A novel multivariate approach using science-based calibration for direct coating thickness determination in real-time NIR process monitoring |
| |
| Affiliation: | 1. Institute of Pharmacy and Molecular Biotechnology, Department of Pharmaceutical Technology and Biopharmaceutics, University of Heidelberg, Heidelberg, Germany;2. Quality Operations, PAT – Laboratory, Merck Serono, Darmstadt, Germany;1. Biogen Idec, 14 Cambridge Center, Cambridge, MA 02142, USA;2. Niracle LLC, New Town, Pennsylvania, USA;1. Merck & Co., Inc., Kenilworth, New Jersey 07033;2. Lonza Pharma & Biotech, Bend, Oregon 97703;1. Department of Chemical and Biochemical Engineering, Rutgers University, United States;2. Department of Mechanical and Aerospace Engineering, Rutgers University, United States;1. Centre for Pharmaceutical Engineering Science, University of Bradford, BD7 1DP, UK;2. Thermo Fisher Scientific, Stafford House, Boundary Way, Hemel Hempstead, HP2 7GE, UK;1. Laboratory of Pharmaceutical Technology, Department of Pharmacy, C.I.R.M., University of Liège, 4000 Liège, Belgium;2. Laboratory of Analytical Chemistry, Department of Pharmacy, C.I.R.M., University of Liège, 4000 Liège, Belgium |
| |
| Abstract: | This study demonstrates an approach, using science-based calibration (SBC), for direct coating thickness determination on heart-shaped tablets in real-time. Near-Infrared (NIR) spectra were collected during four full industrial pan coating operations. The tablets were coated with a thin hydroxypropyl methylcellulose (HPMC) film up to a film thickness of 28 μm. The application of SBC permits the calibration of the NIR spectral data without using costly determined reference values. This is due to the fact that SBC combines classical methods to estimate the coating signal and statistical methods for the noise estimation. The approach enabled the use of NIR for the measurement of the film thickness increase from around 8 to 28 μm of four independent batches in real-time. The developed model provided a spectroscopic limit of detection for the coating thickness of 0.64 ± 0.03 μm root-mean square (RMS). In the commonly used statistical methods for calibration, such as Partial Least Squares (PLS), sufficiently varying reference values are needed for calibration. For thin non-functional coatings this is a challenge because the quality of the model depends on the accuracy of the selected calibration standards. The obvious and simple approach of SBC eliminates many of the problems associated with the conventional statistical methods and offers an alternative for multivariate calibration. |
| |
| Keywords: | Science-based calibration (SBC) Matched filter Wiener filter Direct coating thickness NIRS PAT Coating process Thin film |
| 本文献已被 ScienceDirect 等数据库收录！ |
|
