Why can’t I reproduce my TLC separation using flash column chromatography?

You have performed your synthesis and now it is time to purify the reaction mix. You have used thin-layer chromatography (TLC) and see a separation but when you try to purify with flash column chromatography, you can’t get the target compound separated from an impurity. So, what is happening (or isn’t happening)?

In this post I will give some input on why some separations are not transferrable from TLC.

So, what is going on in this scenario?  Well, the primary factor influencing chromatographic separations is selectivity so if you are not achieving your separation goal, the separation selectivity must be improved.

Ok, so how do you improve selectivity? By addressing the factors which influence selectivity – solvent choice, media choice, gradient slope. As I have discussed previously, solvent choice is critical for separation optimization, especially with normal-phase silica purifications.

If methanol is used as the polar modifier, then it may be too protic for the flash method even though your compound may be separated by TLC. The reason for this particular problem is that with TLC, the developing solvents migrate at different rates with polar solvents being more retained than the non-polar solvent.  With flash column chromatography, this also occurs but since the flash column is typically pre-equilibrated prior to running the gradient (TLC plates are used dry) the actual flash chromatography gradient is different than that seen using TLC.

So, should you then use a non-equilibrated silica column for flash when using a methanolic solvent?  Though an option, the amount of heat generated during purification with a non-equilibrated column causes even bigger issues (solvent heating, retention loss, separation loss) so I do not recommend this approach. What I do suggest is looking at an alternative to methanol.

In a previous post I have discussed using acetonitrile as a replacement for methanol.  While both are polar, acetonitrile is aprotic which means a TLC method using it will have a better chance of successful transfer to flash than methanol.

To emphasize this point, Figure 1 shows the TLC results of three sample components (a mix of methyl + butyl paraben and 4-methyl-4(5)-nitroimidazole)separated on silica plates with 10% MeOH/DCM and 20% ACN/DCM. Though both TLC solvent systems separate the three compounds only the ACN/DCM solvent blend separates all three components when mixed together using flash chromatography, Figure 2.

Figure 1. TLC runs comparing separation capabilities between DCM-MeOH (left) and DCM-ACN (right). With each TLC plate, a mix of methyl and butyl paraben were spotted on the left and 4-methyl-4(5)-nitroimidazole spotted on the right. Both elution solvent systems separate all compounds.
Figure 2. Comparison of a DCM-MeOH separation (top) and DCM-MeCN separation for a 3-component mix of butyl paraben, methyl paraben, and 4-methyl-4(5)-nitroimidazole. The data shows a selectivity and separation improvement with acetonitrile but co-elution when methanol is used.

So, with the results above in mind, if your TLC method is not reproducible using flash, consider modifying your elution solvents to alter elution selectivity.

Have you experienced a non-transferrable TLC method?  Tell us how you solved the challenge.

For more information on flash method development and optimization, click on this link and register.

Click the link below for access to a survey regarding this blog; I’d love to see your thoughts. http://www.surveygizmo.com/s3/2716481/Flash-Blog-Survey


Published by

Bob Bickler

Technical Specialist, Biotage

9 thoughts on “Why can’t I reproduce my TLC separation using flash column chromatography?”

  1. Most of the flash purifications done here use hexane/EtOAc. Sometimes the TLC looks like a difficult purification so a few TLCs are done using hexane and Et2O or TBME. Sometimes this helps but usually EtOAc is best. Often a change to CH2Cl2 instead of hexane makes a significant improvement and also helps with molecules which are not very soluble. For the polar ones its CH2Cl2 / MeOH, usually a shallow gradient from 1-10% MeOH since as you say TLC is not very good at predicting the correct % MeOH to use. Acetonitrile is sometimes used instead of MeOH but often polar molecules seem to need a protic solvent to avoid long streaking spots. I’ve often wondered about EtOH or nPrOH instead of MeOH and maybe replacing CH2Cl2 with EtOAc, but it’s just too time consuming.

    1. Derek,

      I too have thought about using EtOH and IPA in place of MeOH. Since they are still protic but still polar (and share the same selectivity classification) they are logical alternatives though I do not know how well the TLC data will transfer to flash – something to evaluate!


      1. Bob
        I forgot to mention that when I use a 1-10% MeOH gradient, I equilibrate the column with 70% CH2Cl2/30% MeOH for 2 column volumes then 1.5column volumes at the 1% MeOH startpoint. Just to make sure that the column has experienced susbstantial amounts of MeOH before ramping up the gradient. When I was new to chemistry I was told to do this to speed up equilibration. Then I would equilibrate at the intended isocratic conditions, recycling the solvent over lunch break. Didn’t have gradients in those days.

        1. That’s quite interesting, Derek. I have not heard about equilibrating a silica column with a high methanol content solvent only to replace it with a 1% concentration. How well did this work?


          1. I just do this routinely so I can’t compare. Column has to be equilibrated before use, so I do it my way rather than only with 1% MeOH. Some people here start CH2Cl2/MeOH gradients at 100% CH2Cl2 so equilibration is done without any methanol. I worry that when the gradient starts, initial low amounts of MeOH will be taken up by the column a bit like on a TLC plate then their product elutes abruptly as the first of the methanol leaves the column.

          2. Equilibrating with 100% DCM is what I do since meoh is so polar and will deactivate the silica. The consequence is often as you say – abrupt compound elution as the methanol concentration reaches its compound displacement concentration.


  2. Hi Bob,

    One quick comment. Many times I pre-develop my TLC prior to the actual run. This is especially important when running your separation with modifiers like TEA. Not only does it give you better separation on TLC and but also it’s more informative about what’s going to happen in the column. Although far from perfect, it better simulates conditions during flash chromatography.

    And then, in the non-transferrable TLC method department, it’s the solubility issue which has already been discussed.

    Great blog. Keep up the good work.

    1. Hi Tomasz,

      Your point on pre-developing a plate makes sense as it will partially deactivate the plate and may better predict column chromatography results. For organic bases, have you tried using amine-functionalized TLC plates and flash columns?


      1. Hi again,

        I know these are available from Biotage, but I’ve never had an opportunity to use them. I would argue that deactivating silica with TEA gives you similar results, although reproducibility can be an issue.

Leave a Reply

Your email address will not be published. Required fields are marked *