On the inherent data fitting problems encountered in modeling retention behavior of analytes with dual retention mechanism

Comparison of isocratic retention models for hydrophilic interaction liquid chromatographic separation of native and fluorescently labeled oligosaccharides

“…(3)) led to the statistical insignificance of some of the regression parameters (mostly parameter B, which is the slope of the dependency; Table S1), probably due to the low number of isocratic experiments in the low-and high-water concentration region. As the least-square fitting of semilogarithmic scale of dependency may produce errors at high k-values [23], we have tested also the regression of the data in arithmetic scale, but with no improvement of the significance of Table 1 Column types, dimensions, manufacturers and hold-up volumes corrected for extra-column contributions, VM,corr, determined using acenaphthene as non-retained marker with the standard deviation of five consecutive measurements in parentheses. regression parameters.…”

Recent applications of chemometrics in one‐ and two‐dimensional chromatography

“…Figure 18 illustrates the simultaneous assessment of retention parameters for two different two-parameter models, the adsorption model for IEX and the LSS model for RPLC, based on two comprehensive two-dimensional chromatograms. Although the retention equations are well established for the conventional LC modes (RPLC, NPLC, and IEX), there is still considerable discussion about the most-suitable (often non-linear) models for more recent retention mechanisms, such as HILIC [227,228,[231][232][233][234][235][236][237], SFC [233,238], and hydrophobic interaction chromatography (HIC) [239][240][241].…”

“…Simulation of elution profiles in LC: Gradient conditions, and with mismatching injection and mobile phase solvents LC 2016 [288] Data fitting problems encountered in modelling retention behaviour of analytes with dual retention mechanisms LC 2015 [233] Prediction of retention time in high-resolution anti-doping screening data using ANNs LC 2013 [210] Linear gradient prediction algorithm for stationary phase optimized selectivity LC LC 2010 [204] Simulation of elution profiles in LC: Investigation of the injection solvent in the second dimension LC × LC 2017 [289] Sorption of organic compounds on black carbon LFER 2018 [246] Application of hydrogen bonding calculations, LFER LFER 2002 [244] Effect of temperature on retention using RP-LC LFER, van 't Hoff 2019 [247] Optimization of ANNs for modelling impurities retention in micellar LC MLC 2011 [209] Retention modelling in NP-LC and RP-LC -Adsorption model NP-LC, RP-LC 2000 [225] Applications of polyparameter LFER in environmental chemistry Review, LFER 2014 [248] Gradient retention time predictions for suspect screening RP-LC 2016 [217] Improved RP gradient retention modelling, Neue-Kuss model RP-LC 2010 [230] Nonlinear retention relationships in RP-LC, Neue-Kuss model RP-LC 2006 [229] Possibilities of retention modelling and computer-assisted method development in SFC SFC 2015 [238] Peak tracking Title…”

Prediction of gradient retention data for hydrophilic interaction liquid chromatographic separation of native and fluorescently labeled oligosaccharides

“…(3) and it has to be solved by a numerical integration. On the other hand, the integration of the Neue-Kuss model can be solved analytically and therefore it can be preferred among the other retention models [23], although it provides slightly less accurate predictions in HILIC in comparison to the mixed-mode model [20]. It is also worth mentioning that the Neue-Kuss model was developed for simplifying of the prediction process and its parameters are therefore not directly related to the physico-chemical parameters.…”

“…Shorter and steeper gradients may exhibit higher contribution of nonthermodynamic sources affecting the gradient profile [26], which was however not observed in this case.By comparing deviations calculated for the predictions based on the parameters obtained from isocratic (Figure 1B) and gradient (Figure 1C) retention data, it is evident that prediction based on the gradient retention parameters are more accurate (deviations from -4.2 % to -0.75 %). The large errors of the prediction based on isocratic data for very steep gradients can be experienced in HILIC due to the fitting errors of the model used[23]. The shows the arrangement of the results for the mobile phase containing 30 mmolL -1 ammonium acetate.…”