negative peak for CO on methanizer/FID

[geigor]

HI everybody
I'm running since many years an Agilent 6890 with FID to measure ambient air for CH4 and CO, (not CO2) through a methanizer converter. Time by time CO conversion efficiency seems to decrease, with 100ppb of CO giving a null signal (baseline flat) while I see, on a blank run (injecting N2 carrier gas or zero air), a deep negative peak exactly at the CO retention time. I've run several experiment to test for leaks, or column bleed but I can't find a solution: I still can not figure out the reason for the negative peak on the FID. Sample is injected by a sample loop; separation occur in a dual column system to avoid eluition of contaminant (heavier compounds) on main column (by Unibeads1S as a precolumn and Mol SIeve5A as analytical)
I've googled a lot on the forum and also on the web, but I've not find any report concerning a negative peak.
Could it be a matter of NiCat poisoned?
Any idea?

[tlahren]

HI everybody...

DISCLAIMER: I have never used a methanizer on a GC/FID setup before. However, I was going to suggest that the catalyst may be the problem as you describe the "conversion rate" of CO goes down in time. I used to run Organic/Elemental Carbon analysis and this particular machine used a methanizer to reduce CO2 to CH4 to detect on an FID. The Methanizer catalyst usually had to be replaced annually.

I assume that as non-converted CO elutes and reaches the FID it is not detected it could interfere with your baseline signal as a "negative" drop.

Look here under "Disadvantages" and "Alternative Solutions."
https://en.wikipedia.org/wiki/Flame_ionization_detector