retention time shift with Ion pairing LC

[dlliu]

Dear all,

There is a puzzle that I can't really explain for my IP-HPLC, in which I use 5mM NFPA (nonafluoropentanoic acid) as IP agent to analyze underivatized amino acids. I found the retention times get faster and faster over time (as opposed to shifting to slower r.t. as reported by other researchers), and I was wondering what causes this. Is NFPA so corrosive that it gradually eats away my column packing materials? I'm using C18 capillary column, and ACN as solvent B. At the end of the day, I flush the column with 100% ACN for 30 min. After ACN flushing, r.t. doesn't come back; it still gets faster and faster.

I have measured the flowrate from the column and it has been pretty stable. The room temperature is within 20-25 C.

Any insights would be greatly appreciated.

DL

[Chris Pohl]

DL,

Based on the fact that you are unable to recover your lost retention with solvent treatment, it sounds like you may be degrading your stationary phase at low pH. You're ion pair reagent is a pretty strong acid and at your use concentration it's probably causing damage due to acid hydrolysis. Because you will have significant concentrations of your ion pair reagent on the surface of the stationary phase, the stationary phase pH will likely be significantly lower than your mobile phase pH, exacerbating the problem. I would suggest moderating the pH with the addition of some sodium hydroxide if that is feasible.

[Kostas Petritis]

DL,

Two possible solutions,

1) You do not leave your column to equilibrate enough.... there are ways to see the exact time your column needs to be equilibrated. I would try to equilibrate for half an hour (which is a lot for this IP agent) and see the retention times.

2) As Chris indicated, your column might not be adequate for use with perfluorocarboxylic acids. From my experience, non-endcapped columns can be damaged easier that end-capped columns.

Do you mind sharing column chemistry, column dimensions and flow rates with us?

[Uwe Neue]

Anything else in your mobile phase? I was thinking of a buffer, such as a formic acid or phosphate buffer...

[dlliu]

Thank you all for the comments.

I have only NFPA and water (from Burdict&Jackson, HPLC grade) as my solvent A. Before I analyze my sample, I always alow the column to equilibrate with 5 mM NFPA for at least one hour (I even tried a whole morning one day but the retention time issue still remains).

My column is Hypersil GOLD capillary from Thermo, 10 cm x 0.3 micron I.D. with flowrate of 5 uL/min.

I should also mention that last week I filtered solvent A and the retention time miraculously rolled back... but over time r.t. gets faster again....
I filtered again this week but this time the miracle is no longer there.


DL

[Kostas Petritis]

DL,

As you indicate, filtration of your mobile phase shouldn't have to do something with your analytes retention times.

The column you use is also a new generation chromatographic packing so with good pH stability so it shouldn't degrade... Furthermore, the fact that you were able to get your retention time back last week implies that the column should still be OK.

Here are some more thoughts:

1) Try to flash your column with adequate amounts of ACN everytime you finish your analysis as if your sample is pretty dirty you might accumulate impurities over time that occupy your C18 sites diminushing your retention times overtime...

2) I do not think that it is column dewatting as you say that you have a gradual decrease of the retention times while I would expect a sudden decrease if dewatting was the case. Furthermore, the use of ion-pairing agents decrease the possibility of dewatting... so that shouldn't be the case.

3) One hour is really a lot of time for NFPA, you could drop it to half an hour... (assuming that your system does not have large delay volume). I can also tell you how you can determine how much time is enough time for the column equilibration...

Anyway, from what you told us impurity accumulation would be my guess... What about your backpressure? Is it the same? Do you have any proteins in your sample and if yes do you take the appropriate steps to get rid of them?

[dlliu]

Thank you for your prompt response.

Yes 1 hour sounds a lot for column equilibrium, however if I try to drop it to half of the time, the column doesn't seem to be well equilibrated (i.e. the polar AAs don't retain well enough). I'm not sure if this is due to the nature of the column? or I should just measure the conductivity as a way to determine if the column is ready, as you indicated in your paper.

So far I'm only injecting pure AAs, that is, AAs dissolved in 5mM NFPA. The backpressure is fairly stable, except that I notice that it slightly increases over time (<4% increase though). Whenever I see the pressure creeping up, I know I'm getting faster r.t.; and that's always true.

I also have the experience of having a brand new Hypersil column that barely does chromatography with the IP agent so I have to return it, even they come from the same batch (same lot #). Is it normal (or not surprising) to receive columns that they don't behave consistently?


Thank you very much for your time.
DL

[Kostas Petritis]

Hmmm,

Kind of strange. I guess the best thing to do would be to use the Merck Purospher RP-18 column as I found out that it gives the best results (although I never tested this Hypersil GOLD).

In the Thermo Website they claim that the Hypersil GOLD column offer excellent peak shapes etc etc. Maybe is one of these column chemistries where you have residual positive charges so it will provide good peak shapes for basic compounds. If this is the case (maybe someone knows if this is true or not), things would be pretty complicated when on top of this you use negatively charged ion pairing reagents, as you would have all kind of interaction going on there... I would also expect such a column to have longer equilibration times and inconsistent results when used with opposite charge ion-pairing reagents (but again it is just an assumption that this particular column has residual positive charges)...

Also this (small) correlation between back pressure and analyte retention time is strange and could indicate some hardware (pump) problem but you said that you already checked the flow rates...

You could check the re-equilibration time, either with a conductivity detector as I indicate in the paper or with a UV at low wavelenghts.

[Uwe Neue]

I hate to get on my standard bandwagon, but I think it is necessary again.

In true (isocratic) ion-pair chromatography, the ion-pair reagent is adsorbed on the surface. The amount on the surface is substantial, in the order of 1 mmol for a 15 cm x 4.6 mm column. At an ion-pair concentration of 5 mM, it takes at least 200 mL to load up the column with the ion-pair reagent (ball-park figures). At 1 mL/min this makes for an equilibration of about 4 hours before the column is ready. If you work at a higher ion-pair concentration in the mobile phase, you will get there faster. Also, if the ion-pair concetration on the surface is lower, you will get there faster as well, but in all cases, there is an appreciable equilibration time.

You remove the stuff from the surface, if you wash the column with acetonitrile for storage. It is not a good idea to remove the ion-pair reagent from the column for overnight storage. Of course, you can reequilibrate the column every day again, just be prepared to spend your first hour or two in your cafeteria doing nothing...

In gradients, you are never in equilibrium with ion-pair chromatography. If you are executing your gradients in a reproducible way, you may get reproducible results, but only if you are lucky.

The final story is related to my question above. You will be better off separating the control of the pH from the column equilibration with the ion-pair reagent. I suggest to use a buffer that is independent of the ion-pair reagent.

[Kostas Petritis]

Uwe,

The amount that will be adsorbed on the surface depends on the hydrophobicity of the ion-pairing reagent. I have previously tested several perfluorinated carboxylic acids equilibration times by monitoring the breakthrough curves and found that up to NFPA the equilibration times are pretty short (at least for the columns I tested).

It must be noted that I was purging first the system up to the fitting that connects to the column, then connected the column and starting monitoring, so there was not effect of mobile phase mixing due to delay volumes that you would always experience in gradient chromatography (and it depends from system to system).

Typical curves can be found in J. Chromatogr. A 870, 2000, 245 (see Fig 1). See also table 2 in J, Chromatogr. A 833, 1999, 147... Finally there is a figure if I remember well with overlapping breakthrough curves for different concentrations of the same ion pairing reagent [see JOURNAL OF LIQUID CHROMATOGRAPHY & RELATED TECHNOLOGIES 23 (16): 2433-2446 2000].

As a result, I was always flushing the columns with ACN even overnight...

[Uwe Neue]

OK Kostas, you could be right that it is still small enough to not cause an equilibration problem.
How about using a real buffer to control retention? We still need to explain his problem, or better, make it go away...

[Kostas Petritis]

The problem with buffering the mobile phase is that whatever base you'll use (even NH3) you'll make your mobile phase from volatile to not volatile (see LC-GC Europe 15 (2002) 98-102), limiting the LC detectors you could use to almost nothing. Furthermore, you want to be at the side of pH where the amino acids are positively charged in order to retain them through this mechanism and using a strong acid in high concentrations is as acid as you can go (most underivatized amino acids pKa1 is between 2 and 3).

I was quite succesful in the past in obtaining very reproducible results by just using perfluorinated carboxylic acids without any buffering and have published on it i.e. validated method, see Chromatographia 2004, 60, 293. So I am quite confident that you can obtain pretty reproducible retention times with the right choice of column and trouble free pumps...

What do you think of the possibility of the specific column having residual positive charges in order to obtain good peak shapes for basic compounds? If this is the case, you might had some problems when using ion-pairing chromatography (and longer equilibration times due to the electrostatic interactions between the ion-pairing reagent and the column's positive charges). In that case, switching to another column could solve the problem? Of course, this is just an assumption...

[Uwe Neue]

I haven't looked at the Hypersil Gold. So I can't tell you anything about it. However, if there would be positive sites, a brief flushing with a higher concentration of the ion-pair reagent up front would make them go away.

His first guess was that the ligand is splitting off. This would happen only with a monofunctional ligand (or with the endcapping), and a di- or trifunctional bonding may improve the situation. But this will cost money...

[dlliu]

Thank you very much for the valuable comments.

Other than the shifted r.t., the quantitation so far is good from this column. Pretty good linear correlations were obtained (except for some AAs, such as Gln, Cys, but R^2 for these are ~0..

When you said flushing your column with ACN overnight, do you mean flushing continuously? do you use the full flowrate, or a fraction of it? I only flush my column for 20-30 min so I'm not sure if it's sufficient.

DL

[dlliu]

Oops, I mean the R^2 for Cys and Gln is ~ 0.8.