With a series of successful couplings in hand, we returned to the question of mechanism and the hypothesis of glycosyl triflates as the key reaction intermediates. A donor 44, lacking any diastereotopic benzyl hydrogens, was prepared and its 1H-NMR spectrum recorded in CD2Cl2 in the presence of DTBMP at —78°C. Cold (—78°C) Tf2O was then added, and a rapidly recorded spectrum showed the sulfoxide to have been completely consumed and converted to a new carbohydrate species . This new substance was characterized in the 1H-NMR spectrum by its anomeric proton, a broad singlet, which resonated at 8 6.20. In the 13C-NMR spectrum the anomeric carbon had a chemical shift 8 of 104.6 and a 1/CH coupling of 184.5 Hz. These data indicated that a strongly electron-withdrawing group was covalently linked through oxygen to the a position of the mannopyranose ring, providing strong support for the a-mannosyl triflate (45) hypothesis. Confirmatory evidence was obtained when treatment of the mannsoyl bromide 46 with AgOTf and DTBMP in CD2Cl2 gave indistinguishable spectra . This experiment also provided strong support for Schuerch's earlier hypothesis that glycosyl triflates were formed at low temperature upon treatment of glycosyl halides with AgOTf (see above). Addition of methanol at —78°C to these NMR tube experiments resulted in the very rapid formation of mannosides (47) with high ^-selectivity, in full agreement with the general mechanism proposed (Scheme 11) .
A tetra-O-methylmannosyl sulfoxide 48 was prepared as a surrogate for the unselective tetrabenzyl donor 42. Again, low-temperature 1H and 13C NMR experiments indicated clean formation of a covalently oxygen-linked intermediate and, again, the same spectra were obtained going out from the bromide and AgOTf .
A first clue to reasons underlying the differing selectivity of the tetramethyl (or benzyl) and 4,6-benzylidene series was obtained in the course of attempts to record a CH-gated, coupled 13C-NMR spectrum of 49. Substantial decomposition occurred
over the acquisition time, unlike the case of the 4,6-benzylidene donor (44), which prevented us from obtaining the 1JCH coupling constant. This difference in stability was confirmed by a series of variable temperature experiments, which revealed the 4,6-benzylidene protected triflate (45) to decompose around — 10°C, whereas its tetramethyl congener (49) did so some 20 degrees lower (—30°C). These latter observations provide strong support for the notion that the a/^-selectivity in these couplings is a function of the equilibrium between the covalently bound triflate and the ion pair. The less stable the triflate, as reflected in the lower decomposition temperature, the greater the population of the ion pair and the greater the likelihood that a-mannosylation will occur through a Curtin-Hammett type of kinetic scheme. The increased stability of the 4,6-benzylidene protected triflates may be rationalized in terms of Fraser-Reid's concept of a torsionally disarming protecting group . In effect, the sofa conformation of the oxacarbenium ion imposes a twist and torsional strain on the acetal ring, which increases the energy of the oxacarbenium ion with respect to that of the covalently bound triflate. This effect is not present in the per-ether protected systems.
Was this article helpful?