Wethman et al. warn of the use of sodium methoxide in Suzuki reactions in an article published on Chemrxiv (DOI). The research is a collaboration between Bristol Myers Squibb (must be the people with the failing Suzuki) and Scripps Research (the people with the fancy analytical equipment). Case at hand: large variability in Suzuki batches with the ultimate culprit variability in the quality of commercial solid sodium methoxide. In a model reaction the yield was to found to vary from 92% to 8%. Surprisingly it is not possible to visually tell the difference between the good sodium methoxide batch and the bad one, a white powder is a white powder.
Analysis of the degraded base proved difficult. Carbon NMR only uncovered two unexpected impurities sodium formate and sodium carbonate. Detection of the expected impurity, sodium hydroxide (from reaction with water) by conventional methods failed. Ultimately it was found with solid state sodium NMR that sodium hydroxide was not present at all. In a proposed degradation mechanism, reaction of sodium methoxide with carbon dioxide from exposure to the atmosphere (water does not come into play) gives sodium methyl carbonate, nucleophilic substitution with methoxide then gives dimethyl ether and sodium carbonate. The formation of the formate is not explained by this mechanism.
How much wide-spread is the use of this particular base and used in this form? Instead of the commercial solid, you can buy methanol solutions. To prepare it yourself simply add sodium to methanol (may not be suitable in industrial settings). The base may not be the first choice in Suzuki reactions, 10 random Suzuki reaction reactions in the orgsynth.org collection rely on other bases. In fact, for two of the three reactions studied in this article a base other than a sodium alkoxide (tripotassium phosphate and sodium hydroxide(?)) was found to have the superior yield.