Retention Time drift: catecholamines with ion pair

[Hollow]

Hi all

We've got another issue I would like to ask for inputs.

We're trying to establish catecholamine analytics with ion pair reagent on C18 column, as published in varius application notes.

After the first few trials which looked very promising, the retention time of dopamine decreases with about 0.5 min for every injection, starting at about 22 min.
Does anybody have an explanation of this phenomenon and probably also have a solution on how to solve it?

I searched the old forum archive and saw, that similar issues were observed some time ago but didn't find a solution.

The decrease shows good correlation with almost every parameter, time, no of injections or volume of eluent (maybe we didn't too few variation)

I don't think it's due to incomplete column equilibration, because the retention time doesn't stabilize even after >100 column volumes of eluent pumped. Also the column is thermostated at 40°C, so thermal fluctuations are not likely.
The "fault" we've done is to flush the column with 100% MeOH to clean the column from unpolar sample compounds and for storage the column. Then after a week and reequilibrating with the ion pair reagent, we got these drifting retention times...

our chromatographic system:
Col: Symmetry 4.6x75mm, 3.5µm
Eluent: Na-Octansulfonate 5 mM/PO4-buffer 85mM, pH3/MeOH 5%, isocratic
Flow: 1 ml/min
Temp: 40°C
Pressure: about 2700 psi
Inj Vol: 10-20 µl
sample: dissolved in 0.1 M HCl, same effect if diluted or dissolved in buffer.
Areas are not affected by the drift.

Any suggestions?

Could it be possible to be some "dewetting" effect due to the high water content on the C18 column? (>we have to do some experiments on this)

Are we disturbing the ion pair equilibrium with each injections? (> but longer cycle times doesn't improve/restore the retention time)


PS: before we used ion pair we already tried:
- conventionell reversed phase C18 on Atlantis T3 with 100% water
>good selectivity with standards but not with real sample (matrix)
-- SPE clean up with ion exchange didn't solve the problem.

- Atlantis HILIC: no satisfactory separation/reproducibility/peak shape

[Kostas Petritis]

It sounds like a disturbance of the column equilibrium, although it looks like you do everything right. I would look on the details of the method. For example, do you use a loop wash or syringe wash that may introduce higher amount of organic solvent and thus disturbing the equilibrium?

[Hollow]

Hi Kostas

thank for your fast reply.
we use a Waters Alliance and yes we also switched the wash solvent to less MeOH than usual, think it was 10% MeOH (don't know exactly anymore, have to ask my co-worker) but haven't seen any clear differences a.f.a.i.k.

But I will discuss it with my co-worker again.

[Hollow]

Update:

a) phenomen persists also with a flow of 1.5 ml/min


b) retention time is not affected after turning off the flow for 60 min. After restoring the flow, the retention time was the same as it was before stopping the flow but then began to decrease again...
So dewetting seems not to be the an issue.
We also switched the wash solvent to MeOH 5%, as it is on initial.

Moreover it seems to be injection related.

Any hints for troubleshooting?

[Teodor]

I am new at LC, so I hope dis doesn't sound stupid, but maybe the problem is with your column. Did you try another column of the same type, or different analytes on this column, and check if the problem persists?

[Hollow]

Hi teodor

thank you, any suggestions are welcome.

we now dissolved the sample in mobile phase > same effect

after cleaning we run a PQ mix which gave indifferent results from that before running the ion pair reagent on it, so column is likely to be ok (in case of stationary phase).

The weirdest thing is, that the very first injections had good RT reproducibilty.

[Uwe Neue]

None of the things that you are doing strike me as being a problem for the column. Does your heater have solvent preheating capability, and are you using it?

[Hollow]

Hi Uwe

we now tried an older Polarity dC18 3.5µm, 3.9x100

same effect...

RT is decreasing by 0.2 min with each Inj (very regular, RT started at 17.8 min)

About preheating:
we use the Alliancne column compartment and have about 20-30 cm of steel tubing inside the oven before entering the column.

Waters sales person today suggested to switch from silica based material to polymer like XBridge (or Xterra).
Could this be a worthy try?

[Kostas Petritis]

I do not think that changing columns will change anything. It looks to me that you are introducing something during your injection that affects the column equilibrium.

I would assume that once the column gets cleaned and re-equilibrated you achieve the initial retention times? What other compounds in your knowledge are in your mixture (anything too hydrophobic)? Can you bypass the autosampler and inject through a manual injection valve?

[Uwe Neue]

I also do not think that the problem is with the column. pH and temperature are sufficiently mild, the surfactant should not play a role with respect to stability of the column. This why I was asking about the thermal equilibration, but you are using the standard setup, and this should be OK. The use of the older column with the same effect also indicates that something else is happening. The next question would then be the quality of your reagents, such as traces in the surfactant that build up on the column, and related things. The XBridge column is more stable than the Symmetry column, but I am not convinced that the problem is in the column.

[Hollow]

by this time, the "sample" is still a reference standard, only dopamine, and we do observe these phenomenon...
So by now, no real sample matrix on the column (we only had it in the beginning of the development).

About the surfactant impurities:
would they show a behaviour like this?
I mean, it doesn't really matter how much eluent we pumped, but it correlates with the no. of injections.
But we will have a closer look to the reagent used.

Or coudl these impurities be in the sample itself and build up with each injection?

(test to do: injection of a different batch of dopamine or even another catecholamine)

What else experiments are likely to guide us out of this mist?

Would it be a good idea to add a non ionic compound (like some paraben or phenon) to the "sample" and see if its retention also drifts?
(idea: track the problem to ion pair equilibrium or column build up?)

[Hollow]

I do not think that changing columns will change anything. It looks to me that you are introducing something during your injection that affects the column equilibrium.

me too... but what?

like in my previous post above: sample by now is just a single compound, "pure" dopamine. I think we have to have a close look on the impurities of the dopamine we use.

I would assume that once the column gets cleaned and re-equilibrated you achieve the initial retention times?

yes, retention time will be restored but begins to drift again.

Bypassing the autosampler is not possible.
We just could switch the whole method to another HPLC.

[Kostas Petritis]

If the dopamine retention time is independent to the equilibration time and only depends on the # of injections, then your mobile phase additives (i.e. ion pairing reagents etc) are pure enough.

I suggested by-passing the autosampler for trouble shooting reasons. If by using a manual injection valve the retention time of dopamine does not change, then the problem is with your autosampler or the way in general you inject/wash etc. Or maybe try to inject 3 times your mobile phase, then inject dopamine; did the retention time decreased 1.5-2 minutes instead of 0.5? That should again direct you towards the autosampler/wash etc.

If the autosampler/way you inject is not your problem, then the problem is in what you inject. Next step would be to dilute your dopamine sample around 10 times by using your mobile phase as the diluent. If the problem is impurities in your dopamine, then the dilution should minimize the effect (you will still see a decrease in the retention time but it will be much less pronounced).

[Hollow]

Dear all

just to close this thread and give you a feedback.

Finally we did some more tests as suggested and found that we probably missinterpreted (miss-evaluated) some of the previous runs.

After we had changed almost all variables (3 HPLCs, columns etc) without improvement, we had to re-think on the parts which were constant in all these experiment... the NaOH used for pH adjustment and the Na-OSA... or in other words the mobile phase additives.

So we bought again a new LOT of the higher priced Na-OSA (as used in the very first runs) and what a surprise... the drift was gone!

Conclusion:
By comparing the CoA and price of the Na-OSA of the same supplier, we deceided to use the lower priced reagent which had almost the same purity as the high priced (99.8 vs. 100.9%) but then the problems began as described above.
So the higher price is justified in this case.

Thank you all for your assistance on this tricky but learnfull troubleshooting session.

[lmh]

Thanks for an informative thread, and thanks for posting the conclusion.

I must just remember to make sure I always by 100.9% pure chemicals...