Chromatography Forum

Hello,

We are seeing something puzzling on our LC-MS/MS runs (triple quad). Our batches begin with a calibration curve, then we inject samples, with a check standard (bracketed by blanks) every 10 samples. The calibration curve has excellent linearity and the blanks are indeed blank.

What I find puzzling is that the same sample, in the same vial, gives a certain response during the calibration samples--say, 50,000 cts/sec--and then, for the check-standard injections later in the run, it consistently gives a much lower response--say, 30,000 cts/sec. The same vial is being injected for the calibration and for the checks. It is not steadily drifting, like the first injection gives 50,000, the next gives 45,000, the next 40,000, etc. No, instead, the calibration gives one response, and the checks give a different response, consistently--50,000 cts/sec during calibration, then 30,000, 30,000, 30,000, 30,000 cts/sec for the checks.

We have tried equilibrating the instrument by running ~10 blanks at the beginning of the run, to ensure it is fully warmed up. But somehow, the instrument seems to have a telepathic ability to know whether the vial is being injected for calibration purposes (in which case, it measures 50,000 cts/sec) or for check-standard purposes (in which case, it measures 30,000 cts/sec). I do not believe our instrument has telepathy and is trying to sabotage us, so I do not understand how this is possible.

Am I missing something? Any help whatsoever would be appreciated. Thank you!

I'm sorry if it sounds stupid, but did you check the injection volumes?
Sure you use the exact same acquisition method?
Applied some correction factor in the software?
Double-check your acquisition batch, maybe some entries in an invisible column?

Well, it's not a stupid question in the sense that I wonder if our instrument is delivering consistent injection volumes. At the level of applying the same method, the same injection volume, etc., in the software, yes: we are doing everything the same in the two cases.

I am thinking of selecting a more simple and dependable analyte and seeing what the raw area response is as a function of injection volume, above and below our current 5 microliters. Maybe the autosampler is the issue?

If a faulty autosampler was the culprit, you would see random variations of the response and not a systematic differece between Cal and control samples. A contemporary sampler should be able to inject 5 µL reliably.

I still think it could be some kind of software problem.

What is the response you get if you perform a single injection, outside of a sample batch?
Do you observe the same behavior if you use a different standard as control?
If possible, check visually if the Cals and controls are really injected from the same vial.

Hi,

The others have good ideas, I've got a different angle:

Have you tried equilibrating in the middle of the run before doing the check standard injection? Have you tried repeat injections of the check standard in the middle of the run (5-6 injections rather than 2-3)? Is the internal standard changing in the same manner? If this is a multiple analyte method, are the other analytes responding the same way?

The last time I had this problem, I traced it to lingering sample matrix on the column. It took several "clean" injections to clear the column and restore area response in the check standard--I figured that out by repeatedly injecting it after a batch 10-20 times where I could see the area creep back up to normal. For that project, it didn't affect accuracy, precision or LLOQ so I was able to proceed without much difficulty.

Good luck to you!