Towards real-time process monitoring and control of physical properties of tablets using terahertz spectroscopy

• The developed interface enables automated THz-TDS* measurements. • The developed algorithm sorts THz-TDS* data and detects outliers automatically. • 1 st and 2 nd step enable monitoring of physical attributes of pharmaceutical tablets. • Considerations for controlling pharmaceutical tableting using THz-TDS* The modernisation of pharmaceutical manufacturing requires the implementation of process analysers for real-time assessment of both the process and product quality, as part of a process analytical technology (PAT) strategy. To meet the high standards for data integrity, it is essential that process analysers are integrated with a suitable sampling interface and workflow to ensure a reproducible sample presentation and robust data processing strategy. Terahertz time-domain spectroscopy (THz-TDS) with a reflection setup has recently been introduced as a potential process analyser for a fast assessment of tablet thickness, porosity, and mass, simultaneously. This study explores the potential of automated THz-TDS measurements, including the testing of a novel sampling interface prototype coupled with an algorithm for data treatment to select and group tablet measurements from background noise. Sampling interface and algorithm reliably presented, identified, and subsequently analysed THz-TDS signals of tablets. Comparable model errors were observed between the in-line interface and non-automated off-line manual measurements, with root mean square errors (RMSEs) of ≤ 0.012 mm, ≤ 1.23 %, ≤ 1.3 mg for tablet thickness, porosity, and mass, respectively. Additionally, the potential for off-line calibration, ongoing model maintenance of the analytical procedure, and integration within a PAT framework for monitoring and controlling tablet compression are discussed. The proposed automated tablet sampling strategy for in-line THz-TDS measurements proved to be feasible, enabling measurement speeds exceeding one tablet per second. While this strategy is specifically aimed at monitoring tablet compression, the developed algorithm is considered applicable in a broader range of automated THz-TDS measurement strategies.