Chemistry - Reagent choice in the formation of Wittig reagent

Solution 1:

I am fairly sure that for both, there are practicality issues at hand.

This is almost certainly so for the choice of phosphine. Even though PPh3 is absolutely terrible from an atom economy point of view, and even though PPh3O can be a real pain to remove after the reaction is done, at least it's a nice white solid which doesn't smell. On the other hand, the simpler alternative PMe3 is a volatile, pyrophoric, and extremely stinky liquid. PPh3 is also substantially cheaper than other alternative phosphines. (There are similar considerations for the Mitsunobu reaction which I have some experience with.)

As for the base, $\ce{BuLi}$ is just one of the most accessible strong bases to use. I don't think you can buy dimsyl sodium from the average chemical supplier. My suspicion is that the main reason why dimsyl anion is somewhat associated with the sulfur ylids is because E. J. Corey and his postdoc M. Chaykovsky popularised them together (e.g., Ref.1). Also note that $\ce{NaH/DMSO}$ mixtures are associated with significant explosion hazards (Ref.2).

Others with more experience will be able to give fuller answers.


  1. E. J. Corey, Michael Chaykovsky, “Methylsulfinyl Carbanion ($\ce{CH3-SO-CH2-}$). Formation and Applications to Organic Synthesis,” J. Am. Chem. Soc. 1965, 87(6), 1345–1353 (
  2. Qiang Yang, Min Sheng, James J. Henkelis, Siyu Tu, Eric Wiensch, Honglu Zhang, Yiqun Zhang, Craig Tucker, David E. Ejeh, “Explosion Hazards of Sodium Hydride in Dimethyl Sulfoxide, N,N-Dimethylformamide, and N,N-Dimethylacetamide,” Org. Process Res. Dev. 2019, 23(10), 2210–2217 (

Solution 2:

Wittig reactions can be performed with all sorts of phosphanes. The resulting phosphonium salts can be deprotonated by all sorts of bases, as long as the base is basic enough.

Triphenylphosphane has the advantage that its phosphonium salt can only be deprotonated on one carbon atom: the three ipso-phenyl carbon atoms are quarternary and thus cannot be deprotonated. In addition, it is extremely cheap and easy to handle. However, the three phenyl groups also make it somewhat bulky and the Wittig reaction may not be as effective as using other phosphanes.

Trimethylphosphane is another candidate that eliminates all problems of bulkiness. However, it comes with its own set of problems like being a volatile, pyrophoric and toxic liquid that reacts with ambient oxygen to form $\ce{OPMe3}$, meaning that it must be kept under inert gas.

An in-between choice is tributylphosphane. While it is still unpleasant, requires inert gas handling and is flammable, it is less toxic and less susceptible to oxidation than trimethylphosphane. For an example of tributylphosphane being used in total synthesis, check out Gieseler and Kalesse’s synthesis of angiolam.[1]

As for the choice of base, it depends a lot on the compound in question. As you can tell in the same paper by Gieseler and Kalesse (experimental procedures in the supplementary information should be accessible free of charge), they use $\ce{KOtBu}$ as a base to convert the tributylphosphanyl salt into the corresponding ylide – but they have an activated Wittig reaction thanks to the neighbouring carbonyl group. For unactivated Wittigs, BuLi may be your only choice but if it is not, a milder base will typically cause less side-reactions in a sensitive substrate.

Which base is most convenient to use often depends on the laboratory in question. If there is a constant influx of students that have to titrate BuLi to learn the method, and you have large stocks prepared because you need it often, BuLi might be your method of choice. If you just established 234ff-base as a new named compound in your laboratory, you will probably have ample amounts of that and a strong desire to try that one first. And so on.

[1]: M. T. Gieseler, M. Kalesse, Org. Lett. 2014, 16, 548–551. DOI: 10.1021/ol403423r.