Discrimination between linear and non-linear models for retention indices of polycyclic aromatic hydrocarbons in the so-called Lee’s scale Biljana S ˇ krbic ´ * , Natas ˇa Djuris ˇic ´-Mladenovic ´, Jelena Cvejanov Faculty of Technology, University of Novi Sad, Bulevar cara Lazara 1, 21000 Novi Sad, Serbia and Montenegro Received 28 October 2003; received in revised form 12 December 2003; accepted 14 January 2004 Available online 15 April 2004 Abstract Detailed statistical analysis is presented to describe the retention indices of polycyclic aromatic hydrocarbons (PAHs) expressed in so- called Lee’s scale as a function of boiling point, molecular weight and connectivity index. As a training set, experimental retention data of parent and alkylated PAHs obtained on SE-52 phase is used for correlation. On the basis of several statistical characteristics (multiple correlation coefficient, residual mean square, residuals, and the sequential probability ratio) quadratic with two variables and combined equations were chosen for describing the retention data on slightly polar stationary phases. As prediction sets, literature experimental retention indices expressed in so-called Lee’s scale obtained on DB-5 slightly polar stationary phase are applied for comparison. D 2004 Elsevier B.V. All rights reserved. Keywords: Polycyclic aromatic hydrocarbons; Retention index; Lee’s scale; Quantitative structure– retention relationships 1. Introduction Polycyclic aromatic hydrocarbons (PAHs) represent a very important group of chemical carcinogens, especially those with higher number of aromatic rings in their structure, such as benzo(a)pyrene, dibenzo(ah)anthracene, benzo(b)fluoran- thene, etc. Identification and determination of polycyclic aromatic hydrocarbons are very important for the investiga- tion of the environmental pollution level [1], as well as for the evaluation and control of the overall public health hazard, therefore as training sets, these compounds were chosen. High-resolution gas chromatography is a commonly ap- plied method for the analysis of polycyclic aromatic hydro- carbons in various matrices. The most widely used stationary phase for such analysis is methylsilicone with 5% phenyl groups. Among the presentations of gas-chromatographic data for polycyclic aromatic compounds, the temperature- programmed retention index [2] in so-called Lee’s scale [3] based on PAH internal standards (naphthalene, phenanthrene, chrysene, picene) is preferred, because of its much better reproducibility compared to other forms of retention data. The selection of the standard compounds as references in Lee’s index scale was based on the following: a plot of elution temperature as a function of retention index for the four standards approaches linearity; these compounds are quite soluble in the usual organic solvents and are generally commercially available; and test of reliability of the retention indices under different temperature programming rates shows average 95% confidence limits for the series of PAH retention indices (209) calculated from different temperature program- ming rates (2 jC/min and 4 jC/min) are F 0.16 i.u. [3]. PAHs retention indices are calculated by using a linear relationship suggested by van Den Dool and Kratz [2] for linear-temperature-programmed GC: I ¼ 100  T RðsubstanceÞ  T RðC z Þ T RðC zþ1 Þ  T RðC z Þ þ 100  z ð1Þ where T R(substance) is the retention time or elution tempera- ture of the substance for which the retention index is to be determined, T R(C z ) and T R(C z+1 ) are the retention times or elution temperatures for the PAH internal standards pro- posed by Lee et al. [3], which bracket the substance of interest, and z is the number of rings in the PAH standard that elutes prior to the substance of interest. Correlations of the chromatographic retention with mo- lecular properties such as connectivity index, molecular 0169-7439/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.chemolab.2004.01.011 * Corresponding author. Tel.: +381-21-450288; fax: +381-21-450413. E-mail address: biljana@tehnol.ns.ac.yu (B. S ˇ krbic ´). www.elsevier.com/locate/chemolab Chemometrics and Intelligent Laboratory Systems 72 (2004) 167 – 171