Reproducibility Issue on Thermo Ultra Fast Trace GC

[tlahren]

I am running a mix of PCBs (52,153,155 & 209) and perchloro-terphenyls as well as deca-PBDE and deca-PBDEthane on a Thermo UF Trace GC using a 10m × 0.18mm × 0.1µm Rxi-5SilMS column. I have been having reproducibility issues for the PCBs primarily on both a PTV injector and a SSL injector both in SL mode. The PCBs can vary up to almost 10× in area between a series of repeat injections. I have c13 labeled analogs of each compound and the response differences of those don't match those of the native compounds either. I have tried over 4 weeks or more changing every GC parameter for aquisition with nothing yielding a clue as to what the problem could be. Has anyone had a similar experience or give me some advice for something else to try?

Thanks,

[DSP007]

1) The device is technically good? Syringe, the injector septum, the cost of gas
2) No errors in the preparation samples ?
3) "skipping" through a column there? No overload ?
On thin columns used SSL 1:10 or more

[Peter Apps]

What detector are you using ?

Peter

[tlahren]

I am using a Thermo High Resolution magnetic sector mass spec (DFS or Dual Focusing Sector). I have tried a 1:10 split, i've tried 0.5 µL injections, the MS is in great shape and no signs of any problems, I've tried two brand new syringes and multiple septa as I have been having this issue with both the PTV and SSL injectors. Thanks for your replies.

any suggestions?

[DSP007]

0.5 mkl in 10 mkl surgine ?
10x - is a order number (1,10,100,1000) or 10%?
If 10% - RSD explainable. A very small stroke pluger in sugine

[tlahren]

I am using a 10 µL syringe and I have, in some cases, a full order of magnitued (10) difference in area between injections. e.g., 100,000 counts to 1,000,000 counts for area on repeat injections of the same standard.

[Peter Apps]

The peaks coming off your short, narrow, thin film column are extremely sharp; perhaps too sharp for the scan rate of the MS. How many scans do you get across the peaks ?

Peter

[tlahren]

Peter,
I thought about this initially but realized now I never fully looked into it. I guess this may be part of the issue as for the PCBs I am only getting about 7 scans per peak. Right now I am struggling with my DFS so I am trying to see if I can increase the scans/s. The DFS is kind of finicky in that if I go too high it seems to not want to find the lock masses for my high res method. They kind of disappear. If I figure it out i'll post what I see.

Thanks,

[Peter Apps]

Seven points across a peak is dangerously low - if the middle point misses the peak apex it can make a huge difference to measured peak area (an order of magnitude is worse than I owuld expect though). I think that there is an inevitable trade-off between speed and MS resolution; if you want fast separations you have to use a fast MS (probably TOF) which gives only integer resolution. If you want high res MS you have to have lower scan rates, which requires the peak widths that you get with conventional separations.

Peter

[tlahren]

Last night I tried a faster scan rate. It turns out there must have been a miscommunication between me and the service engineers because I was changing something called a tune scan speed (Hz) and we found through trial and error that I was only getting 1 Hz scans when I would change it from 2 to 4 Hz. Turns out this value doesn't affect the aquisition scan rate. That is edited in the MID (SIM) method and we were never told about that value. I thought it was odd that a system this high tech and sophisticated would only have a scan rate of 1 Hz. I am about to check the data from last night and see if it made a difference. I guess we here had all made the assumtion that the scan rate was faster because it makes no sense to use only 1 Hz scan rate. Thanks for you input Peter.