LC-ICPMS Peak Splitting

[jdlh199]

I have a little problem that is driving me crazy! I am using an ion pairing reverse phase method for a metal speciation study with detection by UV 245nm (High level) and/or ICPMS (low level). This method uses a Symmetry C8 column (150 x 4.6mm) with an eluant consisting of 5mM EDTA, 1mL/L TBAH and 10% MeOH adjusted to pH7 with nitric acid (isocratic at 0.8 ml/min with a 30ul sample injection). The data looks great with UV detection, but I'm having some peak splitting issues when connecting to ICPMS. Here's some chromatograms of what I should be seeing:

Figure 1: Good Blank by LC-UV
Figure 2: Good Standard by LC-UV
Figure 3: Good Blank by LC-ICPMS
Figure 4: Good Standard by ICPMS

Everything works perfectly when the waste from the UV really does go to waste. As soon as I connect the waste from the UV to the ICPMS, I get this frequent peak splitting effect:

Figure 5: Bad Blank by UV
Figure 6: Bad Standard by ICPMS

This peak splitting doesn't happen with every sample, but is random. If I run 10 replicates from the same vial, 5 might be split peaks, 5 might be okay. Also, if I bypass the UV and connect the column to the ICPMS, I get the same effect (but with no UV trace for comparison!).

The fact that these split peaks are not consistently observed in the same sample suggests it's not really in the sample, and the fact that split peaks are observed in both the analyte and the peaks observed in blanks suggests that it is not an ICPMS nebulisation/sample intake issue. Plus, the fact that I am getting split peaks in the UV as well as the ICPMS suggests that it is a physical effect occuring between the autosampler and the UV detector that is a direct result of connecting the flow to the ICPMS.

We have a number of other LC-ICPMS methods but I have never encountered this before!

Any comments?

Jon Le Huray

Elemental Research Inc.
North Vancouver, BC
Canada

[Kostas Petritis]

Is the split of the peaks happen at the same way consistenly (i.e. is the ratio of small to big peak always the same)? Have you tried to change the tubing that leads to ICP-MS with a lower diameter one?

[jdlh199]

Without integrating the data, yes it does appear as if the ratio of the small first peak to the second big peak is always the same. So for the split peak samples, it looks like 80% or so of the sample lags behind the earlier 20%. But then this is happening with the blank peaks as well, so it's not just the analyte that is doing this. Woudn't that suggest this splitting is happening within the column as the result of connecting to ICPMS?

I'm using 0.005" ID red PEEK tubing from the column/UV to the teflon ICPMS nebuliser tubing which is 0.2mm ID.

Thinking it may be something to do with this I tried changing to a larger/narrower bore ICPMS teflon line, to no effect. I've also tried increasing the neb gas flow, to no effect.

I'm sure this is something simple, I just need to figure out what it is!

Jon

[jdlh199]

I'm still working on this and would appreciate any comments!!

I've tried switching nebulisers on the ICPMS from a PFA-ST neb to a Mira Mist, with no change.

Another point I've noticed is that once the peak splitting begins it remains even when the ICPMS is then disconnected. This means that the connection of the column waste to the ICPMS sample introduction system is having some kind of effect on the column which remains until the column is given a good clean with 100% solvent and reequilibrated with running buffer.

The only thing I can think of is the effect of the vacuum at the ICPMS nebuliser/spray chamber. Maybe I'll try ultrasonic nebulisation and see if that makes a difference.

It's just weird that I've never noticed this with any other LC-ICPMS methods we use.

[jdlh199]

OK now here's a little conundrum for you (if anybody's reading!). I ran a LC-UV run, 30 samples in all with no split peaks at all. Conected to ICPMS, and the first sample had split peaks. So I've just switched to a brand new column, equilibrated it with running buffer for an hour and injected a blank. Guess what... split peaks. Now this is getting weird!

I tried removing the column, so just straight PEEK tubing all the way to the UV detector with no ICPMS. Guess what? Split peaks. I'm getting a little bald patch on my scalp from all the head scratching!

So the column is not the problem. The autosampler maybe? The PEEK lines? I'm going to have a good clean out, change all the lines, and try again!

I know it's not a problem with the pumps, since this happens only with this method on two different HPLC units. And it only starts to happen once the ICPMS is connected! And it's not the result of something screwing up the column because it's just done this on a brand new column...

[tom jupille]

You have done a lot of work to narrow down the problem. If it were my problem, the next thing I would do would be to hook up the UV->ICP transfer line to the UV detector, but not connect the ICP. Then I would add some additional capillary restrictor tubing to mimic any additional back pressure generated by the ICP. If you get splitting with either of these, then a possible solution may be to split the flow stream after the column and run the two detectors in parallel instead of in series.

If you don't get splitting, then I'm stumped as well.

[jdlh199]

But why would the split peaks remain even after I then disconnect the ICPMS and go back to straight LC-UV

I even tried going all the way back to basics by injecting 30ul of acetonitrile using an isocratic eluant of DIW 1 ml/min with no column, just a straight pump connected to an autosampler with a UV detector and I still get split peaks.

I've swapped pumps, columns, lines, cleaned everything, reprepped buffers, samples etc etc etc... and nothing!

Next thing to try is throwing the unit out the window and going home early!

[tom jupille]

But why would the split peaks remain even after I then disconnect the ICPMS and go back to straight LC-UV

Oops, I missed that part.

Usually, when you see the same problem on all the peaks in a chromatogram, the odds are that the problem occurred when the peaks were still together (i.e., somewhere from the head of the column upstream). You've eliminated the column, since you get splitting even without a column, but then again, you've eliminated every possible cause, so the problem must be imaginary.

The only redeeming feature is that you now have a "consistent failure mode" (a system that gives a split peak with no column and only a UV detector). Time to start replacing bits in that system until you can make the problem go away, then add back the column and finally the ICP.

[HW Mueller]

A double injection? Not easy to do without knowing.

[jdlh199]

It's not a double injection.

I've ruled out the column and the pump since I've changed pumps and removed the column. Since it's happening with no column it must be a problem with either the autosampler injection valve, a line blockage somewhere, or a problem with the UV detector pathway.

Let's see what today brings....

[jdlh199]

OK we're finally getting somewhere. The link below is a screenshot of 10 sequential injections (30ul sample injection using a 50ul injection loop) of 20% MeOH using a DIW eluant flowing at 0.8mL/min. There is no column, so any peaks from the sample should come straight through at just past time=0. Detection is at 245nm by UV.

Figure 1

With every injection there is an initial peak (which cannot be seen at this scale) at around 10 seconds, which I am thinking is where the peak should be with no column. But I am also getting an occasional large peak at around 30 seconds. I am thinking that this is some kind of contamination that is being washed through every so often. Perhaps in the autosampler injector or injector valve.

It's not seen with every injection (as with the split peaks above) and when it does occur it is about 30 seconds after the peak of interest (also as with the split peaks above!). Time for a good clean of the autosampler!!

[HW Mueller]

The link produced a jumbled pic, but still: Could be air.

[jdlh199]

I think when viewing it in an internet browser you have to click the zoom button in the bottom right of the image to get it to full size.

Here are some more anyway:

Injections of 10% Acetone with diw eluant isocratic 0.8ml/min

Fig 1 - good
Fig 2 - bad

Injections of 50ppb cesium in DIW with detection by ICPMS.

Figure 3 - good
Figure 4 - bad

So it's definitely not the UV detector.

Also... I just switched to a 2nd autosampler, and it still does it....

So it's not the autosampler, it's not the UV detector, it's not the pump, it's not the injector... HELP!

The only thing that is common to all is the PeakNet method program that controls the instruments. So let's rewrite that and see what happens...

[jdlh199]

OH MY GOD!!!! I am so happy that I have found the problem that I have no shame in admitting it was totally my own fault. It seems it was a programming error. If anybody is interested you see the problem in these few lines of the screwed up program:

Old Program (split peaks)
CutSegmentVolume = 10
.....
Wait SamplePrep
.....
Load
.....
Inject
.....
End

So we have have "WaitSamplePrep" before "Load". So my guess is that the first smaller peak is the initial 10ul cutsegmentvolume, and the second peak is the complete or partial actual sample that we wanted to inject!

I didn't even think of this as a potential cause, it was only after I had tried to fix everything else that could possibly be the cause... Oh well, at least all the instruments are clean now!

And a fresh new method file!!

Thanks anyways guys....

[tom jupille]

It could have been worse. My career is a good example of the proverb "Experience gained is in direct proportion to equipment ruined!"

And thanks for getting back to us with the solution!