科研之友
微信
新浪微博
Facebook
分享链接
摘要
An approach for the efficient development of near-infrared (NIR) methods for blend analysis using simulated real blending conditions is described. Spectra used to calculate a model were collected from small volume pharmaceutical blends in a die subject to well-controlled forces. Blend end-point detection by NIR spectroscopy is widely used. Strategic deployment of analytical technology is critical to companies seeking to utilize quality by design (QbD) and process analytical technology (PAT) concepts. However, pilot and commercial scale experiments using near-infrared spectroscopy (NIRS) are often limited due to resources. Specifically, limitations to the quantity of active pharmaceutical ingredients (APIs) available during development stage can dissuade early PAT method development. Such experiments are often necessary due to substantial differences between NIR data collected from small-scale (laboratory) blend systems and larger scale equipment. In the present work, data collected from small samples of pharmaceutical powders placed in an Instron testing station under carefully controlled forces matched conditions of the same powders in a blender. Use of this strategy for development of NIR-based blend end-point methods requires minimal quantities of pharmaceutical powders. Principal component analysis (PCA), two-dimensional (2D) correlation spectroscopy analysis, and Hotelling's T-2 ellipse results demonstrated that NIR spectra obtained in the small die matched NIR spectra in a blender when forces to the die were carefully applied. This strategy has the potential to significantly reduce the cost of development of NIR methods for blending.
- 出版日期2015-3
