| dc.description.abstract | Ambient volatile organic compounds (VOCs) are not only toxic at high concentration, but also act as precursors of ozone and secondary organic aerosols (SOA). Monitoring ambient VOCs often requires the step of pre-concentration prior to gas chromatographic (GC) analysis. Sorbent adsorption and thermal desorption (TD) in the process of pre-concentration inevitably result in a certain degree of peak tailing and asymmetry affecting qualitative and quantitative results.
In this study, we used a self-built cryogen-free TD device to test for a series of sorption materials including activated carbons, mesoporous silicates, and Zn porous materials. Special attention was paid to ethane, ethylene, propane, and propylene because of their extremely low boiling points, which easily results in pronounced breakthrough problem and hence low recovery. Of all the tested porous materials, the activated carbon materials were found to exhibit better performance than Si and Zn materials. Although they can trap the widest range of VOCs, but still showed insufficient efficiency on C2 compounds, even trapping at low temperature (-40℃) and high inlet pressure (40 psi). As a result, the commercial Carbosieve SШ, which is a microporous sorbent, was added in the sorbet bed to improve the C2 recovery.This dual sorbent formulation of activated carbon materials plus Carbosieve SШ was able to effectively trap the full range of C2-C12 compounds with desired linearity (R2> 0.99) and reproducibility (RSD < 4%).
The use of activated carbon materials as sorbents also produced much narrower and more symmetric peaks than the the use of other materials. Even so, slight peak tailing still existed. To diagnose the cause of tailing, the Deans swtich method was adopted to slice the desorption peak for the difference in composition. Compound discrimination were found between different slices and higher boiling VOCs tended to reside more in the later slices than earlier ones.
In addition to the aforementioned efforts of sorbent selection and TD peak diagnosis, other TD variables, such as the TD heating rate and flow splitting, were also tuned to optimize peak shape and symmetry. | en_US |