hypothetically speaking - pH matching pKa

[neuger]

If you have matched a mobile phase's pH with a drug's pKa, theoretically making half of the drug ionized and half unionized, would you expect to see two peaks - one earlier representing the ionized and one later representing the unionized state?

[PJ8]

Not sure about theoretically but what actually happens is a little of everything...

I've seen examples that just look like bad peak shape with shoulder/fronts/tails and example where there are two distinct peaks (though never baseline resolved because of the equilibrium.) It very much depends on your analyte (and in some cases changes injection to injection!)

Work at least 2 pH units from pKa and save yourself a world of hassle!

[neuger]

I always try to be above or below as the rule states, but with so many functional groups, it can be hard to get the pKa right.

I have seen the shoulder/split thing too, I just never really thought of it that way before.

[PJ8]

Splitting's not necessarily pKa, could be a approx. million other things (most commonly too strong a diluent I suspect, or a knackered column.) That's the problem with chromatography there's always a million things it could be. I guess you have to look at other factors to suggest which one.

I had a method that was close to the pKa of the analyte (but it was a complex pharmaceutical molecule so we weren't sure of the analyte and it wasn't easy to estimate.) We saw the bad peak shape (not an actual seperate peak, but a kind of lumpy shoulder), but what we also saw was retention time variability each time we prepared solvent because small differences in volume of TFA added (or amount of TFA evapourated) made the peaks move.

Also the impurities moved around with small changes in TFA amount, so we saw slight RRT differences. The imps had similar pKa to the API but not identical so movement was different for each peak. In redeveloping the method we moved from pH1 ish with TFA to pH5 with a buffer and got much more stable chromatography. We did however have to repeat the LC-MS impurity identification because the imps swapped round so much.

[gbalock]

I developed a method for amines at high pH. One of the amines had a phenolic OH that was ripped off by the pH 10 buffer and I got two peaks.

[Uwe Neue]

No guys....

I have run tons of compounds around their pKa. There is no peak splitting due to this. The equilibrium with the hydrogen ions in the mobile phase is VERY fast, much too fast to get peak splitting from the ionic form and the non-ionic form of the same molecule.

There can be peak splitting, if there is a second equilibrium associated with the ionization, or with the hydrogen ions in the mobile phase. Such a secondary equilibrium can be a structural rearrangement of the molecule, which IS slow. But this is a feature that has primarily nothing to do with the ionization per se.

If you have ever gotten two peaks, they were due to other things, but NOT due the protonated and the unprotonated form of the molecule.

[Uwe Neue]

Just wanted to add one more thing:

The requirement to run a method away from the pK of the analyte is a common prejudice, but it lacks to be supported by facts. I have seen demonstrations of bad peak shape for a compound with a pKa of 4.5, but the experiment was run with phosphate buffers which have NO buffering capability whatsoever at pH 4.5. Rerunning the exact same experiment with an acetate buffer, there was NO peak distortion whatsoever at pH 4.5.

What counts is the quality (or lack thereof) of the buffering capacity of your buffer, if you are working close to the pK of your analyte. If you have a good buffer, you do not have a problem with peak shape (unless there are silanol interactions, which are an entirely different story). If you have a reproducible way to prepare your buffer, you do not have a problem with the reproducibility of retention either.

In addition, with the shifts in pK and pH with the addition of the organic solvent, you have most of the time no clue where your method sits with respect to the relationship between pK of your analytes and mobile phase pH. Make reproducible buffers, and there are no problems (as was the solution of PJ8 above).