Two HILIC Questions

[adam]

Hello All

I wanted to pose two questions regarding HILIC analysis. The first has to do with buffer solubility. We are currently doing some work and one of our mobile phases is 100% ACN (there is a reason that it has to be 100%....long story). The separation does not work so well unless there is a significant amount of buffer present (at least 10 mM), but most buffers have limited solubility in pure ACN. The two buffers I am aware of that would be the best bet here are ammonium formate and ammonium acetate. Does anyone have any idea which would be more soluble in ACN, or if there is another better alternative.

My second question is completely unrelated. I understand that in a HILIC system the time frame of ion-exchange can be very slow. Does this suggest that when we do HILIC we should use a diluent that puts the API in the same state as it will be in the mobile phase. Here's what I'm thinking. Suppose we inject something in it's neutral form, but upon injection it wants to convert to the acid or base form. If the time frame of the conversion is slow we may get poor chromatography since it is in one form for part of the separation, and in another form for the rest of the separation (this usually leads to smeared peaks).

Thanks very much for any feedback.

[Petrus Hemstrom]

For the first question are you looking to start a gradient at 100 % AcN or do you only need to have the sample in 100% AcN?
Starting a gradient at 100% AcN will be difficult to get equilibration and stable chromatography.

I have never had any problems with pH if the sample solvent is pure acetonitrile or pure water no matter if the eluent pH is 3 or 11 but in a strongly buffered sample solvent is not a good idea to have a mismatch between the pH of the sample and the eluent.
Generally acid base reactions are fast (in aqueous systems at least, dont really know in pure AcN)

[Andy Alpert]

Actually there are two aspects to the ion-exchange issue. The first is the kinetics for a charged analyte to bind and then dissociate from a charge on the surface. This is pretty much the same in HILIC and regular ion-exchange. The second is the rate at which a charged analyte can exchange counterions during the chromatography (which seems to be what Petrus is addressing and what Adam is apprehensive about). Adam is correct to be concerned. Take a look at Fig. 14 in Anal. Chem. 80 (2008) 62, per the following link: http://pubs.acs.org/doi/pdf/10.1021/ac070997p
This is an extreme case of the kind that Adam described; a charged analyte that elutes in HILIC in two well-separated peaks with a continuum between them. One peak corresponds to the analyte with a counterion different from the one in the mobile phase while the other corresponds to the analyte molecules that exchanged their counterions for the one in the mobile phase at the start of the migration through the column. The two ion pairs differ in polarity (which is something that HILIC is sensitive to) and hence migrate through the column at different rates. The continuum consists of molecules of the analyte that started with one counterion and exchanged it for the other counterion sometime during the migration. A less severe case would feature a badly skewed peak rather than two separate peaks (Adam's "smear").

The example in the figure was obtained with 10 mM salt in the mobile phase, a common concentration in HILIC. Obviously counterion exchange is quite slow under those circumstances. If the concentration in the mobile phase had been, say, 100 mM, then the exchange of counterions would have been 10x faster and the problem less severe. A more convenient solution would be to make sure that charged analytes are paired with the same counterions as the one(s) in the mobile phase before injection. If that's inconvenient to implement, then I'd recommend adding to the sample a solution that contains a lot of the salt that's in the mobile phase. That way, the counterion exchange can be completed prior to injection.

Incidentally, that 2008 paper provides a number of reasons why some HILIC analyses won't go well unless there's some salt in the mobile phases.

Regarding Adam's other issue: No, don't try dissolving a salt in 100% ACN. Either it won't dissolve or you might get phase separation, per Per-Åke Albertsson's work. You would be much better off if the reservoir for Mobile Phase A contained the solution that you actually plan to use for Mobile Phase A, i.e., with some water. That should alleviate your solubility problem. It will also preclude any problems from trying to mix, online, the low-viscosity ACN with the higher viscosity Mobile Phase B, and will permit any degassing to occur offline. Please answer the following questions:
1) Why do you have to start with 100% ACN in the reservoir? Let's hear your long story.
2) Salt selection: Which is more important: volatility or transparency at low wavelengths?

[adam]

Andy, thanks for your response.

Something you said caught my attention. You mentioned that your paper discusses a number of reasons why HILIC does not work as well without some salt in the mobile phase. Now I've seen several 'guidebooks' that say to use acid or salt, depending on the situation. So are you saying that you prefer to use salt in all cases?

Is there no salt with more organic solubility than the two I mentioned.

Thanks again

[Andy Alpert]

By no means. There are applications where an unbuffered acid is the best additive. An example is Wen Ding's paper in 2009 in which he added TFA. This formed hydrophobic ion pairs with the basic residues, rendering them appreciably less hydrophilic and tuning down their normally prominent contribution to retention in HILIC. The result was to tune up the percentage contribution of carbohydrate sidechains to retention. By this means he was able to isolate glycopeptides selectively from a tryptic digest. That said, TFA decreases retention of peptides in HILIC and is to be avoided unless you're trying to do something selective like that. As a gross generalization, salts afford better results and less problems than do unbuffered acids for certain classes of applications such as peptide separations. That concerns the chromatography. If your analytical method requires a volatile mobile phase, then there are additional concerns besides the chromatography.

Whatever you do add, add enough of it to substantially titrate the charged sites on the surface with a layer of counterions. With a particularly highly charged surface, 20 mM salt overall in the mobile phase is usually sufficient.

Ammonium salts are not among the most organic-soluble ones. The analogous salts with triethylamine (triethylammonium acetate; etc.) would be appreciably more soluble in high-ACN solvents. I like to use triethylammonium phosphate because it's transparent at 220 nm (prepare it by adding triethylamine to phosphoric acid in water until you reach the pH you want). Alternatively, you can make the anion the organic-soluble part of the salt. One of my favorite salts for the purpose is the sodium salt of methylphosphonic acid. Be advised that selectivity will be quite different with different salts. That paper from 2008 that I cited in my last posting in this thread has Table 1 comparing retention times obtained with triethylammonium phosphate and with sodium methylphosphonate. With charged analytes, these can be remarkably different.

[adam]

Thanks. This is quite helpful. Just to be completely excruciating, let me follow up once more and see if I've got it right:

- Triethyl ammonium acetate --- volatile but not highly transparent at lower UV wavelengths

- Triethyl ammonium phosphate --- Not volative but transparent at 220 nm

- Sodium Methylphosphonate --- Not volatile but reasonably transparent

Last question: I tend to use the salt forms and just weigh them in to get, for example, 20 mM. I do this because it is easy and I generally don't feel like I need to have a buffer system. But when putting these ingredients into solvent systems with only small amounts of water is somewhat different and...I'm wondering if there is something I don't fully appreciate. Should I be adding some acid or base as a general rule??

Any feedback is much appreciated!!

[lmh]

Andy, I just wanted to say thanks for that remarkable link. I had a peak almost exactly like fig 14 in Hilic a while back, and no one could work out what was going on.

It was rather a sad story at the time: the bad peak was one analyte in a whole set of related things. Unfortunately client, an academic who had asked me to do the work for free on the understanding there would be many paid-for samples when his grant came in, promptly gave all my results and methods to a friend in another lab, and asked if he could improve on the separations of the remaining peaks (unfair, since given a full write-up he could hardly do worse!). Friend of course informed me of what client had done, but I still regard this as deeply ungentlemanly, and frankly theft of IP (our guest-book clearly states that visitors must not disclose anything they learn while working in/visiting our labs without explicit permission). As a result I'm now torn about whether to pass on your reference to friend!

Incidentally, when trying out different Hilic methods, I make up an aqueous buffer at 10 times the intended concentration, and then hand-mix this with acetonitrile/water to make buffers A and B for the aqueous and organic ends of the gradient, checking of course for precipitation. I often make concentrations slightly beyond what I think I'll need, in case the gradient needs modification to deal with very late/very early eluters. I'm not very theory-aware, so unfortunately for me it's normally a matter of dissolving the sample in something as organic as possible (but now I will think about counter-ion/buffer!), and running a wide gradient to see what happens...

[Andy Alpert]

To LMH: Do pass on that link to your friend. I paid good money to Analytical Chemistry so that article could be downloaded for free.

To Adam: LMH's method is what I do as well. I prepare the buffer as a stock solution, sufficiently concentrated so that I can prepare a solution that's mostly organic and still have the concentration of salt I want. If I'm aiming for a pH at or below the pKa of the salt, then I dissolve a weighed amount of the acid in water and then add the base, with stirring, until I reach the pH I want. Then filter, pour into a volumetric flask, and dilute to the calibration mark. Thereafter I don't bother measuring or adjusting the pH, especially after using the stock solution to prepare a solution that's mostly organic (in which the dissociation constants of electrolytes will shift, as will the measured pH). I express the salt concentration as the concentration in the resulting overall solution, not just the concentration in the portion that was aqueous. Be sure to specify that you're doing this when writing up a method so others can reproduce your results.
Your summary about the properties of the salts is correct. Do use HPLC-grade H3PO4 and a really pure grade of triethylamine. As for methylphosphonic acid: Sadly, nobody makes an HPLC grade of this at present. The reagent from Fluka used to be the purest until their acquisition by Sigma-Aldrich.