“Reductive ozonolysis” via a new fragmentation of carbonyl oxides

This account describes the development of methodologies for ‘reductive’ ozonolysis, the direct ozonolytic conversion of alkenes into carbonyl groups without the intermediacy of 1,2,4-trioxolanes (ozonides). Ozonolysis of alkenes in the presence of DMSO produces a mixture of aldehyde and ozonide. The combination of DMSO and Et3N results in improved yields of carbonyls but still leaves unacceptable levels of residual ozonides; similar results are obtained using secondary or tertiary amines in the absence of DMSO. The infl uence of amines is believed to result from conversion to the corresponding N-oxides; ozonolysis in the presence of amine N-oxides effi ciently suppresses ozonide formation, generating high yields of aldehydes. The reactions with amine oxides are hypothesized to involve an unprecedented trapping of carbonyl oxides to generate a zwitterionic adduct, which fragments to produce the desired carbonyl group, an amine, and O2. 10748 SCHWARTZ, RAIBLE, MOTT, & DUSSAULT IN TETRAHEDRON 62 (2006) Ozonides possess a dangerous combination of kinetic stability and thermochemical instability; they are typically isolable yet often capable of spontaneous and dangerously exothermic decomposition reactions.2 Our goal was to develop methodology that would avoid generation of ozonides or other peroxides, and instead directly deliver the desired carbonyl products. Our approach required a reagent capable of intercepting the primary ozonide, the carbonyl oxide, or the ozonide (1,2,4-trioxolane), yet compatible with ozone, one of the strongest oxidants in organic chemistry. Ozonides appeared too stable to be the targets of such an approach. Primary ozonides (1,2,3-trioxolanes) have been generated at very low temperature and separately reacted with strong nucleophiles, but this process has not been accomplished in the presence of ozone.14 This leaves carbonyl oxides, the most reactive intermediates in an ozonolysis, as the most logical targets for in situ capture. 2. Results and discussion Our initial approach focused on cycloaddition of carbonyl oxides with X==O reagents (Fig. 2). An optimal trapping reagent would be a readily available and reactive dipolarophile containing a central atom (X) in an incompletely oxidized state. The derived heteroozonides would be expected to undergo internal fragmentation with liberation of O==X==O and a carbonyl group, achieving net oxidation of the X==O reagent and net reduction of the carbonyl oxide. Literature reports suggested that sulfi nyl dipolarophiles reduce carbonyl oxides, presumably via intermediate 3-thia-1,2,4trioxolanes.15 Moreover, electron rich carbonyl oxides preferentially oxidize sulfoxides (to sulfones), even in the presence of a sulfi de.16 A similar strategy has recently been applied to the reduction of persulfoxides with aryl selenoxides.17 Our investigations began with dimethyl sulfoxide (DMSO). Whereas ozonolysis of decene provides a nearly quantitative yield of isolated ozonide (3-octyl-1,2,4-trioxolane),18 the same reaction in the presence of 2.0 equiv of DMSO generated a mixture of aldehyde and ozonide in which the former was predominant (Table 1). While these results were intriguing, we were unable to fi nd conditions able to effectively suppress ozonide formation. For example, the use of 5 equiv of DMSO offered little improvement in yield of aldehyde,19 while attempts to employ even larger amounts of reagent resulted in phase separation or freezing. The addition of protic nucleophiles provided an opportunity to test the role of the carbonyl oxide in the DMSO-promoted reductions (Table 2). The presence of methanol resulted in the formation of hydroperoxyacetal at the expense of aldehyde. The same effect was observed to a lesser extent for isopropanol, as would be expected based upon the reported rates of trapping by primary and secondary alcohols.10 and 12 The DMSO-mediated reduction was unaffected by the addition of a proton donor (HOAc), but was actively suppressed by Sc(OTf)3. Although we had hoped that the Lewis acid might serve to bring together the reactants, the results suggest that the Sc+3 is simply sequestering the sulfoxide. In contrast, ozonolysis at −78 °C in the presence of both DMSO and Et3N achieved a noticeable improvement in the yield of aldehyde (Table 3); an even better yield was obtained upon reaction at 0 °C. The formation of aldehyde appeared to be enhanced by trace moisture; performing the reaction with deliberate exclusion of water (including drying the incoming stream of O3/O2 through a −78 °C U-tube), resulted in a reduced yield. For reasons that would later become clear, the use of excess Et3N slowed the reaction and resulted in the isolation of recovered decene (not shown). The combination of DMSO and Et3N provides a useful protocol for syntheses of aldehydes and ketones (Table 4). To our surprise, a control reaction investigating ozonolysis in the presence of Et3N furnished better yields of nonanal than had been obtained with DMSO (Table 5). The amine-promoted reduction appeared general for secondary and tertiary amines; primary amines, which react with carbonyl oxides to form oxaziridines, were not investigated.20 The use of anhydrous conditions again resulted in a decreased yield of aldehyde. “REDUCTIVE OZONOLYSIS” VIA A NEW FRAGMENTATION OF CARBONYL OXIDES 10749 The sole precedent for this process was a report describing isolation of adipaldehyde upon ozonolysis of cyclohexene in the presence of Et3N. 21 The reduction of carbonyl oxides by pyridine has been reported and later refuted.22 However, several observations led us to question the role of the amines. First, as had been previously observed during the experiments with DMSO/Et3N, the use of excess amine slowed consumption of alkene. Second, directing the gaseous stream of O3/O2 onto or into a CH2Cl2 solution of alkene and amine resulted in intense fuming, which persisted for a period proportional to the amount of amine. Similar fuming was observed for ozonolysis of solutions of Et3N or N-methylmorpholine (NMM); in contrast, no fuming was observed when a stream of ozone was directed onto or into a solution of decene. Moreover, monitoring (TLC or NMR of quenched aliquots) of the ozonolysis of mixtures of amine and alkene detected very little formation of aldehyde or ozonide until after fuming had ceased. Third, ozonolysis of a solution of amine, followed by addition of decene and continued ozonolysis, produced a mixture of aldehyde and ozonide. These results suggested the intermediacy of N-oxides. The ozonolysis of tertiary amines is known to furnish both Noxides and products of side chain cleavage, the latter process accounting for our observation of acetaldehyde in the crude products from reactions employing Et3N. 23 Furthermore, the ratio of N-oxide formation to side chain cleavage is enhanced in the presence of a proton donor, accounting for the infl uence of moisture on the reactions involving amines. The role of N-oxides was explicitly tested by ozonolysis of 1-decene in the presence of commercial N-methylmorpholine-N-oxide (NMMO). Reaction proceeded without fuming to furnish exclusively nonanal (Table 6).24 Predominant formation of aldehyde was also observed for reactions in the presence of DABCO-N-oxide and pyridine N-oxide. The latter reduction, while complicated by the formation of intensely colored byproducts, is noteworthy given the very limited amount of reduction observed in the presence of pyridine. The intermediacy of carbonyl oxides in these reactions was supported by a simple set of competition reactions. The products obtained from ozonolysis of a CH2Cl2 solution of decene were compared under three sets of conditions: (1) no additives; (2) addition of stoichiometric MeOH; and (3) addition of stoichiometric amounts of both MeOH and NMMO (Table 7). The results demonstrate competition between the amine oxide and the alcohol for capture of the intermediate nonanal-O-oxide.25 Furthermore, 1-methoxydecene, which generates the same carbonyl oxide but cannot easily form an ozonide, also produces nonanal as the major product in the presence of NMMO.1