Application of through-vial impedance spectroscopy (TVIS) to the development of the freezing, annealing and primary-drying stages of a pharmaceutical freeze-drying cycle

Background: Freeze drying process has been practised in number of biopharmaceutical manufacturing where the robustness and the efficiency of the process relying on the understanding of the process attributes such as nucleation temperature, solidification time and primary drying end point. For instance, nucleation time and solidification end point during the freezing stage of the freeze drying process influence the ice crystal size and morphology hence impact the drying rates. However, the nucleation temperature determined by the invasive probes which are being used currently in both laboratory and industrial scale impact the natural nucleation process and mostly provide early nucleation time and further they are incapable of determining the solidification time accurately. Moreover, there is a requirement of determining the primary drying end point ensuring the batch has completed the ice sublimation before advancing the freeze drying cycle into the secondary drying stage. Current primary drying batch end point determination techniques such as comparative pressure measurement and the individual vial primary drying end point using thermocouples provide nebulous estimates under the impacts of the soak period and moisture desorption. Further, their applications are limited to one stage of the drying cycle being unable to monitor the whole process. Through vial impedance spectroscopy (TVIS) as a relatively novel single vial non-invasive technique, has demonstrated its ability to determine the process attributes involved both in the freezing and drying stages and can be used as an in-process monitoring tool for the whole freeze drying cycle leading to optimize the overall freeze drying process with improved quality. Aim: (1) The main focus of the project is to develop a data analysis routine to determine the primary drying end points using the TVIS technology. (2) The secondary target is to broaden the understanding of the TVIS by applying the technique to more complex formulations, e.g. IgG. TVIS applications on determination of ice nucleation temperature, ice solidification time, glass transition temperature, devitrification and annealing end point will be investigated. The specific characteristics of the materials observed via TVIS technology such as unfrozen fraction and moisture desorption will be explained. (3) Thirdly, a method to measure the temperature inside the TVIS vial will be investigated since currently it is relying on the temperature measured by the thermocouples placed adjacent to the TVIS vial. (4) Finally, the quality of the impedance analysers which are being currently used to generate the TVIS data will be investigated. Methodologies: (A) Lyo cycles were designed representing amorphous (5%w/v sucrose), crystalline (5%w/v mannitol) materials and complex formulations (1%w/v and 15%w/v IgG) to characterise the freezing, annealing and primary drying stages applying the TVIS technology. Telstar LyoBeta laboratory scale freeze dryer along with the Pirani gauge and the capacitance manometer and Sciospec TVIS impedance spectrometer (NIBSC) have been used for acquiring freeze drying and TVIS data respectively. A thermocouple-containing vial has been placed adjacent to each TVIS vial and the capacitance spectra across the frequency range of 10 Hz to 1 MHz were recorded for every 2 min for each TVIS vial. The LyoView software-version 2019 was used to extract the TVIS parameters; C'(10 Hz), C'(100 kHz), 〖C″〗_PEAK and F_PEAK. In order to characterise the primary drying stage numerically in TVIS technology, two methods have been established using the high frequency real part capacitance profile; C'(100 kHz), where method 1 detects the first time point which starts to give consecutive ten negative or positive residuals and method 2 detects the first time point deviates from 3.3 root mean squared error (RMSE) in residual vs time plots. (B) With the hope of predicting the temperature of the sample inside the TVIS vial, the temperature sensitivity of the glass wall was determined for the data acquired by broad band dielectric impedance spectrometer (BDS) for a TVIS vial with modified electrode attachment where the electrode in one side of the standard 10 ml TVIS vial has been divided into three electrodes system. Temperature coefficients and the uncertainty of real and imaginary part capacitance data were analysed for selected low frequencies; 0.1 Hz, 1 Hz, 10 Hz, 100 Hz, 1 kHz and 10 kHz (C) Taking RMSE as a quality determination parameter, spectra for ultrapure water at room temperature generated by three main impedance analysers which are BDS, DMU-LyoDEA system and the Sciospec TVIS systems (three versions; which are referred as DMU, GEA and NIBSC) were compared. In the first part of the study a fitting model was applied for the full spectra from 10 Hz to 200 kHz generated by four analysers (BDS, DMU-LyoDEA, GEA and NIBSC Sciospec impedance spectrometers) and the second part of the study a different fitting model (third order polynomial function) was applied for the frequency range of 10 Hz to 400 Hz for the three versions of the Sciospec TVIS spectrometers along with the DMU-LyoDEA system. The comparison in the second part involves the different configuration parts such as cabling and junction boxes. Results and discussion: (1) A numerical methodology has been established to apply in C^' (100 kHz) profile to estimate the time period of ice mass linear reduction and sublimation end points. The predicted sublimation end points were validated against comparative pressure measurements. The proposed method allows to get an early estimation of the sublimation end point where one should not have to wait longer hours after passing the end point to confirm. According to the analysed data, the uncertainty of predicting the primary drying end point using the proposed method for the core vial containing crystalline sample is ±3 min (method 1) and for the amorphous sample it is ± 1.7 h (method 1) (2) The C'(10 Hz), 〖C″〗_PEAK, F_PEAK parameters are capable of determining nucleation temperatures and C'(100 kHz) parameter to determine the solidification time irrespective of the sample type. The C'(100 kHz) parameter has the ability to differentiate the in-process glass transition event from the devitrification. Further this parameter can also distinguish the ice sublimation from the moisture desorption at the end of the sublimation. Further, the three potentials of TVIS technology to predict the (a) annealing end points using 〖C″〗_PEAK and logF_PEAK parameters (b) system instability (c) the batch end point from the core vial have been recognised qualitatively. (3) the attempt of predicting the sample temperature inside the TVIS vial via glass wall temperature measurements was unsuccessful due to the fringing field lines propagate into the vial thus affected by the sample inside the vial giving more complex spectra (4) DMU-LyoDEA system was found as the instrument which generates the best quality spectra by showing RMSE value of 0.005 for the full spectra and 0.001 for the low frequency (10 Hz to 400 Hz) part of the TVIS real part capacitance window.