Lingering 'memory effect' in HILIC column

[Ken]

Hi all, I have a slight question, hope you all dont mind to share your views.

Is it possible to have a lingering 'memory effect' in a HILIC column even after prolong flush - ie 95% water and then 95% ACN and then 95% water again for approcimately 10 hr?

Everytime I have run a batch of app. 80 samples, I started to experience this memory effect in my column for the peak of interest that came out at the app. RT even when I injected a blank sample.

How can I go about removing this problem? There is no issue if I change to a new column ( obviously, I looked like there is some memory effect from the column ).

My gradient: Water + 10mM amm. acetate, ACN + 10mM amm. acetate

Total run time: 10 min

Flow: 0.25mL/min

Column: 150mmX 2.1mm X 3um HILIC column

Gradient conditions: 97 ACN to 80 ACN in 3 min, then 5 ACN for 0.5min and back to 97 ACN for 6.5min. Equilibration time: 5 min

Do advise if you have any idea. Thanks a lot.

[Patrik Appelblad]

Dear Ken,

The "Lingering 'memory effect' in your HILIC column" could be due to accumulation of salt (ions) on the stationary phase, and/or trouble with an improper gradient set-up.

Since, you indicate that there is no problem when you switch to a new column, I would personally try the following:

1. To ensure robustness and proper retention time reproducibility, we do not recommend to use gradients slopes higher than 2%/min. There is nothing wrong to use steeper gradients, but actions is then needed to compensate for the large and quick changes in the mobile phase composition.

In your set-up, you apply a gradient with about 6%/min slope for the first three minutes, and then from 80% to 5% ACN in 0.5 min, followed by a return to initial composition in 6.5 min. The latter, I believe this is a cleaning step, and is not used for elution of compounds of interest, right?

In that case, I bet the amount of stronger solvent is not enough to wash out potentially strongly retained compounds like multi-valent ions. You might need to increase the buffer concentration for clean-up, and regeneration purposes.

Additionally, the re-equilibrium step for 5 minutes, only correspond to approximately 2 column volumes of initial mobile phase. You can do this step at a much higher flowrate (the backpressure should not be very high anyway) allowing minimum 5-10 column volumes for re-equilibrium.

If you use a plain silica HILIC phase, the latter becomes VERY important, as the reformation of the water layer is slower and more complicated on such columns than for a bonded HILIC phase.

[Ken]

Hi Pat, thanks for your reply.

However, I still encounter this problem even after cleaning and regeneration of the column for app. 10 hr ( if you refer to my earlier comment in the post ).

Meaning, after the regeneration and injecting a blank, I experienced such 'effect'.

I am just wondering if my wash solvent for the needle is not appropriate. FYI, the wash solvent is 80% H20 and 20% ACN. Since I am using a HILIC column, perhaps repeated cleaning of the needle using this solvent is not appropriate due to the high aqueous condition that might caused the 'peak of interest' to 'stick' on the aqueous part of the needle.

I will change the wash solvent to 97 ACN; just like the initial gradient condition of my run and will check on the 'effect' then; hopefully this will eliminate the issue.

Any suggestions? Cheers ....

[thohry]

According to me, probably it's due to the ịnector system : loop, syringe, needle, valve etc.. You'd better wash these parts thoroughly.

[HW Mueller]

Ken, if you don´t know whether a column change was nothing but a column change we can not help you.

[Ken]

Hi Mueller, I dont actually get what you are stating there. Can you please explain again? Thanks a lot.

Thohry, for sure the contamination came from the autosampler and no where else as the 'carry over' effect was quite consistent with the RT of the peak of interest, so definitely contamination comes during injection.

I have modified my washing procedure and all; seems like this works; I dont see any carry over but will keep monitoring it.

If there is any other suggestions, do advise me. Thanks a lot.

[HW Mueller]

First you state:
"There is no issue if I change to a new column ( obviously, I looked like there is some memory effect from the column ). "

Then you wonder about the wash solvent of the needle.

Now you say that changing the washing procedure (of what?) solved the problem. I don´t see how I could have helped with this sort of info.

[Ken]

Hi Mueller, sorry for being 'not clear'.

Let me summarise the issue properly.

As mentioned, I'm experiencing a carry over peak that came out at the same RT of my peak of interest; this carry over peak was seen even by injecting a blank. hence confirming some sort of contamination.

I have tried to flush my column using 97% water: ACN and then switching back to 97% ACN: water for approximately 5 hour for each step.

After flushing, I tried to run the blank again and the same carry over peak was observed at the same RT as well.

Thus, I proceed to change to a new column; by using a new column, it seems that the carry over peak was eliminated ie I did not observed the peak anymore.

I am just wondering; how can it be possible for the carry over peak to be still remaining after flushing the column for so long?

And by using a new column, the carry over peak was eliminated. However, I noticed that after repeatedly using the new column, I will experienced the carry over peak again; and this will remained as the older column even after flushing it; meaning, the same problem occurs.

I cant possibly be using new column every time after multiple sample runs ie approximately couple of hundreds of samples.

I have also tried to change to freshly made buffer solutions, washing and rinsing the needle after every injection but still, this doesnt helps at all.

So, I am really curious why this problem seems to linger around?

FYI, I am using HILIC column - 150mm X 2.1mm X 3um; flow rate at 0.25mL/min.

Do advise if you have any idea. Thanks a lot.

[Peter Apps]

First, are you sure that this is carryover, and not simply an interfering contaminant in the samples ? The way to check this is to make an injection of whatever solvent the samples are dissolved in but with no sample (you refer to "blank sample" which I interpret as matrix with no analyte, if you have an interference this will give a peak even with no carryover). If you have carryover you will see a peak, no peak means an interference. If you see a peak inject clean solvent again, the peak will probably get smaller. If you keep injecting clean solvent you will see an exponential decline in peak area.

Assuming that you have carryover, in any HPLC system there are several places besides the column that sample can get stuck and be released in subsequent injections. You need to flush out the whole injector system and do some dummy injections of strong solvents. What happens if you run a gradient with no injection at all ? Are you using a guard column ?, how about changing it ?

Peter

[Ken]

Hi Peter,

I have done both; injecting a blank matrix as well as blank solvent.

On both occasion, the interfering peak were there. The peak area is quite consistent even when multiple injection of the blank solvent.

I did a cleaning and rinsing procedure of the needle by injection of max volume of blank solvent ( in this case, methanol ) multiple times and rinsing the inside of the needle with 97 ACN: 3 water multiple times as well. After these procedures, I proceed to inject the blank solvent again - the interfering peak were still there; there is no reduction in the peak size.

I am using a guard column and changing it does not help at all.

Its just weird considering the interfering peak did not reduce in size even by injection of blank solvents; even if the loop is contaminated; wont it get smaller when I inject blank solvents?

And how can it be possible for the column ( if its really contaminated ) to give me the interfering peak as well after flushing it for a very long time?

For the next step, I will proceed to use the column at another LC system to confirm once and for all that the contamination did not come form the column but rather on the injection side. If necessary, I will use the same chemistries from the current LC to the next one to eliminate any potential causes beside the injection part of the autosampler.

Any advise? Thanks.

[HW Mueller]

The confusion has not abated. What does this quote mean:
"I have modified my washing procedure and all; seems like this works; I dont see any carry over but will keep monitoring it."?

How can the carryover come from anywhere except the column when it disappears as soon as a new column is installed, unless something else was changed?
How can the peak stay constant if you inject consecutive blanks?

You are not telling us something or not noticing a mistake.

It very much looks like you are contaminating your blanks except when you changed to the new column.

[Peter Apps]

Ken

Before you risk contaminating another instrument there is another possibility that you need to eliminate.

Try another type of sample vial, with a different septum.

It is just possible that the contaminant comes from the vials, but that it has a very long retention on the column and so is also a ghost peak. This would explain why it goes away for the first several injections on a new column - it is slowly making its way towards the detector but needs several gradients to get there. Once a system is contaminated like this it will give a ghost peak for every injection. Using pure methanol as injection solvent is probably liberating a plug of contaminant from somewhere in the sytem, can you purge the column with pure methanol ?.

Peter

[HW Mueller]

Peter, wouldn´t a system contamination decrease when you inject consecutive blanks? If there is some sort of a physical action, a jolt or whatever, that loosens some material to produce a peak, it has reduced the deposit. Of course, if Ken only did two consecutive blank injections the decrease might escape detection.

[Peter Apps]

Hans

Yes, a carryover peak should decrease with each clean injection, but whether the decrease is detectable is another question - the decrease has to be bigger than random variation in peak area for a start. I think that you spotted the problem - the system is getting contaminated by the injections - and my guess is that the contaminant is coming from the vials ( I have a vague recollection that this has come up on the forum before) and so a really clean blank is never being injected. Contaminated solvent or water is another possibility of course. This is one of those problems where step by step, one thing at a time troubleshooting is the only way to get to the answer.

Peter

[Ken]

Hi Peter, Hans, thanks for the reply.

1) I did multiple solvent blank injections - consecutively more than 5 injections with no significant decrease in carry over peak intensity.

2) I am using new vials for every injections; so I guess it's not because of the vials rite?

3) I have tried the 'flushed' column with another LC system just now; it turned out to be negative ie no carry over peak observed. We plugged it back to tthis LC and the carry over peak was observed. So, I am pretty sure that we have pin point it to the autosampler.

4) We have purge the autosampler using 90 ACN: 10 H20, we have also prime the line using high aqueous, high organic, concurrently at 4mL/min, still we are not able to remove this carry over peak; the signal remains the same intensity.

5) As mentioned earlier, we have purged the column with 97 ACN for 5 hour followed by 97 water for 5 hr and then back to 97 ACN for 5 hour. Of course, by running the column at different LC which showed negative carry over response, we have eliminated the possibility that this column is contaminated.

Any other advise? I am kinda stuck here because this contamination is quite 'strange' to say. Thanks !!!