HILIC and RP - For Analysis of a Carbohydrate Polymer

[adam]

Hello

We are trying to analyze a carbohydrate polymer (hydroxypropyl methyl cellulose to be exact - also called Hypromellose). We have been trying using both HILIC and reversed phase, using an ELSD.

I have recently come across something in the literature that says in HILIC mode you need to have about 70% ACN to retain a carbohydrate polymer, but that it crashes out above 80% ACN. So basically you have to work between 70 and 80%. So I have a few questions:

- Would it be problematic to use a diluent that was 50/50 water/ACN. It is not very soluble with less than 50% water.

- Is it necessarily a problem if one starts at 95% ACN, for example, and the polymer crashes out. wouldn't it be OK as long as the polymer resolubalizes as the gradient proceeds and the water content increases.

- Any thoughts why this is not working for us in reveresed phase either. We've tried reversed phase starting at 95% water and then increasing the ACN content (a standard reversed phase gradient in other words). But this has not worked well for us either.

Any suggestions would be much appreciated.

Thank You!! Adam

[adam]

I forgot to mention. We used a Dionex mixed mode HILIC column for both the HILIC and Reversed Phase separations.

Thanks again
Adam

[Vlad Orlovsky]

Since you are using Dionex column I will not advice

[Andy Alpert]

Adam: That literature reference you cite is probably my paper: A.J. Alpert et al., J. Chromatogr. A, 676 (1994) 191. The ACN range of 70% (the minimum for getting retention) and 80% (when solubility starts to become a problem) was what we found for simple sugars, simple oligosaccharides, and the type of oligosaccharides that are attached to glycoproteins. If you attach unnatural functional groups such as hydroxypropyl- to a carbohydrate, then you might well extend the range of solubility to higher concentrations of ACN than 80%. You have to ascertain this on a case-by-case basis.

Starting at 95% ACN: No method is rugged if the analyte isn't fully soluble at the beginning. It shouldn't be necessary to use this much ACN to get good retention of an analyte as polar as yours. Also, if your starting mobile phase has, say, 80% ACN, then your sample solvent should have at least 70%. If the disparity in ACN content is greater than about 10%, then you run the risk of some of the sample migrating down the column for a while before being retained. Result: skewed peaks or even multiple peaks for a single analyte. It's analogous to injecting a sample onto a C-18 column when dissolved in a high concentration of ACN.

I recommend that you dissolve your sample in, say, 65% ACN and run it isocratically on your column with the same solvent. Presumably your polymer will elute early in a narrow range. Repeat with 70%. Retention will increase but the polymer will still elute in a reasonable time. You may start to see some separation of the oligomeric sample at this point. Continue this process until you've identified a starting % ACN that you can live with.

Finally: Some research groups and companies produce mixed-mode materials that do have valid applications. However, sometimes a company will develop a material for HILIC in which they have been unable to eliminate some hydrophobic character (perhaps because of constraints from the nature of the monomers required for a polymerization reaction). The resulting materials are sufficiently hydrophobic that at low % ACN they retain analytes via a hydrophobic interaction (the most hydrophilic materials do not do that). The marketing departments of those companies put a spin on the situation by promoting this property as a plus. Personally I take a dim view of having significant hydrophobic property in a material to be used for HILIC. The more hydrophilic a material is, the less %ACN you need to get a given degree of retention of an analyte in the hydrophilic interaction mode. In the case of carbohydrates, that is the difference between success and failure, and was one of the main points of the paper I cited above. At this point you can probably appreciate that in regard to your own application. So: To elaborate on Vlad's remark: If Dionex promoted this column expressly for applications like yours, then you should be contacting their technical service department to help you work out the solution. If they then can't get it to work for you, then you should send it back to them.

[adam]

Much Thanks for the feedback. I think you have identified the issue. I would like to ask a couple more questions:

- I'm wondering about your statement that everything has to be soluble at the starting conditions. Is this really always a requirement. For example, I think you can run PNA's and other non-polar compounds by reversed phase HPLC where you start at 95 or 100% water - where they are not soluble. This is an important issue to address because this is a very common situation in HILIC where the analytes of interest are not soluble in the amount of ACN needed at the beginning of the gradient.

- Secondly I would like to ask: any thoughts why this did not work for us in reversed phase mode either. The analyte should be readily soluble at the starting conditions for reversed phase.

Thanks again.

[adam]

Also, if we are not using an amino column, do we need to use high temperature to prevent anomerization. Maybe this is even more important with a polymer having numerous units that can undergo anomerization.

Thanks!!

[Andy Alpert]

Adam:

Anomerization is an issue only with a carbohydrate with a reducing end. With a polymer like a cellulose derivative, that will be just one residue out of many; small differences between large numbers. Don't worry about it.

Chromatography of an undissolved analyte: If you are seeking a benediction for so perverse a practice, then you are in the wrong forum. Maybe someone did it with PNA's out of determination to make reversed-phase work for an application that wasn't entirely suitable for it. They should be informed that HILIC exists. As for HILIC of analytes that aren't entirely soluble in the % organic solvent used at the start of the gradient:
1) Any analyte that's that polar is probably going to be well-retained at lower % organic solvent than the level that they're starting with. There is a tendency of people to use higher % organic solvent to start with in HILIC than is really necessary. Hardly anything needs more than 90% ACN for good retention if it's going to work in HILIC at all, and I'd certainly try much lower levels for a polar polymer like yours.
2) If you use a more hydrophilic column material, then you can start with a lower % organic solvent and still get good retention. So much for your dual-use column. Some of the materials most heavily marketed for HILIC are actually not the most hydrophilic ones.

Would you please describe just what did happen when you tried reversed-phase with your polymer?

[adam]

In reversed phase we didn't see it at all. It's possible that we are at too low a level. But perhaps you can offer a general comment - is there any reason why reversed phase would not work for this type of application (other than the possibility of insufficient retention).

Thank You

[Andy Alpert]

More details, please. What reversed-phase column did you use, what were the mobile phases, and what was the sample solvent?

[adam]

Sorry for delay.

The column was the same Dionex mixed mode column. It is designed to be run in reversed phase or HILIC (though I understand your not convinced of the latter). We ran from 5% ACN to 90% ACN in water. We tried plain solvents and also with 20 mM ammonium acetate in the mobile phase. The sample was diluted in 50/50 water/methanol and we injected 25 uL.

On a completely different note, I would also like to ask one more HILIC question. This one may seem a bit strange, but I do have a reason for asking. The question is: do water and ACN retain in HILIC. I realize that you generally have water and ACN in the mobile phase, but if I were to inject a molecule of water or ACN in the diluent - and let's say it was labelled somehow (radioisotope or...whatever), would that molecule experience some retention. It seems to me that they would since polar molecules are what HILIC is designed to retain. But I was not 100% sure. I know this may seem a bit out there, but I do have a reason for asking (it relates to a research project we are considering).

Much Thanks again for all of your advice.

[lmh]

(1) This is just a guess, but I would guess that water would be retained, at least a bit, and ACN would not. The reason is that the main mechanism proposed for Hilic is that the column binds a layer of water, in which the analyte partitions. Hilic columns have an affinity for water.

(2) I have heard that it is possible to get good results from Hilic using a gradient that starts at a concentration of acetonitrile where the analyte would, in free solution, precipitate. A column is not a free solution. Since a Hilic column binds water, it has a stationary phase consisting of column and water, and the analyte can be soluble in this local environment while the external mobile phase is quite unfriendly. Well, that's what the Phenomenex person told me!

(3) Have to agree, I don't like bifunctional hydrophobic/Hilic columns. Recently I had this the other way round, with a compound that bound to a reverse phase column at high aqueous and high organic. It took me a long time to understand what was happening, and that only with help from here. The symptom was that the peak came out in the run after where it was injected, but with a very good peak shape, and it continued to so this even when I put a very long high-organic wash in the method. U-shaped elution profiles are very confusing, and I'd never willingly buy a column that did it to me.

(4) Routinely I use 50% ACN as my injection solvent for Hilic of carbohydrates using a Luna NH2 column. When trying out methods for other polar compounds, I start with 50%, but also try up to 90%, looking at (a) peak-area - have I lost sample? (b) peak shape - was my injection solvent washing the peak a long way into the column?

[Andy Alpert]

Yes, water is retained by the surface in HILIC. A polar molecule or stationary phase surface will associate reasonably strongly with water. For example, in K2HPO4.3H20, the H20 is an integral part of the molecule and you will get this composition reproducibly. When a column of our PolyHYDROXYETHYL A (one of the most polar stationary phases for HILIC) is eluted with 80% ACN under conditions where the baseline is elevated, one routinely sees a negative peak at about 4 column volumes beyond the void. We take that to be the elution position of the water that was injected with the sample. The higher the % ACN in the mobile phase, the later the elution position of this water peak.

Following up on lmh's comments: When you have an analyte like this cellulose derivative, that has considerable polar and nonpolar character, then it's perilous to run it on a mixed-mode column that can retain it well at either end of the U-shaped curve. Adam: Dionex sold you this column. It's their duty to support it once it's in the field. Have you asked them about your problem?