Optimum flowrate
Answers

Basically, there is an optimal linear velocity for a given separation. This optimal flow rate depends upon the analyte, conditions and column. Change one of these parameters (e.g., temperature, organic modifier, particle size/chemistry, column dimensions, etc) and you need to reconfirm that you are at the optimal linear velocity for this 'new' separation.
You mention van Deemter curves. Assuming you are familiar with them, you know that there is an Aterm, Bterm and Cterm associated with the van Deemter equation. At very low linear velocities, the Bterm (axial or longitudinal diffusion) can begin to influence the observed plate height (H). The Bterm is affected by analyte k', Diffusion Coefficients (D_{S}, D_{M}) and Obstruction Factors (g_{M}, g_{S}). It is independentof particle size: B = 2g_{M}D_{M} + 2g_{S}D_{S}k'. See (J. H. Knox, J. Chrom. A, 831 (1999) 315) for more information. So, if you increase the flow rate, the Bterm's influence on H decreases; decrease the flow rate, the Bterm's influence on observed H increases; and if you REALLY decrease the flow rate, the Bterm can have a large influence on the observed plate height.
So, to answer your question, if you go too slow, you are eventually running at suboptimal linear velocities and can lose efficiency (increase H) due to the Bterm. Conversely, if you go too fast, you can also lose efficiency but for different reasons (Cterm). Other disadvantages to running at suboptimal flow rates is unnecessarily long analysis times, sensitivity loss (peak height) and possible solvent waste. The only true way to see what the lowest optimal linear velocity is for your separation is to generate your own van Deemter curve for your analyte(s) of interest. You then balance that with what I just mentioned.
Good luck!!
Doug
0