Internet Electronic Journal of Molecular Design - IEJMD, ISSN 1538-6414, CODEN IEJMAT

ABSTRACT - Internet Electron. J. Mol. Des. November 2007, Volume 6, Number 11, 345-362

Quantitative Structure-Retention Relationships for the Gas Chromatographic Retention Indices of Oxygen Containing Organic Compounds Palagiri Tulasamma and Kalluru Subramanyam Reddy Internet Electron. J. Mol. Des. 2007, 6, 345-362

Free: Download the paper in PDF format

Return to Table of Contents

Get Acrobat Reader to view and print the paper

Abstract:
We propose a modified equation for the calculation of valence connectivity indices on the basis of summation over inverse geometric mean terms, ⁿχ^Vm, instead of summation over inverse square root terms, ⁿχ^V. Using the modified connectivity indices along with E-state and boiling points we develop quantitative structure-retention relationships (QSRR) for the gas chromatographic retention indices, I, partial molar enthalpies, ΔH, and partial molar entropies, ΔS, of oxygen containing organic compounds on seven stationary phases. The stationary phases are isosteric, one is apolar and branched alkane 19,24- dioctadecyldotetracontane, (C78), and the other six are polar resulting by replacement of one of the -CH₃ groups of C78 by -OH (POH), -CN (PCN), -SH (PSH), -CF₃ (MTF) and four of the -CH₂CH₃ groups of C78 by (-CH₂CF₃)₄ (TTF) and (-OCH₃)₄ (TMO). Four descriptors (¹χ_P^V, ²χ_P^Vm, E-state and boiling point) are selected from a pool of several descriptors based on stepwise multiple linear regression analysis to obtain the best QSRR models for the three properties. A unified QSRR model has been proposed for each property by combining the data of all the phases and with the use of an additional descriptor, namely the property dependent phase constant (PDPC) to represent the phase. QSRR models for the three retention properties of 18 oxygen containing organic compounds are developed with multiple linear regression using ORIGIN (version 5.0). The QSRR model obtained with ⁿχ_P^Vm descriptors give slightly better predictions in some combinations than with the descriptors ⁿχ_P^V. The predictive descriptors ¹χ_P^V, ²χ_P^Vm, the E-state of oxygen atom and the boiling point values are give better QSRR models in all the seven independent phases as well as for the unified QSRR models. The r_cv values are very close to r in the QSRR models of seven independent phases. The correlation coefficients obtained with unified QSRR are: r(I) = 0.995, r(ΔH) = 0.973, and r(ΔS) = 0.936; and the corresponding leave-one-out cross validation correlation coefficients are r_cv(I) = 0.993, r_cv(ΔH) = 0.965 and r_cv(ΔS) = 0.922. The predictive quality of the three unified QSRR equations proposed for the three properties in this study have been verified considering the retention data of two new stationary phases 1-chloro-18,23-dioctadecylhentetracontane, (PCL) and 18,23-dioctadecyl-7-hentetracontanol (SOH). The predicted results are in satisfactory agreement with the observed values in the two new phases, with maximum standard deviations σ(I) = 7.6, σ(ΔH) = 207.4 and σ(ΔS) = 0.445.

Free: Download the paper in PDF format

Return to Table of Contents

Get Acrobat Reader to view and print the paper