[Babcock_Hall]

We are adding tert-butyldimethylsilyl groups to an aromatic glycoside, following a publshed procedure.  We used 4% (w/v of solution) copper sulfate to try to remove excess lutidine from our synthesis (which also involved an excess of TBDMSOTf and the solvent, which was DCM).  We obtained an emulsion, which broke over a long time as we filtered through Whatman #1.  A milk chocolate colored solid remained on the paper.  On the one hand, I am tempted to increase the copper sulfate concentration to about 25% or so in the future (which we also did in our second wash), in the belief that higher ionic strength disfavors emulsions.  On the other hand after our second copper sulfate wash, we tried a brine wash, and some of the extraction became an emulsion also, which we also broke by filtration.  The procedure did not specify two copper sulfate washes, only one, and it did not suggest a brine step.

We attempted a similar reaction a few months ago, which used an HCl wash, not a copper sulfate wash.  Based on crude mass, I don't believe that we removed all of the by-products.  That is one reason we tried a different literature protocol for this reaction.

Does anyone have any suggestions for ways to avoid the emulsions or any other things we should change?

[Babcock_Hall]

The paper that we followed with the copper sulfate washes was Barrett Anthony, J Organic Chemistry 2000 65(20) 6508-6514.  10.1021/jo000690p.

We patterned an earlier synthesis, the one with a 1 M HCl wash, after the supporting information in Lindner J-P 2011 Chemistry a European Journal 17(15):4090-95. DOI: 10.1002/chem.201003163.

Both papers produced tetra-TBDMS ethers of glycosides from TBDMSOTf and lutidine.  Lindner's paper had more washes, including a brine wash.  Barrett used DCM as the solvent whereas Lindner used diethyl ether.

[rolnor]

I think 1M HCl is a bit acidic for TBDMS-ethers. You can shake with copper sulfate then add Celite and filter through a pad of Celite, then separate. Also if you use chromatography lutidine will be removed I think.

[Babcock_Hall]

Thank you; your replies are always helpful.  Chromatography is the next step, but I was trying to get ride of as much stuff as I could before then.  We will try celite next time, but I think that some DCM will evaporate if we do it under vacuum.

[rolnor]

You will probably wash the Celite so there will be plenty DCM? Dont use full-vacuum when you filter, it just cloggs the filter.

[Babcock_Hall]

We attempted to purify by means of a silica column but did not observe the desired product in any fractions.  My tentative conclusion is that we broke the glycosidic bond somehow.  We are very puzzled. Could prolonged exposure to the copper sulfate solution have this effect?
EDT
We rechecked the calculations for the synthesis, and there was more TBDMSOTf than we intended to add.  This might or might not have been the problem.

[rolnor]

You could have hydrolysis of excess tbdmsotf on the column or in the workup, this make triflic acid so hydrolysis can be quick. You can add a little MeOH+TEA before workup, this will quench the reagent and keep pH high.
Edit: It is actually possible to break/activate glycosidic bonds with tbdsotf so this can be a concern. This is used in nucleoside synthesis, I have made that myself several times. So excess tbdsotf is no good and it could be a problem even if no excess is used.

[phth]

Some suggestions:

Why not just add bicarb and extract with DCM (3 x) wash brine (optional), dry, conc. Should not be a problem with a good vacuum, and it is observable by TLC which may overlap with your product because lutidine moves in polar systems like MeOH/DCM.  Another way is to triturate with ethyl ether or hexanes to remove TBS junk and lutidine.

[kriggy]

Could you use different base? I have good experience with DMF + imidazole as a system for OH silylations:

2-hydroxymethylaniline (123 mg, 1 mmol, 1 eq) was dissolved in dry DMF. Imidazole (163 mg, 2.5
mmol, 2.4 qe) was added followed by TBDPSCl (308 μL, 1.2 mmol, 1.2 eq). Reaction was stirred at
room temperature for 18 hours. Afterwards, water (20 mL) was added and the solution was extracted
with EtOAc (3x 20 mL). Organic extracts were combined and washed with sat. aq. NH4Cl (2x 20 mL)
and brine and dried with MgSO4 and evaporated to yield 270 mg of oil (75%)

[Babcock_Hall]

IIUC lutidine, which is statically hindered, is more often used when there is a trifluoromethanesulfonate leaving group, whereas imidazole, which I believe is more nucleophilic, is used when chloride is the leaving group.  Let me make an educated guess:  Imidazole is a nucleophilic catalyst for TBDMS chloride.  However, one does not want imidazole to displace the trifluoromethanesulfonate group.

We had some problems with this reaction and purification again yesterday, which I may address on a companion thread (unless putting it into this thread would make it easier to follow).  For now, I will just note that after the copper sulfate wash, we had a product that was not homogeneous.  There appeared to be a light brown liquid, a whitish solid, and a purple solid.  Not all of this crude product came into our chromatographic solvent, which was 4% diethyl ether in hexanes.  Perhaps our wash left lutidinium trifluoromethanesulfonate with the organic, not the aqueous layer, and this was a part of the solid. 

I am reconsidering copper sulfate versus HCl washes.  IIRC someone thought that 1 M HCl was too strong for TBDMS groups.  What if we backed off the acid strength, either by going to 0.5 M HCl, or by using a weaker acid, such as KHSO₄ or the sodium form?

[Babcock_Hall]

The reaction in question requires us to add four TBDMS groups to a glycoside.  Could one of these come off during a silica column?  If so, should we try an additive such as TEA, to the elution solvent?  Just thinking off the top of my head.
EDT
After talking to a couple of synthetic chemists, I do not think that migration of the TBDMS group is likely.

[wildfyr]

I don't think all your TBDMS will survive 0.5M Hcl, or even 0.1 M HCl. Beats me on KHSO4.

On the other hand, its pretty stable to silica gel.

[rolnor]

I have commented this before but repeat, four of these bulky PG-s are not easy to get on to a glycoside, if you have trouble doing this I suggest using the SEM-group instead, its much less bulky and can also be cleaved with fluoride, its similar in acid-stability as the TBDMS-group. You can consider this if you repeatedly have a really hard time. The lutidinetriflate is probably easier to wash out if you add some hexane to the mixture. I guess you have checked Greens PG-s in org. synt. it can give you some idea about the SEM-group and general work-up procedures etc.

[Babcock_Hall]

I checked the NMR of the second major pool of product from the silica column, and it looks like the tetra-TBDMS glycoside.  There is a substantial mass that eluted earlier, and I don't know what that substance is (the H-1 NMR did not immediately hit me over the head with its identity).  We plan to move ahead with this synthesis, which is two more steps, and which we did last summer.

The synthesis we ran a few weeks ago might have gone off the rails because too much TBDMSOTf was added, owing to a calculational error.  We have one more glycoside to make, and the starting material is much more expensive than it was for this one.  Let me look into the SEM group and see whether or not glycosides have been protected with it previously.  If they have been, it would leave us with a difficult choice: stick with the TBDMS strategy or move on to the SEM strategy.

[rolnor]

OK, sounds good. The very early fraction is probably TBDMSOH.