INFICON Explorer Portable Gas Chromatograph - Manual (Page 320)

Assay 2 User Notes
318
In addition to container effects, there are a variety of other issues regarding sample
storage, processing and preparation which can compromise the accuracy of the
concentration measurement.
Adsorption
Another potential instrument source of error is sample adsorption by Explorer's
analytical system. If this occurs, it will typically be seen as an increase in peak
area upon repeated sampling of the component, even after the instrument has been
in operation for many hours. The effect is usually concentration dependent, as
higher concentrations will supply the demand of a limited number of adsorption
sites without a significant percentage of sample loss.
Explorer has been designed to minimize adsorption effects by heating the
analytical system. However some compounds may still be problematic at low
concentrations. Each compound has been tested for adsorption effects and the
results are illustrated below. These results were used to generate the precision
specifications for Assay 2 compounds.
The instruments being tested had not previously been exposed to the target
compound. The analyses were performed with typical Assay 2 method parameters.
A bag sample is prepared containing all the Assay 2 target compounds at the
indicated concentrations.
The adsorption test itself consists of repetitive sampling (10 runs) and analysis of the
indicated compound concentration. The sample is typically obtained either directly
from a tank standard or by diluting a tank standard with air to obtain the indicated
concentration. The instrument was calibrated and then the 10 runs were performed.
Most of the Assay 2 compounds are stable in a gas bag at the indicated
concentrations. However, the methanol peak area and height were observed to
decrease over time during this test indicating that the methanol was not stable at the
target concentration in the bag and that therefore this approach to quantitating the
instrument adsorption effects for methanol would not produce accurate results. The
methanol results reported in Figure 31 were generated by sampling a gas stream of
96 ppm methanol produced by dilution from a tank standard.