The Use of Predicted Boiling Points to Review Tentatively Identified Compound (TIC) Data

Abstract

Laboratories which perform organic chemical analysis of samples from hazardous waste sites, using gas chromatography-mass spectrometry (GC/MS) techniques, are required to report the identity of significant unknown compounds (those not present in the calibration mixture, whose GC retention times are therefore not measured). Such "tentatively identified compounds" (TIC's) are identified by the GC/MS data system by comparison of their mass spectra to the 40,000 spectra in the EPA/NIH mass spectral library. A recent article pointed out that the identification of many TIC's is incorrect for a variety of reasons. One of the authors' recommendations to reviewers of TIC data was to compare the boiling points of the spectral library matches to the boiling points of calibrated compounds with similar GC retention times. Experimental studies have shown boiling points to be strongly correlated to GC retention times for semivolatile (bp > 170°C) GC/MS methods. Other parameters, principally molecular connectivity indices, have also been shown to be strongly correlated to GC retention times (and retention indices). These studies have been conducted on single or closely related classes of compounds, whereas the calibration mixture for a typical semivolatile analysis method has compounds from many different chemical classes. The purpose of this article is to: (a) demonstrate how well a single, widely published chemical property (boiling point) may be predicted given the GC retention times and boiling points of a typical heterogeneous calibration mixture, and (b) demonstrate how the predicted boiling point may be used in the initial review of TIC data to eliminate incorrect computer matches and serve as a starting point for more sophisticated retention index analysis.