Mixing processes in anti-solvent crystallisation

  • Russell Miller

Student thesis: Doctoral Thesis

Abstract

Mass transfer phenomena play an important role in many chemical and physical processes, such as crystallisation, and, therefore, it is essential to develop a better fundamental understanding of mass transfer effects in order to design more efficient crystallisation processes. It is aimed to investigate mixing within the context of anti-solvent crystallisation. Two mixing scales are studied, micro-mixing and macro-mixing. The first part of this thesis focuses on diffusive mixing in anti-solvent process and how it relates to turbulent mixing. This is subsequently followed by the characterisation of mixing times in one litre vessels. Through using a combined experimental and CFD approach insight into macroscopic mixing process will be gained for common anti-solvent solvent pairs. At the micro-scale local concentration profiles at interfaces between segregated fluidelements are controlled through diffusion. Consequently diffusion is a significant step of themixing process. This is particularly true for anti-solvent crystallisation in which nucleation outcomes are strongly influenced by localised concentration profiles. Previous work on modelling relied on a Fickian framework where concentration gradients are the driving force for diffusion. This predicts large overshoots in the supersaturation at interfaces between solution and anti-solvent, as is often intuitively expected. In this work, a thermodynamically consistent diffusion model was developed and applied to anti-solvent systems. In this model chemical potential gradients provide a more physically realistic driving force for diffusion. "Non-intuitiveā€ behaviour was predicted for diffusion in highly non-ideal liquid systems. In particular, as solute diffusion towards anti-solvent is severely hindered, it can diffuse against its concentration gradient away from anti-solvent. Furthermore largesupersaturation overshoots above that at the final mixture composition are not found whenthermodynamically consistent approach is used, demonstrating that these overshoots are modelling artefacts and are not expected to be present in physical systems. In addition, for certain conditions, localised liquid-liquid spinodal demixing is predicted to occur during the diffusive mixing process, even when the final mixture composition is outside the liquid-liquid phase separation region. Intermittent spinodal demixing driven by diffusive mixing may provide a novel explanation for differences of nucleation behaviours among various anti-solvents. Further investigation of this phenomenon found that higher anti-solvent content within the system increased the likelihood for LLPS. On the macro-scale, mixing occurs predominately through turbulent mechanisms in which velocity fields act to spatially rearrange fluid elements within the system. Turbulent dissipation leads to the reduction of these elements to the Batchelor length scale, in which diffusion becomes the prevalent mass transfer mechanism. Periodic boundary conditions were used to approximate the case of multiple solution-anti-solvent layers in parallel to give a better representation of diffusive mixing in turbulent systems. A qualitative discussion is offered on the relation of the developed model to other micro-mixing models reported in literature. Macroscopic mixing was investigated through characterising mixing times for a 1litre optimax reactor using a combined experimental and computational fluid dynamics (CFD) approach. Mixing times were then calculated from the resulting conductivity profiles in three ways; the 95% homogenisation method, and through fitting exponential and first order plus dead-time models. The geometry of the system was modelled and simulated on Mstar CFD to predict mixing times. Local and global mixing times were calculated by using a simulated probe, and the tracer concentration relative standard deviation throughout the vessel respectively. Firstly, The predicted variability of mixing from the Mstar simulations for the addition of a tracer to water, including the effects of tracer repeats and addition location is explored. Following this, the experimental results are discussed, with a comparison ofmixing time determination methods shown. Subsequently, the predicted and experimental results are compared. Effects of initial solvent composition on mixing times are then investigated. Lastly, we present the mixing times for the addition of ethanol to water, once more utilising both CFD and experimental conductivity method. Across all mixing time measurements, a wide variance was found to be present, highlighting the inherent variability associated with mixing processes.
Date of Award20 Sept 2023
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsUniversity of Strathclyde & EPSRC (Engineering and Physical Sciences Research Council)
SupervisorJan Sefcik (Supervisor) & Leo Lue (Supervisor)

Cite this

'