Loading...
Modelling and experimental investigation of supercritical CO₂-assisted atomization and particle formation
Citations
Altmetric:
Date
2026-02
Abstract
In recent years, supercritical fluid (SCF) technology has gained an accelerating interest in the pharmaceutical manufacturing sector. The ability of this technology to generate drug particles with enhanced critical quality attributes (CQAs) has garnered research attention towards their employability to address the pressing matter of limited therapeutic efficacy of newly developed active pharmaceutical ingredients (APIs). Producing APIs as nanoparticles is a valid approach to improve their dissolution rates, potentially boosting their bioavailability, especially in BCS Class II drugs where dissolution is the rate-limiting step. Atomization-based SCF technologies offer enhanced potential to produce drug nanoparticles, a feature which when combined with their versatility and adaptability to continuous manufacturing schemes constitute an attractive means of addressing the problem of limited therapeutic efficacy of newly developed drugs. This versatility offers a wide range of relevant critical process parameters (CPPs), the tuning and optimization of which impose financial and environmental burdens on manufacturers. Due to market demand and regulatory aspects, these manufacturers remain in need for process scale-up while reducing waste and maintaining stringently defined product CQAs. Computational fluid dynamics (CFD) modelling of said processes offers opportunities to minimize development costs by reducing the size of required experimental matrices through simulations aiding design and optimization. The special conditions of supercriticality make it difficult to measure thermodynamic and physicochemical quantities throughout the process and add to the complexity of their mathematical modelling. In this project, the purpose was to develop a CFD modelling framework to simulate the supercritical CO2-assisted spray drying process with focus on the atomization region and provide efficient characterisation methodologies and simulation strategies tailored for the purpose of model calibration and validation.
In Chapter 1, an overview on the pathways through which CFD modelling has contributed to the advancement of atomization-based drug particle production processes is provided. Gaps in the literature are highlighted, and recommendations and future directions are provided based on the analysis of the observed trends in numerical modelling of the surveyed processes. In Chapter 2, an Eulerian CFD model to simulate the trans-critical expansion of pure CO2 through a micro-orifice nozzle is developed and experimentally validated. Supersonic jet velocities were reported for inlet conditions ranging 70-120 bar and 313-326 K. Joule-Thomson coefficients of 1-1.7 K/bar were reported, while simulations revealed a barrel shockwave structure that largely influenced the centreline velocity, turbulent kinetic energy, and temperature profiles, which is expected to impact droplet breakup mechanisms. In Chapter 3, the steady-state solution of the CO2 continuum field from the simulations carried out in Chapter 2 is adopted and the model is further expanded to include a Lagrangian Discrete Particle Method (DPM) injection to simulate methanol breakup taking into account gas-liquid coupling and associated droplet transport phenomena. The CFD-DPM model was calibrated and validated against novel experimental data collected via real-time laser diffraction method. For a range of liquid methanol flow rates (1-7 mL/min), reported SMD values were in the range of 2-5 𝜇𝑚 and 1.5-2.7 𝜇𝑚 for an injection pressure range of 70-140 bar. In Chapter 4, the predicted droplet size distributions (DSDs) from simulations carried out in Chapter 3 are adopted and pathways of droplet-to-particle transformation of a spray-dried model API (ketoprofen) are explored experimentally. Experimental particle size measurement was carried out using a custom-built apparatus via in-line real time laser-diffraction characterization and an off-line dynamic light scattering methods. The unravelled pathways of droplet-to-particle transformation informed a set of simplifying assumptions that allowed for analytical estimation of average Dv50 of dried particle sizes in the range of 570-850 nm for feed solution concentrations ranging 50-150 mg/mL, an injection pressure of 110 bar, and a feed solution flow rate of 1 mL/min. Calculated average Dv50 was within 19% error of experimental measurements. In Chapter 5, concluding remarks are presented followed by recommendations and guidelines for future improvements are outlined.
Supervisor
Description
Peer-reviewed
Publisher
University of Limerick