Production of α-siloxycarbenium ions by protonation of photochemically generated α-siloxycarbenes. Formation mechanism and reactivities with nucleophiles

Article abstract of DOI:10.1021/ja9603970

The acyltrimethylsilanes 4-RC₆H₄C(O)SiMe₃ (R = H, Me, MeO) and β-naphthylC(O)SiMe₃, upon photolysis in acetonitrile with 20 ns pulses of 248 nm light from an KrF* excimer laser, give rise to the corresponding α-siloxycarbenes ArC:OSiMe₃, whose absorption spectra (λ(max) between 270 and 310 nm), lifetimes (between 130 and 260 ns), and reactivities with proton donors (ROH, mainly alcohols) are reported. With highly acidic ROH, such as 1,1,1,3,3,3-hexafluoroisopropyl alcohol (HFIP), the reaction is of simple second order, rate constants being in the 10⁹ M^-1 s^-1 range and virtually independent of the nature of the aromatic moiety of the carbene. Isotopic substitution of H in ROH by D has no effect on the rate constant for reaction with carbene. For less acidic alcohols such as, e.g., methanol, the reactivity of the carbenes increases with increasing [ROH]. This behavior is interpreted in terms of reversible adduct formation between carbene and alcohol followed by reaction with further alcohol molecule(s) to give product. On the basis of experiments in the acidic and only weakly nucleophilic solvents 2,2,2-trifluorethanol (TFE) and HFIP, protonation of the carbenes leads to the corresponding carbenium ions, whose absorption spectra (λ(max) between 305 and 355 nm), lifetimes (100 ns-5 μs in TFE), and reactivities with nucleophiles (halides, alcohols, and ethers) are reported. In the solvent HFIP, the reactivities of the carbenium ions with the alcohols and ethers increase with their concentration, in a way analogous to that observed in the reaction of the carbenes with the alcohols. This is explained as resulting from reversible formation of a cation-nucleophile complex followed by reaction of the complex with a second nucleophile molecule which acts as a base. In solvents more basic than HFIP, it is presumably the solvent which serves this function.

Full text of DOI:10.1021/ja9603970

10838
J. Am. Chem. Soc. 1996, 118, 10838-10849
Production of R-Siloxycarbenium Ions by Protonation of
Photochemically Generated R-Siloxycarbenes. Formation
Mechanism and Reactivities with Nucleophiles
Wolfgang Kirmse,1 Michael Guth,1 and Steen Steenken*,2
Contribution from the Fakulta¨t fu¨r Chemie, Ruhr-UniVersita¨t Bochum,
D-44780 Bochum, Germany, and Max-Planck-Institut fu¨r Strahlenchemie,
D-45413 Mu¨lheim, Germany
ReceiVed February 7, 1996X
Abstract: The acyltrimethylsilanes 4-RC6H4C(O)SiMe3 (R ) H, Me, MeO) and â-naphthylC(O)SiMe3, upon photolysis
in acetonitrile with 20 ns pulses of 248 nm light from an KrF* excimer laser, give rise to the corresponding
R-siloxycarbenes ArC:OSiMe3, whose absorption spectra (λmax between 270 and 310 nm), lifetimes (between 130
and 260 ns), and reactivities with proton donors (ROH, mainly alcohols) are reported. With highly acidic ROH,
such as 1,1,1,3,3,3-hexafluoroisopropyl alcohol (HFIP), the reaction is of simple second order, rate constants being
in the 109 M-1 s-1 range and virtually independent of the nature of the aromatic moiety of the carbene. Isotopic
substitution of H in ROH by D has no effect on the rate constant for reaction with carbene. For less acidic alcohols
such as, e.g., methanol, the reactivity of the carbenes increases with increasing [ROH]. This behavior is interpreted
in terms of reversible adduct formation between carbene and alcohol followed by reaction with further alcohol
molecule(s) to give product. On the basis of experiments in the acidic and only weakly nucleophilic solvents 2,2,2-
trifluorethanol (TFE) and HFIP, protonation of the carbenes leads to the corresponding carbenium ions, whose
absorption spectra (λmax between 305 and 355 nm), lifetimes (100 ns-5 µs in TFE), and reactivities with nucleophiles
(halides, alcohols, and ethers) are reported. In the solvent HFIP, the reactivities of the carbenium ions with the
alcohols and ethers increase with their concentration, in a way analogous to that observed in the reaction of the
carbenes with the alcohols. This is explained as resulting from reversible formation of a cation-nucleophile complex
followed by reaction of the complex with a second nucleophile molecule which acts as a base. In solvents more
basic than HFIP, it is presumably the solvent which serves this function.
Introduction
tolysis (eq 2),8,9
The reaction of a series of diarylcarbenes with ROH has
recently been shown to involve protonation (eq 1) yielding the
corresponding carbenium ions (benzhydryl cations) which were
identified using nanosecond optical and conductance detection
methods.3 Experiments with picosecond time resolution4 are
in support of the protonation mechanism suggested,3,5-7 the
lifetime of the parent diphenylcarbenium ion in neat aliphatic
alcohols being in the 40-90 ps range.4
O
.
hν
.
RC SiR′3
RCOSiR′3
(2)
and this method has recently been used to trap phenyltrimeth-
ylsiloxy carbene by reaction with pyridine yielding the corre-
sponding ylide.10 The formation of carbene has been shown9
to proceed via the triplet state of the acylsilane. In the case of
benzoyltrimethylsilane, the lifetime of the triplet in CH2Cl2
solution has been estimated to be 50-100 ns.10 This sets a
lower limit with respect to the lifetime of carbocations that may
be produced by protonation of the carbenes generated from the
acylsilane triplets. It has recently been shown11-13 that lifetimes
of highly reactive carbocations can be quite long in fluorinated
alcohols (due to the solvents’ low nucleophilicity), allowing their
detection on the gnanosecond time scale. Due to the stabilizing
Ar2C: + ROH f Ar2CH+ + RO-
(1)
Benzhydryl cations are obviously highly stabilized, due to
the delocalization of the positive charge over two aromatic rings,
and thus the driving force for cation formation from the carbene
should be considerable. With the less stabilized benzyl systems
there is, in principle, the possibility that protonation of the
carbene by O-H bonds to give the cation is thermodynamically
unfavorable. We have, for this reason, studied a series of para-
substituted phenylcarbenes and a naphthylcarbene carrying the
trimethylsiloxy group. These carbenes can be conveniently
generated from the corresponding acyltrimethylsilanes by pho-
(8) Brook, A. G.; Kivisikk, R.; Legrow, G. E. Can J. Chem. 1965, 43,
1175. Brook, A. G.; Pierce, J. B.; J. Org. Chem. 1965, 30, 2566. Brook,
A. G.; Kucera, H. W.; Pearce, R. Can. J. Chem. 1971, 49, 1618. Duff, J.
M.; Brook, A. G. Can J. Chem. 1973, 51, 2869.
(9) Bourque, R. A.; Davis, P. D.; Dalton, J. C. J. Am. Chem. Soc. 1981,
103, 697.
(10) Perrin, H. M.; White, W. R.; Platz, M. S. Tetrahedron Lett. 1991,
32, 4443. The same method has been used to intercept carboethoxycarbene
(Toscano, J. P.; Platz, M. S.; Nikolaev, V.; Popic, V. J. Am. Chem. Soc.
1994, 116, 8146).
(11) McClelland, R. A.; Mathivanan, N.; Steenken, S. J. Am. Chem. Soc.
1990, 112, 4857.
X Abstract published in AdVance ACS Abstracts, October 1, 1996.
(1) Ruhr-Universita¨t.
(2) Max-Planck-Institut.
(3) Kirmse, W.; Kilian, J.; Steenken, S. J. Am. Chem. Soc. 1990, 112,
6399.
(4) Chateauneuf, J. E. J. Chem. Soc., Chem. Commun. 1991, 1437.
(5) Kirmse, W. Justus Liebigs Ann. Chem. 1963, 666, 9.
(6) Belt, S. T.; Bohne, C.; Charette, G.; Sugamori, S. E.; Scaiano, J. C.
J. Am. Chem. Soc. 1993, 115, 2200.
(12) McClelland, R. A.; Chan, C.; Cozens, F.; Modro, A.; Steenken, S.
Angew. Chem., Int. Ed. Engl. 1991, 30, 1337.
(13) Cozens, F.; Li, J.; McClelland, R. A.; Steenken, S. Angew. Chem.,
Int. Ed. Engl. 1992, 31, 743.
(7) See also Schepp, N. P.; Wirtz, J. J. Am. Chem. Soc. 1994, 116, 11749.
S0002-7863(96)00397-6 CCC: $12.00 © 1996 American Chemical Society

Production of R-Siloxycarbenium Ions
J. Am. Chem. Soc., Vol. 118, No. 44, 1996 10839
effect of heteroatoms at the R-position, siloxy-substituted benzyl
cations can be expected to have lifetimes in excess of that (20
ns) analyzable with the presently available apparatus. With
respect to carbene protonation, fluorinated alcohols have the
additional advantage of a high Bro¨nsted acidity which is likely
to lead to a high rate of conversion of carbene into carbenium
ion, eq 3:
RC:OSiR′3 + R′′OH f RCH+OSiR′3 + R′′O- (3)
For these reasons, by studying the reactions of (photochemi-
cally produced) carbenes with and in acidic alcohols, it was
hoped to gain new information relating not only to these
carbenes and their nucleophilic reactivity but also to the
carbenium ions derived from them by protonation. Since the
carbenes are heteroatom (O) substituted at CR, they can be
expected to be singlet-state species. Mechanistic complications
due to contribution of triplet reactions (from rapid singlet-triplet
interconversion) can thus be regarded as unlikely.
Figure 1. Phosphorescence emission spectrum observed on excitation
(λ ) 250 nm) of PhC(O)SiMe3 (0.2 mM, OD/cm ) 1.55) in a
deoxygenated EtOH glass at 77 K.
Experimental Section
The solvents used were acetonitrile (AN; spectroscopic grade, from
Merck), 2,2,2-trifluoroethanol (TFE; from Aldrich), or 1,1,1,3,3,3-
hexafluoroisopropyl alcohol (HFIP; from Hoechst). TFE and HFIP
were dried with Na2SO4 and then distilled over NaHCO3 (to remove
trace acid impurities). HFIP was additionally zone-refined to a purity
g99.8%.
tions of 0.2-0.5 mM. The solutions were flowed (in some cases, after
bubbling with oxygen or argon) through the 2 × 4 mm Suprasil quartz
cell (flow rates ca. 0.5 mL/min). The light-induced optical transmission
changes were digitized in parallel by Tektronix 7612 and 7912 transient
recorders interfaced with a DEC LSI11/73+ computer which also
process-controlled the apparatus and on-line preanalyzed the data. Final
data analysis was performed on a Microvax I connected to the LSI.
The benzoyltrimethylsilanes 4-RC6H4C(O)SiMe3 (R ) H, Me, MeO)
and â-naphthylC(O)SiMe3 were synthesized as described14 and chro-
matographed to a purity of g99% by preparative HPLC using a 30 cm
Polygosil 60-10 NO2 column (diameter 2.5 cm) and pentane:Et2O 1:1
as eluent. The IR and NMR spectra were in agreement with those14
reported in the literature. The aldehydes 4-RC6H4CHO could not be
completely removed,15 although their concentration could be kept below
0.5%. The purity was checked by GC on capillary columns (35 m
OV17, 140 °C, and 24.5 m OV1, 150 °C). For product analysis
experiments, deoxygenated 15 mM solutions of the silanes in methanol
or 2,2,2-trifluoroethanol (TFE) with 0.01% pyridine added (to neutralize
acid impurities) were photolyzed at 0 °C with the light of a high pressure
Hg lamp (Hanau TQ 150) in a quartz reactor, and they were analyzed
using GC, NMR, and HPLC techniques with authentic compounds for
comparison. The mixed acetals ArCH(OSiMe3)OR were characterized
Results and Discussion
1. Photolysis Products. The products of the stationary
photolysis of deoxygenated 15 mM solutions of the benzoyl-
trimethylsilanes 4-RC6H4C(O)SiMe3 (1, R ) H, Me, MeO) in
water-free methanol were the mixed acetal 4-RC6H4CH(OMe)-
OSiMe3 (>90%), the acetal 4-RC6H4CH(OMe)2, and 4-RC6H4-
CHO (together e 10%). â-NaphthylC(O)SiMe3 (2) behaved
analogously. In TFE, the mixed acetals ArCH(OCH2CF3)-
OSiMe3 and the aldehydes ArCHO were obtained in a molar
ratio ≈ 3:2.
The acetals, whose formation is in agreement with previous
findings,8,9 are thought to arise (see later) from the photochemi-
cally generated carbenes ArC:OSiMe3 by reaction with metha-
nol. Since “reduction products” ArCH2OSiMe3 or Ar(OSiMe3)-
CHCH(OSiMe3)Ar were not found, it is concluded that the
carbenes do not react by H-abstraction. This indicates that they
are in the singlet state.
1
by their H NMR spectra (CDCl3): Ar ) Ph, R ) Me: δ ) 0.17 (s,
9H), 3.2 (s, 3H), 5.8 (s, 1H), 7.1-7.7 (m, 5H); Ar ) Ph, R ) CH2CF3:
δ ) 0.1, (s, 9H), 3.65 (qm, J ) 8.5 Hz, 2H), 5.88 (s, 1H), 7.1-7.7 (m,
5H); Ar ) 4-Me-C6H4, R ) CH2CF3: δ ) 0.17 (s, 9H), 2.33 (s, 3H),
3.9 (qm, J ) 9 Hz, 2H), 6.0 (s, 1H), 7.1-7.5 (m, 4H); Ar ) 4-MeO-
C6H4, R ) Me: δ ) 0.12 (s, 9H), 3.79 (s, 3H), 5.68 (s, 1H), 6.88 (dm,
J ) 9 Hz, 2H), 7.35 (dm, J ) 9 Hz, 2H); Ar ) 4-MeO-C6H4, R )
CH2CF3: δ ) 0.16 (s, 9H), 3.81 (s, 3H), 3.9 (qm, J ) 9 Hz, 2H), 5.99
(s, 1H), 6.93 (dm, J ) 9 Hz, 2H), 7.4 (dm, J ) 9 Hz, 2H); Ar )
â-naphthyl, R ) Me: δ ) 0.14 (s, 9H), 3.31 (s, 3H), 5.9 (s, 1H), 7.4-
7.7 (m, 3H), 7.8-8.0 (m, 4H); Ar ) â-naphthyl, R ) CH2CF3: δ )
0.18 (s, 9H), 4.0 (qm, J ) 9 Hz, 2H), 6.2 (s, 1H), 7.35-8.05 (m, 7H).
IR measurements of solutions of MeOH, t-BuOH, TFE, and HFIP
in CCl4 or CH3CN as solvents were performed with a Bruker IFS66
FTIR spectrometer, using 1-5 mm IR-transparent “Infrasil” quartz and
0.5 mm CaF2 cells.
With the R ) MeO system, photolyses in TFE were also
performed in the presence of 18-20 mM Me4N+Cl- or
n-Bu4N+F- × 3H2O. It was found that Cl- had no effect on
the product ratio, whereas, with F-, the aldehyde 4-MeOC6H4-
CHO was the main product, and only traces of the mixed acetal
4-MeOC6H4CH(OMe)OSiMe3 were detectible (yield e 1%).
The production of aldehyde is explained in terms of reaction
of the siloxycarbenium ion with F- (oxygen displacement
reaction by F-, eq 7), see section 5. This reaction is analogous
to the F--induced fragmentation16 of trimethylsilyl-containing
acetals.
2. Phosphorescence Measurements. Deoxygenated solu-
tions of 1 (R ) H, Me, MeO) and of 2 (optical density at 250
nm ≈ 1.5 per cm) in ethanol or in n-butyronitril at 77 K (where
they exist as glasses) were found to phosphoresce when excited
with light of 250 nm. As an example, in Figure 1 is shown the
phosphorescence spectrum of the parent silane (R ) H). There
are two peaks, one at 509 nm17 and a less intense one at 548
For the laser flash photolysis experiments we used a Lambda Physik
EMG103MSC excimer laser which emitted ≈20 ns, 10-40 mJ pulses
of 248 nm light (KrF*). The solutions containing the silanes had optical
densities (OD) at 248 nm of ≈1.5/cm which correspond to concentra-
(14) Brook, A. G. J. Am. Chem. Soc. 1957, 79, 4373. Brook, A. G.,
Quigley, M. A.; Peddle, G. J. D.; Schwartz, N. V.; Warner, C. M. J. Am.
Chem. Soc. 1960, 82, 5102. Picard, J. P. Calas, R.; Donogues, J.; Duffaut,
N.; Gerval, J.; Lapouyade, P. J. Org. Chem. 1979, 44, 420.
(15) Based on HPLC and on their phosphorescence in EtOH glasses16
at 77 K and their (very weak) triplet-triplet absorption spectra (for a review,
see: Carmichael, I.; Hug, G. L. J. Phys. Chem. Ref. Data 1986, 15, 1) in
acetonitrile, TFE, or HFIP at room temperature (see Figures 2 and 4).
(16) Lipshutz, B. J.; Pegram, J. J. Tetrahedron Lett. 1980, 21, 3343.

10840 J. Am. Chem. Soc., Vol. 118, No. 44, 1996
Kirmse et al.
Figure 2. Time-resolved absorption spectra produced on 248 nm photolysis of PhC(O)SiMe3 1 (0.25 mM) in deoxygenated acetonitrile. The
times indicated refer to the situation after the 20 ns pulse. In the insets are shown the carbonyl triplet of 1 with λmax at 290 nm (a), the buildup of
the carbene at 270 nm (b), the decay of the carbene at 270 nm (c), and the dependence of kobsd for decay of carbene on [HClO4] (d).
nm, and a shoulder at 600 nm. The shorter wavelength peak
(λ1) corresponds to an energy of the triplet, ET, of the silane of
56 kcal/mol.18 On the basis of quenching experiments with
isoprene, the triplet energy of the benzoylsilane R ) H has been
estimated to be e 60 kcal/mol,10 which is in agreement with
our directly measured value.
The phosphorescence lifetime at 77 K, determined by
monitoring the decay of the emission intensity at both wave-
lengths, was found as 0.4 ( 0.1 ms, a value within the range
observed for carbonyl triplets.19 Also in liquid ethanol at ≈170
K could emission be seen, which was strongly quenched by
oxygen. This is in support of the luminescence originating from
the triplet state of the carbonyl compound.
The para-substituted benzoylsilanes also showed phospho-
rescence16 in ethanol glasses at 77 K. The λmax values, triplet
energies (calculated from λ1), and lifetimes (τ) are listed in Table
1. It is interesting that the triplet energies of the acylsilanes
are considerably lower than those (≈74 kcal/mol)19a of the
corresponding acylalkanes.20 Low triplet energies of acylsilanes
have been noted previously.9,10
3. Laser Flash Photolysis (LFP) Experiments in Aceto-
nitrile. a. Identification of Carbenes. The time-resolved
absorption spectra observed on 248 nm photolysis of a deoxy-
genated 0.25 mM solution of the benzoylsilane 1 (R ) H) in
acetonitrile (AN) are presented in Figure 2. Insets a-c illustrate
the observations made during the first 130 ns after the pulse.
As seen in inset a, at ≈10 ns, there is a band with λmax at ≈290
nm, which decays very rapidly (k ≈ 6 × 107 s-1). As this band
disappears, a band with λmax ≈ 270 nm grows in (see inset b),21
indicating that the 270 nm species is the product of the 290 nm
species. The 270 nm species decays on a longer time scale
(see inset c) with k ) 4.5 × 106 s-1. This rate is not influenced
by the introduction of O2 (8.2 mM) to the solution, demonstrat-
ing that the species is probably not a radical or triplet. However,
the species absorbing at 290 nm shows some reactivity with
O2, as concluded from the fact that its lifetime and yield are
reduced by 8.2 mM O2. In the presence of O2, the yield of the
270 nm species is also decreased (to 70% of the value in the
absence of O2), which supports the idea that the 270 nm species
is a product of the 290 nm one.
As seen in the spectrum recorded at 90 ns, in addition to the
main band at 270 nm there is a “shoulder” at ≈ 340 nm and a
broad band centered around 500 nm. The species responsible
for the absorptions at 340 and 500 nm can be efficiently
scavenged by O2. It is identified as the triplet of benzaldehyde,
whose presence as a thermal decomposition product of 1 is
difficult to avoid.22 As can be seen in Figure 2, the benzalde-
hyde triplet is longer lived than the species absorbing at 270
nm.
(17) There was also phosphorescence at lower wavelengths, which, based
on the authentic emission spectra (Go¨rner, H.; Kuhn, H. J. J. Phys. Chem.
1986, 90, 5946), is assigned to the corresponding 4-RC6H4CHO present in
the sample as impurity.
On the basis of its reactivity with O2, the 290 nm species is
identified as the triplet of the benzoylsilane 1; R ) H.23 Its
product, which absorbs at 270 and at ≈500 nm, is assigned as
phenyltrimethylsiloxycarbene 1′ (R ) H), formed9,10 according
(18) The triplet energy of acetyltrimethylsilane has been reported9 to be
61 kcal/mol.
(19) For typical lifetimes of carbonyl triplets at 77 K, see, e.g. (a) Calvert,
J. G.; Pitts, J. N. Photochemistry; Wiley: New York, 1966.
(20) In agreement with the low triplet energies of the acylsilanes is the
observation that the triplet of 1 (R ) H) is not able to H-abstract from the
excellent H-atom donor isopropyl alcohol (used as solvent), as judged by
the observability of the triplet of 1 and by the absence of the protonated
ketyl radical, which has λmax at 310 nm (strong and narrow band) and at
450 nm (weaker and broader band). This radical was generated by reaction
(21) The formation of the 270 nm species appears to be faster than the
decay at 309 nm. This apparent enhancement of rate is an artefact caused
by the almost simultaneously occurring decay of the 270 nm species. This
species has a second, very broad band with a center at ≈500 nm.
(22) The spectra in inset a were recorded with a solution containing less
benzaldehyde than in the case of the main figure.
(23) In agreement with this assignment is the fact that it can also be
scavenged by the triplet quencher 1,3-cyclohexadiene: At 1.3 mM 1,3-
cyclohexadiene the triplet is barely visible, and the yield of the 270 nm
species is reduced to 50%.
of the acylsilane with e- produced by radiolysis of isopropyl alcohol-
solv
water mixtures 3:7 (v/v). Inability of the triplet to abstract H from isopropyl
alcohol was also observed9 in the case of acetyltrimethylsilane.

Production of R-Siloxycarbenium Ions
J. Am. Chem. Soc., Vol. 118, No. 44, 1996 10841
Figure 3. Absorption spectra observed on photolysis of a deoxygenated 0.24 mM solution of 1 (R ) MeO) in AN. Recognizable (from the
time-dependence of the spectra) is the delayed formation of the carbene (λmax ) 310 nm) from the (invisible) carbonyl triplet precursor. In the
insets a and b is shown the situation in the presence of 8.2 mM O2 (the ∆OD scale in inset a is the same as that in the main figure), and in c the
effect of TFE on the decay rate of the carbene as monitored at 310 nm.
Table 1. Spectroscopic (and Lifetime) Properties of Acylsilanes and Their Triplets and of Siloxycarbenes
acylsilanea
λ(absorption)/nm; ꢀ/M-1 cm-1
carbonyl tripletb
ET/kcal mol-1
carbenec
max/nm
Ar
λ
max(emission)/nm
τ
λ
τ/ns
4-RC6H4
R ) H
R ) Me
R ) MeO
â-naphthyl
251; 11370
261; 13070
277; 7070
254; 26900
509; 548
510; 545
496; 535
496; 520; 555
56
56
58
58
0.14 ms;b,d 17 nsc,e
0.72 ms;b,dd e 15 nsc,f
2.2 ms;b,d e 15 nsc,f
270
280
310
300
220; 63a
170
130
260
a In cyclohexane at 20 ( 1 °C. b In deoxygenated EtOH at 77 K. c In deoxygenated acetonitrile at 20 ( 1 °C. d Determined via the phosphorescence
decay. Error limit 10%. e Monitored by the triplet-triplet absorption at ≈300 nm, see text. f Based on the absence of a signal attributable to the
triplet.
to eq 4.24 Based on the fact that 1′ (R ) H) does not react
interpreted as (thermal) regeneration of the acylsilane precursor,
since in the absence of carbene traps the optical density returns
to almost its original level. The λmax values of the absorptions
of the carbenes and of the emissions of the carbonyl triplets
and the lifetimes in AN (room temperature) and ethanol (77 K)
are collected in Table 1. Concerning the carbenes 1′, the λmax
values increase from R ) H to R ) MeO, whereas the lifetimes
decrease in this direction.
with O2, the carbene is in the singlet state.
4-RC6H4C(O)SiMe3T f
1
4-RC6H4C:OSiMe3 k ≈ 6 × 107 s-1 for R ) H (4)
1′
b. Reaction of Carbenes with Proton Donors. It was
found that the carbenes can be scavenged by compounds
containing mobile protons. Addition of increasing concentra-
tions of proton donor led to increasing rates kobsd for decay of
carbene and to increasingly negative ∆OD values after comple-
tion of the reaction. This indicates that the reversion of carbene
into acylsilane (which has absorption at λmax of the carbene) is
prevented by the proton donor. The rate enhancement for decay
of 1′ (R ) H) is shown in Figure 2, inset d, for HClO4, and in
Figure 3 for TFE reacting with 1′ (R ) MeO).
To obtain quantitative data on the reactivity of the carbenes,
i.e., rate constants, the (first order) rates of decay of the carbenes
(kobsd) at their λmax values (see Table 1) were monitored as a
function of the concentration of the proton donor. With the
more acidic proton donors, such as 2,2,2-trifluoroethanol (TFE),
1,1,1,3,3,3-hexafluoroisopropyl alcohol (HFIP), and HClO4, the
observed dependences (see inset d in Figure 2 and inset c in
Figure 3) were found to follow the equation kobsd ) ko + kprot
[proton donor], where ko is the rate of decay of the carbene in
From the decay at 290-310 nm and the buildup at 270 nm, the
rate constant for the transformation reaction 1T f 1′ (R ) H)
is 6 × 107 s-1, which corresponds to a lifetime of the triplet of
17 ns. This measured value in acetonitrile is considerably
shorter than the 50-100 ns estimated10 for CH2Cl2 solutions
but essentially the same as that (19 ns) deduced9 for acetyltri-
methylsilane in acetone solution.
Results similar to those described for 1, R ) H, were found
for the two other benzoylsilanes, R ) Me and MeO, and for
the â-naphthoylsilane 2. In all cases the yields of carbenes were
reduced (see Figure 3) in the presence of O2, which is interpreted
in terms of scavenging of the carbene precursor, the carbonyl
triplet, by O2. However, the lifetimes of the carbenes, which
decay by first order kinetics to OD levels slightly above those
before the pulse (see Figures 2 and 3, insets c and b), were not
effected by O2. The first order decay of carbene can be
(24) Evidence for photoinduced siloxycarbene formation was also
obtained in the case of R,â-epoxysilylketones (Scheller, M. E.; Frei, B.
HelV. Chim. Acta 1992, 75, 69).

10842 J. Am. Chem. Soc., Vol. 118, No. 44, 1996
Kirmse et al.
Table 2. Rate Constants (M-1 s-1) for Reaction of Siloxycarbenes ArC¨ OSiMe3 with Proton Donorsa and for Decay in Oxygen-Saturated
Acetonitrile at 20 ( 1 °C
Ar/λ(observation)
proton donor
pKa (ROH)b,c
C6H5/270 nm
4-MeC6H4/280 nm
4-MeOC6H4/310 nm
â-naphthyl/300 nm
H2O
CH3OH
CH3CH2OH
HOCH2CH2OH
CH3OCH2CH2OH
CH3OCH2CH2OD
ClCH2CH2OH
Cl2CHCH2OH
Cl3CCH2OH
15.7
15.5
15.9
15.1
14.8
9.3 × 107 d
9.2 × 107 d
1.0 × 108 (60 mM)e,f
3.8 × 108 (25 mM)e,f
2.0 × 108 (17 mM)e,f
3.0 × 109 (7 mM)
6.3 × 108 (14 mM)e
6.0 × 108 (16 mM)e
2.0 × 109 (7 mM)
2.1 × 109 (6 mM)
2.0 × 109 (3 mM)
1.5 × 109
1.2 × 108d
3.0 × 108 d
2.7 × 108 d
2.2 × 108d
9.1 × 108 (30 mM)
1.6 × 108 (25 mM)e
1.2 × 108 (20 mM)e
1.1 × 109 g
1.5 × 109 (20 mM)
2.6 × 108 (15 mM)e
2.6 × 108 (20 mM)e
14.31
12.89
12.24
12.8
2.0 × 109 g
1.6 × 109 g
CF3CH2OH (TFE)
(CF3)2CHOH (HFIP)
(CF3)2CHOD (HFIP-OD)
CH3CO2H
1.5 × 109
1.8 × 109
1.0 × 109
1.2 × 109 g
1.1 × 109 g
2.4 × 109
1.5 × 109
9.6 × 108
9.9 × 108
1.0 × 109
2.0 × 109
9.3
1.2 × 109
1.2 × 109
1.3 × 109
1.2 × 109
4.75
7.15
1.1 × 109 g
1.0 × 109 g
CH3CO2Dh
HClO4
4-O2NC6H4OH
CH3CN solvent
1.1 × 109 g
1.1 × 109 g
3.0 × 109
2.6 × 109
∼3 × 109
4.5 × 106 i
5.8 × 106 i
7.5 × 106 i
3.9 × 106 i
a In the case of the linear kobsd vs concentration plots, the error limits are (10%. In the other cases, the k values are approximations based on
the low concentration ranges indicated in the parentheses. The k values were determined at the wavelengths given. b From Dyatkin, B. L.; Mochalina,
E. P.; Knunyants, I. L. Tetrahedron 1965, 21, 2991. c From Ballinger, P.; Long, F. A. J. Am. Chem. Soc. 1960, 82, 795. d The kobsd vs [proton
donor] plot is curved upward. Value reported is from a linear fit in the concentration range 0-0.15 M. e The kobsd vs concentration plot is curved
upward. The value is from a linear fit in the concentration range indicated. f Curvature is not very pronounced. g At [proton donor] > 10 mM, the
kobsd vs concentration plots are curved downward. The numbers given are from the linear parts in the low concentration range. h The water content
of CH3CO2D is 2%. i Rate constant ks for carbene decay in spectroscopic grade CH3CN with a water content <0.1%. Unit s-1
.
Figure 5. Dependence of kobsd for decay in AN of 1′ (R ) H),
monitored at 270 nm, on [2-methoxyethanol]. Main figure, ROH; inset,
ROD.
Figure 4. Dependence of kobsd for decay in AN of 1′ (R ) H),
monitored at 270 nm, on [ethanol]. From the inset it is recognizable
that the dependence can be approximated by a parabola.
donors, the kinetic isotope effect, kH/kD, is thus 1.0. In the case
of 2-methoxyethanol, the reactivity data are less reliable, due
to the curvature of the kobsd vs [ ] plots (see Figure 5). However,
it appears (see Table 2) that the kinetic isotope effect is small
(kH/kD e 1.3).
The rate constants for reaction of HClO4 with 1′ are ≈2.7 ×
109 M-1 s-1, which is almost an order of magnitude lower than
the diffusion limit in acetonitrile (2 × 1010 M-1 s-1).25 In order
to check the possibility that the reactions in this solvent of the
highly polarizing (and therefore probably strongly solvated)
proton might be less than 2 × 1010 M-1 s-1, oxygen-saturated
(to scavenge any radicals) solutions containing 0.3 mM R- or
â-naphthol in acetonitrile were photolyzed with 20 ns pulses
of 248 nm laser light. Since the acidity of the electronically
excited singlet state of naphthol is larger by seven orders of
the absence of the proton donor, and kprot the second order rate
constant for reaction of the carbene with the proton donor. In
these cases the second order rate constants kprot (see Table 2)
are (well) defined. In other cases, such as the reactions of H2O,
methanol, ethanol, or 2-methoxyethanol with the carbenes 1′
and 2′, the kobsd vs concentration plots are curved upward (see
Figures 4 and 5 and Table 2). In other cases, such as the
2-chloroethanols (see Figure 6) and acetic acid, the plots are
curved downward. In these nonlinear cases the numbers given
in Table 2 are approximations based on the initial, quasi-linear
parts and refer only to the concentration ranges indicated in
the footnotes.
In the case of acetic acid, HFIP, and 2-methoxyethanol (see
inset Figure 5), the rate constants (listed in Table 2) were also
measured for reaction of the carbenes 1′ with the O-deuterated
molecules. With HFIP, which has a high reactivity (k ≈ 1.2 ×
109 M-1 s-1), replacement of the alcoholic proton by D has no
influence on the rate constant for reaction with 1′ (R ) H, MeO).
Analogously, O-deuteration does not change the reactivity of
acetic acid with 1′ (R ) H, Me, MeO). With these proton
(25) The rate constant for reaction of the related carbene, benzylchlo-
rocarbene, with hydrogen chloride in isooctane has been measured by Liu,
M. T. H.; Chateauneuf, J. E. J. Chem. Soc., Chem. Commun. 1991, 1575,
to be 4.7 × 109 M-1 s-1, which is also below the diffusion limit in this
solvent (calculated from the viscosity () 0.345 cp) to be 1.9 × 1010 M-1
s-1 at room temperature, using the Smoluchovsky equation).

Production of R-Siloxycarbenium Ions
J. Am. Chem. Soc., Vol. 118, No. 44, 1996 10843
4. LFP Experiments in TFE and HFIP. Formation and
Absorption Bands of Carbenium Ions. If the reaction between
the carbenes and the proton donors proceeds by proton transfer,
carbenium ions are the products. If it is desired to directly detect
these species, the reaction should be conducted such that the
carbenium ions are formed (a) rapidly and (b) in a non-
nucleophilic environment. Conditions (a) and (b) are both
fulfilled by working in acidic and non-nucleophilic solvents such
as TFE and HFIP. In Figure 7 are shown the time-resolved
absorption spectra recorded on 248 nm photolysis of 1, R )
MeO, in TFE. At ≈90 ns after the pulse, a species characterized
by a band with λmax ) 355 nm is observed.31 In the region
250-310 nm, there is a decrease of OD which is caused by
depletion of parent (λmax ) 277 nm, see Table 1). The 355 nm
species decays with first order kinetics (see inset b), the rate (k
) 2.0 × 105 s-1) being insensitive to O2. As this species
disappears, the absorbance at ≈300 nm increases (reflecting
product formation, isosbestic point at 305 nm). The rate of
decay is strongly accelerated by nucleophiles such as N3- (see
inset c) or MeOH (see inset d). On the basis of its nucleophilic
behavior and its nonreactivity with O2, the 355 nm species is
identified as the cation 1′H+, formed by protonation of the
carbene 1′ (R ) MeO), see eq 5.
Analogous observations were made in the case R ) H, where
the species detected (identified as the cation 1′H+, R ) H) had
λmax at 305 nm. Its decay rate (1.0 × 107 s-1) is much larger
than that measured for R ) MeO (2.0 × 105 s-1). A rate much
lower than that in TFE (1.0 × 107 s-1) is observed in the less
nucleophilic32 solvent HFIP (1.0 × 104 s-1). On this basis, the
decay of the cation is assigned to reaction with solvent (rate
constant ks). As in the case of R ) MeO, the rate is not
influenced by oxygen but very strongly by nucleophiles such
as alcohols, ethers, and anions. The experimentally observed
rate kobsd for decay of the species follows in many cases the
equation kobsd ) ks + kNu[nucleophile], with kNu being the
second-order rate constant for reaction with nucleophile. The
bimolecular electrophilic reactivity thus characterized (see Table
3) is in support of the species being a cation, the phenyltri-
methylsiloxycarbenium ion (R ) H), formed by protonation by
HFIP (or TFE) of the carbene, eq 5:
Figure 6. Dependence of kobsd for decay in AN of 1′ (R ) H),
monitored at 270 nm, on [2-chloroethanol]. In the inset it is shown
that at low [2-chloroethanol] the dependence is essentially linear.
magnitude than that of the ground state,26 photolysis leads to
deprotonation which results in a rapid increase in the conduc-
tance of the solution, followed by a slow (second order) decrease
to the prepulse level, this process being due to ion recombination
(protonation of the naphtholate anion). It was found that by
adding HClO4 to the solution, the rate of conductance decrease
after the pulse increased dramatically, with the kinetics changing
from second to first order. The kobsd values for conductance
+
+
decrease were found to follow the equation kobsd ) k0 + kH [H ],
+
from which the protonation rate constants, kH , resulted as 4 ×
1010 and 3.6 × 1010 M-1 s-1 for R- and â-naphtholate,
respectively.27,28 These values are clearly aboVe the diffusion
limit for reaction of neutrals (2 × 1010 M-1 s-1), but they are
in line with the numbers calculated29 for ionic reactants of
opposite charge. From these numbers it is evident that rate
constants for proton transfer reactions in acetonitrile are not
“intrinsically” below diffusion control. The absence of a kinetic
isotope effect in the case of the reactions of HFIP and acetic
acid with the carbenes 1′ and of acetic acid with 2′ is thus
probably not due to these reactions being diffusion controlled
at ≈1 × 109 M-1 s-1 30
.
In agreement with this conclusion is
4-RC6H4C:OSiMe3 + R′OH f
1′
the observation that in the case of 2-methoxyethanol, where the
rate constants are only (2-6) × 108 M-1 s-1 (see Table 2), i.e.,
certainly below diffusion control, there is also little difference
between the O-H and O-D systems.
4-RC6H4CH+OSiMe3 + R′O- (5)
1′H+
While the reaction mechanism for weakly acidic alcohols is
not very clear at the moment (see, however, discussion in section
6), conclusive evidence is available for stronger acids. When
4-nitrophenol was used as the proton donor, the reaction with
1′ (R ) H) resulted in the production of the 4-nitrophenolate
anion, as judged by its characteristic spectrum with λmax at 380
nm. The conclusion is thus that the reaction involves proton
transfer. Due to partly overlapping spectra, the rate constant
could not be determined with a high degree of precision, but
the value is large (≈3 × 109 M-1 s-1, Table 2).
ks
98
4-RC6H4CH+OSiMe3 + R′OH
1′H+
4-RC6H4CH(OSiMe3)OR′ + H+ (6)
R′ ) (CF3)2CH, CF3CH2
The rate of formation of the carbenium ion from the carbene
can be estimated if it is assumed that the rate constant for
reaction 5 in neat HFIP ()9.5 M) is the same as that measured
(1.2 × 109 M-1 s-1, Table 2) in acetonitrile. The result
corresponds to a lifetime of the carbene in HFIP of 90 ps. Since
the observed rate of formation of the cation is much smaller,
(26) For a review, see: Ireland, J. F.; Wyatt, P. A. H. AdV. Phys. Org.
Chem. 1976, 12, 131.
(27) An analogous approach to determine the rate constant for reaction
of H+ with an amine (as a respresentative of a neutral reaction partner)
was unsuccessful.
(28) In aqueous solution, the rate constant for reaction of H+ with
R-naphtholate anion is 5.9 × 1010 M-1 s-1 (Schuler, R. H. J. Phys. Chem.
1992, 96, 7169).
(31) The weak, broad band at ≈400 nm, which is absent in the presence
of O2 (see inset a), is due to the triplet of 4-MeOC6H4CHO, which is present
as an impurity.
(29) See McClelland, R. A.; Kanagasabapathy, V. M.; Banait, N. S.
Steenken, S. J. Am. Chem. Soc. 1991, 113, 1009, and ref 37.
(30) The rate constant for reaction in isooctane of acetic acid with
(32) For quantitative data on the solvolyzing power and the nucleophi-
licity of HFIP, see: Schadt, F. L.; Schleyer, P. v. R.; Bentley, T. W.
Tetrahedron Lett. 1974, 27, 2335. Bentley, T. W.; Clewellyn, G. Progr.
Phys. Org. Chem. 1990, 17, 121.
benzylchlorocarbene is25a 1.0 × 109 M-1 s-1
.

10844 J. Am. Chem. Soc., Vol. 118, No. 44, 1996
Kirmse et al.
Figure 7. Time-resolved absorption spectra observed on photolyzing a deoxygenated 0.16 mM solution of 1 (R ) MeO) in TFE: circles, 90 ns;
squares, 2.6 µs; triangles, 6.7 µs after the pulse. In inset a is shown the situation in the presence of O2, in b the decay of the carbenium ion as
monitored at 355 nm (the decay rate is exactly the same in the absence of O2). The acceleration of the cation decay by N3- and by MeOH is shown
in c and d.
i.e., ≈6 × 107 s-1, this means that the rate-determining step in
the formation of the carbenium ion is the transformation of the
triplet into carbene (eq 4) and not the protonation of the latter.33
is only a small kinetic stabilizing effect associated with replacing
Me with OSiMe3, but as the aromatic becomes less electron
donating the difference becomes larger. In solvolytic studies
of benzyl and 4-methoxybenzyl cations, analogous effects were
reported for a wide range of R substituents.34 Concerning the
effect of solvent, it is evident that the lifetimes of the cations
in HFIP are longer than in TFE by about three orders of
magnitude, as already mentioned for the case of 1′H+, R ) H.
The cations were reacted with a series of nucleophiles (Nu),
alcohols, ethers, and anions. For obtaining the second-order
rate constants kNu, the rate of decay of the cation, kobsd, was
measured at its λmax as a function of [Nu]. In the case of the
solvent TFE, kobsd vs [Nu] plots (for reaction of 1′H+ (R ) Me
and MeO) and of 2′H+) were linear and had ks as the intercept.
The kNu values obtained from the slopes of such plots are
presented in Table 3 in columns 3-5. It is evident that the
reactivities of water and methanol are similar. Among the
alcohols, there is a clear decrease in reactivity from methanol
to tert-butyl alcohol, independent of the nature of the cation.35
Since the basicity/nucleophilicity of the alcohols is not very
different, the decrease of reactivity in the observed direction is
probably due to steric crowding in forming the C-O bond
between the cation and the oxygen of the alcohol in the reaction
in which the mixed acetal9 is formed. Among the ethers
tetrahydrofuran (THF) and 1,4-dioxane, the latter is in all cases
much less reactive, which reflects its considerably lower
nucleophilicity,36 caused by the -I effect of the second oxygen
in the ring.
When the other silanes, 1, R ) Me and 2, were photolyzed
in the solvents HFIP or TFE, transient spectra of the corre-
sponding carbenium ions 1′H+ and 2′H+ were observed. The
lack of reactivity with radical scavengers such as O2 and the
high reactivity with typical nucleophiles such as alcohols, ethers,
and anions are indicative of carbocations. The spectroscopic
and reactivity data of all the cations studied are collected in
Table 3. In the series of the R-trimethylsiloxyphenyl carbenium
ions 1′H+, the substituent in the para position has a strong effect
on the absorption band of the cation: λmax increases from 305
nm for R ) H to 355 nm for R ) MeO. For the â-naphthyl
system, λmax ) 350 nm.
5. Reactivities of Carbenium Ions and Solvent Effects.
Due to the short lifetime (≈100 ns) in TFE of 1′H+ (R ) H),
its reactivity was studied mainly in the less nucleophilic solvent
HFIP. The other cations, 1′H+ (R ) Me, MeO) and 2H+, were
investigated in TFE. Concerning the rate constants for decay
of cation by reaction with solvent, ks, those for 1′H+ decrease
from 1 × 107 to 2 × 105 s-1 and from 1 × 104 to 1 × 102 s-1
in the solvents TFE and HFIP, respectively, i.e., approximately
by two orders of magnitude, in going from R ) H to R ) MeO
(see Table 3). This decrease in reactivity reflects the stabiliza-
tion of the carbenium ions by the electron-donating substituents.
It is interesting that the reactivity of the 4-tolyl- is similar to
that of the â-naphthyl carbenium ion, both with respect to
reaction with solvent (ks) and with the various nucleophiles (kNu,
see below). The rate constants for solvent decay for the cations
ArCH+OSiMe3 are interesting in comparison with previously
reported values for ArCH+Me.34 For Ar ) 4-MeOC6H4 there
(35) Similar differences were seen between the alcohols in their reactivity
with other carbocations, see: (a) Kobayashi, S.; Zhu, Q. Q.; Schnabel, W.
Z. Naturforsch. 1988, 436, 825. (b) Steenken, S.; McClelland, R. A. J.
Am. Chem. Soc. 1990, 112, 9648. (c) Bartl, J.; Steenken, S.; Mayr, H. J.
Am. Chem. Soc. 1991, 7710. (d) Verbeek, J.-M.; Stapper, M.; Krijnen, E.
S.; van Loon, J.-D.; Lodder, G.; Steenken, S. J. Phys. Chem. 1994, 98,
9526.
(36) This is deduced from the lower basicity of dioxane, cf.: Searles,
S.; Tamres, M. Basicity and Complexing Ability of Ethers. In The
Chemistry of the Ether Linkage; Patai, S., Ed.; Interscience: London: 1967;
pp 243-308.
(33) An analogous conclusion was reached10 concerning the rate
determining step in the formation at -40 °C of ylid from triplet 1 (R ) H)
and pyridine via 1′.
(34) Amyes, T. L.; Stevens, I. W.; Richard, J. P. J. Org. Chem. 1993,
58, 6057.

Production of R-Siloxycarbenium Ions
J. Am. Chem. Soc., Vol. 118, No. 44, 1996 10845
Table 3. Rate Constants kNu (M-1 s-1) for Reaction of Carbenium Ions with Nucleophiles (Nu) at 20 ( 1 °C
â-naphthyl
p-RC6H4CH+OSiCMe3; solvent
CH+OSiCMe3
Ph2CH+; solvent
TFEb
R ) H; HFIP
R ) Me; TFE R ) MeO; TFE
solvent TFE
HFIPa
CH3CNc
Nu
λmax ) 305 nm λmax ) 315 nm λmax ) 355 nm λmax ) 350 nm λmax ) 430 nm λmax ) 435 nm λmax ) 435 nm
TFE
HFIP
H2O
MeOH
EtOH
2-PrOH
t-BuOH
HOCH2CH2CH2OH
THF
1 × 107 d
1 × 104 e
1.4 × 106 i
2.8 × 106 i
2.0 × 106 i
8.1 × 105 i
2.1 × 105 i
4.5 × 106 i
2.1 × 106 i
4 × 104 i
2 × 106 e
7 × 103 g
3.8 × 107
5.7 × 107
3.7 × 107
2.0 × 107
8.6 × 106
3.9 × 107
3.0 × 107
4.0 × 106
7.1 × 108
1.8 × 108
1.3 × 108
l
2 × 105 e
1 × 102 g
9.5 × 106
7.8 × 106
6.3 × 106
2.3 × 106
6.8 × 105
1.6 × 106
2.2 × 106
5.0 × 105
1.7 × 108
2.1 × 107
2.2 × 107
1.5 × 107
3.1 × 109
1 × 107 n
4 × 106 e
1 × 103 g
4.2 × 107
4.9 × 107
3.9 × 107
2.3 × 107
1.4 × 107
6.8 × 107
3.9 × 107
3.2 × 106
9.4 × 108
1.3 × 108
5.3 × 107
l
3.0 × 106 d
2.5 × 106 f
5 × 103 g,h
(1-2) × 103 i
5 × 105 i
1.1 × 106
1.0 × 107
1.3 × 107
7.8 × 106
3.2 × 106
1.2 × 108
1.2 × 109
8.5 × 108
3.1 × 108
6.9 × 107
3 × 105 i
(1-2) × 105 i
9 × 104 i
6 × 105 i
4.9 × 104
2.2 × 104
3.8 × 107
8.1 × 108
3.8 × 109
8.6 × 109
5.6 × 108
k
k
1,4-dioxane
1.3 × 106
1.9 × 108
1.9 × 109
5.9 × 109
7.5 × 109
5.3 × 109
F-
3.2 × 107
8.4 × 106
3 × 106
2.0 × 1010
2.0 × 1010
2.1 × 1010
1.9 × 1010
2.1 × 1010 m
Cl-
Br-
I-
1.4 × 106
2.6 × 108
-
N3
4.9 × 109
5.4 × 109
5.2 × 107 n
a Cation produced by photoheterolysis of Ph2CHOH (ref 35). b Cation produced from 1,1,2,2-tetraphenylethane radical cation by C-C fragmentation
(Faria, J. L.; Steenken, S. Unpublished results). c Data from ref 32c. d Rate (ks/s-1) in TFE as solvent. e ks, units s-1 f Solvent rate (ks) in CH3CN.
.
g Rate (ks/s-1) in HFIP as solvent. h Data from Faria, J. L. i The kobsd vs concentration plots are curved upwards, see text. The reported second order
k values are therefore upper limits, obtained from the tangent in the concentration range 0-0.25 M. k Reaction is reversible (Bartl, J.; Mayr, H.;
Steenken, S. Unpublished results). l No effect on I- on decay of cation seen up to 0.018 M. m This work. n In HFIP as solvent.
A very interesting phenomenon is that the halides react with
the cations 1′H+ and 2′H+ in the reactivity order F- > Cl-
In fact, azide is the most reactive nucleophile with 1′H+ and
9
2′H+. With kN ≈ (3-5) × 10 M-1 s-1, the reaction in TFE
-
>
3
Br- > I-, i.e., in a way opposite to the halide nucleophilicities.
In order to test whether this “inversion” of normal reactivity
has to do in any way with the fluoroalcohol solvents, the
benzhydryl cation Ph2CH+ was produced by 248 nm photolysis
in HFIP of Ph2CHOH37 or in TFE of tetraphenyloxirane38 or of
1,1,2,2-tetraphenylethane39 and reacted with the nucleophiles
monitoring the cation decay at 430-435 nm. The results are
presented in columns 6 and 7 of Table 3, together with
analogous data obtained35c in acetonitrile. It is clear that in the
solvents TFE and HFIP the reactivity of the halides with Ph2-
CH+ is governed by their nucleophilicity, with the classical order
I- > Br- > Cl- > F-.40 The opposite behavior of 1′H+ and
2′H+ is thus an intrinsic property of these cations and not of
the solvents. Their higher reactivity with the smaller halides
can be explained by assuming (initial) reaction not with the
cationic center CR, but with the silicon atom. Particularly for
Hal ) F, an oxygen-displacement reaction (eq 7) could be
thermodynamically more favorable than attachment to CR, due
to the large Si-F bond energy.
is diffusion controlled.43 However, in HFIP the rate constant
is only 2.6 × 108 (for 1′H+, R ) H), 1 × 107 (for 1′H+, R )
MeO), and 5.2 × 107 M-1 s-1 (for 2′H+). Since the rate
constant for diffusion in HFIP is 4.2 × 109 M-1 s-1 37,44
,
these
low numbers (and the similarly low number for Ph2CH+, see
Table 3) are remarkable. A possible explanation involves
-
decrease of the effective nucleophilicity of the strong base N3
(pKa(N3H) ) 4.75)45 by hydrogen-bonding (quasi-protonation)
from the acidic O-H group of HFIP.
OR
OR
H • • N3H • • –OR
RO
H
H
H • • N3– • • H OR
RO
RO
(8)
RO
H
H
This concept of deactivation by hydrogen bonding to the
nucleophile is also applicable to reaction of cation with alcohols
and with water. E.g., in the case of Ph2CH+, the kNu values
are lower by the factor 40-100 in TFE than in acetonitrile
(where there is no such deactivation), and they are even lower
in HFIP.46 Other examples are 1′H+, where the kNu values for
water or alcohol or ether, measured in TFE, increase in going
F–
-CH+-O-SiMe3 + F–
O + FSiMe3
(7)
-CH+-O-SiMe3
-CH
(41) Reversibility has been found in the addition reaction of Ph3C+ with
Br-, but not with Cl- or F- (McClelland, R. A.; Banait, N.; Steenken, S.
J. Am. Chem. Soc. 1986, 108, 7023). It is interesting that the rate constants
The formation of aldehyde as main product in the presence of
F- (see section 1) could be in support of eq 7. However, Cl-
does not seem to react analogously, based on the fact (see section
1) that Cl- has no effect on the product yield. Therefore, the
apparent decrease in the reactivity from Cl- to I- may result
from an increasing degree of reversibility of the ion combination
reaction.41
for reaction in TFE of the cations with N3- (k ) (3.1-5.4) × 109 M-1 s-1
,
see Table 3) are diffusion controlled, which contrasts with the low values
for I-. This is an argument in favor of reversibility in the reaction with I-,
since I- and N3- have similar nucleophilicities, but differ vastly with respect
to leaving group properties. A reviewer suggested that it is also possible
that the forward rate constants decrease in a way reflecting the decreasing
stability of the C-X bond in going to X ) I.
(42) Richard, J. P.; Jencks, W. P. J. Am. Chem. Soc. 1982, 104, 4689.
Amyes, T. L.; Richard, J. P. J. Chem. Soc. Chem. Commun. 1991, 200 and
references therein.
Due to its exceptionally high reactivity, the strong nucleophile
N3- has been widely used to trap carbocations (“azide clock”).42
(37) Faria, J. L.; Steenken, S. J. Phys. Chem. 1993, 97, 1924.
(38) Kilian, J.; Kirmse, W. Unpublished results. The cation is produced
via Ph2C:.
(43) The diffusion rate in TFE, whose viscosity at 25 °C is 1.780
centipoise (from a) Murto, J.; Heino, E.-L. Suomen Kemistilehti B 1966,
39, 263), is calculated with the Smoluchovsky equation to be 3.7 × 109
(39) The formation of the cation proceeds via the radical cation (Faria,
J. L.; Steenken, S.; McClelland, R. A. Unpublished results.
M-1 s-1
.
(44) Calculated with the Smoluchovski equation from the viscosity of
HFIP of 1.579 centipoise at 25° C (from a) Murto, J.; Kivinen, A.; Lindell,
E. Suomen Kemistilehti B 1970, 43, 28).
(45) Cotton, F. A.; Wilkinson, G. Advances in Inorganic Chemistry; John
Wiley and Sons: New York, 1980; p 421.
-
(40) The fact that in acetonitrile there is no difference in the kNu values
for the different halides (see column 8 of Table 3) is due to the reactions
being diffusion controlled. For a discussion of diffusion control in
carbocation reactions, see ref 29 and ref 37.

10846 J. Am. Chem. Soc., Vol. 118, No. 44, 1996
Kirmse et al.
in cation reactivity with Nu in going from acetonitrile via TFE
to HFIP, as described above, is evidence for strong solute-
solVent interaction. An even stronger argument against oligo-
merization being the reason for the curvature is the fact that in
the case of the ethers THF and 1,4-dioxane, the kobsd vs [Nu]
plots are also “curved upward” (see Figure 8 and Table 3).
Obviously, dimerization via H-bonds, as assumed in the case
of alcohols, is not possible with ethers. This type of mechanism
is therefore not applicable. For this reason, a mechanism that
does not involve (rely on) di- or oligomerization is required to
explain the curvature in the kobsd vs [Nu] plots. We therefore
propose an alternative concept which is based on the very weak
basicity of HFIP. In the reaction of a cation C+ with ROH (R
) H, alkyl) yielding C-OR, a proton is released which has to
be transferred to a base. In many cases the solvent will be able
to serve as this base. However, if the solvent has a very low
basicity, C-O bond formation will not go to completion unless
a second ROH takes part serving as the base.52 The most likely
way as to how this can happen is shown below:
Figure 8. Dependence on [THF] of kobsd for decay of 1′H+ in HFIP.
The mechanism involves an equilibrium between the cation and
the alcohol and their addition product,53 the O-protonated ether,
with kf being the second-order rate constant for adduct formation
and kr the first order one for reversal into reactants. The reverse
step corresponds to what happens in the H+-catalyzed hydrolysis
of ethers. Decomposition of the protonated ether into product
(second-order rate constant kd) requires transfer of the proton
to a second alcohol molecule.
Figure 9. Results of the computer simulation of eq 9 in terms of kobsd
for cation decay as a function of [ROH]. The reaction with solvent
(eq 6) was included, taking ks ) 105 s-1. It was assumed that kf ) 108
M-1 s-1 and kd ) 109 M-1 s-1. The rate constant kr was varied: The
squares are for kr ) 107, the circles for 109 s-1
.
In order to test whether this mechanism gives rise to the
experimentally observed curved kobsd vs [Nu] plots, computer
simulations were performed. In addition to eq 9, the reaction
of cation with solvent (rate constant ks) was taken into account
and assumed to be reversible with a ratio of 10:1 for the forward
and reverse directions. The dependence on time of the
concentration of the cation was then calculated for various
concentrations of Nu, while the ratio [Nu]:[cation] was always
kept g100, such that pseudo-first order conditions were fulfilled
with respect to the reaction of the cation (and of the adduct, eq
9). The rate constants kf, kr, and kd were then systematically
varied (ks was taken as 105 s-1), and the resulting calculated
concentration-time profiles were analyzed in terms of first order
kinetics for the decay of the cation54 (as is done with the
experimentally obtained concentration-time profiles). An im-
from R ) MeO to the less stabilizing Me, and a further increase
is expected on going to R ) H. However, for R ) H the values
are considerably lower, since the k values now refer to reaction
in the much stronger H-donor HFIP.47
With respect to the solvent HFIP, a particularly interesting
phenomenon is that for reaction of cations 1′H+ (R ) H) and
Ph2CH+ with water, the alcohols, and ethers, the kobsd values
for cation decay are not proportional to [Nu] but follow the
relation kobsd ) ks + k[Nu]n, with n ) 1.2-2.48 As an example
for this behavior is shown the system 1′H+ with THF (Figure
8).
The case n > 1 suggests that more than one molecules of
Nu are involved in the reaction with the cation. Nonlinear
dependences of kobsd on [Nu] have been previously observed
for reaction of Ph2CH+ with methanol, ethanol, and 2-propanol
in 1,2-dichloroethane.49 For an explanation, di- or oligomer-
ization by intermolecular hydrogen bonding of the alcohols in
this very nonpolar, nonprotic solvent was invoked.49-51 In
contrast, in an alcoholic solvent such as HFIP, solute-solute
association by hydrogen bonding is very unlikely, due to heavy
competition by H-bonding with solvent. In fact, the decrease
(49) Sujdak, R. J.; Jones, R. L.; Dorfman, L. M. J. Am. Chem. Soc. 1976,
98, 4875.
(50) The crucial assumption involved in this concept is that the dimers
or oligomers, formed by hydrogen bonding between the oxygens, are more
nucleophilic than the monomeric alcohols where the only nucleophilic center
of the molecule, the oxygen atom, is not involved in hydrogen bonding.
(51) An analogous concept was later applied to reactions of carbenes
with alcohols, see ref 67.
(52) Interestingly, it was already noted by Dorfman49 that in the case of
participation of a second alcohol molecule, the leaving group from the
protonated ether is not the proton but an alkoxonium ion, a “bonded,
stabilized state of the proton”.
(46) The values in HFIP are upper limits, due to the upward curvature
of the kobsd vs [] plots. For a discussion see text.
(53) Evidence for an equilibrium is the observation that the decay of
the cations in HFIP on the microsecond timescale does not quite go to the
zero-line in the absence of nucleophiles, but it does in their presence, due
to depletion of the cations by reaction with the nucleophiles.
(47) There is also a polarity change in going from HFIP (ꢀ ) 16.6)43a to
TFE (ꢀ ) 26.7)44a to acetonitrile (ꢀ ) 37.5).
(48) A similar observation has been previously made in HFIP.11,12

Production of R-Siloxycarbenium Ions
J. Am. Chem. Soc., Vol. 118, No. 44, 1996 10847
portant result of the simulations is that the lifetime of the
complex has to be very short, in order to get upward curved
kobsd-[ ] dependences. E.g., if it is assumed that kf ) 108 M-1
s-1 and kd ) 109 M-1 s-1, kr must be >107 s-1, smaller values
giving linear dependences. This means is that quasi-termo-
lecularity is required in order for curved-up behavior to be
seen: With a complex which is too long-lived, the condition
of quasi-simultaneity of the interaction of two Nu molecules
with cation is not fulfilled. The computer calculations thus show
that the experimental data are compatible with the proposed
mechanism. Another relevant result is that curved-down kobsd
vs [Nu] plots were not achievable on the basis of eq 9.
with MeOH(OD) and EtOH(OD),10 respectively.61,62 Since, as
reported in section 3b, the reactions of the siloxycarbenes with
the proton donors are very probably not diffusion-controlled
(certainly not in the case of 2-methoxyethanol), a kinetic isotope
effect of 1 excludes a linear transition state for transfer of the
proton63 fom O to the carbenic C,64 i.e.
The mechanism explains the upward curved kobsd vs [Nu] plots
also for THF and 1,4-dioxane. In addition to the equilibrium
postulate,55 it invokes the participation of a second ether
molecule which reacts as a Lewis base, i.e., by nucleophilic
attack at the (incipient) carbocationic center produced by ring-
opening of the oxonium ion (nucleophilic assistance by the
second ether molecule). This is essentially the mechanism of
cationic polymerization of ethers.56 The difference between eqs
9 and 10 is that in eq 9 the second molecule acts as a Bro¨nsted
base, whereas in eq 10 it functions as a Lewis base.
However, if the carbene approaches in a direction orthogonal
to the O-H bond as in eq 12, an isotope effect close to 1 is
expected:
As seen from the entries in Table 2, the kprot values for TFE,
HFIP, and HClO4 (obtained from the slopes of the linear kobsd
vs [proton donor] plots) are essentially invariant of the nature
of the carbene.66 Since the stability of the corresponding
carbenium ions (as deduced from their lifetimes, see section 5
A prediction that can be made on the basis of the mechanisms
9 and 10 is that curved kobsd vs [Nu] dependences should also
be observed in solvents more basic than HFIP provided the
cations are less electrophilic (more “stable”, i.e., longer lived
in the usual nucleophilic solvents) than the siloxycarbenium ions
discussed here.57 This situation is, however, not yet reached
with Ph2CH+ in TFE (see Table 3 for some rate constants).
6. Comments on the Reactivity of Carbenes with Alco-
hols. As already mentioned in section 3b, in the reaction in
acetonitrile of the carbenes 1′ with the proton donors 2-meth-
oxyethanol, HFIP, and acetic acid, O-deuteration does not
decrease the reactivity of the donors. Analogous observations
have been made previously. E.g., with 9-xanthylidene, kH/kD
) 1.13 for reaction with tert-butyl alcohol-OH/OD),58 for 3,6-
dimethoxyfluorenylidene, kH/kD ) 1.17 for reaction with MeOH/
OD,59 for (MeO)2C:, kH/kD ) 1.0 for reaction with AcOH/
AcOD,60 and for 1′ (R ) H), kH/kD ) 1.6 and 1.9 for reaction
(61) Kinetic isotope effects larger than those quoted above have also
been reported in the literature. E.g., in ref 60 is given kH/kD ) 3.3 for
reaction of (MeO)2C: with MeOH(D). However, the rate constants were
taken from kobsd vs [MeOH(D)] plots with negatiVe intercepts60 (which
physically make no sense). This indicates that the kobsd-[MeOH(D)]
dependences are really curVed upward, as seen in the case of MeOH(OD)
reaction with the closely related carbene MeOC:Me, where kH/kD (as taken
from the low concentration range) appears to be 1.60 The curved plots
indicate a dependence of the reactivity on alcohol concentration, see text.
Concerning the kH/kD numbers of 1.6 and 1.9 for 1′ (R ) H),10 these were
determined by an indirect method, possibly at a single concentration of
alcohol. If the dependence on concentration of the reactivity was different
for MeOH and MeOD, this would lead to concentration-dependent kH/kD
numbers. A further example is the MeOC:F reaction with AcOH(OD) for
which a kH/kD value of 1.95 was reported.68b It is not, however, clear to
which extent kobsd-[AcOH(OD)] dependences were determined.
(62) As an explanation for the low kinetic isotope effects, the argument
was invoked that the reactions were (close to) diffusion controlled.58-60
However, the reported rate constants are lower by the factor g10 than kdiff
,
as calculated using the Smoluchovsky equation.
(63) Lowry, T. H.; Richardson, K. S. Mechanism and Theory in Organic
Chemistry, 2nd ed.; Harper and Row: New York, 1981; pp 205-212. See
also, e.g.: Westheimer, F. H. Chem. ReV. 1961, 61, 265.
(54) The exponential fit to the calculated concentration-time profiles was
always very good, and the same is true for most of the experimentally
obserVed ones.
(55) The reversibility of cation reaction with ether has been shown for
the reaction of Ph2CH+ with THF (Bartl, J. Dissertation, Universita¨t zu
Lu¨beck, Germany, 1990).
(56) For a discussion of relevant mechanistic problems see, e.g.: Kirmse,
W., Jansen, U. Chem. Ber. 1985, 118, 2607 and references cited therein.
(57) For a given cation and ROH, with increasing basicity of the solvent,
ROH will be replaced by solvent molecules. This leads to a decrease (loss)
of curvature in the kobsd-[] dependences.
(58) Lapin, S. C.; Schuster, G. B. J. Am. Chem. Soc. 1985, 107, 4243.
The rate constants were determined by an indirect method.
(59) Chuang, C.; Lapin, S. C.; Schrock, A. K.; Schuster, G. B. J. Am.
Chem. Soc. 1985, 107, 4238.
(64) The solvent kinetic isotope effect for protonation on C of CN- is
4.5 (Bednar, R. A.; Jencks, W. P. J. Am. Chem. Soc. 1985, 107, 7117). A
further example is the protonation by H2O of the electron adduct of
guanosine where kH/kD ) 8.0 (Candeias, L. P.; Wolf, P.; O’Neill, P.;
Steenken, S. J. Phys. Chem. 1992, 96, 10302). For a review on solvent
kinetic isotope effects, see: Kresge, A. J.; More O’Ferrall, R. A.; Powell,
M. F. Solvent Isotope Effects, Fractionation Factors and Mechanisms of
Proton Transfer Reactions. In Isotopes in Organic Chemistry; Buncel, E.,
Lee, C. C., Eds.; Elsevier: Vol. 7, pp 177-273.
(65) Obviously, an ylid mechanism is excluded due to the absence of
reasonable ways by which an ylid could be converted into the experimentally
observed carbenium ion.
(66) However, in the case of ethylene glycol and 2-methoxyethanol, there
is an increase by the factor 3-4 in going from R ) H to R ) MeO (see
Table 2), indicating a small contribution of proton transfer to C: in the
transition state.
(60) Moss, R. A.; Shen, S.; Wlostowski, M. Tetrahedron Lett. 1988, 29,
6417.

10848 J. Am. Chem. Soc., Vol. 118, No. 44, 1996
Kirmse et al.
and Table 3) increases considerably in going from 1, R ) H to
R ) MeO and to 2, the independence of the kprot values on the
Bro¨nsted basicity of the carbenes indicates that in the transition
state for carbene protonation a carbenium ion nature is only
weakly developed. This essentially excludes eq 12. The
independence on the nature of R also excludes an ylide-type
transition state where there would be negatiVe charge on the
carbenic carbon. However, the experimental observations are
understandable in terms of the carbene attacking not H but the
O-H bond (“insertion”) since in such a transition state there is
little, if any, ion character (see eq 13).
molecule of reactant, whereby a (contact) ion pair is possibly
involved:
It is assumed that the decomposition rate kd of the complex is
determined (the carbenium remaining the same) by the stability
of the (incipient) alkoxide anion for which the pKa of the alcohol
is a measure. For weakly acidic alcohols, complex decomposi-
tion needs assistance by a second alcohol which stabilizes the
(incipient) anion by H-bonding. Comparison of this concept
with that involving the cation reaction with alcohols (eq 9)
shows that the mechanisms, although formally analogous, are
chemically reciprocal: with the cations it is the nucleophilicity,
with the carbenes the electrophilicity of the alcohols which is
the determining factor. The fact (see Table 2) that the kobsd vs
[alcohol] plots for 1′ and 2′ are linear for highly acidic alcohols
such as TFE and HFIP can be explained by assuming that kr is
small or negligible and kd large, in contrast to the case of the
less acidic alcohols and water.73 With the weaker proton donors,
the proton transfer from O to the carbene C: is assumed to be
reversible and to go to completion only if a second proton donor
solvates the (incipient) alkoxide anion.74,75 Scheme 1376 is able
to account for the different curvatures observed for different
alcohols by allowing for different sets of rate constants for kf,
kr, and kd. The mechanism is also in agreement with the
observation that the reactivity of methanol is lower in acetonitrile
than in isooctane: In acetonitrile the alcohol O-H protons are
tied up in hydrogen bonding to the nitrile group67 and therefore
only partially available for protonation of the (incipient)
alkoxide.77,78
A further aspect of mechanistic importance already referred
to in footnote 61 has to do with the curVed kobsd vs [proton
donor] plots. In Table 2 it is reported that these plots for the
reaction of the siloxycarbenes 1′ and 2′ with, e.g., water,
methanol, and ethanol are curved upward, i.e., the reactivity of
the proton donor increases with its concentration (see also
Figures 4 and 5). This type of phenomenon was first reported
for reaction of the carbenes 4-RC6H4C:Cl with methanol in
acetonitrile and isooctane,67 and a number of similar cases has
since become known.68,69 The upward curvature was interpreted
on the basis of oligomer formation by hydrogen bonding
between the alcohol molecules and assuming a higher reactivity
of the oligomers than of the monomer to an extent overcom-
pensating the concomitant decrease of concentration.70,71 Con-
cerning the highly polar solvent CH3CN which is an excellent
proton acceptor, dimerization by H-bond formation is not very
likely, due to the strong competition by H-bonding to the solvent
(the nitrogen of CH3CN). In order to test this idea, IR
measurements of solutions of MeOH and of t-BuOH as model
alcohols (and also of TFE and HFIP) in CH3CN were performed.
In contrast to the situation in the solvent CCl4, where the alcohol
di- and oligomers are clearly visible (for MeOH at 3523 (dimer)
and 3342 cm-1 (oligomer), monomer at 3643 cm-1),72 in CH3-
CN only the solvated monomer (for MeOH at 3541 cm-1) was
seen, and absolutely no evidence for dimer or oligomer
formation was found up to alcohol concentrations of 0.4 M.
Thus, in the solvent CH3CN, di- or oligomerization of the
alcohols mentioned above does not take place. For explaining
the “upward behavior” observed with the less acidic alcohols,
a mechanism is therefore suggested which contains the same
formalism as the reaction mechanism proposed for carbocations
and nucleophiles, i.e., an equilibrium formation of a complex
between the reactants followed by assistance by a second
Possibly the strongest, though indirect support for the
reversible complex formation hypothesis comes from work on
diphenylmethylene79 and fluorenylidene,80 where it was shown
that the reactions of these carbenes with alcohols are not one-
step processes81 and that, specifically, there is an equilibrium
in which the carbene is regenerated.82 Furthermore, on the basis
(73) Linear behavior with acidic alcohols has also been observed with
other carbenes, see ref 68c.
(74) In the case of glycols, the electrophilic assistance by the second
OH group can occur intramolecularly. This may be a reason for the linear
kobsd-[] plots and for the considerably enhanced reactivity of ethylene glycol
(67) Griller, D.; Liu, M. T. H.; Scaiano, J. C. J. Am. Chem. Soc. 1982,
104, 5549. Griller, D.; Nazran, A. S.; Scaiano, J. C. Tetrahedron 1985,
1543.
compared to the more acidic 2-methoxyethanol (see Table 2), whereby the
statistical factor (ethylene glycol has two OH) has to be taken into account.
It may also be interesting that in the case of 1′ (R ) MeO), ethylene glycol
is more reactive than the much more acidic HFIP.
(68) (a) Sheridan, R. S.; Moss, R. A.; Wilk, B. K.; Shen, S.; Wlostowski,
M.; Kesselmayer, M. A.; Subramanian, R.; Kmiecik-Lawrynowicz, G.;
Krogh-Jespersen, K. J. Am. Chem. Soc. 1988, 110, 7563. (b) Du, X.-M.;
Fan, H.; Goodman, J. L.; Kesselmayer, M. A.; Krogh-Jespersen, K.; La
Villa, J. A.; Moss, R. A.; Shen, S.; Sheridan, R. S. J. Am. Chem. Soc. 1990,
112, 1920. (c) Vasella, A.; Briner, K.; Soundarajan, N.; Platz, M. S. J.
Org. Chem. 1991, 56, 4741. (d) Moss, R. A.; Shen, S.; Hadel, L. M.;
Kmiecik-Lawrynowicz, G.; Wlostowska, J.; Krogh-Jespersen, K. J. Am.
Chem. Soc. 1987, 109, 4341. (e) Reference 60.
(75) The dimerization hypothesis49,67 involves simultaneous, whereas
Schemes 9 and 13 are based on consecutiVe interaction with two reactants
of the same type. The idea that the function of the second alcohol is to
stabilize by hydrogen bonding the alkoxide resulting from the first alcohol
(rather than to increase the acidity of the first one) can, of course, also be
applied to the dimerization hypothesis. An essential difference between
the two ideas lies in the equilibrium assumed with the consecutive interaction
concept.
(69) Interestingly, the corresponding plots for tert-butyl alcohol were
(76) For the computer calculations, the reversal of the carbene into
acylsilane was taken into account. The most important condition to be
fulfilled for obtaining curved-up kobsd-[] plots is a short lifetime of the
curved downward.67
(70) This concept is analogous to that proposed by Dorfman49 for
reactions of alcohols with carbocations.
complex, i.e., kr should be g108 s-1
.
(71) The crucial assumption involved in this concept is that the dimers
or oligomers, formed by hydrogen bonding between the oxygens, are more
electrophilic than the monomeric alcohols where the only electrophilic center
of the molecule, the OH proton, is not involved in hydrogen bonding. For
methanol, the enhancement of reactivity required to explain the experimental
results on the basis of this concept corresponds to a factor ≈150.67 Cf. ref
50.
(72) For information on alcohol dimer and oligomer formation, see,
e.g.: Frange, B.; Abboud, J. L. M.; Benamou, C.; Bellon, L. J. Org. Chem.
1982, 47, 4553. For the effects of this on the philicity of alcohols, see
Huyskens, P. L. J. Am. Chem. Soc. 1977, 99, 2579.
(77) However, in order to explain the curve-down dependences (see
Figure 6 and Table 2), association phenomena (without overcompensatory
reactivity enhancement) have to be invoked.
(78) A case has been reported68d where the kobsd-[] plot, after an initial
curved-up part, curves down. This can be explained by the effects of di-
and oligomerization overcompensating electrophilic assistance by the second
alcohol molecule.
(79) Bethell, D.; Newall, A. R.; Whittaker, D. J. Chem. Soc. B 1971,
23.
(80) Zupancic, J. J.; Grasse, P. B.; Lapin, S. C.; Schuster, G. B.
Tetrahedron 1985, 41, 1471.

Production of R-Siloxycarbenium Ions
J. Am. Chem. Soc., Vol. 118, No. 44, 1996 10849
of the results with fluorenylidene, it was suggested80 that there
was some “communication” from one alcohol to another in the
reaction sequence and that observed kinetic isotope effects
“might arise from changes in solvation or association equilibria
of the alcohol”. These general ideas can be visualized in terms
of Scheme 13. On this basis, a kinetic isotope effect can be
envisaged as arising under conditions where kd is rate determin-
ing, specifically, the proton transfer from the second alcohol to
the incipient alkoxide anion, involving a linear transition state
RO-..H-OR.
spectroscopic properties and electrophilic reactivity have been
described. With respect to reaction with the halides, there is
an inversion of the classical order, i.e., F- being more reactive
with the carbenium ions than I-. In the weakly basic solvent
HFIP, the reaction with alcohols and ethers is suggested to
proceed via reversible formation of a cation-nucleophile adduct
which proceeds to product (ether and H+ or secondary cation)
upon assistance by a second alcohol molecule which serves as
the proton (or cation) acceptor.
Concerning the mechanism of carbene reaction with ROH
in CH3CN, the observed kinetic isotope effects close to 1 and
a weak dependence of protonation rates on cation stability
indicate an “early” transition state with little proton transfer from
ROH to carbene, and, correspondingly, little charge develop-
ment. With alcohols less acidic than TFE the protonation of
the carbenes is suggested to proceed via reversible formation
of a carbene-alcohol adduct which decomposes into product
by interaction with a second alcohol molecule.
Summary and Conclusions
The carbenes 4-RC6H4C:OSiMe3 (R ) H, Me, MeO) and
â-naphthylC:OSiMe3, formed by rearrangement from the triplet
states of the corresponding acylsilanes ArC(O)SiMe3, have been
characterized spectroscopically and by their reactivity with
proton donors, particularly alcohols. In the solvents TFE and
HFIP, the carbenes give rise to R-siloxycarbenium ions, whose
(81) See also, Warner, P. M.; Chu, I. S. J. Am. Chem. Soc. 1984, 106,
5366; J. Org. Chem. 1984, 49, 3666.
Acknowledgment. We thank Marion Stapper for her com-
petent technical help and J. L. Faria for letting us use some of
the rate constants for reaction of Ph2CH+ determined by him
(see Table 3). M.G. thanks the Deutsche Forschungsgemein-
schaft for a scholarship (DFG-Ste 474/1-1).
(82) In the work of Schuster (ref 80), the alcohol is considered a
nucleophile with respect to fluorenylidene. However, on the basis of LFP
experiments (unpublished results), the reaction leads to 9-fluorenyl cation,
i.e, the alcohol acts as an electrophile. As already noted by Schuster, a
difference of this type in the mechanism does not, however, invalidate his
conclusion as to the reVersibility of the reaction between carbene and
alcohol.
JA9603970