The Thermodynamics of Chromatography - Thermodynamics Basics > Page 10

 

J. Chromatogr. Sci.7(1969)65

Figure 3. Graphs of Log(V'r) against 1/T for Some n-Alkanes Separated on Water-Vapor-Moderated Alumina

  

 

When the temperature of the column is increased, some of the water moderator is desorbed from the surface and the alumina became more active. As a consequence, as the temperature is initially raised, the retention volume of each solute is increased. However, when all the water has been removed from the alumina then the surface takes on a constant interactive character. Subsequent increases in temperature will cause the retention volume to begin to fall in the expected manner. The net result of these two effects are shown in the curves depicted in figure 3. The curves are not linear because the interacting surface changes as the temperature is raised and thus the distribution system also changes. As a consequence of the distribution system not remaining constant, the curves are not van't Hoff curves.

Under certain circumstances a similar situation can occur in LC but the distribution process can often be more confusing. Consider an ethanol/water mixture in equilibrium with a hydrocarbon reversed phase. Ethanol will be adsorbed onto the surface according to the Langmuir isotherm and the surface will be saturated with the alcohol at an ethanol concentration of about 25%v/v. Consequently, if the ethanol concentration significantly exceeds 25%v/v, the surface being covered by ethanol, will not change with increase in temperature. Consequently, a plot of log(V'r) against 1/T for any solute would probably be linear and the curve would represent a van't Hoff curve. However, at ethanol concentrations less than 25%v/v, the surface will be only partly covered with solvent. Consequently, as the distribution coefficient of ethanol will change with temperature so will the surface coverage. It follows that, as the properties of the surface will be temperature dependent, the distribution system will also be temperature dependent, and a graph relating log(V'r) with 1/T for any given solute will not be linear. Thermodynamic measurements made with solvent mixtures in LC must be carried out with caution and data will only be meaningful if the physical nature of the distribution system is not temperature dependent.