Sulfonyl azides, $\ce{RSO2N3}$, are capable of acting as sources of electrophilic azide ("$\ce{N+3}$"), but are also capable of effecting diazo transfer, as shown in the scheme below, taken from ref 1:
In 1990, Evans and coworkers1 showed that the aryl group on the reagent was important in controlling the outcome of the reaction:
$$\begin{array}{ccc} \ce{Ar} & \text{yield of }\textbf{3a-S}\text{ / %} & \text{yield of }\textbf{5a}\text{ / %} \\ \hline \ce{(4-NO2)C6H4} & 15 & 70 \\ \ce{(4-Me)C6H4} & 51 & 26 \\ \ce{(2,4,6-^iPr3)C6H2} & 91 & 0 \\ \end{array}$$
which suggests that $\ce{TrisN3}$ (the third entry) is the best azide transfer reagent.
However, confusingly, it has also been reported that under phase transfer conditions, $\ce{TrisN3}$ outperforms $\ce{TsN3}$ (the second entry) in diazo transfer to ketones.2
With this in mind, Evans concluded that
Collectively, these independent observations underscore the lack of mechanistic sophistication which currently surrounds these reactions.
Since then, have there been any efforts to rationalise the effect of the aryl group on the product distribution? And if so, how can the results above be explained?
References
Evans, D. A.; Britton, T. C.; Ellman, J. A.; Dorow, R. L. The asymmetric synthesis of α-amino acids. Electrophilic azidation of chiral imide enolates, a practical approach to the synthesis of (R)- and (S)-α-azido carboxylic acids. J. Am. Chem. Soc. 1990, 112 (10), 4011–4030. DOI: 10.1021/ja00166a045. Non-paywall version available from the Evans group website.
Lombardo, L.; Mander, L. N. A One-Step Synthesis of Cyclic α-Diazoketones. Synthesis 1980, Vol. 5, 368–369. DOI: 10.1055/s-1980-29022.