[rolnor]

If you mix the cyclopropylGrignard with 1eqv. sulphur and evaporate you have the magnesium salt of the thiol.

[Babcock_Hall]

In one class our best inhibitor has a value of k_inact/K_i near 5000 M^-1 sec^-1 against the target enzyme.  In a different class the value is about 40,000.  A few of the compounds show activity in disk diffusion assays, but I was hoping for a bit more.

[Babcock_Hall]

I set up a test reaction with diethyl phoshonomethyltoluenesulfonate and PhSO₂Na, which was in 20% excess.  I used DMF and the reaction temperature was 80-90 °C.  I took a time point of the reaction after about 14 hours. There are four spots by TLC and three peaks by P-31 NMR. One signal (at 15.2 ppm) and one TLC spot are consistent with diethyl phosphonomethyltosylate standard, and one signal (about 11.5 ppm) and one TLC spot are consistent to authentic product. I think that the reaction is roughly 65% complete. By TLC the starting material and the product elute with fairly close R(f) values in ethyl acetate/hexanes, but I have not searched exhaustively for a solvent system. I would predict that the O-alkylated product would have a more downfield shift than the S-alkylated product, and the unassigned signal is near 24 ppm, but have not yet checked whether or not this is a known compound.

[Babcock_Hall]

To my surprise the second time point (24 hours later) looked essentially the same as the first time point, both by TLC and by P-31 NMR.  Therefore there are probably three products, with just under 50% being the product of interest.  I plan to do a similar, small-scale synthesis with diethyl iodomethylphosphonate next.  If the iodomethyl test reaction worked better but I had to use the tosylate for some reason, I might be tempted to add tetrabutylammonium iodide as a catalyst.

[Babcock_Hall]

The reaction of sodium phenylsulfinate with diethyl iodomethylphosphonate was about 20% complete after four hours at about 90 °C.  I have been performing microextractions using 5% aqueous LiCl versus ethyl acetate fo the TLC time points, with the hope of removing most of the DMF.

[rolnor]

Have you tried the triflate vs the iodo? Maybe that gives non-wanted O-alkylation?
It would be much faster.

[Babcock_Hall]

I have not tried the diethyl triflate yet (I used to make the di-tert-butyl version and then remove the protecting groups).  I agree that it will be faster, and I also agree that it may give less-than-ideal S versus O-alkylation.  The other thing that could be tried is a microwave-based reaction, but I have no personal experience with it.

[rolnor]

Its easy to make the diethyl directly? The alcohol is commersially available I think.

[rolnor]

You can easily make it from parafirmaldehyde+diethylphosphite. The triflate is obtained by treting the alcohol with tfmsaa+DIPEA in Et2O at -78°, then just dekant the solution to remove the DIPE-Triflate salt and evaporate. You dont need to wash with H2O.

https://www.sigmaaldrich.com/SE/en/product/aldrich/392626

[Babcock_Hall]

We have diethyl hydroxymethylphosphonate on hand.  The addition of paraformadehyde to a phosphite can be little finicky, but basically it can be done.  The triflate can be purified via flash chromatography if need be.

[rolnor]

Yes, its a fairly stable triflate. The method I mentioned gives NMR-pure material, you dont need purufication.

[Babcock_Hall]

The second time point (about 16 hours of reaction) of the iodomethylphosphonate indicated that the product peak was smaller and a new, broad peak was seen.  This is surprising, in that a similar reaction worked in the solid phase.  Perhaps the conditions could be improved (fresher solvent, less light, etc.). However, I might try a preliminary reaction with a sulfonyl fluoride instead.

[rolnor]

Is there a risc of error in the analysis of the mixture? It seems strange that the product breaks down? How do you work-up samples? Is is TLC, in that case, how do you dry the spots before developing? Do you have DMF as solvent? You should get iodide-ions as by-product, these can sometimes interfere.

[Babcock_Hall]

For the P-31 NMR samples I simply sampled about 10 µL and dissolved in CDCl₃.  For the TLC, I took 20 µL and extracted with about 50 µL each of 5% aqueous LiCl and ethyl acetate.  The percentage of the peak area at 11.2 ppm went down from the first time point to the second.  Two new, broad peaks are visible that were not visible before, but they both have very low signal-to-noise.  The reaction picked up a brown color with time.  I am not sure what any degradation product might be.

I also spiked the first time point for the TsOCH₂P(O)(OEt)₂ reaction with an internal standard of diethyl phosponomethyltoluenesulfonate, and the peak near 15.2 ppm increased in intensity, confirming its assignment.  What is odd about this reaction is that the first and second second time points had similar ratios of peak areas to each other.  It is also odd that there was still diethyl phosphonomethyltoluenesulfonate left at time point 2; I set up the reaction to have PhSO₂Na in 20% excess.  Perhaps there is water present in this salt, or perhaps my technique was not good.  I have not yet performed an internal standard experiment with authentic product, but I am fairly confident that the peak near 11.2 is the product, based on similar molecules.

[Babcock_Hall]

Post script, One thing that surprised me a little is that I was able to detect diethyl iodomethylphosphonate by UV quenching in TLC.  I am giving some thought to either looking to improve the toluenesulfonate reaction slightly or to trying a microwave-promoted reaction with either leaving group.