GC/MS reproducability

[smithdm]

We re-inject our high control at the end of each batch to make sure the analysis is the same through the end. However, for one of our assays on a separate instrument, the end batch QC quantitates at a lower concentration than it did toward the beginning of the run by about 100-150 ng/mL. It sticks out in my mind because our batches are relatively small (about 40-45 samples including blanks), and it doesn't happen with any of our other assays.

Does anyone know what would cause this to happen?

[Don_Hilton]

How long a list would you like?

Is this other instrument from a different GC/MS manufacturer? But, then again, I've seen differences in stability between instruments from the same manufacturer... Does the autosampler tray sit in a different environment - like directly over the GC oven on one and out away from the GC on the other?

Is there a different exposure of samples to light as they sit on the autosamlers between instruments?

Does the method use an internal standard? Does the intensity of the internal standard change through the sequence? (And I would have QC's spaced a bit more closely through the run because a failed QC leaves everything suspect back to the last good QC.) Are your internal standards labeled compunds or similar compunds to the target analytes. And if similar compunds, how similar?

Are you using the same type of inlet liner on each instrument? (Are you using the same type of GC inlet - including manufacturer - between instruments?)

Are you using the same GC oven and transfer line conditons between instruments? Same ramps, rates and bake out time and temperature between isntruments?

Do you change the GC septume with each batch? And have you looked at the end of the needle on the syringe for a burr.

If you run an MS tune check at the end of the sequence, are the ion ratios for PFTBA the same as befor the sequence (is the mass spec changing in relative sensitivity)?

Perhaps one of those questions will give you an idea of a place to look - or more information to help narrow dwon what is going on.

[wraymogg]

Hi,

Don covered most of all.

Different devices will behave ... differently. For example some older devices will be incapable of high source temperatures, so the ion source will get dirty way faster. This can decrease your sensitivity even inside 45 rounds, depending of the analyte.

Another difference is from diffusion pump vs turbomolecular pump, even on the same device with the same method run. For example the sensitivity is decreasing with diffusion pump on multiple injections, if there is a big buildup in vacuum. Check if the pressure inside is increasing from sample 1 to sample 45. This happens especially if you have dirty samples and pulsed mode.

I presume you don't spike your samples with internal standard, otherwise every change in sample will be compensated by the same change in internal standard.

Regards,
Vlad

[smithdm]

Thanks for the help

Our instruments are by two different manufacturers, but the same models (Agilent or HP GC 6890 and MS 5973) and the autosampler trays are in the same location (out over the mass spec). The light exposure is the same (same room, same vials, overhead fluorescent lighting in the same place over each instrument). We use the same liner for each, and change the liners and septums before each run. The primary instrument is older, but we can run different assays on it, and there is no issue.

We do use labeled internal standards. Do to the nature of the job, we are required to run an end batch QC. I have been running smaller batches of about 35 injections and the problem still exists. But I can re-inject the batch afterwords, and the beginning QC is still in the range at which it should be (same vial, one injection toward the beginning of a run, one injection at the end).

Perhaps it is because the samples are dirtier than other assays. I will check the tune at again at the end of a run.

How do I check the pressure during the run?

Thanks so much.

[Steve Reimer]

5973 MS came with either a turbo pump or a diffusion pump. If you want to put a lot of time into this you could switch the source between the two to see if the problem moves. Even if the old HP was a GPIB and the Agilent is an N version the source parts are the same.

If you ran a lower level control at the end, would it fail? If the problem is targets saturating the detector then the internal standards would not necessarily correct for it (I am assuming the internal standards are at the same level throughout and less than the targets in the high level control). How old is the electron multiplier in the HP? When you do an autotune what is the voltage?

[Don_Hilton]

There is a clue here. The QC, when run at the end of the batch fails, but when reinjected some time later is OK. Sounds like someting in the sampels that hangs around for a while, but goes away over time. Leaves me wondering aobut the column. An older column will have more activity -and may hang onto junk longer. Any difference in ages of the columns? If the coolumn associated with the problem has seen more or dirtier injections, I'd give thought to a new column.

[James_Ball]

Try running the check standard after every 10 injections and see if it trends downward with each run or drops after the first 10 injections then holds steady.

If it drops quickly then holds steady, it could be the analyzer loading up with solvent, which after it sits over night is pumped away and the sensitivity returns. This is especially bad if you are doing purge and trap or injecting aqueous samples.