Jonathan E. F. Fulkerson

Vincent J. Hoon

CS 497 Presentation

Wednesday, April 30, 2003

Consortium for the Advancement of Manufacturing of Pharmaceuticals

Sensors - Real-Time Chemometric Analysis

Advisors: Dr. Ken Morris, Dr. Lynne Taylor

Overview:  Our interdisciplinary research focused on aiding faculty and graduate students working in Purdue’s research group for the Consortium for the Advancement of Manufacturing of Pharmaceuticals (CAMP).  Jonathan has been working with the group since last summer on the Near-Infrared Sensors project, developing Principal Component Analysis (PCA) and Principal Component Regression (PCR) software to prove PCR's applicability to chemometric analysis of NIR spectra in the CAMP projects underway at Purdue involving granulation, coating, and particle characterization.  This semester, Jonathan and Vince verified the PCA software developed during the fall semester against other software packages and worked on software allowing the real-time display and analysis of NIR signals from the sensors in use by CAMP.  The project will continue in Maymester as we integrate the Principal Component Analysis and real-time software packages.

Problem Statement: Can chemometric processes, specifically PCR, be applied to in-line monitoring of full-spectral data obtained through NIR sensors to accurately and reliably distinguish between good and bad granulation batches, determine the end- and over-wet points of binder addition, and predict thickness, hardness, moisture content, particle size, etc.?

Hypothesis:  Chemometric analysis of real-time NIR spectra can reliably distinguish between good and bad granulation batches, determine the end- and over-wet points of binder addition, and predict thickness, hardness, moisture content, and particle size in tablet formation.

Null Hypothesis:  Chemometric analysis of real-time NIR spectra cannot reliably distinguish between good and bad granulation batches, determine the end- and over-wet points of binder addition, and predict thickness, hardness, moisture content, and particle size.

Experiment: We were involved with a variety of CAMP experiments utilizing chemometric analytic techniques including off-line and real-time slope calculation, off-line PCR, artifact reduction and elimination analytic techniques.  Our role involved assisting CAMP researchers in leveraging these analytic methods for use in a wide range of CAMP-sponsored projects and experiments.

Results: Chemometric analysis has proven quite successful through the CAMP projects in which it has been utilized thus far.  The Sensors project was renewed by CAMP last semester, and we anticipate that with the late-May completion of the real-time PCR software, it will continue to enable and enhance Purdue's contributions to CAMP's ongoing mission of improving pharmaceutical processes.

Project Continuation: This summer, Sensors will focus on unifying the real-time analysis software and the project’s chosen Principal Components Package (SIMCA) to enable in-line analysis for ongoing CAMP projects based out of the University of Puerto Rico.