Theoretical Modeling and Near-Infrared Spectroscopy Monitoring of Vial Lyophilization
Doctoral thesis, 2008
The research documented in this thesis was directed towards lyophilization of pharmaceuticals in vials (vial lyophilization) and it was focused on the design and control of this process. Theoretical modeling has considerable potential to guide experimental studies, thereby decreasing development time and improving the possibilities to achieve an optimal and robust process design. There was no existing method available to directly study the lyophilization process within a conventional lyophilizer except for product temperature monitoring. Product temperature can however only give limited information about the drying process and there was therefore a need for a complementary technique that could generate additional information, especially about the desorption process. This was the reason for investigating the feasibility of using NIRS as such a monitoring tool.
The developed two-dimensional axisymmetric, unsteady state model, which was based only on first principles including physical constants, could well simulate both of the studied processes, i.e. heating and cooling of vials without sublimation and sublimation of ice in vials. Also the dynamics of the corner vial could be well modeled by adding heat transfer between the walls in the lyophilizer and the side of the vial. A heat transfer coefficient assessed from experimental data was introduced for that special case. The sublimation process was, apart from the shelf temperature and the chamber pressure, also influenced by the curvature of the bottom of the vial, the position on the shelf and the atmosphere in the freeze-dryer. The atmosphere in the freeze-dryer was dominated by nitrogen in the case of the cooling of vials without sublimation and by water vapor in the case of the sublimation of ice. The fact that the predictions were good for both cases showed that the used theory of gas conduction at low pressures models the effect of the difference in atmosphere well. The curvature of the vial bottom and the heat flux to the sidewalls of the corner vials contributed to a significant radial influence on the heat transfer.
The theoretical model of vial lyophilization generates detailed information about the drying process. The developed NIRS monitoring tool is a very good instrument for further investigation of such process details. It was possible to precisely detect the freezing point, the completeness of the ice formation process and the transition from frozen solution to ice-free material with the developed NIRS monitoring tool. The NIRS data yielded significantly more information about the actual process and essentially explained the observed temperature changes. The rate of the desorption process and the steady-state at which the drying was complete could also be precisely monitored with NIRS. NIRS has therefore been demonstrated to be a viable tool for in-line monitoring of the lyophilization process, both qualitatively and quantitatively. It is possible to investigate the drying process within a sample with NIRS monitoring and new information can be gained that can increase our understanding of the process. Sample presentation and sample selection will be very important for this technique because only a small volume of the sample will be monitored.
Spectral peak area analysis has been shown to be a viable method in near-infrared spectroscopy moisture assays. It was shown that spectral peak area analysis gave similar results as partial least squares regression (PLSR). The spectral peak area analysis method was, however, more robust and less sensitive to sample variations than PLSR. Spectral peak area analysis will facilitate using minimal calibration sets and will be especially suitable for moisture assays used in early formulation development and in-situ process monitoring. Spectral peak area analysis will also facilitate one or a few points moisture analysis calibration of the NIR instrument, using reference samples with known moisture content.
spectral peak area analysis