Unexpected chromatographic behavoir

[RichardB]

I've recently been developing a separation and have come across some chromatographic behaviour which I have not seen before.

I'm using a C18 column with a mobile phase containing (0.17% TFA + 0.1% TEA in H20) + (0.05% TFA in ACN) and running a gradient from 10% to 30% to achieve the separation.

The unexpected behaviour comes when I vary the gradient time in order to optimise the separation. I have four peaks which all elute closely together. The resolution for the four peaks is around 0.8 to 1.2, however the elution order and the separation factor don't vary when the gradient time is varied?

A plot of log K Vs tG represents this very well and all four peaks give nice straight parallel lines all with slopes of 0.0015 and intercepts 0.6829, 0.6547, 0.6129 and 0.5872. To me this would imply that they were separated before I injected them on the column?

[DR]

Just a w.a.g. - add same conc. TEA to the ACN and see what happens... I would suspect that a pretty severe difference in pH between the phases could be a problem for you.

[Uwe Neue]

It is not that terribly unexpected. If you have compounds with very similar overall properties, the slope of the Ln(k) vs. solvent composition is similar. With other words, the compounds move parallel to each other in a gradient. In the extreme case of perfect parallelicity, you will get the best separation under isocratic conditions. You now can change the solvent or the temperature to see if you get any interesting effect that will help you.

[tom jupille]

RichardB, at the risk of sounding pedantic, k' (or K) can't be measured directly from a single gradient run (if you think about it, k' is a function of %B, which changes during the gradient).

If you're applying the isocratic formula (tR/t0 - 1), what you're calculating might be referred to as an "apparent k'", but all it does is obscure the fact that tR is a linear function of tG.

That doesn't affect the conclusions -- Uwe is right.