Adsorption mechanism in RPLC. Effect of the nature of the organic modifier.

The adsorption isotherms of phenol and caffeine were acquired by frontal analysis on two different adsorbents, Kromasil-C18 and Discovery-C18, with two different mobile phases, aqueous solutions of methanol (MeOH/H2O = 40/60 and 30/70, v/v) and aqueous solutions of acetonitrile (MeCN/H2O = 30/70 and 20/80, v/v). The adsorption isotherms are always strictly convex upward in methanol/water solutions. The calculations of the adsorption energy distribution confirm that the adsorption data for phenol are best modeled with the bi-Langmuir and the tri-Langmuir isotherm models for Kromasil-C18 and Discovery-C18, respectively. Because its molecule is larger and excluded from the deepest sites buried in the bonded layer, the adsorption data of caffeine follow bi-Langmuir isotherm model behavior on both adsorbents. In contrast, with acetonitrile/water solutions, the adsorption data of both phenol and caffeine deviate far less from linear behavior. They were best modeled by the sum of a Langmuir and a BET isotherm models. The Langmuir term represents the adsorption of the analyte on the high-energy sites located within the C18 layers and the BET term its adsorption on the low-energy sites and its accumulation in an adsorbed multilayer system of acetonitrile on the bonded alkyl chains. The formation of a complex adsorbed phase containing up to four layers of acetonitrile (with a thickness of 3.4 A each) was confirmed by the excess adsorption isotherm data measured for acetonitrile on Discovery-C18. A simple interpretation of this change in the isotherm curvature at high concentrations when methanol is replaced with acetonitrile as the organic modifier is proposed, based on the structure of the interface between the C18 chains and the bulk mobile phase. This new model accounts for all the experimental observations.