Oxaphosphetanes equilibration

Monocyclic Phosphoranes. - Further studies on the mechanism and stereochemistry of the Wittig reaction have been conducted by a combination of 1H, 13C and 3 P n.m.r.2k 25. The results show that at -18°C both ois and trans diastereomeric oxaphosphetans (e.g. 17 and 18) may be observed and their decomposition to alkenes monitored by n.m.r. Evidence was presented to suggest that during this process oxaphosphetan equilibration involving the siphoning of (17) into (18) occurred in competition with alkene formation. 

By adding a strong base to the cold solution of the oxaphosphetane before it eliminates, die oxaphosphetane equilibrates to die more stable anti isomer and die E olefin is produced upon elimination. This so-called Schlosser modification in conjunction with the normal Wittig reaction enables either the Z or E isomer of the olefin to be prepared selectively. 

Methods A, D, and E suffer from the inherent limitation that they deliberately generate a betaine as the precursor of the oxaphosphetane. Since there is no assurance that the reaction of an ylide with an aldehyde would involve the same ionic intermediate (19,21) these control experiments may provide opportunities for stereochemical equilibration that may not be available to the corresponding Wittig reactions. If the oxaphosphetane generated by methods A or E is stable enough to observe directly, then it is usually possible to distinguish between oxaphosphetane equilibration, betaine equilibration, and other mechanisms for loss of stereochemistry (21 c). However, this is not possible for oxaphosphetanes that contain unsaturated substituents at Cj because oxaphosphetane decomposition is fast at — 78°C (21c). In these examples, method A (like method D or E) can only establish an upper limit for equilibration of all of the conceivable intermediates betaines, betaine lithium halide adducts, oxaphosphetanes, and so on. 

The observations summarized in Table 8 have important preparative consequences. To achieve the highest possible ( )-alkene selectivity in a system that is capable of stereochemical equilibration, it is essential to provide sufficient time for oxaphosphetane equilibration below the decomposition temperature. This is best done by monitoring the diastereomer mixture using NMR methods to establish the temperature thresholds for diastereomer equilibration as well as for alkene formation from the more reactive cis-diastereomer. Once these temperatures are known, equilibration can be allowed to proceed below the temperature for (Z)-alkene formation until the optimum ratio of trans-cis oxaphosphetanes is obtained. Subsequent warming completes the optimized E-selective alkene synthesis in an equilibrating system (Table 7). 

There are two steps to the reaction that define the stereochemical outcome. The first is the intial addition of ylide and carbonyl, with inherent preferences for the fonnation of cis- and tran.r-oxaphosphetane intennediates (104) and (105), and the second is the ability of the intermediates to equilibrate. Maryanoff has studied numerous examples in which the final iE)/ Z) ratio of the alkene (102) produced does not correspond to the initial ratios of oxaphosphetanes (104) and (105) and has termed this phenomenon stereochemical drift .The intermediate oxaphosphetanes are thought to interconvert reversal to reactants (98) and (99), followed by recombination. In this case the final ratio of alkene can be substantially different from the initial addition ratio. 

Figure 2 illustrates several reaction coordinate diagrams that allow exothermic oxaphosphetane formation. Option a (four-center process) and the kinetically equivalent b (transient betaine precursor of the oxaphosphetane two-step mechanism) are consistent with the observation that oxaphosphetanes are formed rapidly and decompose slowly when R = alkyl. Since the barrier AGjJgc decomposition to the alkene is smaller than AGJ y, there will be little reversal or loss of stereochemistry in option a. Reversal should become less likely if the a-substituent R is unsaturated (CH=CH2 or aryl), a situation that would decrease AG g by weakening the P—Cj bond (reaction profile c). If substituents are present that retard the rate of decomposition relative to reversal (as in options d or e), then oxaphosphetane reversal and equilibration of stereochemistry become possible, as discussed in a later section. However, this behavior has not been demonstrated for members of the Ph3P=CHR ylide family in the absence of lithium salts. 

The last item (S) may explain why the independent betaine generation control experiments gave contradictory results. If betaines are high-energy species that lie above the saddle point leading from the ylide to the oxaphosphetane, then experiments that deliberately generate betaines (Schemes 3-5) may encounter pathways for stereochemical equilibration that are not accessible to typical Wittig reactions. 

Several tests are available to determine whether equilibration of stereochemistry occurs in the course of oxaphosphetane decomposition (methods A-E, Scheme 7), but each method has some limitations. In method A, oxaphosphetane diastereomers are prepared independently by deprotonation of the )S-hydroxyphosphonium salts 27 or 28 with base (NaHMDS, NaNHj, KO-tert-Bu, etc.) (20). If each isomer affords a distinct oxaphosphetane 31 or 32 according to NMR analysis (usually, or H), then the solutions are warmed up to the decomposition temperature. Kinetic control is established if stereospecific conversion to the alkenes can be demonstrated from each diastereomer. A less rigorous version of this test is to perform the experiment only with isomer 27, the precursor of the cis-disubstituted oxaphosphetane 31 (21c). All known examples of significant (> 5%) stereochemical equilibration involve 31 and not the trans-disubstituted isomer 32 (20, 21c). A negative equilibration result with the cis diastereomer 31 can be assumed to apply to 32 as well. 

Both of the above teehniques can be reinforeed by erossover experiments (method C) (12-14,20,2Ic,22,23a). An excess of CIC H CHO (ArCHO) is added to the solution below the temperature for oxaphosphetane decomposition. If stereochemical equilibration occurs exclusively by a retro-Wittig process to give the ylide 33, then an excess of the crossover aldehyde must produee the crossover products. Since 33 would be intercepted by excess ArCHO faster than it can recombine with the original aldehyde, the conversion from one oxaphosphetane diastereomer into the other (i.e., from 31 to 32) by way of any retro-Wittig mechanism will be suppressed using method C. However, it is essential to prove that the oxaphosphetane has not already decomposed prior to the addition of the crossover aldehyde. Otherwise, there is the risk of a false-negative crossover result. 

Tables 6 and 7 summarize results from stereochemical equilibration studies performed over the past decade by MaryanofT et al. (22,23), and Vedejs et al. (20, 21c, 39-42). A few other convincing examples are included to expand the scope of the systems covered. Table 6 lists those examples where control experiments establish at least 90% retention of stereochemistry from intermediates-to alkene products. As already discussed, the percentage of equilibration represents the upper limit for loss of stereochemistry from all possible pathways in the control experiments. No attempt has been made to determine whether the minor levels of stereochemical leakage in Table 6 occur at the stage of oxaphosphetanes, betaines, or other potential intermediates. Table 6 includes entries corresponding to all of the principal families of Wittig reagents nonstabilized ylides (entries 1-12, 24, 25, 29, and 30), benzylic ylides (entries 13-17 and 28), allylic ylides (entries 22, 23, 26, and 27), and ester-stabilized ylides (entries 18-21). The corresponding Wittig reactions must take place under dominant kinetic control.

By the same logic, it is unlikely that the stabilized ylide reactions can be influenced by stereochemical equilibration. It would be difficult for any equilibration process to compete with the exceptionally rapid oxaphosphetane decomposition step. Partial loss of stereochemistry does occur in some of the high-temperature control experiments at the betaine stage, as listed in Table 7, but oxaphosphetane decomposition takes place with retention of stereochemistry (Table 6, entries 18-21) (21c). 

Thermal Equilibration of Salt-free cis-Disubstituted Oxaphosphetanes. Only three examples of this process are known. The reaction appears to occur spontaneously when certain oxaphosphetanes are warmed to temperatures near the decomposition point. Spontaneous equilibration is restricted to oxaphosphetanes derived from P-trialkyl ylides and aromatic or tertiary aliphatic aldehydes. A catalyzed process via betaine derivatives is not ruled out, but there is no direct evidence to implicate catalysis. 

Table 11 summarizes many of the representative Wittig reactions of nonstabilized ylides Ph3P=CHR that contain no other functional groups that might influence the stereochemical outcome. The table entries have been compared with relevant control experiments discussed in connection with Tables 6 and 7. In those cases where > 5% catalyzed or spontaneous equilibration of oxaphosphetane stereochemistry appears likely, the stereochemical results are marked by a double asterisk. Entries for the lithium-containing experiments include a rough estimate of the maximum possible lithium ion concentration. However, the estimate assumes that all of the... 

In nearly all cases, (Z)-alkene selectivity is higher for tertiary than for unbranched aliphatic aldehydes. The eombination of a tertiary aldehyde and bulky phosphorus ligands in the ylide usually results in the highest Z E ratios, except for the cases already mentioned where oxaphosphetane intermediates undergo cis-trans equilibration according to control experiments. The kinetic oxaphosphetane ratios follow the general rule that cis selectivity is higher for tertiary than for unbranched aldehydes. If this rule is not reflected in the empirical alkene ratios, then equilibration of intermediates is a distinct possibility. 

Several puzzling entries in Table 14 remain to be explained, including reactions where hydroxylic solvents or alkoxide bases are used (entries 11-14). Betaine reversal was demonstrated under hydroxylic conditions in the original Trippet-Jones experiment (Scheme 4) (13), and it is conceivable that interconversion between oxaphosphetanes and betaines could be fast enough in hydroxylic solvents to allow significant betaine reversal to the ylide and aldehyde in some cases. However, there is no clear evidence to implicate stereochemical equilibration of benzylide-derived Wittig intermediates in ether solvents. 

It is likely that the ( )-alkene selective reactions of anionic ylides are due to equlibration of the betaine lithium halide adduct as discussed earlier. However, the balance is delicate and small structural changes can have surprising consequences. Thus, Corey s stereospecific trisubstituted alkene synthesis via /3-oxido ylides (Table 10) is clearly under dominant kinetic control, even though lithium ion is present and aromatic aldehydes can be used as the substrates (54,55). The only obvious difference between the intermediates of Table 10 and oxido ylide examples such as entry 11 in Table 21 is that the latter must decompose via a disubstituted oxaphosphetane while the stereospecific reactions in Table 10 involve trisubstituted analogues. Apparently, the higher degree of oxaphosphetane substitution favors decomposition relative to equilibration. There are few easy and safe generalizations in this field. Each system must be evaluated in detail before rationales can be recommended. 

---