'dancing' RT

[sergioanis]

hi all,
I'm having an interesting situation here. I was given 3 samples of a dye from different manufactures to check the purity by UV. The dye is IR-783 (more info: http://www.sigmaaldrich.com/catalog/Pro ... KEY&F=SPEC )
I used a very long gradient (had plenty of time) 5-95 ACN:MeOH 1:1 in H2O (0.1M H3PO4) on a C18 column. I found that the RT of the 3 samples were different (up to 0.5 min). I used a different column, different gradient, different organic phase (ACN alone). Still, all of them coming at different RT.
So, I moved to a different instrument (which I just used for another project with no problems). Still, same results.
So, moved to another instrument (have plenty of them . Check the full UV-vis spectrum and they are identical (which doesn't mean the structures are identical). Then, I ran 2 consecutive runs of the same sample and found that the RT were different. Took a vial from some old project and did 2 consecutive runs and got 2 perfect identical chromatograms.
Now I have to conclude that the IR-783 has an erratic behavior? Does anybody seen this type of behavior before?
BTW, although the manufactures claim "90% dye", I don't see any other peak on the UV-vis range. One interesting difference in the manufacture specs: 2 of them have melting point: 209-212C, the other one 256-259C. May be they are different crystallographic forms?, although there is no observable difference in the powder aspect. Anyway, a different crystallographic form shouldn't matter when they are in solution, right?
I really appreciate your thoughts
thanks,
Sergio

[lmh]

Try running a mixed sample containing the dyes from two manufacturers. If there is something in the preparation that is altering the chromatography and giving different retention time, but the dye is fundamentally the same, you will get one peak. If the dyes are actually different structures, you will get two peaks.

[sergioanis]

that's a good point.
I mixed 2 at the time and ran them getting always one peak

[tom jupille]

I ran 2 consecutive runs of the same sample and found that the RT were different.

That's a red flag. If this were my problem, I would run the same sample multiple times (e.g., 5 injections) to get an idea of the retention time repeatability. If this is too time-consuming, the shorten the gradient time (longer is not necessarily better). A reasonable value for a 90% range would be about 25 x t0 (about 20 minutes for a 150 x 4.6mm column at 2 mL/min).

That melting point difference is also unusual. I agree with you that it *could* be different crystal forms (which would end up being the same in solution), and although my organic chemistry skills are lamentable, I can't visualize anything complexing with that structure that would not affect the spectrum.

[Consumer Products Guy]

I'd increase the equilibration time between injections, at least until I could rule that out as being a factor...

[sergioanis]

I double the eq time without any change in the outcome (10 min for a 3u 4.6x100 column with a flow of 1ml/min; I think it's more than enough). Unfortunately, I can't keep 'playing' with this sample any more (at least for the next 2 weeks).
thanks to both of you for your input
have a great weekend

[Camisotro]

Do you have pKa information for the dye?

If your mobile phase pH is near a pKa of the analyte, then small changes in pH will create large differences in % ionization which in turn affects retention behaviour. If you can adjust the mobile phase to a region where the % ionization of your analyte does not change much with respect to pH, that would stabilize retention behaviour - if this is what the problem was in the first place. =)

Edit: Looked at the structure... looks like the sulfonic acid pKas are low so that both would be fully ionized at any pH above, say, 2 or 3. The conjugated nitrogen base however would probably not protonate at any chromatographic pH.

I see your mobile phase is 0.1 M phosphoric acid. So it's quite possible your pH is in the "trouble zone" where the degree of sulfonic acid ionization is quite variable. The only way to get this compound in a "neutral" (zwitterionic) form would be to protonate one (but not the other) sulfonic acid to cancel out the quaternary nitrogen. But both sulfonic acid groups probably have around the same pKa so this would be a challenge. If you raise the pH to, say, 7, then it's a singly-charged anion that won't retain on C18 (but try it out just to be sure that's what you see).

This suggests that maybe your entire column chemistry just isn't right for this compound. Perhaps you need a column designed to retain anions and a mobile phase to suit it. Or alternately you might try an ion-pairing reagent designed for acidic samples and use that in the mobile phase at a more or less neutral pH.

[sergioanis]

I don't have the pka but I was told that is very acid, in fact I wasn't sure if RP would work.
I thought that when the pka is close to the ph you get a bad peak shape (which is not this case) but I didn't know/remember that the RT could be influenced too. I just came back to chromatography after a few years into lab automation and NMR, so I'm a little rusted
Unfortunately, I wont be able to tested because: I have other projects and none to delegate, I got the results the chemist needed (even with the variability in the RT since I was able to rule out that they were different compounds), and the chemist already retreived his samples and I didn't realize that I didn't keep anything (my mistake).
So, we'll have to keep this in a 'theorethical' mode.
thanks,

[Camisotro]

Fair enough. It provided an interesting thought experiment for me in any case. Glad the results were acceptable.

[HW Mueller]

A theoretical mode would be that, practically, one can not protonate a sulfonic acid in a HPLC environment, and that a good buffer will allow all pH which are compatible with the column.