On-Line Blend Monitoring of Pharmaceutical Materials Using Multiple NIR Sensors
- Type: Conference Presentation
- Skill Level:
You will be able to download and print a certificate for PDH credits once the content has been viewed. If you have already viewed this content, please click here to login.
In recent years, powder blend monitoring and control systems using process analytical technologies (PAT) have become increasingly popular in the pharmaceutical industry. Among them, spectroscopic techniques allow for real-time evaluation of blend homogeneity and powder behaviors necessary for both formulation and process scientists to gain process understanding without invasive sampling to ultimately improve manufacturing quality. Near infrared spectroscopy (NIRS) has predominantly been used for blend homogeneity monitoring and control since it can predict in real time concentrations of drugs and excipients with reasonably low detection and quantification limits. At small scales, the use of a single sensor might be sufficient to sample a representative fraction of a blend container, but at larger scales, it might be necessary to employ multiple sensors. Indeed, the scale of scrutiny of one sensor might not be large enough to provide an accurate estimate of the homogeneity of a blend in large scale manufacturing facilities. However, as additional sensors are added, an increased number of parameters must be optimized and a more complex decision strategy must be implemented. It is conceivable that one sensor might indicate the blend is homogeneous, while a second sensor might not have reached that same homogeneity state due to blender loading discrepancies, blender shape, multivariate model differences, spectrometer differences, etc. The present study proposes to investigate the implementation of blend monitoring using two NIR sensors on a V-blender. In a first part, using a six-component formulation, the effect of prediction model error and spectrometer differences on blend end point determination will be studied. Discrepancies between sensor outputs will be evaluated and methods to limit their impact presented to ease the decision to stop blending. In a second part, the effect of loading patterns will be studied on a three-component formulation and their impact on the blend end point will be investigated. The objective of this study is to provide practitioners with a set of tools to ease the implementation of multiple sensors during the blending process. PAT calibration model quality and its impact on the blend end point will be assessed as well as the need for calibration transfer or standardization methods to limit the effect of discrepancies between PAT sensors.