I had good results with chloroform-methanol-ammonia mixtures for purifying amines and other polar tailing stuff on silica. Some people claim that anhydrous ammonia in methanol instead of aqueous conc. ammonia works even better – but I personally found no difference. Few good ratios of CHCl3-MeOH-conc.NH4OH are here: (ordered by increasing polarity)
80:35:4 (this is quite polar already – useful for strong bases and free aminoacids)
Apart from basic compounds, these solvent systems are also good for purifying free carboxylic acids and phospholipids. The acids are eluted as ammonium salts from the column – so one has to liberate the purified acids from their ammonium salt if the next step is amide coupling.
These mixtures were originally developed for TLCs and they work for columns also; it is important to slurry-pack the column in the same solvent mixture in this case. Gradient elution is usaly unnecessary because an ammonia gradient naturaly forms on the column during the course of elution.
One has to make sure that the used chloroform is not acidic – to avoid ammonium chloride and urea in the product. Replacing the chloroform in these mixtures with dichloromethane does not work very well here.
Ammonia sticks to silica quite strongly. When one is using ninhydrin detection with this system on TLC, it is important to bake the TLCs or use heat gun until ammonia smell is no longer perceptible. (The ammonia on silica surprisingly does not interfere with the Dragendorff stain so heating the TLC before detection is unnecessary)