Search

A new method for estimating the thermodynamic parameters of ΔH(T0), ΔS(T0), and ΔCP for use in thermodynamic modeling of GC×GC separations has been developed. The method is an alternative to the traditional isothermal separations required to fit a three-parameter thermodynamic model to retention data. Herein, a non-linear optimization technique is used to estimate the parameters from a series of temperature-programmed separations using the Nelder-Mead simplex algorithm. With this method, the time required to obtain estimates of thermodynamic parameters a series of analytes is significantly reduced. This new method allows for precise predictions of retention time with the average error being only 0.2s for 1D separations. Predictions for GC×GC separations were also in agreement with experimental measurements; having an average relative error of 0.37% for (1)tr and 2.1% for (2)tr., (Copyright © 2013 Elsevier B.V. All rights reserved.)

Thermodynamic parameters of ΔH(T0), ΔS(T0), and ΔCP for 156 compounds comprising alkanes, alkyl halides and alcohols were determined for a 5% phenyl 95% methyl stationary phase. The determination of thermodynamic parameters relies on a Nelder-Mead simplex optimization to rapidly obtain the parameters. Two methodologies of external and leave one out cross validations were applied to assess the robustness of the estimations of thermodynamic parameters. The largest absolute errors in predicted retention time across all temperature ramps and all compounds were 1.5 and 0.3s for external and internal sets, respectively. The possibility of an in silico extension of the thermodynamic library was tested using a quantitative structure-retention relationship (QSRR) methodology. The estimated thermodynamic parameters were utilized to develop QSRR models. Individual partial least squares (PLS) models were developed for each of the three classes of the molecules. R² values for the test sets of all models across all temperature ramps were larger than 0.99 and the average of relative errors in retention time predictions of the test sets for alkanes, alcohols, and alkyl halides were 1.8%, 2.4%, and 2.5%, respectively. [ABSTRACT FROM AUTHOR]

Highlights: [•] A new method for normalizing thermodynamic data between columns with varying dimensions. [•] Employs the use of simple standards and techniques available to any laboratory. [•] Provides accurate estimates for analyte retention between columns. [•] Average error in predicted retention across columns was 1.6s. [ABSTRACT FROM AUTHOR]

Thermodynamic modeling of retention times in gas chromatography depends on the accurate estimation of thermodynamic parameters. Previous research has used manual injections of samples with coinjection of a dead time marker to obtain accurate measurements of the retention factor of analytes. Ideally this process would be automated. Herein an approach is presented by which thermodynamic parameters can be estimated both autonomously and accurately. This method also allows for a consistent estimation of thermodynamic parameters regardless of factors such as data system delays and the nature of the void time marker employed. Ignoring these factors can lead to significant errors in the prediction of retention times when using thermodynamic models. [ABSTRACT FROM AUTHOR]

Abstract: We present an investigation into the influence of carrier gas on the thermodynamics governing a capillary gas chromatographic separation. Thermodynamic parameters are estimated for a series of alkanes and alcohols on three common stationary phases using helium, hydrogen, and nitrogen carrier gases. It is shown that the substitution of carrier gases for one another results in a change in the thermodynamic parameters governing the separation. The effect of the carrier gas on the thermodynamic parameters is large enough to compromise the accuracy of the retention time calculations based on thermodynamic parameters collected in a carrier gas other than the one actually in use in a specific gas chromatographic system. A possible kinetic explanation for these observations is also investigated. [Copyright &y& Elsevier]