(657e) Characterization and CFD Modelling of Supercritical CO2-Assisted Spray Drying for Drug Particle Production

Improving therapeutic efficacy of newly developed drugs remains a major challenge for the pharmaceutical industry [1]. Given the highly time-and-cost demanding nature of experimental optimization, especially when dealing with expensive or sensitive pharmaceutical materials, process modelling comes at the core of transformation of these manufacturing techniques towards more sustainable and smart schemes that go along with Industry 4.0 design principles. The need for such progress is further driven by growing concerns over drug shortages and medicine efficacy, paralleled with increasing demands to reduce environmental impacts of pharmaceutical manufacturing. Atomization-based pharmaceutical manufacturing techniques are particularly popular due to their ability to generate high degrees of supersaturation leading to formation of fine drug particles with improved solubility and bioavailability [3]. Supercritical fluids (SCFs) have been employed to enhance atomization and favor formation of smaller drug particles owing to their distinct solvation power and enhanced mixing potential [2]. In particular, supercritical carbon dioxide (scCO₂) has been the fluid of choice in pharmaceutical research because of its abundance, low cost, and low environmental impact [3]. A multitude of parameters contribute to the quality of the final atomized pharmaceutical products resulting from SCF-based atomization including high-pressure-nozzle design, drying chamber geometry, and operating pressures and temperatures. In this context, our work is oriented towards advancing atomization-based drug particle manufacturing technologies through detailed analysis of the scCO₂-assisted spray drying process. In specific, and due to limited research efforts reported on the topic, we investigate the high pressure trans-critical operation of atomization nozzles and aim to characterize its key features while attempting to link those features to the final products of supercritical spray drying. In parallel to experimental characterization, we develop a finite volume CFD (Computational Fluid Dynamics) model using Ansys Fluent (Version 2023 R1) to describe the scCO₂-assisted atomization process, starting from pure CO₂ trans-critical expansion and proceeding to simulate the expansion process of a ternary mixture of an active pharmaceutical ingredient (API) dissolved in a solvent (API-Solvent-CO₂). We used the Peng-Robinson real gas equation to capture the thermodynamic gas properties throughout the process and employed a pressure-velocity coupled solver, while adopting the realizable k-ε model to describe turbulence. Model verification was achieved by simulating CO₂ choking flow phenomenon, while experimental validation was held by means of mass flow rate measurement and spray characterization through thermal imaging, followed by image processing/analysis. For supercritical pre-expansion conditions and a nozzle orifice size of 50-80 µm, we report a nominal spray angle range of 10-25⁰ and an outlet CO₂ mass flow rate range of 1.7 – 9.9 g/min. Experimental results matched predictions with a maximum error of 10%. In future work, we seek to model the scCO₂-assisted ternary mixture atomization process and validate results by means of phase doppler anemometry (PDA) and to explore the effects of relevant process parameters on the atomization phenomenon and ultimately on the characteristics of manufactured drug particles.

Materials and Methods

As shown in Figure 1, CO₂ is cooled down to -8°C using a PID-controlled recirculating chiller with a temperature stability of +/- 0.5°C (Julabo F250) to increase its density and facilitate its transport by the pumping unit. The CO₂ pumping unit is a linear piston pump (SFE Dose HPPC 400) with a flow range of 1 – 225 g/min and a pressure range of 1-400 bar. CO₂ is pressurized to 120 bar then heated to 50°C by means of a PID-controlled batch water bath (Grant JBA5). The API solution was prepared at ambient conditions (i.e., T = 19°C, P = 1 bar). The solution was pumped into the atomization nozzle via a high-pressure HPLC pump with a flow range of 0.001-5 ml/min and an operating pressure range of 0-600 bar (Agilent 1260 Infinity II). The spray dryer was a lab-scale custom-built stainless-steel vessel with dimensions of 60 cm in length and a maximum cross-sectional diameter of 15 cm. Particles were collected for characterization on a glass micro-fiber filter. The atomization nozzle assembly and the drying chamber walls were heated to the required temperatures according to the experiment at hand using a PID-controlled heating element from Siemens/Schneider Electric. The mass flow rate of CO₂ at the high-pressure injection side was measured by a Coriolis mass flow meter (CODA KMO Series) with a flow range of 0-3000 g/hr and a pressure rating of 272 bar. Volumetric flow rate measurement was held at the low-pressure side of the spray dryer using a turbine wheel flow meter with a flow range of 10-50 L/min, a maximum pressure drop of 0.1 PSI, and a pressure rating of 6.8 bar. A white surface with an emissivity of 0.68 was installed at a fixed distance from the expansion nozzle to estimate the spray impact area at different operating conditions. Thermal imaging was performed using an infrared camera with a spectral range of 7.5-13 µm, a resolution of 160 x 120 pixels, a thermal sensitivity of 0.06 °C, and an image frequency of 9 Hz (FLIR e6390). Thermal images were analyzed using ImageJ java-based software (ImageJ 1.53t, NIH) and Matlab (Version 2022a).

As shown in Figure 2, the simulation domain was 2D axisymmetric and discretized into a grid of 450,000 cells, which was based on a mesh sensitivity analysis performed before proceeding with simulations. A higher mesh resolution was implemented in the vicinity of the nozzle orifice to accommodate for severe envisaged flow disturbances, while maximum growth factor was maintained to 1.2 throughout the mesh. Mesh was generated using ICEM software while maintaining an orthogonality factor of at least 0.98 in >95% of the domain. Boundary conditions were set to a pressure inlet at the atomizer, and a pressure outlet at the domain’s outlet, where inlet pressure varied according to the simulation at hand, while outlet pressure was set to ambient pressure (1 bar). Equations of continuity, momentum, and energy were solved in steady-state using a pressure-velocity coupled algorithm applying second-order spatial discretization scheme. The overall domain dimensions were 100 mm (L) x 80 mm (W). Turbulence model was chosen as realizable k-ε model in line with previous studies on similar applications. Simulations were carried out on a desktop computer with performance specifications as follows: Core i7 10700 CPU/2.9 GHz processor.

Results

As shown in Figure 3, for an injection pressure range of 100-150 bar, the mass flow rate of CO₂ increased with decreasing pressure ratio (PR) until it reached a stable value near PR = 0.5, while the choked mass flow rate at 150 bar (16.5 g/min) was higher than that at 100 bar (10.9 g/min). Figure 4 shows the measured mass flow rate values at the spray dryer’s outlet when using a nozzle orifice diameter of 50 and 80 µm, and assuming a density of 1.87 g/L for CO₂ at 1 bar. Figure 5 shows the measured spray angles assuming different thresholds for the thermal images binarizing process. These values were used as guidelines to validate the CFD model.

Conclusion

A 2D steady axisymmetric CFD model was developed to simulate the trans-critical expansion of CO₂ for a pharmaceutical atomization-based manufacturing application. The model was verified and experimentally validated with an acceptable error range. The data presented herein is part of an ongoing work wherein scCO₂-assisted API solution expansion is being modelled and experimentally verified against PDA observations. A parametric study is to be carried out on the effect of atomization parameters on final drug particle characteristics, while further development of the model will be part of another analysis on exploring 3D and transient effects on supercritical atomization modelling.

References

[1] Sun, D., Gao, W., Hu, H., & Zhou, S. (2022). Why 90% of clinical drug development fails and how to improve it? Acta Pharmaceutica Sinica B.

[2] Padrela, L., Rodrigues, M. A., Duarte, A., Dias, A. M., Braga, M. E., & de Sousa, H. C. (2018). Supercritical carbon dioxide-based technologies for the production of drug nanoparticles/nanocrystals–a comprehensive review. Advanced drug delivery reviews, 131, 22-78.

[3] Chakravarty, P., Famili, A., Nagapudi, K., & Al-Sayah, M. A. (2019). Using supercritical fluid technology as a green alternative during the preparation of drug delivery systems. Pharmaceutics, 11(12), 629.