinetics was established to computationally interpret mixed semi-batch and batch as well as continuous antisolvent crystallization experiments under consideration. Standard method of moment and lognormal distribution with moments-related parameters σ and µ was utilized to simulate PSD. A lumped parameter, namely mixing coefficient C_mix was proposed to characterize the influence of non-ideal mixing on effective supersaturation and consequently PSD of produced crystals. The effectiveness of C_mix was examined in mixed semi-batch and batch operation under a short residence time of the loop (~200 s) as well as continuous operation under relatively large residence times of the loop (~6.5 and ~13.5 min). An effort was made to develop a relation between mixing coefficient C_mix with other process parameters such as residence time, degree of supersaturation, configurations of devices as well as seeds attributes. This thesis performed experimental and computational investigations of antisolvent crystallization of paracetamol in methanol-water solution using fluidic devices without moving parts. Two key points were observed from this research. Firstly, it was found that designed loop setup with the fluidic oscillator exhibited good mixing performance in continuous antisolvent crystallization. In addition, it was observed that established PBE-based models coupling with mixing coefficient C_mix can simulate PSD of manufactured paracetamol crystals in mixed semi-batch and batch as well as continuous crystallization experiments under different operating conditions. The presented experimental setup, approach, results, and computational models will be useful for expanding applications of fluidic devices without moving parts in antisolvent crystallization domains.