pH gradients have been used but have never been very popular. One reason is that they are difficult to control. If you think about the "titration curve" shape of a retention versus pH plot, nothing happens until you get close to the pKa of the analyte and then things change very quickly.
I'm not sure that the baseline would be any better than with a solvent gradient; many buffers have fairly high UV cutoffs, and many compounds have chromophores that are coupled to ionizable groups, so that absorbance spectra change with pH.