by
lmh » Thu Oct 24, 2019 4:04 pm
Hm, I'll keep an open mind, but I'll admit I'm starting with an element of suspicion about how easy it will be to set up new methods using this. I'm worried because I don't understand properly how it works.
From the column's perspective, the analyte is badly retained, the IP reagent strongly retained, so if the two are exactly co-injected, the analyte ought to whizz off down the column and never meet the IP reagent. I think this is what's happening in fig 2a and 2b. In order to be retained, the analyte has to be in the same place as the IP reagent, which means we have to get IP reagent onto the column as fast or faster than the analyte. Perhaps it happens by overloading the column with IP reagent, so there's a front of IP reagent moving down the column at the speed of the solvent front, smearing IP reagent onto the column a bit like a knife smears peanut butter onto a slice of bread? This would contain a region of column that would tend to slow down the analyte.
The difficulty is in fig 2d onwards, where increasing dosage of IP reagent doesn't affect the retention time once a threshold has been reached. The peanut-butter-smear model suggests that there would be a band of column doing all the retention, and the longer the band, the longer the retention time, up to the stage where the column is saturated start to finish, and excess IP reagent starts to elute off the end with the solvent front - but we know they never reached this point because they were monitoring IP reagent too.
I worry a little because if there is this smear, then given that IP reagent moves only slowly down the column, while analyte moves much faster, there is a risk that any analyte peak that gets ahead could encounter the front of the wave of IP reagent, and suddenly whizz off. Perhaps this explains the broadening in Fig 2c?
But because of this, I'd like to see proof that no analyte managed to escape. I can imagine a situation where the bulk of the peak behaves itself, but a portion manages to get lost either by hitting the column early enough that it never saw any IP reagent, or by being so far on in the smear that it escapes early. On the good side, I can't see any obvious fronting, or an injection peak. On the bad side, I can't help noticing that in fig 4, the peaks in the presence of IP reagent look a bit smaller than those without; dark blue and purple are, if anything, narrower when longer-retained by IP reagent, and they're half the height. Red is less clear, it's the same height and I can't judge its width. Most things ionize better if they're in higher organic concentration, not worse. Of course ion pair reagents decrease ionization, but the IP reagent isn't supposed to be eluting as early as the peaks. Or does some creep off throughout the runs, just the bulk of it coming out in the wash?
Might work, but I suspect the does of IP reagent necessary (if the smear-idea has any truth) will vary between columns (obviously) and possibly depend on the effectiveness of mixing and diffusion between the autosampler and the head of the column, as well as on flow-rate, and it may be a balancing act between getting enough to achieve retention, and not so much that it starts to leak before the wash. I'd love to know how much safety-room there is between these limits. It'd be good to know how reproducible it is, both RT between runs on the same system, between days on the same system, and between different labs... interesting idea that I'll watch, but won't try out until I've seen someone else get results from it, without contaminating their MS! (I'm a coward!)
Interesting link, thanks Gaetan.