UPLC columns not stable (solved!!)

[Mattias]

This question is somewhat related to my post about comparison of Acquity and Kinetex columns. I hope someone can help me about what is going on!

History:

I had developed an UPLC method using a Acquity BEH RP-Shield column (1.7 µm, 2.1 mm x 100 mm) and and a weak acetonitrile gradient, going from 20 to 25% acetonitrile in 5 mM ammonium acetate pH 5.2 (over 10 minutes). Temperature: 50°C.

The problem with this method was that the Acquity column did not offer a stable separation. I could see a continuous loss of retention of the acidic compounds. The retention of the neutral compounds didn't change, which lead to co-elution after about 24 hours of usage of the column. The column seemed to have anion-exchange properties that was washed of during usage. This was confirmed by the late Uwe (thank you for all your help!). The official standpoint from Waters was that I had been very unlucky to see this effect (which I find strange, due to the loss of retention counting in minutes).

I was very glad too see that I could replace the Waters column with a Phenomenex Kinetex column (only a slight pH change to pH 4.6 needed). The Kinetex column also offered a fantastic improvement of plate count and peak symmetry, which was quite surprising. The problem was that this column was even more unstable, using the given conditions. After only a few hours use, some peaks even shifted elution order.

I have a lot of questions.....

- Has anyone seen the same thing?
- What are these anion-exchange groups (residues from column manufacturing acc. to Uwe)?
- Are there any UPLC columns that do not show this behaviour?

I am under quite a lot of pressure about this, since I was supposed to transfer this method to QC now. The method would simply not work in a QC environment. The obvious solution is to lower the pH so all components are neutral, but it doesn't work. All peak then elute at the same time. HELP!

[Mattias]

I was just thinking...

Could it be possible that what I am seeing is not a loss of anion-exchange, but instead an increase of ion-exclusion. That more and more negative groups (silanols?) are exposed during usage - reducing the retention time of the negative compounds?

[Lusby43]

I find the BEH columns are usually quite stable, especially in the conditions you are using. I'd be more suspiscious of pH drift, your buffer is very weak for acetate, try 10 or even 20 mM and see if the retention is more stable. The fact that neutrals are not affected supports this conclusion.

[tom jupille]

Any time I see retention variation of this sort and I'm running within +/- 1.5 pH units or so of analyte pKa, I suspecct pH control. That hypothesis is easy to confirm or eliminate: if it's a mobile phase pH problem, retention should change again when a new batch of mobile phase is prepared.

Another possibility is that anionic CRUD from your samples is accumulating on the column (which might then result in anion exclusion). Since the deterioration is happening quickly, you can set up an experiment to determine if the problem is mobile phase / column deterioration on the one hand or contamination on the other hand. It's described on our web site:
http://www.lcresources.com/resources/TSWiz/hs390.htm
If the problem *is* sample related, then the fix is straightforward: a guard cartridge or better sample workup.

If the problem is, in fact, deterioration of the column, then things are not so straightforward and you might have to start over again, perhaps with a different column at a lower pH, or perhaps ion-pair chromatography or one of the mixed-mode columns.

[Mattias]

Thank you for your answers!

Monday morning here, and I have been pumping mobile phase through a new column over the weekend (did one injection Friday, giving good retention of all compounds). I can see from the first injection after the weekend that I have lost a lot of retention of the acidic compounds. So the loss of anion-exchange/increase of cation-exchange groups has nothing to do with the injection of samples at least. It happens just by pumping mobile phase though the column.

I will now prepare new mobile phase and see if I get back my original separation. The pH of the old buffer has not changed, so I doubt that it will have any effect.

Possible solutions??:
- Test the Kinetex XB-C18 phase (supposed to have more silanol protection than the C18 phase)
- Add heptane sulfonic acid to the mobile phase to have a controlled anion-exclusion effect
- Test a Primesep 100 column to get a reproducible anion-exclusion effect (but then I will not get any UPLC advantages - it only comes with 5 µm particles)

[Mattias]

Mystery solved!

I had some discussions with a column expert at Phenomenex, amd his theory was that I was seeing ion-exclusion from the silanol groups. I tested to add some (2.5 mM) octane-sulfonic acid to the mobile phase and all of a sudden all chromatograms looked the same. I can now take a new column or a column used for hundreds of injections - and still get the same separation! That was impossible before. The only drawback is that the method is not longer MS-compatible, but this is an easy sacrifice to get a stable method.

Learning points:
- Even top-of-the-line UPLC columns contain quite a lot of residual silanol activity
- The end-capping can apparantly easily be washed off, despite friendly conditions (pH 4-5). This applies also for the Waters BEH column (i.e. don't trust all the claims of the catalogues)
-

[tom jupille]

Thanks for the update!

[Mattias]

Thanks Tom!

I guess some of you may wonder why I added a negative ion-pair reagent, since I didn't want the anion-exclusion effect (that was increasing during usage). This was a mistake that turned out very well.... The negative ion-pair reagent pushed out the acidic compounds to an area in the chromatogram that was empty before. So now I have a completely different elution order, with huge peak spacing (we are talking about 9 peaks that elutes within 8 minutes, with a lowest resolution of 4!).

And the anion-exclusion effect of the octane-sulfonic acid is way stronger than the effect of silanols. I even doubt that the acidic (negatively charged) compounds ever get a chance to see these silanol groups (?). They can be how many as they want.

[Vlad Orlovsky]

My guess is that when you added 2.5 mmol of sulfonic acid your pH dropped below pka of your acidic analyte. Did you check pH of the mobile phase? Or it stayed the same because you added sodium salt? Also you increased "buffer concentration" by 50%. Also may be you increased hydrophobicity of your column because sulfonate is on the column tide to your basic sites.
P.s. we have primesep a and primesep d columns packed with 2 um silica gel

[Mattias]

Hi Vlad,

I actually pH adjusted the 5mM ammonium acetate with the octane sulfonic acid (instead of using acetic acid) - to a pH around 5 (estimated conc 2.5 mM).

If I use a low pH mobile phase for this application, all nine peak elutes in one big blob (i.e. I really need to use the differences of pKa between the compounds). I have not been able to test a higher ion strength before, since the Acquity system gets really noisy above 5 mM acetate (at 220 nm). (I don't know if my ion strength is that much higher than before I added reagent?)

The ion pair reagent makes the acidic compounds elute much earlier, making the time span between the first and last compound larger (giving better opportunities).

I can see that this application would probably work really well on a negatively charged Primesep column (such as Primesep A), without the use of ion-pair reagent. I cannot see the 2 µm version on your website - is it on the market now?