Changing flow direction during injection

[Mattias]

Hi,

This is just a crazy and not well thought through idea:

What would happen if I injected a compound with a gradient and then reversed the flow before the compound eluted and started another gradient in the other direction of the column?

The compound would at least experience more "cm of column" per analysis. I assume the separation would get very complex though. Has this been done before?

[lmh]

You can try it, at least with paper chromatography. Dot your felt-tip pen on the paper, let it run for a bit, dry the paper, and put the paper back in the tank upside down.
In isocratic conditions, what will probably happen is that the fast movers will head back off towards the start of the column, moving fast, but starting further away, while the slow movers will move more slowly. Unfortunately the likely result is that everything will arrive back at the start at the same time, but smeared out.
In gradient conditions, it's a bit messier, because the second gradient (backwards) is starting with a non-equilibrated column. But there is still a risk that you're washing things that have already passed the peak of interest back towards the peak, and you're washing the peak of interest back towards stuff from which it has already separated.
It's good to think about these crazy ideas. I was wondering some while ago about a hypothetical circular column, and whether it would have infinite length (because you could keep the analyte going round it for as long as you want). Of course it doesn't achieve the miracle I'd hoped, because after a while the fast-moving analyte will catch up with slow-moving things. You could indeed achieve enormous resolution, but not enormous resolution over a long gradient; peak capacity remains very finite! Before I worked that out, I'd sketched a way to operate a circular column which was quite cute. But it wasn't all wasted, because I realized the only thing it was good for was purification of a single peak that's hard to separate from a very near coeluter. And in fact the whole idea has been done already, had I bothered to Google. The simplest way to achieve the same effect is to collect the peak and reinject it. The place where this makes sense is preparative chromatography of a peak that overlaps with a contaminant, in which case you want to siphon off the part that didn't overlap, and reinject the rest of the peak, and repeat until yield is satisfactory. So crazy ideas sometimes bump into reality.

[Mattias]

Thank you! Your idea is as crazy as mine!

I will test the approach now with two compunds that are hard to separate with reversed-phase.

With a low pH of the mobile phase, the impurity elutes in the tail on the main peak. With a neutral pH of the mobile phase it elutes just before.

Thus, with low pH of the mobile phase, the impurity will move slighty slower than the main peak and will be slightly behind the main peak in the column.

But when I change flow direction it will instead be slightly ahead the main peak in the new direction. Then I start a new gradient with a higher pH where the impurity moves slightly faster than the main peak.

In my blurred theory, this should give more separation. And the tailing of the peaks should be lower since they will be refocused in the other direction?

The baseline will probably look very interesting though...

[Vlad Orlovsky]

I think that this will be a mess. When are you going to make decision on reversing the flow? What will happened to impurities which are eluting earlier or later after your target compound? I would not expect your peak to have a better shape, in my opinion it will get worse due to additional tailing, stopping the flow and restarting it. I don't understand the reason for doing this. Do you have issue with separating impurities? There are other ways to improve separation as you know

[Mattias]

Hi Vlad,

To be honest, I was slightly bored and want to test something completely crazy. There is usually a good reason why nobody else is doing this...

I will post chromatograms!