Modeling Limits of Detection in Headspace GC/MS

The limits of detection (LOD) for headspace Gas Chromatography-Mass Spectrometry (GC/MS) of a diverse series of compounds were accurately predicted using a simple linear regression model incorporating Henry's law constants and a polarity descriptor. LODs were predicted with an R 2 of 0.9154 and a standard error of 1184 pmol/L (approximately 17 mg/L) over a range of LODs from 2-585 mg/L using Henry's law constants and dipole moment. This study has important practical applications, as experimental LOD studies are often timeconsuming. The ability to estimate LOD values from existing data provides a rapid means of approximating the method sensitivity. A simple rule of thumb derived from this study is that for compounds with comparable molecular weights, nonpolar hydrocarbons have the lowest limit of detection, polar aprotic solvents have intermediate LODs, and polar protic solvents have the highest LODs. These generalizations are directly related to the ability of the species of interest to interact with water molecules. Henry's constants capture the bulk of the variations in sensitivity for this data set, while dipole moment captures the non-ideal interaction effects that are not resolved by Henry's law constants alone.