Sometimes we had some troubles with choosing the right column, its just a fact of trial-and-error in our lab.

I guess there must be some theory behind this stuff, for example: there are 2 peaks not properly separated and if we see the structure formule of both peaks we can say that we have to use a column with more selectivity or less polarity.

Is this right or is it just a trial-and-error?
If this is not, are there on-line courses or books or where can I learn this stuff?

Take a training course. Stuff like this depends on whether the difference is a nonpolar or polar difference. There are some basic charts, also in suppliers catalogs, about column selection; or call a supplier tech line, oftentimes they'll let you try a column for free, faster than waiting for them to actually try it out for you. Oftentimes we'll start out here on RP-18 using columns already used in-house. For added selectivity, we've added some THF to replace ACN. We've also used "intrinsically base-deactivated" columns which have the property of reversing elution order for some compounds which really helped us out.

My partner, Dr. Yuri Belov would be able to answer any specific questions for you regarding this matter. We are a small company that produces custom and standard HPLC columns, and whether it is with a column or just some advice, I am sure we could help.

ChromBA
814-867-4110

Assuming a reversed-phase system, there are really only six ways to change selectivity:
- change mobile phase strength (isocratic %B or gradient time)
- change temperature
- change mobile phase organic solvent (ACN, MeOH, THF)
- change pH
- put in additives
- change column chemistry.

Unless you are set up with a range of columns and appropriate switching valves, changing columns is the least convenient of these options. I would try the others first. Basically:
- Try changing %B (I'm assuming you are running an isocratic separation) by about 5% and see if the relative spacing of the peaks improves.
- If not, then go back to the original mobile phase and increase the temperature 20 degrees or so.
- If the spacing still doesn't improve try substituting MeOH or THF for the ACN (or vice versa) in your mobile phase (remember that the polarities are somewhat different, so it takes more MeOH or less THF to match the retention of ACN; there are nomograms for this in most of the standard textbooks, or e-mail me and I'll send you one).
- If your compounds have ionizable groups, moving the pH closer to the pKa of an ionizable group may help (this takes some effort, if you're not near a pKa, then pH has almost no effect; near the pKa, a small shift in pH can make retention -- and sometimes selectivity -- change dramatically).
- Additives that interact differentially with your analytes or compete with the analytes for active sites on the stationary can change selectivity as well.
- As I mentioned at the beginning, if you have column switching, then having a number of different columns on the system can make changing columns very convenient (perhaps more so than something like pH or additive changes). If you have to manually connect/disconnnect columns, this becomes the last thing to try (and then repeat all the others on the new column).

Aside from column changes, only one or two experiments are required for each variable; the entire process can be carried out in an afternoon if you plan your experiments ahead of time.

The best reference to all of this is the "Practical HPLC Method Development" book by Snyder, Glajch, and Kirkland. We just happen to sell it at LC Resources:
http://www.lcresources.com/resources/resbooks.html

;-)

-- Tom Jupille / LC Resources