Appraisal of an Empirical Model for Simulation of Retention from Structure in Temperature-Programmed Gas Chromatography

Abstract

The solvation parameter model and response surface methodology are evaluated for the prediction of retention in temperature-programmed gas chromatography. A large and varied group of compounds were separated at a constant flow rate on three columns of different selectivity (DB-1701, DB-210 and EC-Wax) with initial temperature in the range 60–120 °C and program rates of 1–15 °C min⁻¹. The solvation parameter model provides an acceptable fit to the experimental retention factors independent of column identity, initial temperature and program rate. The system coefficients of the solvation parameter model are shown to fit a second order program rate model of the form system coefficient = a_o + a₁X + a₂X² where X is the program rate (°C min⁻¹) and a_o, a₁ and a₂ are fitting coefficients with no physical significance. Response surface methodology was used to derive empirical models to predict system coefficients with program rate and initial temperature as variables. These models explain the experimental data quite well but are local models that depend on the average properties of the solutes used for their derivation. Since they dependent on solute identity, these models are unsuitable for the general prediction of retention from structure, but may prove useful for estimating retention associated with variation in experimental variables for a defined group of compounds.