polar/nonpolar solvent mixture-TCD vs FID linear response

[cbd]

Hello everyone,

Can someone help me to better understand why I get a linear calibration curve with a TCD versus an FID? I am trying to get a calibration curve over the entire composition range for a 2 solvent mixture (one polar solvent, one nonpolar solvent) and am using both FID and TCD (two different instruments). For TCD, I do the separation on a DB-1 column, Tinj=200C, Toven=30C ramp to 200C, Tdet=250C. For FID, I am using an HP-5 column, but am using the same operating conditions. I am operating in split-mode, and the split ratios are similar. I inject 0.2 microL manually for all injections.

Why is the calibration curve using the response of the TCD linear (over the entire composition range) and the FID response is non-linear? For the TCD, both area% vs weight fraction, and area vs. weight fraction are linear. For FID, the area% vs. weight fraction polar solvent is non-linear (parabolic, concave upward; area% vs. weight fraction nonpolar solvent is concave downward). For FID the area vs. weight fraction also appears non-linear, but because of higher variability in injection volume, it is not that obvious.

I think it may be related to the changing volume due to volume expansion/contraction of the solvent mixture. Why/how could this affect the FID and/or TCD response and the calibration curve.

Many thanks for your help,
Christina

[Don_Hilton]

Area % as a function of weight percent will be a straight line only if the response factors for the two compunds are equal. Apparently in the TCD system, they are close enough. In the FID system, they response factors differ sufficiently to make the curve show.

You can demonstrate this in a spread sheet. Build two columns for proportions of A&B (they will, of course sum to 1 for each line in the table.) Then make two column headings that are numeric values - these will be your response fators. Computer the "area" for each peak by multiplying the corresponding response factor. Finally compute area percent for each row in the table and plot that row against the corresponding column in the first two columns you created. Change the response factors and watch the line dance. if one response factors differes from the other by even a factor of two, the line clearly bends.