The separation of nine un-ionized samples was studied as a function of temperature (T) and gradient steepness (b). Selectivity values Δlog α∗ were obtained for 160 compounds, ranging from nonpolar hydrocarbons to very polar drugs. Selectivity varied markedly with sample type: nonpolar compounds such as aromatic hydrocarbons and fatty acid methyl esters generally showed only modest changes in band spacing as temperature or gradient steepness was varied. More polar samples exhibited larger changes in α (Δlog α∗) when temperature and/or gradient steepness wee changed, but the largest values of Δlog α∗ for these non-ionized samples are less than the average value of Δlog α∗ for the ionized samples of Part III [1]. Poly-functional silane (“polymeric”) columns exhibit slightly increased b- and/or T-selectivity for some samples.

The DrylabG software (LC Resources, Lafayette, CA, USA) was applied for optimization of HPLC resolution of some nucleosides in the reversed-phase systems. Two preliminary runs based on linear gradient range of acetonitrile (ACN) from 0% (pure buffer) to 20% and of methanol (MeOH) from 0% to 50% are shown to be satisfactory for optimization of the resolution. A good agreement between simulated and experimental chromatograms was observed.

A single gradient elution run with water (A) to acetonitrile (B) as mobile phase can be used to estimate preferred conditions for subsequent method development experiments based on RP-HPLC. For a broad range of sample types, that includes both very hydrophilic and hydrophobic compounds, it was found that isocratic retention is given by log k ≈ log kw −4.2ϕ, where ϕ = 0.01 %B. An initial gradient run allows values of log kw to be estimated for each compound in the sample, which then permits compound retention to be approximated as a function of either isocratic or gradient experimental conditions.

The use of an initial gradient run in this way provides a rational basis for the subsequent development of a final HPLC method. The present approach is based on a wide range of sample types and different reversed-phase columns; for this reason it is expected to be reasonably general and accurate. 

Initial method development experiments for both neutral and ionic samples are best carried out with reversed-phase high-performance liquid chromatography (HPLC) using acetonitrile-methanol-buffer mobile phases. The preceding paper (Part I) suggests the use of an acetonitrile-buffer gradient to start method development. In this paper, optimum conditions for this first separation are discussed. Sequential method development experiments plus computer simulations are then used to obtain a final HPLC method. In this connection, we have examined how many experiments are required for reliable predictions of ternary-solvent retention. Three experiments are sufficient to predict isoeluotropic retention for methanol-acetonitrile-buffer ternaries where solvent strength does not vary, but five experiments are required for ternaries that contain tetrahydrofuran.