Distribution-centric 3-parameter thermodynamic models of partition gas chromatography

- Known 3-parameter Clarke-Glew thermodynamic model is unsuitable for complex samples.
- An alternative 3-parameter model suitable for complex samples proposed.
- Advantages and properties of the new model are discussed.

If both parameters (the entropy, ΔS, and the enthalpy, ΔH) of the classic van't Hoff model of dependence of distribution coefficients (K) of analytes on temperature (T) are treated as the temperature-independent constants then the accuracy of the model is known to be insufficient for the needed accuracy of retention time prediction. A more accurate 3-parameter Clarke-Glew model offers a way to treat ΔS and ΔH as functions, ΔS(T) and ΔH(T), of T. A known T‐centric construction of these functions is based on relating them to the reference values (ΔSref and ΔHref) corresponding to a predetermined reference temperature (Tref). Choosing a single Tref for all analytes in a complex sample or in a large database might lead to practically irrelevant values of ΔSref and ΔHref for those analytes that have too small or too large retention factors at Tref. Breaking all analytes in several subsets each with its own Tref leads to discontinuities in the analyte parameters. These problems are avoided in the K‐centric modeling where ΔS(T) and ΔS(T) and other analyte parameters are described in relation to their values corresponding to a predetermined reference distribution coefficient (KRef) − the same for all analytes. In this report, the mathematics of the K‐centric modeling are described and the properties of several types of K‐centric parameters are discussed. It has been shown that the earlier introduced characteristic parameters of the analyte-column interaction (the characteristic temperature, Tchar, and the characteristic thermal constant, θchar) are a special chromatographically convenient case of the K‐centric parameters. Transformations of T‐centric parameters into K‐centric ones and vice-versa as well as the transformations of one set of K‐centric parameters into another set and vice-versa are described.