Applications for Through-Vial Impedance Spectroscopy (TVIS) in the Development of Pharmaceutical Freeze-Drying Processes

In recent years, through-vial impedance spectroscopy (TVIS) has been identified as a process analytical technology for the determination of critical process parameters within individual vials, at user-defined locations across a lab-scale freeze dryer. Much of the work to date, which is collated in Muhammad Sohail Arshad’s (2014) PhD thesis, employed a guard electrode assembly attached to a standard 10 mL type-I tubular glass vial and focused on the dielectric loss peak during (1) the freezing steps (demonstrating the impact of the filling volume and the influence of sucrose on the crystallisation of mannitol), (2) the annealing steps (establishing a method for the measurement of the glass transition temperature) and (3) the primary drying steps (determining the primary drying rate and the end point), with some observations on the manifestation of the collapse event. These studies were restricted to the investigation of the dielectric loss peak associated with the relatively low conductivity solutions sucrose, mannitol and maltodextrin whereas the more conductive solutions (i.e. those solutions spiked with electrolytes) were not investigated. Also, no attempt was made to measure product temperature and the determination of drying rate was impossible with this vial design. Therefore, the full potential of TVIS technology was not realised.

In this research programme, the TVIS measurement vial was re-designed in order to further develop and refine applications for TVIS technology across the freezing, annealing and drying stages of the lyophilisation cycle, with new methods being developed for the determination of (1) nucleation onset and ice-growth end point (this could be applied to the more conductive solutions containing high levels of salt), (2) the glass transition temperature, (3) the sublimation rate and (4) the ice temperatures at the sublimation interface and at the base of the vial, from which the heat transfer coefficient, in-vial product collapse event, and the dry layer resistance could be determined. In order to characterise phase transition (1) and (2), it was necessary to look at the characteristics of the real part capacitance (i.e. dielectric storage) in high and low frequency region (relative to the characteristic relaxation frequency of the ice) instead of simply relying on the characteristic of the dielectric loss peak in the imaginary part capacitance.

In chapter 6, the design of the TVIS measurement vial was revisited in order to establish whether it was possible to improve the quality and applicability of the TVIS spectrum by studying the factors that affect measurement response, namely (1) guard vs non-guard electrode designs, (2) electrode dimensions (with the height of the electrode increasing from 5 mm to 15 mm), (3) the position of the electrode above the base of the vial and (4) electrode attachment techniques (i.e. adhesive copper foil, sputter coating, tape wrap and glass surface treatment). A standard vial with 10 mL of nominal capacity and a non-guard-type electrode pair with dimensions of 19 x 10 mm, positioned at a distance of 3 mm from the base of the vial, is the suggested geometry for the measurement vial. This design can then be used as the standard vial for tracking the progression of the product during the various stages of the freeze-drying process – freezing, annealing and primary drying. No additional dielectric processes were observed from the adhesive of the copper foil electrode, and it is acceptable to use this adaptable method for investigating different vial designs in the future.

In chapter 7, a universal method for determining the ice nucleation temperature of conductive solutions was developed. When cooling 5% sucrose solutions with different salt concentrations (0%, 0.26% and 0.55% NaCl), it was clear that the Maxwell-Wagner (MW) relaxation process of the liquid state of these conductive solutions (i.e. the sample containing salt) remained outside of the TVIS measurement window (even at the sub-zero temperatures that the solutions reaches prior to nucleation). It was only after the ice started to form that a relaxation peak emerged within the experimental frequency window (owing to the presence of an ice phase within the vial). Therefore, an approach based on the MW relaxation peak could not be used to determine the liquid to solid phase transition (ice nucleation). Instead, at low frequency (i.e. 10 Hz), the real part capacitance was found to be more useful for indicating the point of nucleation, given the sensitivity of the MW relation peak to the temperature of the solution. Another interesting observation was that at a frequency above ice relaxation (i.e. 0.2 MHz), the real part, which has almost no temperature dependence, could identify the point where the solidification process was completed. Then, the ice growth period could be calculated from the difference in time between the onset and end point of the ice crystallisation process. Notably, based on this particular study, the ice formation period increases with increased salt concentration.

In chapter 8, different facets of TVIS response were investigated in order to provide an insight into the underpinning of physical mechanisms related to phase change and the structural modification of the product contained in a conventional 10 mL TVIS vial during the freezing and annealing steps. Aqueous solutions of 5% w/v sucrose were treated to three different cooling–heating cycle experiments with annealing/holding temperatures of −35 °C, −32 °C and −10 °C. Each experimental solution was annealed three times and two key observations were made. The first was that the drift in the amplitude of ice relaxation peak (C_PEAK^″) during the holding period was more pronounced at higher annealing temperatures, especially when the annealing was performed above its glass transition temperature, T_g^' (as determined by DSC). This observation was explained by the alteration of the ice crystal structure during the recrystallisation process, which increases the conductivity of mobile charges and results in greater charge accumulation at the glass wall (which is a factor that indirectly impacts the magnitude of the ice relaxation peak). The second observation was that by heating the frozen sucrose solution through its glass transition (to –10 °C), it was possible to demonstrate an inflection in the temperature profile of C^'(0.2 MHz), which coincided with the T_g^' of the sucrose solution. Unlike the structure alteration, this discontinuity of the real part might be a consequence of the temperature dependence of the unfrozen phase rather than the ice relaxation process, and it was therefore proposed that this transition provides an opportunity to develop a new method for measuring the in-situ glass transition of the freeze-concentrated phase.

In chapter 9, a method was developed for predicting the sublimation rate and ice interface temperature, based on the peak amplitude (C_PEAK^″) and the peak frequency (F_PEAK) of ice relaxation in a frozen aqueous solution of 5% w/w lactose. Dried layer resistance (����) could be determined from these critical process parameters. The results showed a relationship between the sublimation rate and the microstructure of the dried matrix, particularly when the ice temperature at the sublimation front reached the critical product temperature. It may therefore be concluded that this method could be used for the routine determination of in-vial collapse temperature and whether the collapse temperatures determined by the freeze-drying microscope are relevant to the process scale.

In the last chapter, the TVIS measurement vial was once again modified with the development of a dual-pair electrode system for predicting ice temperature at the sublimation front and at the base of the ice cylinder. From these two temperatures and the sublimation rate (as predicted using the method described in chapter 9), it was possible to estimate the single-vial heat transfer coefficient using a method first defined by Pikal and his co-workers (1984). The value of the heat transfer coefficient for this vial type was comparable to that found in published work (Tchessalov, 2017).

In conclusion, this work highlights the prospective applications of broadband (10 Hz to 1 MHz) impedance measurement, known as through-vial impedance spectroscopy, for investigating the in-vial characteristics of a solution within a freeze dryer and during the freezing, re-heating and sublimation phases. A range of physical phenomena was inferred from data extracted from both the real and imaginary parts of the TVIS spectrum and then used to deliver a comprehensive set of process parameters that will be valuable in the development and optimisation of future freeze-drying processes. Further work should consider a wider range of solutions containing small molecule and biological drugs in particular, and the extension of the applications to the secondary drying phase. Then it may be said that the work has come full circle by expanding on the earliest work on TVIS-type methodologies that was undertaken by Phe Suherman and collated in her PhD thesis of 1991.