Tailing as a function of stationary phase chain length

[adam]

We have noticed on many occasions that going from a C18 phase to a phenyl phase can significantly reduce tailing. The reason for this I completely understand. Tailing normally occurs because only a certain fraction of analyte molecules can get down to the silica surface and interact with the silanols. The fraction of molecules that is affected in this way experiences extra retention - hence the classic tailing profile: where a fraction of molecules lags behind the rest of the peak. When one uses a phenyl phase there is a much shorter ligand. Therefore, all of the molecules have equal access to the silanols; and hence the symmetry of the peak is not effected.

My question is that I've been finding recently that a C8 column doesn't work as well as a phenyl column in this regard (and in some cases can make the problem worse). Could it be that a C8 column is still too long and that you have to go to a shorter ligand to eliminate tailing.

Does anyone have any ideas about this?

Any papers to recommend?

Much thanks in advance.

[Mark Tracy]

Unless the C8 and phenyl columns are on the same silica substrate and use similar bonding techniques, it is too hard to make a general conclusion. End-capping also makes a difference. All other factors being equal, C8 chains pack better than C18 so the maximum surface concentration (in moles/m^2) is greater. One can always bond for less than maximum coverage. Not sure about C8 vs. phenyl, but I suspect phenyl is sterically bulkier.

Just to toss out an alternative hypothesis... In hydrophobic nanopores, water behaves more like a vapor than a liquid. Loosely packed hydrocarbon chains permit more water vapor to permeate to the silica surface than tightly packed chains. Sparsely packed hydrocarbon chains would allow liquid water to wet the silica substrate. This would affect base tailing.

[Uwe Neue]

If your theory is correct, you can prove it with a low-level C18 coating that is fully endcapped. Try Atlantis T3!

[adam]

Is that my theory?? I thought that was a well established thing.

Are you saying that you do not agree with my synopsis?

I am very interested in getting a better handle on this topic. With all the variations in chain length that are available today (C8, C4, C1, C30, ect) it would be extremely usefull to understand the relationship between chain length and tailing.

Much Thanks

[Uwe Neue]

Don't know if this is your theory... I have never heard that phenyls are better with respect to tailing, but I definitely do not know everything...

What I do know is that with an identical preparation procedure, a C8 gives a marginally smaller tailing value than a C18. I have always attributed this to the slightly higher coating that you get with a C8. This would not be true for a phenyl, which will tend to have a smaller maximum coating level due to steric hindrance. On the other hand, endcapping may be more successful (= less hindered) with a phenyl than with a C18.

Honestly, you can try to demonstrate your point with the Atlantis T3. The coating is less than half of a standard C18, and the endcapping is very good.

[adam]

Uwe: Let me ask you this question.

When you do your testing, does the XTerra RP8 (a C8 column that also contains a polar embedded group) show more tailing that the XTerra MS8 (a C8 column that does not have a polar embedded group).

If I'm right, I'll explain later what my logic is.

Thanks

[Uwe Neue]

A packing with an embedded polar group will always give less tailing for bases, but for the XTerra packing the difference is barely perceptable. We have attributed this to the fact that the polar group strongly binds water and therefore forms a shield above the silanols. This is the reason why we called the older form of such a packing SymmetryShield. Other proposals for the mechanism have been brought forward as well, but for me the binding of water was always the most logical explanation.

[adam]

OK. For whatever it's worth, here is my current theory.

We all know tailing has to do with interactions with the silanols on the silica surface. So the way I figure it, the trick to minimizing tailing is either to minimize these interactios or to maximize them.

So, for example, a C18 phase with a polar embedded group creates a situation where it is difficult for the analytes to reach the base silica. Hence we see minimal tailing. At the other extreme, if we have a C8 or a phenyl column (with no polar embedded group) the analytes will have a high degree of access to the silica such that (if my theory is right) all analytes have equal access to the silanols and, hence, are equally effected - retention time is therefore effected but we see little peak tailing.

Finally, the things to avoid would be the in-between situations: such as a C18 phase with no polar embedded phase or a C8 phase that does have a polar embedded phase. Because in cases like these the analytes have "moderate" access to the base silica: such that only a fraction of the analytes may be affected.

If this were true it would mean that when we use a C8 column we would not want it to be polar embedded.

Maybe I am wrong. But this logic would explain some of the things I have seen in the lab.

[Uwe Neue]

Well designed polar embedded phases with a C8 chain work like a charm. So this can not be the problem.

Also, a long, long time ago, I investigated what makes tailing go away with various levels of C18 coating (no endcapping). You needed to go down to a coating level of about 1/6 of the standard coating, before the tailing disappeared.

You have a point though about multiple interactions. You may want to read my paper on „The combined effect of silanols and the reversed-phase ligand on the retention of positively charged analytes“, U. D. Neue, A. Méndez, K. Tran. P. W. Carr, J. Chromatogr. A 1063 (2005), 35-45. In this paper, I have pointed out that different interactions do not add up, but multiply with each other - at least partially.

This does not appear to link with tailing though.

[HW Mueller]

Uwe,
do you still recall what made the tailing go away in going to 1/6 coverage? Was it that the compound disappeared by being hung up on the column or because you changed conditions like pH and/or ionic strength?

[Victor]

Adam, I found a paper which describes what you are seeing in terms of the reduction in tailing with decreasing chain length. However, this paper does not offer much additional information as to why this happens, other than the increased coverage generally achieved with shorter ligands, and perhaps the less restricted access of buffer ions and solutes to silanol groups in shorter chain length phases.

see J. Chromatogr. A 844 (1999) 23

[Uwe Neue]

Hans, the tailing disappeared with the low coating, not the compound (a tricyclic antidepressant). The entire experiment was run with a phosphate buffer at pH 7, no change in solvent composition, if memory serves me well.

People use silica successfully for the analysis of basic compounds in HILIC. For bases, HILIC may just be a label, and the dominant mechanism is ion-exchange.

[HW Mueller]

OK, the peak probably just came out later. I just wanted to know whether the change was really due to the change of the stationary phase, and not due to a change in the mobile phase.
Or, in other words, it could be that adam´s conclusions ("theory" on chain length) are based on comparing results of a column with optimal mobile phase conditions with columns that where off as far as optimization was concerned?

[Uwe Neue]

Hmm... The McCalley paper indeed shows more tailing for bases on a C8 compared to a C18... I suppose that I need to do a bit more homework on this...

[HW Mueller]

A quick look at the McCalley paper seems to corroborate what Victor related: A better symmetry for the peaks in the SHORTER chain columns is attributed to higher coverage, thus LESS contact to silanols (by the way, why are SiO- not called silanolates?).
There were also less diffs in symmetry when MeOH was in the mobile phase as compared to ACN, also less diffs in k (between the columns). I have my qualms about comparing symmetry of low k peaks with high k peaks. If the same amount of analyte is always injected (didn´t check, yet, whether this was the case here) then early eluting peaks will be much higher than the late ones. Can one assume that tailing and main peak would change proportionally?