Delay between UV and MS chromatogram

[arnabroy81]

I am observing a delay of about 0.2 minutes between the UV and the corresponding MS chromatogram. Is it acceptable for publication? See the chromatogram.

Please advise.

[Gaetan Glauser]

0.2min delay seems a bit too much to me, I would typically expect 0.02-0.06 min between UV and MS under standard configuration. Possible explanations for this unusual delay:

-The flow rate is too low low or the tubing used to connect UV and MS detectors too wide. One might expect some diffusion and broadening of the MS peak, which doesn't seem to be the case here.

-The peaks detected in UV and MS are due to different compounds.

[dblux_]

I am observing a delay of about 0.2 minutes between the UV and the corresponding MS chromatogram. Is it acceptable for publication? See the chromatogram.

Please advise.

Check if your program allows signals to be time shifted.

[Rndirk]

On a sidenote, you want at least to double the amount of data points on the yellow one, whichever detector that is. It will look much nicer.

[arnabroy81]

0.2min delay seems a bit too much to me, I would typically expect 0.02-0.06 min between UV and MS under standard configuration. Possible explanations for this unusual delay:

-The flow rate is too low low or the tubing used to connect UV and MS detectors too wide. One might expect some diffusion and broadening of the MS peak, which doesn't seem to be the case here.

-The peaks detected in UV and MS are due to different compounds.

I'll give full details of my system configuration.
MS: ESI-QTOF operating in Auto-MSMS in negative mode.
LC: micro-LC. Buffer A: Water, Buffer B: acetonitrile. Column: C18, 150 x 4.6 mm, 5 micron. Flow: 200 uL/min. Gradient method.

The connector tube that I am using is red color coded (0.005" ID) and the length is around 2 feet.

[arnabroy81]

I am observing a delay of about 0.2 minutes between the UV and the corresponding MS chromatogram. Is it acceptable for publication? See the chromatogram.

Please advise.

Check if your program allows signals to be time shifted.

Thank you for the suggestion. The program do allow to shift the MS chromatogram towards right side but not the UV. I cannot shift the MS to the left or the UV.

[arnabroy81]

On a sidenote, you want at least to double the amount of data points on the yellow one, whichever detector that is. It will look much nicer.

Thanks for your suggestion. This is actually an extracted ion chromatogram of the expected mass and this is the maximum I got fineness I got. I can smooth it though. I that okay?

[Rndirk]

If you don't need to integrate it for quantitative work, it's OK i guess. Just saying that if you decrease the cycle time (=increase cycles/second) next time you perform it, you will have a nicer peak

See also this discussion for more info

[arnabroy81]

If you don't need to integrate it for quantitative work, it's OK i guess. Just saying that if you decrease the cycle time (=increase cycles/second) next time you perform it, you will have a nicer peak

See also this discussion for more info

Thank you very much Rndirk. I read the link you posted. It has good information on dwell time and cycle time. Now, in my Bruker software there are two options in Auto-MSMS mode.

1. Number of precursors
2. Cycle time

From the discussion, I am thinking that these values are way too much in my instrument!!!

What do I do? Should I just decrease the Cycle time and check? Or should I increase/decrease the number of precursors? Any advise?

[Rndirk]

Not sure what this 'auto-msms' mode does. But with a cycle time of 3 seconds, you collect 1 data point every 3 seconds. Set it to 1 second and compare the result.

[lmh]

(1) The delay is a little long, but not necessarily incorrect; you're running quite a slow flow-rate. At 200uL/min, 0.2min is 40uL delay volume. 60cm of red tubing is 7uL, and unfortunately MS systems will often have extra volume in additional tubing between wherever you attach the tube from the PDA and in the spray needle itself. Ideally it shouldn't add up to 33uL, but sadly it's not impossible.
(2) I wouldn't recommend time-shifting the chromatograms to make the delay look as though it doesn't exist. Many of us expect the delay, and if we don't see it, will be suspicious of whether the peaks are the same chemical. If you do choose to do any time-shifting, then you will need to state what you've done in the text of any article, and will have to justify the time-shift value you choose (for example by basing it on a standard containing few enough peaks that you can be absolutely sure which peak is which).
(3) and no, definitely, don't on any account smooth the MS data to conceal the small number of spectra collected during each chromatographic peak. Putting a Gaussian filter through a single spike will convert it into a nice Gaussian peak, but that doesn't mean the underlying data were Gaussian. Using smoothing to conceal the peak shape doesn't improve your measurement of the peak shape, and probably counts as an unethical manipulation of data (you just don't want to go there!). If you're not quantifying using the MS data, then not knowing the exact shape of the peak doesn't matter; if you're using MS data purely for qualitative confirmation of identity, then it only matters if you have nearly-isobaric compounds eluting so close to one another (e.g. isomers) that it becomes important to know exactly which of two very close peaks corresponds to a particular MS/MS spectrum (and this is a horrible situation to be in). Yes, you probably want nicer data next time, and it's worth exploring how you can reduce your MS cycle time, but I've seen far worse peaks published, so you don't need to worry unduly.

I am a big fan of smoothing, because it makes integrators work more reliably. It can reduce the risk of the starts and ends of small peaks being influenced randomly by noise. My feeling is that smoothing is a much lesser evil than manual integration (Smoothing is applied to all runs in the same (documented) way, and shouldn't introduce human bias, while manual integration is by definition unique to each sample and depends on human gut feeling). But smoothing should only be used to counter genuine noise. It should never be used to make a signal look a different shape or imply more measurements than actually exist.

[arnabroy81]

Not sure what this 'auto-msms' mode does. But with a cycle time of 3 seconds, you collect 1 data point every 3 seconds. Set it to 1 second and compare the result.

The Auto-MSMS mode picks ions automatically based on intensities and fragments them. In the method we can set the number of maximum precursors that can be selected for fragmentation, that is 'No. of Precursors'. I spoke to the service engineer. He said I can increase the Spectra Rate, which is currently set at 1 Hz. That means one cycle per second.

Generally what is the Spectra Rate do you use? I think there is a cumulative effect if I keep the Spectra Rate slow... more ions get accumulated and the MSMS fragments show up in more intensity. So, what should be an optimum Spectra Rate? And how far I can go?

[arnabroy81]

(1) The delay is a little long, but not necessarily incorrect; you're running quite a slow flow-rate. At 200uL/min, 0.2min is 40uL delay volume. 60cm of red tubing is 7uL, and unfortunately MS systems will often have extra volume in additional tubing between wherever you attach the tube from the PDA and in the spray needle itself. Ideally it shouldn't add up to 33uL, but sadly it's not impossible.
(2) I wouldn't recommend time-shifting the chromatograms to make the delay look as though it doesn't exist. Many of us expect the delay, and if we don't see it, will be suspicious of whether the peaks are the same chemical. If you do choose to do any time-shifting, then you will need to state what you've done in the text of any article, and will have to justify the time-shift value you choose (for example by basing it on a standard containing few enough peaks that you can be absolutely sure which peak is which).
(3) and no, definitely, don't on any account smooth the MS data to conceal the small number of spectra collected during each chromatographic peak. Putting a Gaussian filter through a single spike will convert it into a nice Gaussian peak, but that doesn't mean the underlying data were Gaussian. Using smoothing to conceal the peak shape doesn't improve your measurement of the peak shape, and probably counts as an unethical manipulation of data (you just don't want to go there!). If you're not quantifying using the MS data, then not knowing the exact shape of the peak doesn't matter; if you're using MS data purely for qualitative confirmation of identity, then it only matters if you have nearly-isobaric compounds eluting so close to one another (e.g. isomers) that it becomes important to know exactly which of two very close peaks corresponds to a particular MS/MS spectrum (and this is a horrible situation to be in). Yes, you probably want nicer data next time, and it's worth exploring how you can reduce your MS cycle time, but I've seen far worse peaks published, so you don't need to worry unduly.

I am a big fan of smoothing, because it makes integrators work more reliably. It can reduce the risk of the starts and ends of small peaks being influenced randomly by noise. My feeling is that smoothing is a much lesser evil than manual integration (Smoothing is applied to all runs in the same (documented) way, and shouldn't introduce human bias, while manual integration is by definition unique to each sample and depends on human gut feeling). But smoothing should only be used to counter genuine noise. It should never be used to make a signal look a different shape or imply more measurements than actually exist.

Thanks a ton for your thoughts, lmh.

1. Yeah, I know the slow flow rate is one of my drawback with this configuration. I am working on it following your previous suggestion on splitters.

2. Our concern was whether this can be published. Thank you for clearing our doubt.

3. Okay, so, we can either work on increasing the data points or report as it is.