[Babcock_Hall]

We have two impure samples of an aromatic glycoside.  In both cases we used eight equivalents of tetrabutylammonium fluoride (TBAF) to remove tert-butyldimethylsilyl (TBDMS) groups from the four oxygen atoms of the glycosidic portion of the molecule.  Our first two attempts at using Dowex-50/calcium carbonate to remove the TBA cations and to protonate the oxygen atoms went well, with over 99% being removed.  This protocol was found in a 2007 J. Org. Chem. paper (9(4):723-726) written by Kaburagi and Kishi (10.1021/ol063113h).  Our second two attempts removed roughly 90% of the TBA, which still leaves roughly one equivalent of TBA present.  The TBA is easily observed in the H-1 NMR spectrum.  We do not know why there was this decline in efficiency.  We are not sure of the identity of the anion paired with TBA (it might be fluoride or bicarbonate, I suppose).  Right now my thinking is to repeat the Kaburagi-Kishi step on one sample, perhaps doubling the amounts of both the Dowex-50 resin and calcium carbonate. 

If this attempt fails, we can try reverse phase flash chromatography on a Teledyne ISCO system on one or both samples.  Chapter 4 of their book has an example (Figure 60 on page 69 of the fifth edition of Effective Organic Compound Purification, which is Teledyne's book) in which two aromatic glycosides are purified nearly to baseline using water and acetonitrile as the mobile phase, but few details are given.  We presently have a 26-gram C18 reverse phase column, and the samples are very roughly in the range of 100 milligrams, going on memory.  We have no experience in preparative RP-chromatography, and I am not sure how to detect TBA during a chromatographic run (which are reasons to favor trying the Kaburagi-Kishi method first).  We did obtain some information from Teledyne on changing from normal to reverse phase and back again.

Is there any point in doing preliminary TLC on RP plates to get a sense of what solvent ratio will move our compound?  I never had much luck with them.  Can anyone inform us about possible pitfalls that we might encounter in preparative RP chromatography?  Thanks for any help or suggestions.

[MOTOBALL]

You may recall from my response to one of your earlier posts, that I am a strong proponent of the use of TLC with silica gel/microscope slides to monitor carbohydrate reactions.

I would start with EtOAc/Hexane (2:1, v/v) and modify the mobile phase from there.
I used to aim for an Rf value of about 0.2-0.3 (IIRC) from a single run, and then perform three runs (letting the slide dry each time) before spraying with 5% conc H2SO4 / EtOH and heating to char the organics. This will give you a picture of how the reaction mixture will elute from a Silica column.
Use that mobile phase for dry-column chromatography.

Good Luck!

[Babcock_Hall]

I no longer see the TBDMS portions of the molecule (the protecting groups) by NMR, and there are four hydroxyl groups in the product.  I strongly doubt that this product will move on silica or be soluble in ethyl acetate/hexane mixtures.  Thank you for reminding me about sulfuric acid as a detection method.

[MOTOBALL]

You are correct Re solubility and lack of movement on silica gel; I had misunderstood the status of your product.

Suitable developing phases could be,

1) EtOAc/EtOH/H2O (10:3:2 v/v/v)

or

2) 45:5:3

Regards.

[wildfyr]

Water on acidic silica gel seems bad for TBDMS. Definitely do 2D TLC first.

[MOTOBALL]

Water on acidic silica gel seems bad for TBDMS. Definitely do 2D TLC first.

Certainly, but the original objective by the poster was to remove those groups!

[Babcock_Hall]

The TBDMS groups are gone.  The problem is that the (Bu)₄N⁺ cations are partially but not completely removed; we started with eight equivalents of TBAF and have removed about 90% of them, based upon H-1 integrations.  The identity of the anion is uncertain at this point.

[MOTOBALL]

Can I suggest that you run TBAF with either one of the 2 suggested aqueous phases just to determine how it behaves?

Regards

[RedViper9]

Re: RP Chromatography on Combiflash

Typically Teledyne ISCO Combiflash systems are equipped with UV detectors. Unless your ISCO system is equipped with a Refractometer, Evaporative Light Scattering Detector (ELSD, rare) or mass spectrometer (more rare) you should not expect to visualize the tetrabutyl ammonium component. Still, assuming your target is not itself an ammonium salt, the expectation is that there will be some significant separation of the two components on C18.

One major problem may be mixed fractions due to poor peak shapes (tailing). If you want information on the separation before hand, you might try getting some C18 TLC plates or running analytical HPLC to get an idea of the retention and peak shapes.

Be sure to collect your waste stream in a large beaker rather than dumping it straight to solvent waste. Otherwise, a bad column or accidentally using the wrong solvent as the weak phase may add months of work to your isolation...

Some general recommendations if you're going in blind and want maximum info from one run:
- Use an RP column of appropriate size to the sample per manufacturer recommendations.
- Dry load your sample onto an appropriate amount of Celite.
- Run a scouting gradient, the system will suggest a generic gradient based on the size of the column you use.
- Use the collect by volume option.
- Individually check the fractions which light up by UV for TBA contamination.
- If they're clean, great you're done.
- If there are still mixed fractions, start checking fractions ahead (or behind as appropriate) of your target to determine whether you're seeing peak tailing or you just happen to have similar RF.
- ISCO RP columns can be reusable, so if the first separation fails adjust gradients to get what you need.

Remember a Combiflash is simply an automated method of running flash columns. The instrument only gives two real benefits vs. traditional glass columns: 1) More continuous pressure 2) Smaller gradient steps. That is significant for a tight separation, but will not give you HPLC levels of resolution.

[RedViper9]

I just read in one of your previous posts that you were working with compounds consisting of a glycoside tethered to a phosphate. Would these two happen to be such compounds?

If so, the fact that you're getting stuck at one equivalent of tetrabutyl ammonium suggests to me that you're isolating the TBA salt of your desired compound. In this case I'd most likely focus efforts on different cation exchange resins.

RP chromatography on it's own won't separate the two salt components. Even if the ammonium salt and phosphate salt are separate salts RP may struggle. You can try adding acids to compete the ammonium off, but I'd expect you'd need trifluoroacetic acid to get any strong effects and I'm not sure the ISCO will stand up to that. You can try Preperative HPLC if available. I've resorted to multiple injections on analytical HPLC with manual fraction collection before, but that's not a great way to spend an evening.

Ion exchange chromatography will work, but automated systems are still rare in academic settings.

[Babcock_Hall]

Hi RedViper9,

First, thank you very much for your suggestions concerning the Teledyne CombiFlash; we can also ask for some help at our institution.  Second, we are working on two glycosides which are made using very different synthetic strategies.  One has a phosphonate diester as an intermediate, but this portion of the molecule will be discarded in a Horner Wadsworth Emmons reaction.  We still have a few more steps to go, but I could see reverse phase used after the final deprotection step.

The present synthesis used four TBDMS groups on the four carbohydrate oxygen atoms (at carbons 2, 3, 4, and 6).  The final step was the use of 8 equivalents of TBAF to remove these protecting groups.  We have performed this step several times with only one failure.  Then we followed the procedure of Kaburagi and Kishi (J. Org. Chem. 2007) to remove the remaining fluoride ions and the tetrabutylammonium ions, some of which may be pairing with carbonate or bicarbonate anions.  In brief calcium ions form CaF, and the Dowex-50 provides protons to oxygen and binds to the TBA ions.  This step was partially successful, removing roughly 90% of the TBA ions by H-1 NMR (it has previously been more successful in our hands).  Our job now is to remove the majority of the remaining TBA ions.  Our product is electrically neutral, but I would describe it as amphipathic, having an aromatic portion and a glycosidic portion.

If I had a charged product, I might be tempted to try low pressure ion exchange using TEAB, which is volatile.   

[clarkstill]

Can you use a different fluoride source that doesn't leave the NBu4 to deal with? HF.Pyridine?

[Babcock_Hall]

clarkstill,

Yes, that is an alternative but it would require us to repeat a three-step synthesis, and I am short on person power.  However, we have previously been able to get to better than 99% removal on two occasions when we synthesized a structural isomer.  We presently have two samples of the desired glycoside, both of which have about 10% remaining TBAX (where X is the anion). 

[rolnor]

Tetrabutylammonium will stick very hard to RP-gel, it will not move. Just 50%MeOH/H2O will do the trick I think. Load the column with pure MeOH, then wash a little with 50%MeOH/H2O before loading the sample.

[rolnor]

It should be Pack the column with pure MeOH, not Load the Column with pure MeOH,