When carrying out HPLC method development, it is often necessary to vary the relative retention of the sample (values of alpha) by changing some experimental variable, e.g., solvent type, pH, etc. The choice of which variable will be most suitable for a change in selectivity depends on two conflicting goals: (a) the attainment of maximum changes in alpha for the better control of resolution and (b) the avoidance of practical problems associated with the use of a given variable to optimize selectivity. This study provides a quantitative evaluation of different variables for their effect on selectivity (alpha). Various practical problems which must be balanced against this ability of a variable to change value of alpha are also discussed. The selection of any two variables for their simultaneous use in controlling alpha is also examined.

It has been shown previously that computer simulation based on two initial experiments can predict separation in reversed-phase gradient elution as a function of gradient conditions (gradient steepness, gradient range and gradient shape) and column conditions (column length, flow-rate and particle size). The present study extends this capability for changes in temperature. Four initial experiments (two different gradient times, two different temperatures) provide input data that allow predictions of separation as a function of temperature as well as gradient and column conditions. A semi-empirical relationship, tR=a+bT, is able to relate gradient retention time tR to column temperature T (other conditions constant). The accuracy of this approach has been evaluated for 102 solutes and a variety of experimental conditions, including the use of five different HPLC instruments (four different models).

A change in temperature (T) or gradient steepness (b) can result in changes in reversed-phase selectivity (α). The magnitude of these changes in α will vary with other separation conditions (column, pH, etc.) and with sample type. In this paper, selectivity changes as a function of T and b are discussed and a simple treatment that allows changes in selectivity to be compared quantitatively for different samples and HPLC conditions is developed. Following papers in this series will apply this theory to arrive at conclusions concerning the use of temperature and gradient steepness in HPLC method development. The present treatment assumes that gradient-steepness selectivity (measured by the parameter S) does not change significantly with temperature. Data for a wide range of compound types and conditions are provided in support of this assumption.

The ability of temperature and gradient steepness to change band spacing has been investigated for several ionizable samples that include 8 substituted benzoic acids, 9 substituted anilines, 22 basic drugs, 9 structurally-related herbicide impurities, 7 chlorophylls and 72 peptides and proteins. Mobile phase pH was also varied to determine the effect of sample ionization on temperature and gradient-steepness selectivity.