I analyze nucleotides with tetrabutylammonium ions as ion-pairing agent on a C18 column and I have made an observation that the resolution of the peaks often decreases when I use buffers that I presume are more soluble in the stationary phase and wonder if there is a general trend that others have observed as well. I use a column which is supposed to have nearly no silanol activity (Kromanik Sunshell). Here are som examples of my observations:

-Phosphate buffers generally give good narrow peaks in this system but the plate number is slightly lower (reduced by 25%) when I use ammonium phosphate than potassium or sodium phoshate. Since I assume ammonium to a larger extent can penetrate into the stationary phase whereas the other two stays in the mobile phase, I was thinking this could be a reason for the lower performance with ammonium phosphate.

-When the pH is decreased to 3.4 (I normally use pH 6) some of the peaks get much broader (plate number decreased by more than 50%), whereas most get a bit broader (one peak became more narrow, though). Here I was thinking that a small proportion of the phosphate will be in the form of H3PO4 that can penetrate into the stationary phase layer and disturb the chromatography of some of the peaks.

-When I use potassium acetate (pH 5), the chromatography gets horrible (10% plate number or less). In this case, I assume that the proportion that is in the form of HAc can penetrate into the stationary phase and cause the problem. The problem is partially solved by mixing potassium phosphate 1:1 into the acetate buffer (same final pH) but the peaks are still much worse than with phosphate alone.

My question is basically if my observations has been observed by others and if the idea to try to use buffers that are fully ionized and have as little solubility as possible in the stationary phase is what is recommended by others as well.