Why so much salt in HILIC?

[Mattias]

I am about to set up a method using a ZIC-HILIC column.

The supplier (Merck) sent me a starting point for method development that says 70% acetonitrile in 100 mM ammonium acetate. I could not really get a good motive why the buffer concentration was so high. Do you know?

[Gerhard Kratz]

Reason is that this zwitter ionic HILIC column needs high amount of salt. Not so good for MS detection. I'm sure people from Sequant can explain it more detailed. Other HILIC columns have different modification groups and starting point is 70% ACN and 30% Water. I would recommend to test 3 different HILIC brands from 3 different manufacturers.
Good luck.

[Andy Alpert]

You should probably read Merck's instructions to mean 70% ACN/ 30% 100 mM ammonium acetate. The net concentration of ammonium acetate would then be 30 mM overall. Rationale: Below 20 mM salt in the mobile phase, the terminal sulfonate groups in the ZIC-HILIC ligand are not titrated and the material exhibits cation-exchange characteristics. At 5 mM salt or less, the cation-exchange capacity exceeds that of PolySULFOETHYL A, a material that's made to function as a cation-exchanger. The result is electrostatic attraction or repulsion of charged analytes. If your analytes aren't charged, then you only need 5-10 mM salt in the mobile phase, not 30 mM. If you want references to these properties, then let me know. I will say that all silica-based HILIC materials exhibit charge effects to some extent with salt concentrations below 20 mM, an observation that's well documented in the literature. It's more severe with some materials than with others.

[Vlad Orlovsky]

ZIC-HILIC does not have any cation- or anion exchange properties, because quaternary amine and sulfonate are two close to each other. It only can show slight ion-exchange properties with another quat or diguat or sulfonate/disulfonate. In order to test ion-exchange properties of ZIC-HILIC you need to run it with low ACN concentration. You can inject ionic compounds and use no buffer/additives in the mobile phase. Ionic compounds most of mono-ionic compounds will come in the void, or retention will come from residual silanols, since column is "acidic".
In our HILIC columns we addressed the issue of additional ionic interactions (both cation- and anion-exchange) and our HILIC/ion-exchange columns show ionic properties. This allows you to use less ACN which leads to better solubility of polar analytes in HILIC mobile phases.

http://www.sielc.com/Application-HPLC-S ... isc-N.html
http://www.sielc.com/HPLC%20Separation% ... ic%20Acids
http://www.sielc.com/Application-Simult ... olumn.html

[Andy Alpert]

Vlad:
This is what one might think by inspection of the functional group of the ZIC-HILIC material, which has an amine and a sulfonic acid in the same ligand, a zwitterionic combination. Certainly SeQuant advertised the neutrality of their material for the first couple of years, since zwitterions are supposed to be neutral. They quit claiming this once the evidence piled up in the literature that in fact the two charged centers do act independently under some conditions. References:
1) Y. Takegawa et al., J. Sep. Sci. 29 (2006) 2533: At 5 mM salt, glycans with (-) charge from sialic acid elute earlier from a ZIC-HILIC column than do neutral glycans, due to electrostatic repulsion. 20 mM salt shields the repulsion and the sialylated glycans then elute later than the neutral glycans.
2) Y. Guo and S. Gaiki, J. Chromatogr. A, 1074 (2005) 71: Increasing the salt conc. in HILIC from 5 to 20 mM causes a decrease in retention of (-) charged analytes on an amino column (by knocking out electrostatic attraction) but increases retention on ZIC-HILIC and several other materials (which presumably all exhibit some (-) charge themselves. Again, the salt shields electrostatic repulsion.
3) Y. Guo, S. Sriniviasan, and S. Gaiki, Chromatographia 66 (2007) 223: Very nice graph showing the increase in retention of a (-) charged analyte on ZIC-HILIC, silica, and two other materials between 5-20 mM salt and a levelling off of retention above that.
4) A.J. Alpert, Anal. Chem. 80 (2008) 62: Extensive discussion of electrostatic repulsion effects in HILIC, including the conclusion that it takes about 20 mM salt to form a complete electrical double layer on a stationary phase with some charge and thereby shield electrostatic effects [cf. the Merck recommendation that started this thread].
5) N.P. Dinh et al., J. Chromatogr. A, 1218 (2011) 5880: At 5 mM salt in HILIC, ZIC-HILIC has a higher cation-exchange capacity than does PolySULFOETHYL A, which is designed to be a cation-exchange material. At 20 mM salt PolySULFOETHYL A is still a cation-exchanger but ZIC-HILIC now does act neutral. This highlights the conditionality for titrating the sulfonate groups in ZIC-HILIC.
6) D.V. McCalley, J. Chromatogr. A, 1217 (2010) 3408: Similar observations about ZIC-HILIC acting as a cation-exchanger at low salt concentrations but not at higher ones.

The consensus about this subject is that all silica-based materials exhibit some (-) charge in HILIC under some conditions (pH > 4; < 20 mM salt) unless they have amine groups in the coating. This doesn't mean that HILIC is not useful; it just means that you have to take them into account and use more salt in the mobile phase if you don't want them to be present.

[Mattias]

Thank you for your replies, now I fully understand the recommendations! Reading between the lines, the ZIC-HILIC column seems not fully up to date. For many applications, 30 mM of NH4OAc would not be possible to use.

In my case this was not a problem since my molecule does not bear any charges. I could reduce the NH4OAc concentration to 10 mM. I got very nice chromatograms with a gradient from 95 - 80% acetonitrile.

My intention is to use the HILIC as a 2nd ID test. For a 2nd ID test the separation principle must be different from the first ID test. I have been in some discussions with Sequant that claims that HILIC is orthogonal to reversed-phase. Since the elution order is more less inversed, it is by definition not orthogonal...

But orthogonal enough for a 2nd ID test, I assume.