I've been trying to reproduce a published gradient method for the determination of certain vitamers in human tissue samples. The two solutions used in this gradient are: A) 33 mM phosphoric acid with 1 mM heptane sulfonic acid and 1.2% isopropanol B) 33 mM phosphoric acid with 10% isopropanol. The gradient starts off as 100% A for 2 minutes then goes to 100% B at 27 minutes. Ex = 287 nm, Em = 393 nm. At approximately 20 minutes into this gradient a very large peak occurs which just happens to correspond to one of the more important vitamins of interest (this peak appears in both samples and HPLC grade water samples). After exhaustively trying to find the source of the problem (peak shows up on both our old instruments and a brand spankin new one as well as a new column and with HPLC grade reagents, so its not an instrument or solvent contamination problem), i've discovered that the ion pairing reagent in buffer A is the source of the peak (I've tried heptane sulfonic acid, octane sulfonic acid, and HFBA all with the same results (but different elution times)). Analysis of the analyte using the same two solutions, yet excluding any ion pairing reagent, results in a clean baseline (i.e. no ion pairing peak). Unfortuneately the use of the ion pairing reagent is essential to this analysis as the analyte of interest is basic and will elute extremely early (with the other non-retained compounds from the tissue sample) if it is omitted. My question here is two fold: First has anyone seen this before and what is causing it. Secondly does anyone have any logical way to circumvent this ion-pairing problem?

We have seen considerable problems (peaks) in changing a mobile phase from one containing ion-pairing reagents to another one without. (Some even suggest never to use a column for non-i-p MP if it was once used for ion-pairing). A higher pH should give higher retention (less ionization). Also, the newer "aqu", etc., columns usually retain polar substances better.

The simple answer is to use the ion pairing reagent in both mobile phases, then the only difference is the propanol content which dictates the elution.
Also, check the column can handle the very acidic pH, would a phosphate buffer be more suitable?
HW Mueller mentions dedicating the column to ion-pairing only, which I would certainly endorse, if the column can be thought of as being in dynamic equilibrium, one must have the same concentration of I-P reagent, or one gets what you have found. I suggest phosphate buffer, say pH 3, and acetonitrile, rather than propanol. Try and match the solution the sample is dissolved in to the starting mobile phase.
Good luck
Bryan Wallwork