This paper describes two new retention models for predicting retention under different reversed-phase liquid chromatography (RPLC) conditions. The first one is a global linear solvation energy relationship (LSER) that expresses retention as a function of both solute LSER descriptors and mobile phase composition. The second is a so-called "typical-conditions model" that expresses retention under a given chromatographic condition as a linear function of retention under different so-called "typical" conditions. The global LSER was derived by combining the local LSER model and the linear solvent strength theory (LSST) of RPLC. Compared to local LSER and the LSST models, the global LSER model requires far fewer retention measurements for calibrating the model when different solutes and different mobile phase compositions are involved. Its fitting performance is equal to the local LSER model but worse than that of LSST. The poor fit of the global LSER results primarily from the local LSER model and not from the LSST model. The typical-conditions model (TCM) was developed based on a concept of multivariate space that is conceptually compatible with LSER. However, no LSER descriptors are used in the TCM approach. The number of input conditions needed in the typical-conditions model is determined by the chemical diversity of the solutes and the conditions involved. Principal component analysis (PCA) and iterative key set factor analysis (IKSFA) were used to find the number of typical conditions needed for a given data set. Compared to LSER, LSST, and global LSER, the typical-conditions model is more precise and requires fewer retention measurements for calibrating the model when different solutes and different stationary and/or mobile phases are involved.
- Linear solvation energy relationship
- Linear solvent strength theory
- Retention models
- Retention prediction
- Typical-conditions model