The retention behavior of the Metformin is much the same on both the bonded phase and bare Silica columns, that is, aqueous normal phase with retention as the acetonitrile percentage rises above 50%. Both columns exhibit little band broadening, but both also exhibit some tailing of the Metformin peak. The similarity of these responses above suggests that the same underlying mechanism govern this retention response.

This suggests that the silica hydride surface, which is the common factor to both these columns, is a major selectivity factor for the aqueous normal phase response of Metformin.

For Glyburide being analyzed in the same injection, there is a radically different response between the bare TYPE-C Silica and the bonded phase.

With the bonded UDC-Cholesterol phase, this compound exhibits little band broadening and is symmetrical, plus it is significantly retained at 50% acetonitrile. It is even retained later that the solvent front at 80% acetonitrile(to aid comparison, we have repeated this slide.

But for the bare Type C Silica the results are quite different. At over 30% acetonitrile the Glyburide elutes on the solvent front. When retained at 27% acetonitrile, even at retention time of 1.7 minutes very significant band broadening occurs, however the peak is symmetrical.  

This suggests that the bare Type C Silica whilst exhibiting aqueous reverse phase response for Glyburide, does this at a radically lower percentage of acetonitrile than the TYPE-C UDC-Cholesterol. In chromatography it can be generally assumed that if mechanisms work at a 50% organic solvent difference that one would totally dominate the other.  

It would thus appear that for the TYPE-C based UDC-Cholesterol, the Glyburide retention is mainly a bonded phase effect. In the bare TYPE-C Silica, a much weaker reverse phase effect can be noted based on the silica hydride surface.