[Babcock_Hall]

We prepared the trichloromethylimidate of tetraacetylglucopyranoside (the Schmidt donor) and an alcohol with the intent of forming a glycosidic bond.  We followed a protocol that had used BF₃-etherate as the Lewis acid promoter and ran the reaction in the presence of 4 angstrom sieves.  Dry DCM was the solvent.  Following the suggestion of a colleague, we worked on a small scale and tried to isolate the steps of the reaction: quenching, extraction, and purification.  We followed the disappearance of the Schmidt donor by TLC.  There was a small excess of alcohol to Schmidt donor.  We initially stored the reaction at -80 °C, then we quenched with TEA, which was the only deliberate deviation from the protocol that we were following.  Those authors used aqueous sodium bicarbonate as the quench (other workers used TEA).  The reason that we used TEA was to isolate the quench step from the extraction step; however, if I had this to do over again, I would just go with bicarbonate and not worry about isolating the quench step from the extraction step.

We removed the solvent and dried in vacuo.  We took H-1 and P-31 NMR data.  The H-1 suggested that the integral in the acetyl methyl region was about 1/3 of the expected value.  In the starting material, the hydrogen at C-1 is a doublet near 6.6 ppm.  We saw a broad signal in the same area.  The NH signal of the Schmidt donor is no longer present.  There are two populations of triplets in the general vicinity of where I would expect OCH₂CH₃ to show up.  The P-31 NMR data suggested the presence of three populations of molecules.

In my informal survey of conditions in various papers, I found large variations in temperature, time, and mole ratios of Schmidt donor to alcohol to BF₃3-etherate.  I am not sure about what to try next.  I can supply more details if that would be helpful.

[rolnor]

The Schmidt-donor will dissapear even if it does not react with the alcohol beacause you have BF3 present so this is no good way to monitor the reaction. Have you considered that you can get a mixture of alpha and beta coupling?

[Babcock_Hall]

How would you monitor the reaction?  Is there any point in doing a mini-extraction on what we have and taking another NMR?

There are two populations of triplets (presumably -OCH₂CH₃) separated by 0.15 ppm.  In the presumed product they are 10 atoms away from C-1 of the glucopyranoside.  Their combined area is twice that of the four CH₃C(O)- protecting groups, when their area should be half.  One possible explanation is that we had poor control of stoichiometry, owing to small masses or other reasons.

[rolnor]

You could look at the dissapering of the alcohol. The raction can probably take some time, you have steric bulk from PG-s and low temperature. If it is working as intended I think TLC would give a clear result, mainly the product.

[Babcock_Hall]

We attempted a second, small-scale Schmidt reaction with BF₃-etherate as the acid promoter, monitoring the loss of the alcohol acceptor.  I do not recall anything unusual about the TLC results.  We checked the product by NMR.  The P-31 NMR suggested one major and one minor product.  The H-1 NMR was disappointing.  The signal from the acetyl methyl groups was only about half of what we expected, based on the integral of the methyl groups from the alcohol. 

We are continuing to review the literature.  We plan to perform a reaction in the presence of TMSOTf as the acid promoter.

[rolnor]

TMSOTf is good, its not really a Lewis acid so sensitive groups are spared.
Are you planning to use LC-MS? You are potentially wasting time, effort and money by running only TLC…

[Babcock_Hall]

For our first reaction, our use of TLC was suboptimal because we followed the donor, not the acceptor.  For our second reaction we followed the acceptor (the alcohol).  We don't have walk-up LC/MS, but we might be able to run it on a more limited basis.  What is the advantage of LC/MS over TLC, and would you follow the reaction this way by quenching a small portion of the reaction to generate the LC/MS sample?

[kriggy]

Well the advantage is that you see mass of the product and can better identify whats going on. I would do the quench for LCMS just for the sake of preventing column damage

For the TLC... dont you spot all stuff thats in the reaction?

[Babcock_Hall]

Yes, but in the past we chose TLC conditions to observe the loss of one reactant or the other.  We could run two plates, choosing the best solvent to observe each reactant, if that is what you are thinking.  One problem that we have to overcome is that the ratio between the trichloroacetimidate (the glycosyl donor) and the alcohol (the acceptor) varies a good deal from one protocol to the next.  Although frequently the ratio is close to 1:1, either reactant can be limiting.

I think that our initial choice to follow the glycosyl donor might have been misleading; perhaps the donor reacted with the acid promoter (BF₃-etherate) but not necessarily the acceptor.  Our latest attempt used TMSOTf as the acid promoter, and the TLC looked promising.  However, we won't know until next week some time.

[Babcock_Hall]

Update:  We have been running several small scale reactions.  We have been trying variations in temperature or the mole ratio of TMSOTf.  We have yet to see obvious signs of success.  I am obtaining review articles on Schmidt chemistry.

Using NMR to study the pooled fractions from one reaction and purification, we found a small amount of product and a large amount of something else, that may be the ether created from two molecules of alcohol.  I consulted with someone much more knowledgeable about carbohydrate chemistry than I am, and he thought that this side product was the the ether, but the two hydrogen atoms that are closest to the central oxygen atom of a putative ether in our molecule are more downfield than they should be if we were making this ether.  However, I have been unable to come up with an entirely satisfactory alternative structure.  The -OCH₂- group apparently overlaps with a signal from another part of the molecule, and it has a chemical shift that is more consistent with being part of an ester.

[rolnor]

You can have a thio-glycoside and activate this with methyl triflate, I think thats even more selective than Schmidt. It would be really nice to see your compounds structure to be able to give you advise. I guess I need to sign som documents of secrecy for that, as I understand this is commercial research.

[Babcock_Hall]

Our donor is the trichloroacetimidate of 2,3,4,6-tetraacetylglucopyranoside.  In the acceptor there is a phosphorus that is distant to the primary alcohol group, and we can see small differences in the P-31 shifts of various pools.  The alcohol is not hindered.
EDT
One way to explain the H-1 NMR data is if an acetyl group somehow migrated onto the alcohol.  I don't know what the mechanism would be.

[rolnor]

You can mix the protected methylglucopyranoside with TMSBr to get the bromo-sugar and then react this with the alcohol+1,1eqv. TEA in DCM.

[Babcock_Hall]

There are two other potential nucleophiles on our 1° alcohol, a sulfide and the oxygen from a phosphonate.  I am working out alternative synthetic paths.  We may also try a simpler alcohol to see that our conditions are good, as per a suggestion at ChemPros.

[rolnor]

Yes, the sulphide could be a problem, oxidize it to sulphone, use NaIO4/MeOH/H2O. The phosphonate oxygen is not nucleopphilic.