Premium
At‐Line Raman Spectroscopy Determination of Tablet Mass Gains during the Tablet Coating Process
Author(s) -
Kim Jaejin,
Lim YoungIl,
Han Janghee,
Woo YoungAh
Publication year - 2018
Publication title -
bulletin of the korean chemical society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.237
H-Index - 59
ISSN - 1229-5949
DOI - 10.1002/bkcs.11482
Subject(s) - coating , raman spectroscopy , process validation , normalization (sociology) , calibration , materials science , film coating , analytical chemistry (journal) , chromatography , mathematics , chemistry , optics , composite material , statistics , physics , sociology , anthropology , verification and validation
The coating process in the manufacturing of solid dosage forms is important for optimizing therapeutic efficacy because the coating is vital for controlling the drug release rate. The coating is most commonly applied by spraying a solution containing the coating material to form a film. In this study, core tablet samples were collected at regular intervals from four batches made using a scale‐up manufacturing coating process, and the Raman spectra were obtained to determine the mass increase of the coated tablets. To acquire the spectra, a wide area illumination scheme was used to sample a large area (28.3 mm 2 ) with a 785‐nm laser. Additionally, we established a calibration model using the Raman spectra of three tablet batches and verified the accuracy of the model by predicting the tablet weights of a fourth batch. All spectra were preprocessed with baseline correction and normalization. The spectral range was 1700 to 300 cm −1 , and partial least squares analysis was performed to establish a model with full cross‐validation. The accuracy was assessed by comparing the difference between the predicted weights using the model and that measured on an analytical balance, which was the reference test method. Standard error of cross‐validation (SECV) values and standard error of prediction (SEP) values for the fourth batch were 0.0023 and 0.0020 g, respectively, showing high accuracy. We established four optimal models that encompassed all combinations of different batches in a workplace environment and verified them by predicting each of the remaining batches.