Principal component analysis of nonlinear chromatography.

Area of Science:

• Analytical Chemistry
• Chromatography Science

Background:

• Nonlinear chromatography under overload conditions presents modeling challenges.
• Traditional methods often require extensive physicochemical data and first-principle models.

Purpose of the Study:

• To apply Principal Component Analysis (PCA) for modeling nonlinear chromatography.
• To evaluate PCA's effectiveness across different stationary-phase chemistries and column geometries.
• To explore correlations between PCA scores and sample mass for predictive modeling.

Main Methods:

• Loading crude erythromycin samples across a 10-fold range onto columns with polystyrene and methacrylate stationary phases.
• Utilizing Principal Component Analysis (PCA) to analyze elution profiles and data variance.
• Correlating principal component scores with sample mass and comparing scores between different columns.

Main Results:

• PCA models accounted for over 98% of the original variance in the chromatographic data.
• Excellent modeling of complete chromatograms was achieved without first-principle models.
• Consistent correlations between sample mass and principal component scores were observed across different column types.
• Linear relationships with high correlation coefficients were found when comparing principal component scores between columns.

Conclusions:

• Principal component analysis is a powerful tool for modeling nonlinear chromatography under overload conditions.
• PCA offers a robust alternative to complex first-principle models, simplifying data analysis.
• The observed correlations suggest significant potential for predictive modeling and inter-column comparisons in chromatography.