About TFA

[syx]

Thank you. Now, I could follow the conversation here… At least I have found 2 substances which have same abbreviation: TEA for triethylamine and triethanolamine. I had clash with formulation department about TEA.

Back to topic, if we need ion-pair property beside the acidity we could use perchloric acid or methane sulfonic acid. Which one better and how much the common concentration used? 0.1%?
If methane sulfonic acid (what is its abbreviation?) is used, could it be easily removed from the column?

[HW Mueller]

(Have been gone for a while)
Mark´s example is typical of what I have seen on such matters. Using the same org./aqu ratio would seem to indicate that either one or the other, or both were not optimized

[Mark Tracy]

I did get ambitious, and compared formic, phosphoric, methanesulfonic, perchloric, trifluoroacetic, pentafluoropropionic and heptafluorobutyric acids. The gradient elution was 5-52% acetonitrile over 15 min on a 4.6x150 mm 300A C18 column. The acid content was 0.01%, 0.03% or 0.1% (except phosphoric 0.015, 0.045, 0.15%). The column had previously been used for peptide analysis at low pH, so all the end-capping was gone. The test probes were the Sigma test mix H2016, and a mix of angiotensin variants; about 10µg/injection. The results were a bit complex, but a few general observations:

1- Ion pairing strength: H3PO4 ~ H2CO2 ~ CH3SO3H < TFA < HClO4 < PFPA < HFBA

2- Strong acids gave better peaks shapes that weak acids. The amount injected was relatively large, and "shark-fin" peaks in the formic acid eluents suggest mass overload.

3- Higher acid concentrations gave sharper peaks than lower concentrations. Again, a tendency toward "shark-fin" peaks at low acid.

4- Increasing acidity tended to make early peaks later and late peaks earlier.

5- Baseline irregularities: (214 nm) H3PO4 ~ CH3SO3H ~ HClO4 < TFA < H2CO2 < PFPA << HFBA

If I had more time to spend, I would investigate mixed acid systems. I've learned a few things from this exercise and I hope it is helpful going forward.

[Bill Tindall]

What is "ion pairing strength" and how is it measured?

[Mark Tracy]

The way I am defining "ion pairing strength" is retention time shift of the peptides relative to phosphoric acid. This is simply a ranking, not a numerical measure. For instance, for val-tyr-val and 0.1% v/v acid the retention times are:
H3PO4: 6.7
HCO2H: 6.6
CH3SO3H: 6.9
TFA: 7.5
HClO4: 7.7
PFPA: 8.4
HFBA: 9.3

The ordering is somewhat substance dependent, but this is a "consensus" ordering.

[HW Mueller]

Mark, if I understand this correctly: If one fiddles with the components (including org. modifier) until one gets an optimum for each conceivable combination one might get a completely different picture. In other words, TFA can probably be avoided in ~~99.9% of the cases?

[Victor]

Well done Mark for doing this work!

I do not see this as confirming that TFA can be avoided at all costs. The mass spectrometrists have to eliminate the non-volatile buffers such as phosphate (I guess methane sulfonic acid also?). It seems that formic acid can give mass overload problems (from Mark's study). Longer chain fluorinated acids appear to give more problems with baseline instability in UV , and there have been some doubts about the persistence of HFBA in MS systems (although see Kostas' post above).

The results of Mark's study seem to agree mostly with the article pointed out by HWM in LCGC May 2005 in the cases where similar eluent were used.

[HW Mueller]

In still different words (one example): It would not be surprising if one can get similar results with formic and TFA if one optimized both systems independently. For instance, instead of comparing the two acids at the same mol% (they will produce different pH) and at the same relative org. mod. amount (the elution power will be different) one could use the two acids at the same hydrogen ion activity and a bit more organic with the formic and possibly get nearly identical results?
Or stated completely differently: HPLC seems to be still at such a crude state that just about anything can be replaced by something else.

[Victor]

HWM- you are right to question the bases of these comparisons which are often rather strange.

Conventionally, 0.1% TFA or 0.1% formic acid (by weight) are used for peptide separations. However, these do not have the same molar concentration (TFA has about one third the molar concentration). However, the pH of these solutions is not grossly different. I do not think it is the pH difference of these solutions which is contributing to the differences Mark has shown. I think it is more likely to be the strength of these acids (as Mark already mentioned) i.e. their degree of dissociation which contributes to their ability to produce good peak shapes with basic peptides like angiotensin. I hope I have not contradicted myself here in that acid strength and pH are obviously related. However, I do not feel that a drop in pH of 0.5 units at a pH below 3 will make a huge difference to the surface of a modern RP column, or to the state of ionisation of a basic peptide like angiotensin.

[Mark Tracy]

I also compared 0.01, 0.03 and 0.10% (v/v) acids. The change in concentration did not change selectivity much for phosphoric, formic or methanesulfonic acids, although higher concentrations did improve peak shapes considerably. None of the non-ion-pairing acids fully resolved met-enkephalin from angiotensin-II, whereas all the ion-pairing acids did. The selectivity change versus acid concentration was much more noticeable for the ion-pairing acids, even reversing the elution order for one pair (leu-enkephalin & angiotensin-II).

[Bill Tindall]

This discussion could take the award for most intellectually stimulating of the year. First let me admit that I don't know anything about chromatography of peptides. But I presume these in question contain amines and carboxylic acids in addition to the assortment of amide bonds? And therefore keeping the acid undissociated and the amine protonated is desired?
In the concentrations of acetonitrile/water you explored there is little likelyhood of ion pairing in solution, at least for conventional ions (again peptide chemistry is unknown to me).

For a given molarity of acid there is going to be a substantial divergence of pH among these acids as the organic modifier increases(that is pH in the mobile phase where it counts, not in the water). I have not messed much in acetonitrile but in mostly methanol one could expect perchloric and the sulfonic acids to be vastly stronger than phosphoric or formic. Not to say that this fact has anytihing to do with explaining the retention data. Which leads to the question, what is the explanation for the retention time vs acid data, especially perchloric acid? Yes, "ion pairing" was the explanation, but fundamentally what is going on? It does not seem likely that ion pairs in solution at these high water levels, and therefore dielectric constant values, is the variable driving these differences.

[Victor]

Bill,

I am not sure that keeping the acid undissociated and the amine protonated is actually desired, but perhaps it is one of the better alternatives or the lesser of the evils.
At acid pH, there is less likely to be trouble from ionised silanol groups from the phase. Also, acid pH gives the best sensitivity in +tive electrospray MS: however, it is not impossible to work at higher pH on both of these counts.
I am absolutely sure that you are right in that ion -pairing in the mobile phases used for these analyses (with relatively low [ACN]) is going to be rather limited. Similarly, the divergence of the pH in the aqueous-organic mixture from that in aqueous solution is not drastic. However, the big question is, do those ion pairs that form transfer to the stationary phase? Is the peak shape markedly influenced by this ion pairing in the stationary phase, or is the good peak shape Mark has found mostly influenced by the higher ionic strength of strong acids (like perchloric acid or TFA) rather than their ion pair effect?

[Mark Tracy]

First of all, if any of you would like to repeat or extend the experiments that I did, I can give more experimental detail. There are several experiments that come to mind. Unfortunately, I have some other things that need my attention and my instruments this month.

One experiment worth doing is to use methanesulfonic acid to fix the [H+] while varying the concentration of ion-pairing acid. Perhaps add sulfuric acid to the comparisons. The ionic strength could be controlled independently with sodium sulfate. Various ratios of ion-pairing acid versus its sodium salt would allow pH to be adjusted indepently.

I used a gradient that is somewhat typical of a peptide mapping experiment. For more quantitative comparisons, an isocratic method would be preferred.

Different peptide probes have a different sensitivity to the ion-pairing effect; comparing the number of basic residues to selectivity would be helpful to understand the process.

A polymeric reversed-phase column might be useful to separate silanol effects. The C18 column I used had some history in this application, so any end-capping was long gone.

I have been using the term "ion-pairing" loosely since I don't have enough evidence to prefer a more specific mechanism.

Are there any grad students looking for a project?

[HW Mueller]

Also one would need to have a look at the organic modifier: which one is best with which acid, which org. modifier concentration is best with what pH, of which acid, etc.? The grad student would need to have lots of patience and time. If I look at Bill´s contribution it would seem there is not too much in prediction via calcs here . . . trial and error.

[Kostas Petritis]

I have played around in the past a lot with different ion pairing reagents for the separation of amino acids and peptides and I have tried more or less what it has been suggested here (i.e. polymeric columns (or even porous graphitic carbon which also does not have residual silanols, fluorinated columns, end-capped and not end-capped different bondings, different organic modifiers, buffering etc

So I have published some papers on this. The studies are not extremely systematic and of course you always publish whatever worked best (always for the sake of simplicity).

For example:
Petritis et al. Ion-pair reversed-phase liquid chromatography-electrospray mass spectrometry for the analysis of underivatized small peptides
JOURNAL OF CHROMATOGRAPHY A 957 (2): 173-185 MAY 31 2002

Petritis et al. Ion-pair reversed-phase liquid chromatography for determination of polar underivatized amino acids using perfluorinated carboxylic acids as ion pairing agent
JOURNAL OF CHROMATOGRAPHY A 833 (2): 147-155 FEB 19 1999

Also in reply to Hans comment, you can predict the peptide retention time in IPC. You may see:
Petritis et al. Use of artificial neural networks for the accurate prediction of peptide liquid chromatography elution times in proteome analyses
ANALYTICAL CHEMISTRY 75 (5): 1039-1048 MAR 1 2003

I have also submitted another article which is using complete peptide sequence information, and several other peptide descriptors resulting in much better retention time prediction (100% better).

Finally as I said before, most of the work proposed here has already been done, so I would suggest ambitious graduate students to wait before starting working on these

In my next message I copy paste titles and abstracts of published work that address part of what is been discussed... (I prefer to keep the messages short otherwise even I am bored to read through them...).