Marcus Skogevall1, Per-Ola Quist2, Mats Josefson3, Alma Söderström1, Damir Ivanisevic4,
Kerstin Nolkrantz5, and Lars Karlsson1
1Inhalation Pharmaceutical Development, Pharmaceutical Technology & Development, Operations, AstraZeneca, Gothenburg, Sweden
2Science & Innovation, Global Medicines Development, Operations, AstraZeneca, Södertälje, Sweden
3Oral Product Development, Pharmaceutical Technology & Development, Operations, AstraZeneca, Gothenburg, Sweden
4Technology Formulation & NPI, Operations, AstraZeneca, Södertälje, Sweden
5Technical Operations, Science & Innovation, Pharmaceutical Technology & Development, Operations, AstraZeneca, Gothenburg, Sweden
Summary
A novel non-invasive Laser Doppler Vibrometry (LDV)-based, and Machine Learning (ML) driven Process Analytical Technology (PAT) method has been developed, enabling continuous monitoring of a commercial scale spray drying process. The approach has been progressed from feasibility studies of acoustic emission (sound) measurements in laboratory scale, via further evolvement on a smaller scale industrial type of spray dryer, while using acoustic emission and vibrometry in parallel, and finally by confirming the LDV applicability and industrialization potential on the large-scale commercial manufacturing unit. It was confirmed that the LDV-ML procedure has, as indeed was shown earlier for acoustics at laboratory scale, the capability to provide early detection of process deviations that significantly impact quality. It was demonstrated that there is a correlation of vibration spectra to both critical quality attributes, e.g., particle size, and, as well as to process parameters, such as relevant flows and temperatures. The next step in development and validation of the proposed method will be performed via vibration measurements during the manufacturing of several commercial scale batches.