Chromatography Forum

Oops, I'm the world's worst chromatographer.

For ages I've had a variety of typical flow-rates and scout gradients using a 100*2mm 3u C18 column which I might use as a starting point on a novel analyte. The last couple of days I've been using a fairly wide gradient, 5-95% acetonitrile (versus 0.1% formic acid; this is LC-MS) over 20 min at 280uL/min, followed by 5 min at 95% acetonitrile because some of my analytes seemed to be strongly retained and I wanted to make sure they were really gone.

The peaks are coming out in the run after the one where they were injected. I know this happens in isocratic chromatography, but I haven't seen it in gradient chromatography until now. The strange part is that they don't come out particularly late; they're eluting at about 11min into the run, i.e. 10 min into the gradient given the dead-volume on the system. This definitely isn't autosampler needle carry over, as the peaks don't appear in the run where they were injected, only the subsequent run (there are no peaks in a further blank run; if I inject (1)blank; (2)sample; (3)blank; (4)blank, the only run with the expected peaks is (3)).

I always assumed that analytes were more-or-less stationary on the stationary phase until the gradient reached a percentage where they begin to partition into the mobile phase; then there's a period where they're partitioning and moving at less than full mobile-phase speed (separation is now behaving like an isocratic run) but as the gradient continues, they get to a stage where they're partitioned entirely in the mobile phase. As a result, I'd expect a compound that is capable of moving at all at only half-way up my gradient to be legging it along the column at full speed when running at 95% acetonitrile, so I can't see how on earth these peaks were retained for 5 minutes at 95%, but then able to leave the column at only 11min = 50% in the next run.

My only possible explanation is that they are sticking to the column by two reasons; hydrophobic interactions and something completely different. A relatively low percentage of acetonitrile is overcoming the hydrophobic interactions, but the other interaction is constant, so my analytes never partition fully off the stationary phase; instead they spend the greater part of the gradient behaving in an isocratic way. I don't know if this is possible? The compounds are nod-factors (http://en.wikipedia.org/wiki/Nod_factor) so they are very multifunctional; they've got a sulphate group, a long fatty chain, and a row of glucosamine sugars. I don't really believe my explanation because I'd expect the long fatty chain to act as a hydrophobic anchor and need quite a high percentage of acetonitrile to elute, and there's no particular reason why a sulphate should stick to a Luna C18 column, or a glucosamine for that matter, is there??

Incidentally, I don't think I'm alone in finding these things weird. Looking at literature, the solvbent mixes that people seem to need to elute them vary by about 50% organic between references using very similar columns, which worries me.

Any explanations/comments gratefully received...

If your RP column has polar-embedded groups and or/residual silanol, there is a chance that when you move above 80% ACN you enter HILIC/ion-exchange mode for your compounds. Some of the compounds will demonstrate U-shape retention relatively to amount of ACN. So at higher ACN you mind have stronger interaction (95% vs. 80%). Also if your compounds are ionic you might be enhancing "ion-entrapment" (my term) at higher organic.

Hi Vlad, thanks for that, yes, that explanation makes sense. So there's a fairly limited period when the analyte is free to move, before it gets into being retained hilic-style, and is fixed again until the next run (U-shaped retention)? I've overcome the problem at the moment by using a half-length column with increased flow, but keeping the original length of gradient (both time and percentage) so that more solvent passes down the column at each percentage, and the analytes don't have so far to move. I'm now getting all the expected analytes in the same run as the injection! Separation isn't marvellously good, but it's adequate for the current job, and peak-shape is reasonable, so I think I'll regard this as having been a useful learning opportunity, send the data off, and hope I don't have to optimise it...
Thanks again!

try to use stronger acid and reduce end point of ACN by 10%. Also you can try to replace 10% ACN with 10% of THF. Both of these actions might help in reducing retention of late eluters.

thanks again, I'll give it a go.

Do you normally run at 280uL/min?

What HPLC system are you using?

Are you using an injector program?

Hi lmh

Several years ago when I earned my daily crust doing HPLC we had a method for a large planar molecule whose name I forget that by accident was found to separate nicely from whatever else was there using isocratic acetonitrile as mobile phase. Even a little bit of water and everything eluted together more or less at the dead time. This was also on a Luna C18 column, so U-shaped or opposite-the-expected retention might be causing your problem.

How fast are you going back to initial conditions after the end of the run - and does that make any difference to the subsequent retention of the Nods ? (some sexy chemical signalling going on there, thanks for the link) - if retention is U-shaped, on the way back to initial conditions there should be a mobile phase composition that gets the peaks moving again, and then stops tham as the aqueous gets too high for them to move.

Peter

I went back to starting conditions over 1min, which is slow for me (usually I zip back as fast as I dare; someone once told me modern C18s don't need a gentle transition). From what you say, it sounds as though slow could be misleading, because I could have a peak eluting early in that second gradient, and think that it's strongly-hydrophobically-retained, whereas actually it was just coincidence that its U-shaped elution profile just got it nearly to the end of the column before organic got too high for it to be mobile.

Using a short column and high flow-rate I seem to be getting away with this now. Unfortunately I've got very limited amounts of sample or standard, so I can't play around much to see how U-shaped the profile actually is. I wish I could - it'd be quite educational.