Hello,
I have recently tried scaling different HPLC analyses between 250 x 4.6mm and 150 x 2.0mm columns, using a scaling factor for the flow rate based on the formula flow2=flow1 x length2/length1 x (radius2/radius1)^2. By doing this, I get the same retention times when scaling the flow rates appropriately for the 2 columns, but peak widths are no narrower and resolution is compromised for the narrower bore columns (I am observing loading limits for the narrower column and keeping injection volumes at 5 uL for both). I thought this might be due to extra column dead volume, but even at the same flow rates (the lower rates), the resolution is better on the larger columns (with markedly longer run times). My question is this: is there an effect of increased cross sectional area of the stationary phase in a given column diameter that accounts for improved peak resolution in wider-bore columns? If not, what else might be the reason for the lower resolution in the narrower-bore column?

thanks
Tony

You got several things that are happening. The scaling that you have applied - scaling of the flow rate by column volume - will get you the same retention times (definitely in isocratic analysis, with some potential differences in gradient analysis, see below). In order to get the same resolution, you also need to get the same peak width. This is not so straightforward for several reasons, for one thing, the column plate count is not likely to scale that easily (it would scale this way only for very large velocities, i.e. short run times, which are not likely on a 25 cm column). The second effect is the increased extra-column bandspreading due to the fact that the column volume is much smaller. You can tell if this is an important factor, if your plate count increases with the retention time (as a rule of thumb). I bet that part of the problem is there. The extra-column bandspreading can be reduced by using shorter connection tubing, smaller tubing i.d., a smaller detector cell, and a smaller injection volume.

If you are working with a gradient analysis, some of this extra-column bandspreading is going away (the pre-column bandspreading). Now, your column performance could be much closer. You still have some differences in the time that the gradient needs to reach the column, which depends on the gradient delay volume of your instrument. If you were running gradients, your chances of getting the same resolution on both columns is higher, but the retention could have shifted somewhat, depending on your instrument.

If you want to know how all of this works, I recommend the chapter on "Column Design" in my book on "HPLC Columns". It covers alot of these things.