[spirochete]

When figuring out what solvent/s to use for column chromatography, I read that it's best to find solvent or mixture of solvents that will give an RF value of 0.25-0.35 for the desired compound, and have it separated from it's nearest neighbor by 0.2.

But my understanding is that when running a column one usually starts very non polar and slowly increases polarity, and then figures out where their product is through TLC, or other means if needed. 

How does one use the information gained by the initial TLC to optimize the column procedure? 

Also I understand Rf values on a TLC plate, but could somebody explain or give me a link to what an Rf value from a column means?

[azmanam]

start here (subscription may be required)

http://pubs.acs.org/doi/abs/10.1021/jo00408a041

[spirochete]

Wow so you just use whatever solvent mixture gives the right Rf.  It just requires careful control of flow rate and strict following of the procedure.

Under what conditions would you choose to vary the polarity of the eluting solvent?  If an Rf of 0.35 can't be achieved?  Or if the spots are too close on even the nicest TLC plate?

[azmanam]

Using the Still paper, I rarely have to use gradient elution.  In the column I ran yesterday, however, the product gives Rf 0.35 in 4% acetone:hexanes, and the major (and desired!) byproduct gives Rf 0.35 in 20% acetone:hexanes, so I ran 4% until the product was off (as monitored by TLC) then switched to 20% to get the byproduct.  The alternative is when the two compounds just don't give good separation at 0.35.  First thing I would do is slow the Rf down to 0.2 or so and lengthen the column past 6 inches.  If that didn't work I'd try other solvents. or perhaps gradient elution.

this might also help
http://www.chemistry-blog.com/2008/10/29/oh-and-what-do-you-study-in-grad-school/