The question of C- vs. O-silylation of ketenes: Electrophilic triethylsilylation of diphenylketene

  1. G. K. Surya Prakash,
  2. Chulsung Bae,
  3. Golam Rasul, and
  4. George A. Olah
  1. Loker Hydrocarbon Research Institute and Department of Chemistry, University of Southern California, University Park, Los Angeles, CA 90089-1661
  1. Contributed by George A. Olah, March 10, 2005

Abstract

Electrophilic triethylsilylation of diphenylketene leads to exclusive C-silylation giving the diphenyl(triethylsilyl)acetyl cation in the solution phase even though density functional theory calculations at the B3LYP/6-311+G* level indicate that the O-silylation of diphenylketene is preferred over C-silylation by 5.4 kcal/mol in the gas phase. On the other hand, in the case of the parent ketene, similar density functional theory calculations show that C-silylation is preferred over O-silylation by 8.2 kcal/mol.

Ketenes are unusual alkenes as well as carbonyl compounds that are reactive toward both electrophiles and nucleophiles. From 13C NMR chemical shift studies it has been shown that ketenes are highly polarized molecules with the Cβ and oxygen atoms bearing a substantial negative charge and Cα bearing the positive charge (ref. 1; for a review, see ref. 2). Electrophiles attack at Cβ or oxygen centers from above or below the plane of the ketenes, and nucleophiles attack at Cα in plane. Protonation of ketenes has been the subject of numerous experimental and theoretical studies. Theoretical studies (3-8) show that the protonation at Cβ of ketene 1a (to form acylium ion 2a in Eq. 1) is more favorable than protonation at oxygen (to form O-protonated ion) by 43 kcal/mol. In solution, the site of proton addition to ketenes has been mostly determined from kinetic studies of acid-catalyzed hydration (9-13). We have previously shown that protonation of diphenylketene and of di-tert-butylketene in Magic Acid solution at low temperature resulted in the formation of corresponding acylium ions 2b and 2c (Eq. 1), as indicated by the 13C NMR signals for CO+ at δ13C 154.7 and 154.1, respectively (14).

Long lived trialkylsilyl cations are still elusive in the condensed phase due to their kinetic instability. The in situ generation of trialkylsilyl cations in solution, by hydride transfer/abstraction from organohydrosilanes by the triphenylmethyl (trityl) cation (so-called “Corey hydride transfer”) (15), has been adopted for the preparation of a number of silylated onium ions (16-21). However, electrophilic silylation of ketenes has not yet been reported. Thus far, silylation of ketenes employing silicon compounds has been known to produce only silyl enol ethers, formed by nucleophilic addition of a leaving group, such as CN- or N- 3, at Cα followed by O-silylation (22, 23). Herein, we report a theoretical and experimental study of electrophilic silylation of ketenes in solution. In principle, the incipient silyl cation may react with the Cβ of ketene to give β-silylacylium ion or with oxygen to give silylcarboxonium ion due to strong Si-O bond formation (Eq. 2).

Experimental Methods

The preparation of β-silyl carbenium ion 5 was as follows. Diphenylketene (0.13 mmol) and triethylsilane (25 mg, 0.22 mmol) in dry CD2Cl2 (0.25 ml) were loaded into a dry, 5-mm NMR tube in a glove box under an argon atmosphere. The NMR tube was taken out of the glove box and then cooled by using dry ice/acetone to -78°C still under argon. A solution of Ph3C+ B(C6F5)- 4 (101 mg, 0.11 mmol) in dry CD2Cl2 (0.5 ml), prepared in the glove box and cooled to -78°C, was then added to the NMR tube. The NMR tube was sealed at -78°C, and the solution was rapidly stirred by using a vortex stirrer. The obtained dark-red solution was subjected to NMR analysis at -78°C. 1H NMR (300 MHz, CD2Cl2) δ 0.94, 1.03, 7.13 ∼ 7.59 (m); 13C NMR (75 MHz, CD2Cl2) δ 4.3, 5.9, 59.3, 128.4, 130.1, 130.2, 133.2, 187.1; 29Si NMR (59.6 MHz, CD2Cl2) δ 49.6. The 1H and 29Si NMR spectra of triethylsilylated ketene 5 at -78°C are shown in Figs. 2 and 3, respectively.

Fig. 2.

Proton-decoupled 13C NMR spectrum (75 MHz) of 5 in CD2Cl2 at -78°C.


Fig. 3.

29Si NMR spectrum (60 MHz) of 5 in CD2Cl2 at -78°C.


Results and Discussion

We have initially carried out density functional theory calculations using the program gaussian 98 (24) to ascertain the relative stability and structural information of Cβ- and O-silylated isomers of each of 1a and 1b. We have fully optimized the structures at the B3LYP/6-311+G* level. Optimized geometries are depicted in Fig. 1. Two possible structures of trimethylsilylated ketene, i.e., 3a and 4a, were calculated. Both were found to be stable minima on the potential energy surface as indicated by the frequency calculations at the B3LYP/6-311+G*//B3LYP/6-311+G* level. It was found that C-silylated 3a was more stable than O-silylated 4a by 8.2 kcal/mol at the B3LYP/6-311+G*//B3LYP/6-311+G* + ZPE [zero point vibrational energies, which were scaled by a factor of 0.98 (25)] level (Table 1). We have also calculated the structures and energetics of Cβ- and O-trimethylsilylated diphenylketene 3b and 4b, respectively, at the same level of theory. However, unlike silylation of ketene 1a, C-silylated 3b is in fact 5.4 kcal/mol less stable than O-silylated 4b as shown in Table 1. Interestingly, the cation 3b shows a relatively long Si-Cβ bond distance of 2.121 Å (Fig. 1), indicating the involvement of substantial Si-C hyperconjugation. In contrast, in the case of protonated diphenylketene, the same level of calculations showed that the oxygen protonated form is 31.6 kcal/mol less stable than the Cβ-protonated form. To evaluate the effect of substituents on the relative stability of Cβ- and O-silylated isomers of ketenes, we calculated the energies of both isomers with different substituents (3d-3f and 4d-4f in Eq. 2). At the same level of density functional theory calculation, 3d is more stable than 4d by 4.1 kcal/mol. However, in the case of ketenes with both substituents bulkier than H, the O-silylated isomers were found to be more stable: 4e is more stable than 3e by 0.3 kcal/mol, and 4f is more stable than 3f by 1.9 kcal/mol (Table 1). We have also computed 13C and 29Si NMR chemical shifts of 3b and 4b using the IGLO method (26) at the II′ level using B3LYP/6-311+G* geometries (Table 2). Huzinaga (27) Gaussian lobes were used as follows. Basis II′: Si, 11s 7p 2d contracted to [51111111, 211111, 11], d exponent = 1.4 and 0.35; C, O: 9s 5p 1d contracted to [51111, 2111, 1], d exponent = 1.0; H: 3s contracted to [21]. The 13C and 29Si NMR chemical shifts were referenced to tetramethylsilane (calculated absolute shift, i.e., σ(Si) = 379.3 and σ(C) = 196.8). Whereas the δ29Si of 3b was calculated to be 44.7, the δ29Si of 4b was computed to be 94.7.

Fig. 1.

B3LYP/6-311+G*-calculated structures of trimethylsilylated ketenes 3-4.


View this table:
Table 1. Total energies, ZPE, and relative energies
View this table:
Table 2. Experimental and computed 13C and 29Si NMR chemical shifts of diphenylketene, protonated diphenylketene, and silylated diphenylketene

To resolve these conflicting theoretical results between ketene and diphenylketene, we silylated the latter in solution with the in situ-generated incipient triethylsilyl cation. The 1H and 29Si NMR spectra are shown in Figs. 2 and 3. The 13C and 29Si NMR chemical shifts of the silylated diphenylketene 5 in solution are summarized in Table 2. The silylation of diphenylketene thus afforded a single isomer, Cβ-silylated diphenylketene 5, as indicated by 1H, 13C, and 29Si NMR spectroscopy. The isomer is stable at temperatures up to -20°C. The selected 13C NMR data are compiled in Table 2. The peak at δ13C 187.1 ppm is assigned to the CO+ carbon. Comparison of the 13C NMR data for silylated diphenylketene 5 and protonated diphenylketene 2b is informative concerning the charge distribution of the acylium ions. The reported δ13C of the CO+ in 2b is 154.7 ppm, consistent with the typical acylium ion resonance structures (28). Thus, the chemical shift of the CO+ carbon in 5 is deshielded relative to that in 2b by 32.4 ppm. The peak at δ13C 59.3 ppm is assigned to the Cβ of 5. This resonance is also deshielded relative to the corresponding carbon peak of 2b by 10.4 ppm. These results indicate enhanced charge delocalization in 5 compared with 2b due to β-silyl hyperconjugation (29-33). Whereas ion 2b is primarily stabilized by delocalization involving oxygen (oxonium ion character), the stabilization of ion 5 seemingly involves both the oxygen atom and Si-C hyperconjugation (both oxonium ion and ketene-like character; the latter results in deshielding of the acylium ion carbon). The 29Si NMR spectrum of 5 showed only a single resonance at δ 49.7 ppm, substantially deshielded from that of Et3SiH (δ -0.2 ppm in CD2Cl2). This value also suggests some transfer of positive charge to silicon through hyperconjugation. The calculated 13C NMR chemical shifts of CO+ and Cβ carbons of 3b are 174.3 and 34.2 ppm, respectively. These are substantially more shielded by 12.8 and 25.1 ppm, respectively, than the experimentally observed values of 187.1 and 59.3 ppm in 5. The discrepancy between experimental and calculated values may be improved by using correlated level calculations such as the GIAO-MP2 method. However, GIAO-MP2 calculations using gaussian 98 (24) are presently limited to only small molecules (limits also strongly depend on molecular asymmetry). On the other hand, that the δ29Si of 3b computed to be 44.7 agrees very well with the experimental value of 49.6 in 5.

In conclusion, we have shown that silylation of diphenylketene by reacting diphenylketene with in situ-generated electrophilic incipient triethylsilyl cation results in exclusive formation of Cβ-silylated ion 5. This result is in contrast with results of density functional theory calculations relating to the idealized gas phase, indicating that solvation may play a major role in stabilizing the Cβ-silylated ion 5 in the condensed phase. Comparison of the 13C NMR spectra of 5 with that of Cβ-protonated diphenylketene also suggests that the β-silyl group plays a significant role in charge delocalization through β-silyl hyperconjugation, with some positive charge transferred to silicon. The calculated NMR chemical shifts correlate well with the observed experimental chemical shifts.

Acknowledgments

This work was supported by the National Science Foundation.

Footnotes

  • To whom correspondence should be addressed. E-mail: olah{at}usc.edu.

References

« Previous | Next Article »Table of Contents