Mobile phase in HILIC

[Pistek]

Hello,
I use Alltima HP Silica HILIC column for LC/MSMS cystathionine determination. With 85% acetonitrile, cystathionine is retained on the column. Using gradient from 85% acetonitrile to 100% methanol , cystathionine comes out as an acceptably sharp peak.
Questions:
1) I am surprised, that cystathionine is not retained in 100% methanol. Is this effect of methanol on retention common in HILIC chromatography?
2) May I expect some separation of cystathionine from other compounds (TRIS buffer, EDTA, salts, …) present in the assay mixture (diluted cell extract) using these elution profile? I use LCMSMS and i am affraid of possible suppresion effects or contamination of the MS from coeluting compounds.
Thanks & regards

[Patrik Appelblad]

In HILIC mode, water has the strongest eluting strength, followed by methanol and other unbranched aliphatic alcohols. Thus, using 100% methanol, you will not get anything retained on a HILIC column. If you go to www.sequant.com, you can leave your contact details and a copy of the Pratical Guide to HILIC will be sent to you. It is a booklet that gives a thorough introduction to HILIC as well as it exemplifies areas of use via some applications.

[HW Mueller]

This is from another novice regarding HILIC (though I did this some 40 years ago with TLC, none of us were clever enough to give this a new name):
First, there must be H-bonding involved with MeOH as its dipole moment is 1.7 vs 3.92 for ACN, while the respective polarity indices are 5.1 and 5.8, even the dielectric constants of ACN (35.94) is higher than that of MeOH (32.66, data from a Fisher poster).
Second, whether you will separate Tris, etc. will most likely depend on how much is present. If enough is present to spread out (produce a very broad peak, spreading into and byond your analyte peak) chances are that they may even change retention properties. At least that is my experience (not with Tris, but with ionic liquids). Now, since I have almost never seen this sort of thing with reverse phase my feeling is that HILIC is much more sensitive to such influences, in other words, less robust than RP. This seems to corroberate my earlier statement that HILIC can have a lower capacity (overload more easily). I see some great difficulty in proving this as the compounds are not ideal for or don´t work in one of the two methods. However, it is clear to me that the zwitterionic HILIC versions, though better than silica, are still touchy in the discussed respects.

[Einar Ponten]

Yoshida suggested hydrogen bonding to be the principal interaction in HILIC, while partitioning between a non-polar mobile phase and a water enriched layer in the stationary phase has been a more popular "explanation". Clearly, hydrogen bonding affects the water layer and influences the retention, as will the type of functional group responsible for attracting water into the layer. Thus methanol will be stronger than acetonitrile and a silanol -OH will bind water differently from an amino-group. Zwitterions have a unique behavior and water associated with some zwitterions exists more or less as bulk liquid water. Nevertheless, the HILIC mechanism need to be further studied.

If only HILIC mechanism is in action, then I think that a mobile phase composition with more than 50% water should be rarely needed. In gradient runs the major effect will be longer equibrilium times, but most polar compound will already elute at <50% water. When using methanol as the organic solvent this may well be at even lower water ratios. In the case with ion-exchangers (silica, amino etc) then electrostatic interactions may be superimposed and result in more of a mixed-mode retention behavior.

While using (A) 85% acetonitrile/15% water and (B) 100% methanol for a gradient run (for solubility reasons?) then the water content will actually decrease during the gradient. The net effect is hard to figure out.

I suggest:
(A) 85% acetonitrile/15% water (buffer)
(B) 60% acetonitrile/40% water (buffer)
Then you will cover the range where most of the resolution is achieved and separation of other components in your sample is likely.

(An interesting experiment....in case you need the methanol...
(A) 85% acetonitrile/10% methanol /5% water (buffer)
(B) 60% acetonitrile/35% methanol /5% water (buffer)


Yoshida, T., J. Biochem. Biophys. Meth. 2004, 60, 265-280.

[Uwe Neue]

I played with this mechanism first in 1973, trying to separate sugars and other things. I worked with silica, and polyamide-coated silica, and all kinds of mixtures of water, methanol, methylene chloride, ethylacetate and related things. No matter what you did, it was good to have water in the system, and if you had a true partitioning, it worked best. Otherwise, it did not make a lot of a difference what the specific solvents were or what the stationary phase was. Mostly based on this, I concluded that the underlying mechanism was some kind of a partioning mechanism. When I looked later in 1975 at sugar separations on an amino column, the stuff looked overall very similar, with a few little twists though.

For other compounds and different stationary phases, one can of course have added mechanisms, such as ion-exchange with the silanols or amino functions, if an amino column is used. This helps in the usefulness of a phase.

[Pistek]

Thank you all for interesting discussion.

In fact, I tried different gradients, also very similar to gradients suggested by Einar. I used methanol because I expected better signal in MS with methanol than with water. I forgot to say, that I added 0.1% formic acid (FA) in both A and B because it improves MS signal. Unfortunately, whatever gradient I tried, there was signal suppression at higher volumes of injected sample (i.e. above 5ul of sample injected on 50*2,1mm column). This might be due to lower capacity of HILIC as noticed by HW.

The point important to me was : which gradient should be more effective in separation of my compound (cystathionine) from interfering compounds causing suppression (buffer, salts, metabolites etc. present in assay mixture) -

1) A= 85%Acetonitril/15%Methanol/0.1% FA
B = 60% AcN/ 40% MeOH/ 0.1% FA
or

2) A= 85%Acetonitril/15%Water/0.1% FA
B = 60% AcN/ 40% Water / 0.1% FA

or something in between as suggested by Einar.

Thank you again for the contributions to this topic, I appreciate that I have learned something about mechanisms in HILIC.

[Einar Ponten]

The acidic condition, while adding formic acid, also minimise dissociation of silanolic groups and the ion-exchange interaction with the amino groups. I suppose that influence peak shape and retention?

We have not tested the use of non-aqueous (acetonitrile/methanol) eluents and I would speculate that the column will be more prone to overloading in absense of water in the mobile phase (no stationary phase water layer).

Consequently, I vote for the 2nd alternative with water (or as I suggested in post above keeping water ratio constant), Uwe's experience also suggest this acc to his last post.

Please let us know when you have tested.

[HW Mueller]

Does anybody have an "educated guess" (or better) on why water has this mentioned advantage (am especially interested in that overloading phenomenon). Could the postulated layer of H2O on silica be more stable, thicker.... than for instance that of MeOH?
I am presently grabbling with repeatability. It seems that even the H2O layer, or whatever, is very easily influenced by ions or polar substances. One can "chase these out" quite nicely with high ionic strength solutions, but then one has other ions there. Since I use phosphate buffers in the mobile phase (analytes are prolines) I had quite some success with "restoring" columns (for instance after ionic liquids were on them), regarding peak shape and retention time, by flushing them with high concentration of phophate. But some retention shifts are still a puzzle to me.

[Victor]

HWM- I do not have a problem with the notion that there is a thicker layer of water on a silica surface in HILIC than there would be if methanol is used. In normal phase chromatography, water is a stronger solvent than methanol and thus will be more likely to associate with the silica surface and with itself by hydrogen bonding.

I am interested in any views as to the amount of water held on a surface as a function of the % acetonitrile in the mobile phase. Clearly if 100% acetonitrile is used, there is no water layer. If the water layer is established at say 95% acetonitrile-5% water, what happens to the held water layer as the mobile phase is changed say to 80% acetonitrile-20% water? At some value the stationary phase must become saturated with water?

HWM- I am surprised you are using phosphate buffers. Don't you have problems of solubility, or are you using relatively low concentrations of organic modifier in your work?

[HW Mueller]

Victor, I didn´t mean to express doubts on that "notion", rather I just wondered whether there is evidence that this is behind the difference(s) to MeOH regarding overloading, maybe robustness.
On phosphates: There were strange solubility problems, especially since the modifyer must be EtOH. It sometimes took weeks for a precipitate to appear, other times they were there over night. The way to go was to lower the phosphate concentration, the results still are better than with other buffers.

[Einar Ponten]

Victor; I imagine that the stationary phase is saturated already at some few percent 3-5% of water in the mobile phase, especially when mixed with the non H-bonding acetonitrile.

[Uwe Neue]

I am not aware of any detailed studies, but I bet that a monolayer of water is accomplished very rapidly, even with as little as 1% water in acetonitrile. I am also not sure that the water layer thickness increases a lot as a function of the solvent composition. I would guess that once a layer has been established, its "thickness" does not change. However, different ligands on the surface are likely to bind water to different levels.

[HW Mueller]

I am surprised that I havn´t seen more helpful theory on this, that´s why I asked above. Water is confusing and fascinating. I remember that we put silica TCL plates in an oven at ~120° to drive off the water. Also water of crystallization stays in the crystal usually even more tenaciously. Yet when you dissolve such a crystal (for instance, NaH2PO4.2H2O) in water, the water molecules held by Na+ exchange very rapidly. Maybe the water on the silica surface has a status between that of water of crystallization and that of water held by an ion in solution. So the role of buffer ions on this system may be very complex..... no wonder I see such strange things.
Has anybody seen anything on whether water is salted out onto the silica by salts?