[rolnor]

That sounds very strange, in my experience even TBDS-groups are completely deprotected by TBAF in a few minutes. TMS-ethers are unstable even to moist air.

[Babcock_Hall]

In the first table (p. 197), the substrate is C₁₂H₂₅OTBDMS.  In a different table on page 198, they give 20 minutes for TBAF in THF, to deprotect the TBDMS group.  At a couple of places in the text, they mention the use of acetic acid or ammonium fluoride to buffer the basicity of TBAF.

[rolnor]

Its possible that they did not monitor the reaction untill 20minutes had passed. Acetic acid seems very good, mayby TBAF can also be used in combination with HF-Pyridine as buffer?

[Babcock_Hall]

Our budget is quite limited; preparative HPLC is not something we can consider just yet.  Even ordering more reagents is tricky, especially as we near the end of the fiscal year.  If we could push our yield to 50% with TBDSCl, that might be enough; that is why was thinking about adding DMAP.  AgNO₃ is something I will keep in mind when I have a compatible substrate, but I don't believe that our present one is.  I am continuing to look into TBDMSOTf protocols and will try to order some, if possible.  Right now the calcium carbonate/Dowex-50 protocol looks like a contender in the Desilylation derby.

[Babcock_Hall]

https://doi.org/10.1016/j.carres.2005.05.016

I just read a number of references such as the one above that concern adding TBDMS groups to glycosides.  It seems as if TBDMSOTf/ludidine is much more likely to succeed than TBDMSCl/imidazole is, and I am in the process of ordering some.  Although I have not looked at every protocol in detail, only one used copper sulfate as one of the wash steps.  This was probably to help in the removal of the lutidine.

[rolnor]

That seems very good. Also you can heat, maybe reflux in DCE or DMF if it is really sluggish to get all the PG on. If you have a good pump and heat to 50°C you will remove lutidine I think.

[rolnor]

When imidazole is used as base and catalyst, the reacting species will be the silyl-imidazolide regardless what silylating reagent is used, that means that imidazole is no good with the triflate, the triflate itself is more reactive then the imidazolide so lutidine is better.

[Babcock_Hall]

Good point about imidazole.  I wanted to discuss the choice of deprotection/clean-up method.  Along with an aromatic ring our product will have four free hydroxyl groups (being an aromatic glucoside), making it fairly polar.  The technique in the 2007 Organic Letters paper (10.1021/ol063113h) does not use an aqueous work-up.  Four of the entries in their table are glycosides based upon monosaccharides.  I don't have a strong opinion on whether or not this is a better approach versus versus the ones we have already mentioned.

"General workup procedure for removal of TBAF residue (Table 1, entry 1): A 10 mL flask containing a stirring bar was charged with substrate
(110 mg, 0.169 mmol). TBAF (1.0 M in THF, 1.35 mL, 1.35 mmol) was added via syringe. After the solution was stirred at room temperature for 4
h (no SM nor partially deprotected intermediate detected by TLC), CaCO₃ (280 mg), DOWEX 50WX8-400. (840 mg, used as supplied), and MeOH
(2.0 mL) were added. The suspension was stirred at room temperature for 1 h. All insoluble materials were removed by filtration through a pad of Celite, and the filter cake was washed with MeOH thoroughly. The combined filtrates were evaporated to dryness under reduced pressure to give the crude product (36.3 mg, 111%) as a white solid. 1H NMR spectra of the crude product showed that 99.3% of tetrabutylammonium salt was removed by these operations."

[rolnor]

The workup seems very good, clever.

[Babcock_Hall]

In the 1987 paper that I cited, their Method B used catalytic DMAP (8% by mole), along with 2 equivalents of TBDMSCl, and triethylamine was the base.  Their method A used a variable number of equivalents of TBDMSCl.  I have not yet compared Methods A and B at the same mole ratio of TBDMSCl to methylglycopyranoside.

I also wanted to draw attention to the failure to observe any tetra-silylation with methyl galactopyranosides as substrates.  The three substituents on C3 (-OH), C4(-OH), and C5(-CH₂OH) are all on the same side of the ring, in contrast to glucopyranosides, where the substituent at C4 is on the opposite side of the ring.  This is consistent with a steric problem, although I would hesitate to call it strong evidence.

https://doi.org/10.1002/ejoc.201200277
Last week I found a reference from 2010 which also obtained 0% tetrasilylation for a galactoside derivative, when using TBDMSCl/imidazole.  Galactose is epimeric with respect to glucose at carbon-4, and they obtained disilylation and some trisilylation.  This gives me more confidence in the results from the 1987 paper, although I was already pretty convinced.  The second group switched to TBDMSOTf.

[Babcock_Hall]

https://pubs.acs.org/doi/10.1021/ol063113h
We tried the work-up from the paper I mentioned above, which appeared in Organic Letters 2007.  After one removes the TBDMS groups with TBAF, one has the problem of how to get rid of excess TBAF.  Sometimes an aqueous work-up is not possible, owing to high water solubility of the product.  Basically the method is to use Dowex-50 and CaCO₃.  The Dowex-50 provides protons and binds the tetrabutylammonium ions, and the calcium forms a precipitate with fluoride.  We removed a little over 99% of the tetrabutylammonium groups.

[Babcock_Hall]

The question remains of the best way to remove the remaining tetrabutylammonium fluoride from the desired product (an aromatic glycoside that does not have charged groups. Whether or not the counter-ion is still fluoride is open to question, but the presence of the methyl group from the TBA cation is clear.  We have roughly 30 mg.

I have been thinking about various reverse-phase methods.  As an alternative, one idea that just occurred to me is mixed-bed ion exchange resin, such as Dowex 501.  I would think in terms of swirling a solution with the ion-exchange beads, filtering, and rinsing to recover as much sample as possible.  It is possible to follow the deionization of urea using a conductivity meeter, although our volume is probably too small for a standard meter.  Does anyone have any thoughts or suggestions?

[rolnor]

I think if you have say 1% TBAF left you have very little ions left considering the high mw of TBAF?

[Babcock_Hall]

Yet the original TBAF was in eightfold molar excess over the glycoside (2-fold per -OH group in the final product).  Therefore, 99% removal still means 0.08 mole TBAF per glycoside.  The questions are whether or not the remaining TBAF would interfere with the intended purpose and whether or not the material in its present form would be considered pure enough for publication.  On the former question, I can make a plausible case for interference.

I may try Dowex-501 on the NMR sample, which I still have.  Centrifuging the ion-exchange beads is another alternative.

[Babcock_Hall]

Using the NMR sample, I tried some old Dowex 501, which I first rinsed with a small portion of D₂O.  The good news is that the triplet assigned to the methyl group in the tetrabutylammonium ion is about 40% of the area it used to be.  The bad news is that there is now a sharp singlet in the H-1 spectrum, at 2.16 ppm.  This is not far from acetone in water.  I had just been cleaning NMR tubes prior to performing this experiment, and it is possible that I used a tube that was not dry.