Chrysene-d12 inherently erratic?

[whdees]

Good morning, all....

I have the most baffling problem, I am tearing out what little hair I have left. I am analyzing pesticides by 525.2, on a Thermo ISQ. I get very nice chromatography, good peak shapes, great linearity, excellent repeatability, and good performance from all of my compounds - except for Chrysene-d12, which shows random drops in response of 40 to 60 percent. It doesn't matter if the sample is an extracted drinking water or a standard freshly made from a new stock standard. Here is the oddest part - re-injection of the same solution from the same vial under the same conditions, sample or standard, will usually give excellent results, with the Chrysene-d12 coming in right where it should be. The fact that this compound is one of my internal standards makes this an even more pressing problem... totally bum-fuzzled here - any ideas?

Thermo Trace 1310, ISQ, 1uL autoinjection

30M x .25mm column in great shape
injector temp 305 (it was 285...raising it didn't seem to make a difference)
init temp 34 C, hold for .5 minutes
16 C/min to 300, hold for 3.5 minutes

MS transfer line 305 C, MS is running great, everything looks terrific - except this one compound!

[JMB]

You have not said that all of the analytes also drop out when the chrysene-d12 drops out, so presumably not a faulty injection.

Do any of the analytes that bracket the chrysene-d12 also drop out ??

Erratic behavior strongly suggests an intermittent electrical fault--- power supplies to detector/ transfer line/oven controller etc. may be failing at a certain condition (voltage, temp. etc) or a short-circuit somewhere.

[kubowicz.tomasz]

Hello

I get very nice chromatography, good peak shapes, great linearity, excellent repeatability, and good performance from all of my compounds - except for Chrysene-d12, which shows random drops in response of 40 to 60 percent.

So how is it possible that you have good linearity and repeatability if Chrysene is your internal standard and calculations are based on ISTD?

Regards

Tomasz Kubowicz

[whdees]

Hello

I get very nice chromatography, good peak shapes, great linearity, excellent repeatability, and good performance from all of my compounds - except for Chrysene-d12, which shows random drops in response of 40 to 60 percent.

So how is it possible that you have good linearity and repeatability if Chrysene is your internal standard and calculations are based on ISTD?

Regards

Tomasz Kubowicz

basing my observations on raw peak areas of the individual compounds they look great. if I change to external calibrations they look great. if I look at the RSD of a series of runs, the RSDs are very tight - usually <5%...the fact that the Chrysene-d12 is an IS, and shows random drop outs, is indeed the problem...

[whdees]

You have not said that all of the analytes also drop out when the chrysene-d12 drops out, so presumably not a faulty injection.

Do any of the analytes that bracket the chrysene-d12 also drop out ??

Erratic behavior strongly suggests an intermittent electrical fault--- power supplies to detector/ transfer line/oven controller etc. may be failing at a certain condition (voltage, temp. etc) or a short-circuit somewhere.

no, just the Chrysene-d12 drops out, the rest of the compounds in the run look fine. that is why this is so puzzling to me. It simply doesn't make sense...especially when re-injection of the same cal standard or sample extract may give excellent results for all compounds, including the Chrysene-d12...there are no ghost peaks, no noise spikes, no indications of carrier flow interruptions, nothing that gives me any hint of why this is happening...

[kubowicz.tomasz]

Hello

If for compounds calibration you're using target ion (for IS) you need to make sure that integration for this specific ion is correct. When you click "integrate" on SIM chromatogram it is integration for total SIM (target ion+ 2 or 3 qualifiers). So in theory you can have perfect integration for total SIM and very poor integration for target ion (it can cause problem with final results).

Regards

Tomasz Kubowicz

[whdees]

Hello

If for compounds calibration you're using target ion (for IS) you need to make sure that integration for this specific ion is correct. When you click "integrate" on SIM chromatogram it is integration for total SIM (target ion+ 2 or 3 qualifiers). So in theory you can have perfect integration for total SIM and very poor integration for target ion (it can cause problem with final results).

Regards

Tomasz Kubowicz

an excellent point, thanks... I'll check that now.

EDIT;
this turns out to not be the problem...the instrument is seeing the correct masses...also, the Total Ion Chromatogram is proportional... but it was a good idea, thanks!

[JMB]

Apologies if the following is already well-known to you, but when you plot the profile of chrysene-d12, you see a nice smooth, Gaussian peak which is variously at 100% or ~50% of expected area. However, this is not what the detector sees; the data (maybe 10-20 points across the peak) is recorded as a table of m/z vs Intensity for chrysene-d12. If you could intercept the data before the software manipulates it, you could plot the data yourself as a stick graph. If you then joined the tops of the adjacent 10-20 sticks with a straight edge, you would get a sharp-pointed, multi-sided approximation to a Gaussian curve. If you further took an eraser and a pencil to eliminate sharp points and unnatural angles you would get a smoother Gaussian curve. You could repeat this exercise multiple times on the same dataset until a perfectly smooth curve resulted.

This is essentially what the software does before the data system releases to you the smoothed plot (complete with area counts) from which you take your numbers. I am not familiar with the data system software on your MS, but in general there are at least two parameters such as baseline (noise) level and smoothing functions which are applied to generate the smoothed profile. These are very powerful, and changing these parameters on the same dataset can change the area counts for the analyte. The automatic smoothing software can conceal a multitude of sins such as spikes, and signal "drop-out".

Are you able to re-plot the failed d12 data as a stick graph (either centroided or non-centroided) rather than the smoothed curve ?
Are there any abnormalities apparent ?
Can I assume that you ARE measuring peak areas, rather than peak heights ?
How is the mass spec data acquisition set up ? Full range scan continuously (e.g. m/z 50-450,1 sec), different mass range at different RT's etc ??

[Peter Apps]

I do not know any of the specifics of the method, but I note that the column starting temperature is low (and not a round number), and in general solvent effects are among the few things that can generate inconsistent results for one peak out of several on a chromatogram.

What solvent are you using, how much do you inject and are you operating splitless (I would guess splitless for a residue method) ? How much are you able to change the column initial temperature within the dictates of the method ? If the problem is intermittent just 1 degree increase might be enough to prevent it.

The peak distortions caused by solvent effects can compromise integration (rising and/or falling peak slopes are changed) or if you have very tight SIM windows the retention shifts from a solvent effect can carry part of your peak outside the window.

Peter

[X]

You may need to reduce the injection temp. and MS transfer line temp., I had no problem with 250 C for the injector and 280 C for the transfer line.

[James_Ball]

Normally here I believe we start 525 runs at 50c.

If you look at the spectra do you see a single mass peak for the quant ion, or do you get multiple mass peaks that are 1 amu apart?

We had a problem once where if the extract was too acidic, you would lose deuteriums while on the column and you would get a nice looking peak but the quant ion was low in abundance because we were losing deuteriums and by the time it was at the analyzer you actually were looking at multiple molecules ( d10, d9, d8 ect). This can happen if your extract is too acidic.

[whdees]

Thermo ISQ on a Trace 1310, XCalibur for data acquisition, scanning 40-450, 0.2 secs, measuring peak areas...was able to plot out my raw data in "stick" form in XCalibur....the data I saw agreed with the processed data coming out of TraceFinder 3.2...

can adjust injection port temps and transfer line temps as I see fit, consistent with good performance... have always had very nice results shooting BNAEs and other extractable semivolatiles at 34 C with the injection port right around 300 C, using split/splitless injection (hold splitless for 1 minute, then split ratio of 33:1), solvent is ethyl acetate, 1 uL. with a BP of 77 C, I do have plenty of room for adjustment. Peter, you mentioned that my oven starting temp is not a round number - is this a general rule of thumb? I've never heard of this...

The injector and transfer line temps actually started out somewhat lower. This is a new instrument from a different vendor, so I was being conservative. Raising the transfer line and injection port temp may have helped the situation - I need more data to be able to draw a conclusion one way or the other. We'll see how it goes.

James, you mention the possibility of there being too much acid in the process - you may have a very good point. I need to double check the strength and amount of acid used in the extraction. OTOH, I am also seeing this problem with calibration standards straight out of the bottle...

I expect this to be an on-going process of testing and tweaking one parameter at a time. Unfortunately, I am far behind my projected time line for my present work load, so I do foresee some late nights in the future... many thanks for the good ideas, I appreciate any help I can get.

[JMB]

Here is the oddest part - re-injection of the same solution from the same vial under the same conditions, sample or standard, will usually give excellent results, with the Chrysene-d12 coming in right where it should be.

I think that this result would rule out any issue with sample prep.

My preference is still for an unknown electrical issue, but having said that..........it only affects the -d12, not the other analytes close by (i.e. RT's). You are scanning ~2000 m/z per sec---is the -d12 at a relatively high loading---does the EM crash because it cannot handle the sample response quickly enough---wish I knew more about electronics and data acquisition rates !!!

[James_Ball]

Here is the oddest part - re-injection of the same solution from the same vial under the same conditions, sample or standard, will usually give excellent results, with the Chrysene-d12 coming in right where it should be.

I think that this result would rule out any issue with sample prep.

My preference is still for an unknown electrical issue, but having said that..........it only affects the -d12, not the other analytes close by (i.e. RT's). You are scanning ~2000 m/z per sec---is the -d12 at a relatively high loading---does the EM crash because it cannot handle the sample response quickly enough---wish I knew more about electronics and data acquisition rates !!!

I usually set mine to scan at least 10 scans across a peak, 20 is better if you can scan that fast or have wider peaks. Too few scans across a peak can give mass discrimination since the abundance of analyte coming off the column at the beginning of the scan is different from the abundance at the end of the scan. It can cause some weird results if you only have 3 or 4 scans per peak. At 0.2 seconds per scan a 1 second wide peak would only have 5 scans, if the peaks are at least 2 seconds wide then you should be ok.

Since you are seeing it with both extracted samples and diluted standards then it is probably not the amount of acid in the samples. Since EPA525 normally uses extracted standards I did not know if you had injected any non-extracted ones.

I don't know how the Thermo software is set up, but on Agilent the default for finding a peak at a certain mass is -0.3 to +0.7 amu such that if you are looking for mass 237 anything detected from 236.7 to 237.7 would be called mass 237, if the target analyte is at mass 236.7 and drifts a little negative in mass then you will have a peak that is much smaller than it should be because the software thinks it is mass 236 instead of 237. I always have to check mass assignment in the tune and in the quant software to be sure I am not missing some abundance of my target masses.

[whdees]

I'm seeing 14 to 16 scans per typical peak, so there is plenty of data to work with. the mass assigned for Chrysene-d12 is 240.16 - it looks for the assigned mass, +/- .5 and is impressively stable in that regard. I have a hunch this is going to turn out to be something really basic that I just overlooked, and I think Peter already touched on it, as did X and you, James...the oven temp at injection. I have been extracting into methylene chloride for a couple of decades now, and a low initial temp was always optimal, but these days most of my work is in the 500 series, where the BNAEs and 525 pesticides end up in ethyl acetate, which has a much higher boiling point... I am looking at initial temps in the 50 to 60 range, and the peak shapes (which I thought were pretty good to start with) seem to be even cleaner and more symmetrical than before. I will know more as I collect more data, but things seem to be improving (keeping fingers crossed on that).

old dogs can learn new tricks, but sometimes it takes longer for us to catch on