Conditions for the optimal analysis of volatile organic compounds in air with sorbent tube sampling and liquid standard calibration: demonstration of solvent effect

Abstract

The combined use of sorbent tubes (ST) and thermal desorption (TD) has become the common practice for the trace-level analysis of gaseous volatile organic compounds (VOCs). In this research, the potential bias in VOC analysis due to the solvent introduced into the system as a liquid standard (LS) is examined in three stages by analyzing LSs of 19 VOCs in methanol solvent against a three-bed ST (Tenax TA, Carbopack B, and Carboxen 1000). In experimental stage 1, LS made at four concentration levels (between 10 and 150 ng μL(-1)) were each analyzed at four injection volumes (1, 2, 5, and 10 μL) based on a vaporization method. In experimental stage 2, calibration was also conducted by direct injection over an extended concentration range at two volumes, 1 and 10 μL. In experimental stage 3, the response factors (RF) of a single analyte mass were compared across the four injection volumes and between two injection methods. These results were analyzed to explore the complex relationship between variables such as LS volume, target/solvent chemical type, sorbent strength, and prepurge condition. There was no change in the ST/TD performance up to 2 μL of LS. However, as the injection volume increased up to 5 μL, a notable shift in RF and retention time occurred (e.g., for benzene and methyl ethyl ketone). At the maximum injection volume (10 μL), a significant reduction in sensitivity is evident for all compounds, e.g., 50 % drops relative to 1 μL injection. As such, the TD performance tends to deteriorate with increasing volume of methanol initially loaded on the ST. Although the dominant fraction of solvent was removed by two prepurge steps, residue caught in the strong sorbent fraction is still found to exert an effect on the subsequent analysis, e.g., delayed retention, sensitivity reduction, or disappearance of certain compounds.