Particle size distribution (PSD) is often considered as critical material attribute for active pharmaceutical ingredients (APIs), and the need for regular evaluation stands as an important quality control parameter in the pharmaceutical industry. Near-infrared (NIR) spectroscopy, used routinely for API identification, was introduced as analytical tool for simultaneous determination of particle size of ibuprofen. The demonstrated potential was highlighted by the development of rapid, robust, and noninvasive method coupled with multivariate data analysis (MVA), which can be easily transferred in QC laboratories for routine analysis. Principal component analysis (PCA) and partial least squares (PLS) regression analyses were performed on a calibration set of 61 ibuprofen samples, which differed in their median particle size Dv(50). The score scatterplots revealed evident clustering of ibuprofen samples according to their particle size, as well as occurrence of a distinctive outlying group of ibuprofen samples originating from one manufacturer. Further testing by means of mid-infrared spectroscopy, X-ray powder diffraction, and particle morphology analysis pinpointed particle morphology being responsible for the observed outlying group. Consequently, PLS class modeling based on particle morphology was introduced, which delivered two separate PLS regression models: one for blade-like ibuprofen crystals and another for irregular plate-like ibuprofen crystals. The former regression model exhibited high correlation coefficients and satisfactory predictive power (R2X = 0.999, R2Y = 0.917, Q2 = 0.901), whereas the latter demonstrated lower statistical indicators (R2X = 0.99, R2Y = 0.72, Q2 = 0.55). Additionally, the study underlines the importance of particle shape evaluation and sample classification according to particle morphology similarity prior to building NIRS-based regression models for PSD determination.
Read full abstract