The 0.1% TFA Revolution—A Sign of Chromatographic Laziness?

Oct 10, 2017
Volume 13, Issue 14, pg 11–13

Photo Credit: sirtravelalot/Shutterstock.comIncognito wonders if chromatographers are just plain lazy. 


Most of us working with high performance liquid chromatography (HPLC) will have a method that uses 0.1% trifluoroacetic acid (TFA) or 0.1% formic acid (and perhaps a conjugate base if you are very lucky).

I’ve seen a huge proliferation in these mobile phase additives in recent years, which I’m told have become popular because of their compatibility with liquid chromatography–mass spectrometry (LC–MS) and because they are more volatile and do not foul the ion source. So why do the vast majority of the methods I see that use these additives have ultraviolet (UV) detection? I admit that perhaps some (the minority I suspect) may have been initially developed or validated using MS detection for peak identification or tracking; however, I have an alternative theory regarding their proliferation—laziness combined with good fortune.

You will no doubt be aware that ionogenic analytes have a range of eluent pH values (typically +/- 2 pH units from their pKa value) over which their degree of ionization in solution will change, and accordingly so will their hydrophobicity and hence their retention times. The rate of change of degree of ionization (and hence hydrophobicity) will be greatest as one approaches the pKa value, and one tends to avoid eluent pH values close to the analyte pKa because small variations in eluent pH (poorly prepared eluents, acidified as a result of ingress of CO2 on standing, loss of volatile additives) will result in large changes in retention and, perhaps, resolution within a method. That is, the method will be less robust than we might have hoped.

Most of you will also be aware that at pH 2.1 (approximately the pH of 0.1% v/v TFA), most basic analytes will be fully protonated (charged) and most acidic analytes will be fully protonated (uncharged). Of course, for stronger acids and very weak bases this may not be true, but let’s work with this as a general rule of thumb. What I’m driving at here is that pH 2.1 is well away from the pKa of most analytes that we might encounter and as such, with all of our bases and acids fully protonated, there is a good chance that from the perspective of retention time and selectivity change as a result of small pH variations, we are pretty much in the clear, no matter what happens to our analytes. In fact, this also relates to why we frequently use these reagents to modify eluent pH without the use of their conjugate bases (ammonium acetate or formate for example) to form a “buffered” solution. There is no need to worry about small changes in pH because our methods are rock steady and, therefore, we don’t need to buffer. By the way, it really isn’t good to refer to 0.1% TFA as a buffer—it really isn’t, and neither is 0.1% formic acid.

And here is where we need the stroke of good fortune.

Early on in my career it would have been unthinkable to analyze fully protonated basic analytes at pH 2.1. Even if by some good fortune you got them to retain on the column, their peak tailing would be horrendous because they experienced severe secondary interactions with the anionic silica surface silanol species, and at pH 2.1 the bonded phase really wouldn’t remain attached to the silica substrate for long.

The good fortune that we have now is that column manufacturers have really upped their game and provided us with new generations of silica that cause very little peak tailing and are more resistant to pH degradation, whilst also giving us new phases, which are much better at retaining more highly hydrophilic species. The use of polar endcapping reagents, embedded polar functional groups within stationary phase ligands, dipolar phases such as pentafluoro phenyl, and a host of other “tricks” such as low ligand density phases means that an ionized basic moiety doesn’t spell the end for a reversed phase approach to the analysis of fully protonated basic analytes. 

pH robustness and the ability to retain more polar analytes with good peak shape—what’s not to love about 0.1% TFA? Well quite a lot actually.

First, I believe the approach of using low (TFA, formic acid) or high (ammonia) eluent pH values removes one major selectivity variable from our developer’s armoury. We really only have three or four selectivity variables to play with: type of organic modifier (typically limited to methanol or acetonitrile), gradient steepness, stationary phase chemistry, and temperature (which is rarely used in practice). pH has traditionally been a major development variable to adjust the separation selectivity of ionogenic analytes, but maybe I should face the fact that it’s easier (lazier?) to screen a whole bunch of stationary phases on an automated column switching HPLC system than to have to make up a bunch of eluents at different pH and experiment with the selectivity, even though this can also be done quite easily in an automated fashion with modern instruments. 

I still believe that the possibility to “finesse” a separation using pH is much more effective that hunting to find the ideal stationary phase and then playing with the gradient steepness to perfect the selectivity. It’s possible to develop very robust methods using pH control between 2 and 10, but perhaps modern method developers lack the skills or knowledge to determine analyte pKa values and then fashion a separation using an eluent pH that isn’t too close to any of them. Or perhaps developers assume that the guys in the “routine” laboratory are inept and they need to find new ways of preventing them screwing up, and eliminating pH as a susceptible variable is just one of them? In my darker moments, I also wonder if the proliferation of hydrophilic interaction liquid chromatography (HILIC) as a technique for polar analyte retention isn’t in some way related to an assumption that analytes which do not retain well in reversed phase mode couldn’t be made to do so with a little pH control to ion supress them. Have we lost focus on ion suppression as a chromatographic technique and turned too readily to HILIC mode instead?

I know some of you will be thinking about the ion pairing nature of TFA and that ionized bases will be in their neutral “ion-paired” form in solution, and therefore retention in reversed phase will not be an issue. However, unless we know something of the fundamental mechanism of ion pair chromatography, this can also lead to non-robustness of the method because the nature of the ion pair and its concentration in solution are variables that can also affect retention relative to non‑ionogenic analytes and therefore selectivity may be affected. Further, whilst TFA is a strong ion pair, what of additives such as formic acid, which is a much weaker ion pair—can we rely on a robust ion pairing mechanism in this case? 

There is also the further complication of TFA adsorption onto various parts of the HPLC system including any Teflon tubing, metal parts, the tubing inside the online degasser, and of course the stationary phase at low percentages of organic. As the UV cut off for TFA is relatively high (210 nm), one may see a changing baseline position during gradient analysis as the TFA concentration changes (unless constant ionic strength is maintained) or as TFA is released from the stationary phase as the eluotropic strength of the mobile phase increases. TFA also causes ion suppression (reduced sensitivity) in positive ion electrospray LC–MS. As a caveat, the reason that formic acid is preferred for LC–MS is because of its lower ion pairing strength, which allows it to “decouple” from the analyte ion during liberation into the gas phase in the atmospheric pressure ionization (API) interface. 

As TFA, formic acid, and ammonia are moderately volatile, we risk losing them from the eluent on standing, and therefore run the risk that pH and ion pair concentration (depending upon the mechanism of retention) may change and lead to robustness issues.

Of course, much of what I say above is tongue in cheek and deliberately provocative to allow readers to reflect on the changing nature of chromatographic science. Why would we not take advantage of more modern stationary phases and harness the automation available to us to develop more “generic” methods to improve robustness? My point is this: Please don’t forget that pH control is, and always will be, a primary variable for the optimization of selectivity when separating ionogenic analytes, but one does need insight and good practice to take this approach. I’m very much in favour of saving time in method development and making methods more user friendly—it allows me more thinking time for the separations where these generic approaches don’t work!


Contact author: Incognito
E-mail: [email protected]

lorem ipsum