Comparison of Chromatographic Stationary Phases Using a Bayesian-Based Multilevel Model

Area of Science:

• Analytical Chemistry
• Chromatography

Background:

• Comparing reversed-phase high-performance liquid chromatography (RP-HPLC) stationary phases is crucial for method development.
• Existing methods for phase comparison can be time-consuming and require extensive experimentation.

Purpose of the Study:

• To develop and validate a Bayesian multilevel model for comparing five RP-HPLC stationary phases.
• To characterize the differences between stationary phases based on analyte retention behavior.
• To provide a data-driven approach for selecting optimal chromatographic conditions.

Main Methods:

• Utilized a Bayesian multilevel model based on chromatographic retention principles.
• Analyzed a large dataset of retention times from gradient RP-HPLC experiments.
• Included 300 small analytes across various pH, solvent (methanol, acetonitrile), temperature, and gradient conditions.
• Employed mass spectrometry for analyte detection.

Main Results:

• Successfully characterized between-column differences in chromatographic parameters for neutral, acidic, and basic analytes.
• The model provided an interpretable summary of stationary-phase properties.
• Demonstrated the model's utility in decision-making for selecting chromatographic conditions.

Conclusions:

• The proposed Bayesian modeling approach offers an effective alternative for comparing RP-HPLC stationary phases.
• This method aids in optimizing chromatographic conditions using limited experimental data.
• Provides valuable insights into stationary phase selectivity and performance.