Relative response factors for H2 and O2 using GC-TCD

[starfry]

Hi There,
We have to analyze hydrogen and oxygen concetration. I am using GC-TCD, Column: MS-5A, carried gas :N2.
The Relative response factors for H2 determined in our system was about 312. But we don't know is the value correct ? Is there any literature for the
Relative response factors for H2 using GC-TCD? I'v serached google and ISI
but found nothing about that.
Any comments would be appreciated.

[chromatographer1]

Starfry

The relative response depends upon your detector parameters. You will find thermal conductivity values in the literature however. Handbook of Chemistry and Physics CRC Press page E-3

Calibration? try injecting air at a fixed volume and then determine the mole response to oxygen on your detector at your specified conditions.

Then inject pure hydrogen or a mix of hydrogen where the mole % of the mix is known and calibrate the mole response of hydrogen.

Sound simple enough?

Rodney George
rgeorge@sial.com

[Les]

When calibrating my TCD, the hydrogen curve may not be linear. I've used a couple of GCs with differing responses based on the range setting, the integration parameters, and the injection volume. I had one GC that required a quadratic curve while another used a linear curve.
Rely on calibration, and uniform injection technique rather than on literature response factors.

[starfry]

Thanks a lot for your informations ^_^
By the way, is there other method to measure hydrogen concetration ?
In my experiment, the producion rate of hydrogen was about 1ml/hr, the production is so small that we could not measure continuously (the loop of our GC is 5ml). Thus, we use gas collection with displacement of water to collect hydrgen. After several hours, till the collection gas was about 8 ml, about 1µl gas was then injected to GC by a syringe. However, the hydrogen peak detected in GC is much lower than theoretical value..

[chromatographer1]

Have you considered the amount of hydrogen that could have been dissolved in the water and perhaps was lost?