Solvent reorganization controls the rate of proton transfer from neat alcohol solvents to singlet diphenylcarbene

Article abstract of DOI:10.1021/ja017485r

Femtosecond transient absorption spectroscopy was used to study singlet diphenylcarbene generated by photodissociation of diphenyldiazomethane with a UV pulse at 266 nm. Absorption by singlet diphenylcarbene was detected and characterized for the first time. Similar band shapes were observed in acetonitrile and in cyclohexane with λ_max ≈ 370 nm. The singlet absorption decays by intersystem crossing to triplet diphenylcarbene at rates that agree with previous measurements. The singlet absorption band is completely formed 1 ps after the pump pulse. It is preceded by a strong and broad absorption band, which is tentatively assigned to excited-state absorption by a singlet diazo excited state. In neat alcohol solvents the growth and decay of the diphenylmethyl cation was observed. This species is formed by proton transfer from an alcohol molecule to singlet diphenylcarbene. Since a shell of solvent molecules surrounds each nascent carbene, the intrinsic rate of protonation in the absence of diffusion could be measured. In methanol, proton transfer occurs with a time constant of 9.0 ps, making this the fastest known intermolecular protontransfer reaction to carbon. In O-deuterated methanol proton transfer occurs in 15.0 ps. Slower rates were observed in the longer alcohols. The protonation times correlate reasonably well with solvation times in these alcohols, suggesting that solvent fluctuations are the rate-limiting step. In all alcohols studied, the carbocations decay on a somewhat slower time scale to yield diphenylalkyl ethers. In methanol and ethanol the rate of decay is determined by reaction with neutral solvent nucleophiles. There is evidence in 2-propanol that geminate reaction within the initial ion pair is faster than reaction with solvent. No isotope effect was observed for the reaction of the diphenylmethyl carbocation in methanol. Using comparative actinometry the quantum yield of protonation was measured. In methanol, the quantum yield of carbocations reaches a maximum value of 0.18 approximately 18 ps after the pump pulse. According to our analysis, 30% of the photoexcited diazo precursor molecules are eventually protonated. Somewhat lower protonation efficiencies are observed in the other alcohols. Because the primary quantum yield for formation of singlet diphenylcarbene is still unknown, the importance of reaction channels that might exist in addition to protonation cannot be determined at present. Singlet carbenes are powerful, photogenerated bases that open new possibilities for fundamental studies of proton transfer in solution.

Full text of DOI:10.1021/ja017485r

Published on Web 05/08/2002
Solvent Reorganization Controls the Rate of Proton Transfer
from Neat Alcohol Solvents to Singlet Diphenylcarbene
Jorge Peon,† Dmitrii Polshakov, and Bern Kohler*
Contribution from the Department of Chemistry, The Ohio State UniVersity, 100 West 18th
AVenue, Columbus, Ohio 43210
Received November 7, 2001
Abstract: Femtosecond transient absorption spectroscopy was used to study singlet diphenylcarbene
generated by photodissociation of diphenyldiazomethane with a UV pulse at 266 nm. Absorption by singlet
diphenylcarbene was detected and characterized for the first time. Similar band shapes were observed in
acetonitrile and in cyclohexane with λmax ≈ 370 nm. The singlet absorption decays by intersystem crossing
to triplet diphenylcarbene at rates that agree with previous measurements. The singlet absorption band is
completely formed 1 ps after the pump pulse. It is preceded by a strong and broad absorption band, which
is tentatively assigned to excited-state absorption by a singlet diazo excited state. In neat alcohol solvents
the growth and decay of the diphenylmethyl cation was observed. This species is formed by proton transfer
from an alcohol molecule to singlet diphenylcarbene. Since a shell of solvent molecules surrounds each
nascent carbene, the intrinsic rate of protonation in the absence of diffusion could be measured. In methanol,
proton transfer occurs with a time constant of 9.0 ps, making this the fastest known intermolecular proton-
transfer reaction to carbon. In O-deuterated methanol proton transfer occurs in 15.0 ps. Slower rates were
observed in the longer alcohols. The protonation times correlate reasonably well with solvation times in
these alcohols, suggesting that solvent fluctuations are the rate-limiting step. In all alcohols studied, the
carbocations decay on a somewhat slower time scale to yield diphenylalkyl ethers. In methanol and ethanol
the rate of decay is determined by reaction with neutral solvent nucleophiles. There is evidence in 2-propanol
that geminate reaction within the initial ion pair is faster than reaction with solvent. No isotope effect was
observed for the reaction of the diphenylmethyl carbocation in methanol. Using comparative actinometry
the quantum yield of protonation was measured. In methanol, the quantum yield of carbocations reaches
a maximum value of 0.18 approximately 18 ps after the pump pulse. According to our analysis, 30% of the
photoexcited diazo precursor molecules are eventually protonated. Somewhat lower protonation efficiencies
are observed in the other alcohols. Because the primary quantum yield for formation of singlet
diphenylcarbene is still unknown, the importance of reaction channels that might exist in addition to
protonation cannot be determined at present. Singlet carbenes are powerful, photogenerated bases that
open new possibilities for fundamental studies of proton transfer in solution.
Scheme 1. Proposed Reaction Pathways for O-H Insertion by
Singlet Diphenylcarbene
Introduction
Diphenylcarbene, the simplest diaryl carbene, is a widely
studied model system for understanding the spin-dependent
reactivity of carbenes.1 The lowest energy state of diphenyl-
carbene (3Ph2C) has triplet spin and is 4-5 kcal mol-1 lower
in energy than the lowest singlet state (1Ph2C). Despite their
energetic proximity, these two spin states undergo markedly
different reactions. 3Ph2C abstracts hydrogen to produce triplet-
polarized radical pairs, while 1Ph2C inserts into O-H bonds of
alcohols to form diphenylalkyl ethers. The mechanism of this
reaction has intrigued chemists for over 40 years. Scheme 1
summarizes the reactive pathways, which have been proposed.
Protonation to yield diphenylmethyl cation (Ph2CH+) and an
alkoxide ion is shown by pathway A. Nucleophilic attack on
the carbocation by the geminate alkoxide ion (pathway A′) or
by a second alcohol molecule (not shown in Scheme 1) then
yields the ether product. Alternatively, electrophilic attack by
* Corresponding author. Telephone: (614) 688-3944. E-mail: kohler@
chemistry.ohio-state.edu.
† Current address: Arthur Amos Noyes Laboratory of Chemical Physics,
California Institute of Technology, Pasadena, CA 91125.
(1) For a review, see: Platz, M. S.; Maloney, V. M. In Kinetics and
Spectroscopy of Carbenes and Biradicals; Platz, M. S., Ed.; Plenum
Press: New York, 1990; pp 239-352.
9
6428
J. AM. CHEM. SOC. 2002, 124, 6428-6438
10.1021/ja017485r CCC: $22.00 © 2002 American Chemical Society

Proton Transfer from an Alcohol to Diphenylcarbene
A R T I C L E S
1Ph2C on an alcohol molecule can produce an intermediate ylide
(pathway B), followed by a 1,2-proton shift (pathway B′). A
third possibility is a concerted pathway (pathway C).
A decade ago, Kirmse et al. provided direct spectroscopic
evidence for the intermediacy of carbocations in reactions of
singlet diarylcarbenes with proton donors.2 Carbocation inter-
mediates have since been observed for reactions between a
number of carbenes and proton donors.3-5 Although these
studies have provided strong evidence in favor of pathway A
in Scheme 1, it has been suggested that protonation is only a
minor reaction channel.6,7 In this report, we present new insights
1
into the O-H insertion reaction of Ph2C, which have been
obtained by femtosecond transient absorption spectroscopy. As
we have shown previously,8 the femtosecond pump-probe
technique allows us to detect even extremely short-lived singlet
carbenes.
The fact that carbenes can deprotonate weak proton donors
such as alcohols reveals their highly basic character. The gas-
3
phase proton affinity of Ph2C was calculated to be 1151 kJ
mol-1.9 Pezacki argued that proton transfer to carbenes is an
intrinsically fast reaction akin to those of normal acids and
bases.10 More kinetic data are required to understand the proton-
transfer dynamics of these unusual carbon bases. Hence, a
second interest in this system is the information it provides about
fundamental aspects of proton-transfer reactions in solution.
Excitation of Ph2CN2 with a femtosecond UV pulse generates
a powerful base on a subpicosecond time scale. Through
spectroscopic detection of 1Ph2C, we have succeeded in directly
measuring the rate of protonation of 1Ph2C in several neat
alcohols for the first time. This has allowed us to study proton-
transfer reactions between the solvent and 1Ph2C in the absence
of diffusion. Our results show that 1Ph2C protonation is among
the fastest intermolecular proton-transfer reactions known.
Furthermore, the rate of this reaction appears to be controlled
by the time required for solvent reorganization.
Figure 1. Transient spectra observed 30 ps (open circles) and 650 ps
(crosses) after photolysis of diphenyldiazomethane with a 180 fs, 266 nm
laser pulse in (a) acetonitrile and (b) cyclohexane. The absorption band
with λmax ≈ 370 nm at 30 ps is assigned to Ph2C.
1
wavelength. Since the total group delay between the shortest and the
longest wavelengths of the continuum is about 2 ps, it was not necessary
to apply group delay (“chirp”) correction to transient spectra recorded
at delay times greater than approximately 10 ps, as all rapid spectral
evolution has ceased by this time. The instrument response was
approximately 220 fs (Gaussian full width at half-maximum (fwhm))
as determined by difference frequency mixing between the third
harmonic and 10 nm portions of the continuum in a thin BBO crystal.
(2) Materials. Diphenyldiazomethane was a generous gift from
Professor Matthew S. Platz of our department. The highest purity
solvents were obtained from Burdick and Jackson. Deuterated methanol
(CH3OD, 99.5% atom purity) was purchased from Aldrich. All solvents
were used as received. Approximately 1 mM solutions were studied in
a free flowing jet 0.4 mm thick. Because the diazo compound is
continuously photolyzed during the experiments, a large, 0.5 L reservoir
was used for recirculating the solutions through the jet. During data
acquisition, small samples of the solution were periodically withdrawn
and UV/vis absorption spectra were recorded. Solutions were replaced
when these spectra indicated significant depletion of the diazo
compound.
Experimental Section
(1) Spectrometer. UV pump pulses (λ ) 266 nm) were obtained
by third harmonic generation using a kilohertz-repetition-rate, regen-
eratively amplified Ti:sapphire laser system. A description of this
instrument and the relevant signal acquisition techniques is available
elsewhere.11-13 Probe pulses between 400 and 750 nm were generated
by focusing the laser fundamental in a water cell, while probe pulses
from 320 to 380 nm were generated in a 1 mm CaF2 window with the
laser second harmonic. Polarizations of the pump and probe pulses were
set to the magic angle. After the sample, the probe beam was passed
through an f ) 0.25 m monochromator to isolate an approximately 10
nm portion of the continuum for single wavelength measurements.
Transient spectra were recorded by scanning the monochromator
Results
(1) Spectroscopic Detection of Singlet Diphenylcarbene.
A strong transient absorption band (λmax ≈ 370 nm) was
observed 30 ps after excitation of diphenyldiazomethane (Ph2-
CN2) by a 150 fs, 266 nm laser pulse in acetonitrile and in
cyclohexane (Figure 1). Single-wavelength transients recorded
in acetonitrile (Figure 2) and cyclohexane (Figure 3) show the
spectral evolution of this band and reveal additional dynamical
features. In both solvents, a spikelike transient absorption signal
is observed at all probe wavelengths near zero delay time (t )
0). This feature shows an instrument-limited rise, followed by
a decay with a time constant of ≈300 fs. An analogous feature
was seen in measurements on Ph2CN2 in methanol (see inset to
Figure 6). This subpicosecond decay is intrinsic to diphenyl-
(2) Kirmse, W.; Kilian, J.; Steenken, S. J. Am. Chem. Soc. 1990, 112, 6399.
(3) Chateauneuf, J. E. J. Chem. Soc., Chem. Commun. 1991, 1437.
(4) Dix, E. J.; Goodman, J. L. J. Phys. Chem. 1994, 98, 12609.
(5) Steenken, S. Pure Appl. Chem. 1998, 70, 2031.
(6) Belt, S. T.; Bohne, C.; Charette, G.; Sugamori, S. E.; Scaiano, J. C. J. Am.
Chem. Soc. 1993, 115, 2200.
(7) Pliego, J. R., Jr.; De Almeida, W. B. J. Phys. Chem. A 1999, 103, 3904.
(8) Hess, G. C.; Kohler, B.; Likhotvorik, I.; Peon, J.; Platz, M. S. J. Am. Chem.
Soc. 2000, 122, 8087.
(9) Pliego, J. R.; DeAlmeida, W. B. J. Chem. Soc., Faraday Trans. 1997, 93,
1881.
(10) Pezacki, J. P. Can. J. Chem. 1999, 77, 1230.
(11) Peon, J.; Hess, G. C.; Pecourt, J.-M. L.; Yuzawa, T.; Kohler, B. J. Phys.
Chem. A 1999, 103, 2460.
(12) Peon, J.; Tan, X.; Hoerner, J. D.; Xia, C.; Luk, Y. F.; Kohler, B. J. Phys.
Chem. A 2001, 105, 5768.
(13) Pecourt, J.-M. L.; Peon, J.; Kohler, B. J. Am. Chem. Soc. 2001, 123, 10370.
9
J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002 6429

A R T I C L E S
Peon et al.
Table 1. Global Fit Parametersa for Transients in Figures 2 and 3
solvent
τ1 /fs
τ2 /ps
τ3 /ps
λ/nm
A
1
A
2
A
3
A
4
CH3CN 270 ( 40 21 ( 2 340 ( 20 320 0.42
0.08 -0.16 0.34
360 0.48
380 0.58
450 0.88
0.04
0.09
0.08
0.40 0.09
0.27 0.05
0.00 0.04
C6H12 280 ( 40 9 ( 1 110 ( 10 320 0.49
0.06 -0.16 0.28
360 0.64
380 0.68
0.05
0.00
0.25 0.06
0.23 0.08
0.12 0.11
0.03 0.07
400 0.76 -0.02
450 0.86 0.03
a The model function was ∆A ) A1 exp(-t/τ1) + A2 exp(-t/τ2) + A3
exp(-t/τ3) + A4. The amplitudes have been scaled such that ∑i|Ai| ) 1.
The time constants, τi, were global parameters in each solvent.
intersystem crossing.15 Using a laser-induced fluorescence
technique Eisenthal and co-workers reported an intersystem
crossing rate of (110 ( 12 ps)-1 in 3-methylpentane, a highly
nonpolar solvent like cyclohexane.16,17 In acetonitrile, they
measured an intersystem crossing rate of (311 ( 22 ps)-1 16,18
.
Figure 2. Single wavelength transients recorded at the indicated probe
wavelengths following 266 nm photolysis of diphenyldiazomethane in
acetonitrile. The solid curves are global nonlinear least-squares fits to the
experimental points, shown as open circles. The fitting parameters are
summarized in Table 1.
The excellent agreement with the decay times from our fits
indicates that the absorption band with λmax ) 370 nm is due
to 1Ph2C. Chergui and co-workers studied Ph2CN2 by the
femtosecond pump-probe technique.19 Their pump wavelength
was the same as in our study, but probing was done at just two
wavelengths: 390 and 780 nm. They reported a time constant
of ≈100 ps in isooctane in good agreement with our result in
cyclohexane.
The rising signals at 320 nm are assigned to the growth of
1
3
3Ph2C by ISC from Ph2C. Absorption by Ph2C was reported
previously in single crystals,20 in low-temperature glasses,21 and
in solution.22 Hadel et al. reported λmax ) 314 nm for 3Ph2C in
cyclohexane at room temperature.22 The signal at 320 nm does
not begin from the baseline after the completion of the
subpicosecond dynamics observed near t ) 0. Since very little
ISC is expected just a few picoseconds after excitation of the
diazo precursor, this suggests that 1Ph2C has considerable
absorption at 320 nm. The offsets at 320 nm are larger than at
any other probe wavelength, consistent with the long lifetime
3
of Ph2C.
(2) Singlet Diphenylcarbene Reactivity in Methanol. The
kinetics of the O-H insertion reaction of 1Ph2C was studied in
several neat alcohols. Figure 4 shows the transient spectra
recorded at four different delay times after photoexcitation of
Ph2CN2 in methanol. The broad and intense spectrum observed
150 fs after the excitation pulse (Figure 4, squares) decays on
a subpicosecond time scale as shown by the spectrum at 800 fs
(Figure 4, crosses). A new absorption band with λmax ≈ 440
nm appears next (Figure 4, filled circles). This band attains
maximum amplitude 18 ps after the pump pulse and then decays
toward the baseline. By 100 ps after the pump pulse there is
Figure 3. Same as Figure 2 for diphenyldiazomethane in cyclohexane.
diazomethane, as it was not observed in control experiments
on solvent blanks. Following the spike, the signal evolves on a
much slower time scale. Decays are observed for probe
wavelengths g360 nm, while a rising signal is observed at 320
nm in both solvents. Transients in each solvent were globally
fit at the probe wavelengths listed in Table 1 to the sum of
three exponentials and a constant offset analytically convoluted
with a Gaussian to model the instrument response function.14
The fits are shown by the solid curves in Figures 2 and 3, and
the best-fit values of the parameters are given in Table 1 for
each solvent.
(15) Langan, J. G.; Sitzmann, E. V.; Eisenthal, K. B. Chem. Phys. Lett. 1984,
110, 521.
(16) The time constant is the reciprocal rate constant reported by Eisenthal and
co-workers. The uncertainty in the time constant was computed by error
propagation from the reported uncertainty in the rate.
(17) Dupuy, C.; Korenowski, G. M.; McAuliffe, M.; Hetherington, W. M. I.;
Eisenthal, K. B. Chem. Phys. Lett. 1981, 77, 272.
We assign the longest time constant (τ3 in Table 1) to
(18) Eisenthal, K. B.; Turro, N. J.; Sitzmann, E. V.; Gould, I. R.; Hefferon, G.;
Langan, J.; Cha, Y. Tetrahedron 1985, 41, 1543.
1
3
intersystem crossing (ISC) from Ph2C to Ph2C. The nearly
3-fold longer time constant in acetonitrile compared to cyclo-
hexane is due to the pronounced effect of solvent polarity on
(19) Portella-Oberli, M. T.; Jeannin, C.; Soep, B.; Zerza, G.; Chergui, M. Chem.
Phys. Lett. 1998, 296, 323.
(20) Closs, G.; Hutchison, C. A. J.; Kohler, B. E. J. Chem. Phys. 1966, 44,
413.
(21) Trozzolo, A. M.; Gibbons, W. A. J. Am. Chem. Soc. 1967, 89, 239.
(22) Hadel, L. M.; Platz, M. S.; Scaiano, J. C. J. Am. Chem. Soc. 1984, 106,
283.
(14) A detailed description of our global fitting procedure can be found in
ref 12.
9
6430 J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002

Proton Transfer from an Alcohol to Diphenylcarbene
A R T I C L E S
Figure 4. Transient spectra for diphenyldiazomethane in methanol 150 fs
(squares), 800 fs (crosses), 15 ps (filled circles), and 100 ps (open circles)
after the 266 nm pump pulse.
Figure 6. Transient absorption at 435 nm in CH3OH (circles) and CH3OD
(triangles). The inset shows the early-time signals.
two solvents are compared in Figure 6. The data were recorded
in back-to-back scans on solutions of Ph2CN2 in CH3OH and
CH3OD. Solutions with equal concentrations of the diazo
precursor gave equal absorbance at the pump wavelength,
making the relative signal magnitudes in Figure 6 significant.
Within our experimental uncertainty, the spikelike signals at
early time have the same amplitude and identical decay times
in both solvents. At longer times, the two transients in Figure
6 have similar decay rates, but as the inset to Figure 6 shows,
the rise times are different in the two solvents. Additionally,
the signal in CH3OD at 435 nm reaches a maximum at a later
time than in CH3OH. Rate constants determined by least-squares
fitting (see below) confirm that the slower rate constant is the
same in both solvents, while the faster step shows a kinetic
isotope effect of kH/kD ) 1.7.
(3) Singlet Diphenylcarbene Reactivity in Other Alcohols.
Femtosecond pump-probe experiments were also carried out
for Ph2CN2 in a series of neat alcohols. Transients recorded at
435 nm are shown in Figure 7. This wavelength is the
approximate maximum of the Ph2CH+ absorption band in each
of the alcohols investigated. The transients in Figure 7 were
normalized such that the subpicosecond decays after t ) 0 have
identical amplitude.
Figure 5. Transient absorption at 435 (circles) and at 360 nm (crosses) in
CH3OH. The 360 nm transient has been offset by 20 ps for clarity.
minimal transient absorption at λ g 370 nm (open circles in
Figure 4).
The asymmetric spectrum with λmax ≈ 435 nm matches the
well-known spectrum of diphenylmethyl cation (Ph2CH+).
Absorption by this carbocation was characterized in superacid
conditions in the 1950s.23 It was later detected transiently in
pulse radiolysis24 and pulsed laser experiments.25-27 The
dramatic growth and decay of Ph2CH+ can be seen clearly in
the single-wavelength scan recorded at 435 nm (circles in Figure
5). The signal rises to a maximum near t ) 18 ps before
decaying back toward the baseline. Figure 5 also shows the
(4) Dynamics in Tetrahydrofuran. No evidence of an
intermediate was seen when Ph2CN2 was photolyzed in tetra-
hydrofuran. The transient absorption at 420 nm is shown in
Figure 8. As in the other solvents, a pronounced, subpicosecond
decay is observed. This is followed by a steadily decaying signal.
By fitting a straight line to the logarithm of the signal at delay
times t g 4 ps, a decay constant of 151 ( 16 ps was estimated.
(5) Transient Absorption Actinometry. Comparative actino-
metric measurements were performed to quantify the importance
1
decay of Ph2C at 360 nm. A small rise with a maximum near
5 ps delay is due to absorption by Ph2CH+ at this wavelength
(see below).
1
of protonation as a decay channel for Ph2C. In these experi-
ments, transient absorption signals from solutions of diphenyl-
diazomethane were compared with signals recorded under
identical experimental conditions from a reference solution
containing benzophenone (BP) in acetonitrile. The concentration
of BP was adjusted to give the same ground-state absorbance
at the pump wavelength of 266 nm as a second solution of
diphenyldiazomethane in methanol. Under these conditions, and
in the absence of multiphoton absorption and optical saturation,
both solutions absorb the same number of photons, and quantum
Deuterium Isotope Effect in Methanol. The effect of
isotopic substitution was studied in normal methanol (CH3OH)
and O-deuterated methanol (CH3OD). Signals at 435 nm in these
(23) Deno, N. C.; Jaruzelski, J. J.; Schriesheim, A. J. Am. Chem. Soc. 1955,
77, 3044.
(24) Wang, Y.; Tria, J. J.; Dorfman, L. M. J. Phys. Chem. 1979, 83, 1946.
(25) Bartl, J.; Steenken, S.; Mayr, H.; McClelland, R. A. J. Am. Chem. Soc.
1990, 112, 6918.
(26) Bartl, J.; Steenken, S.; Mayr, H. J. Am. Chem. Soc. 1991, 113, 7710.
(27) Faria, J. L.; Steenken, S. J. Phys. Chem. 1993, 97, 1924.
9
J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002 6431

A R T I C L E S
Peon et al.
Figure 8. Transient absorption of Ph2CN2 in tetrahydrofuran induced at a
probe wavelength of 420 nm by a pump pulse at 266 nm. The solid line is
the result of a linear least-squares fit to the logarithm of the long time data.
The time constant obtained is 151 ( 16 ps.
Figure 7. Transient absorption at 435 nm following 266 nm photolysis of
diphenyldiazomethane in (a) methanol, (b) ethanol, (c) 1-propanol, and (d)
2-propanol.
yields and extinction coefficients can be estimated using the
relation
∆AX(t,λ)
ꢀX(λ) φX(t) ) ꢀBP(λ′) φBP(t′)
(1)
∆ABP(t′,λ′)
In this equation, ꢀ denotes (molar decadic) absorption coeffi-
cients, φ represents time-dependent quantum yields that account
for population kinetics, and ∆A(t,λ) is the transient absorption
signal at probe wavelength λ, measured at time t. The subscripts
X and BP label the solution under study (Ph2CN2 in methanol)
and the benzophenone/acetonitrile reference solution, respec-
tively.
If more than one species contributes to the signal, then the
product ꢀφ must be replaced by a sum over all absorbing species.
As the time arguments in eq 1 indicate, signals from the
reference system and system X can be compared at different
delay times, and it is not necessary to measure the signals at
the same probe wavelength, as long as the relative spectral
amplitudes are accurate at the two wavelengths. This is achieved
by making the spot size of the probe pulse at the sample several
times smaller than the pump spot. The spectral fidelity of our
transient absorption spectrometer has been demonstrated previ-
ously using the well-characterized spectrum of one-electron
reduced methyl viologen.12 Although this technique has been
frequently employed in the past on slower time scales,28-31 we
believe this is the first application of the comparative method
to femtosecond transient absorption experiments.
Figure 9. Signal comparison of equal absorbance solutions of benzo-
phenone in acetonitrile (triangles) and diphenyldiazomethane in methanol
(open circles). The latter curve is displaced by 20 ps for clarity. The probe
wavelengths are 525 and 435 nm for benzophenone and diphenyldiazo-
methane, respectively.
The signal from benzophenone was recorded at 525 nm at
the peak of its long-wavelength triplet-triplet absorption band.
The signal, which is shown in Figure 9, rises with a single-
exponential lifetime of 9.2 ( 1.2 ps to a constant plateau. The
rise time is in excellent agreement with the value 9.6 ( 0.9 ps
determined previously by femtosecond transient grating spec-
troscopy in acetonitrile.32 This time constant reflects ultrafast
ISC from the optically prepared singlet state. Following the
completion of the rise, a constant transient absorption signal is
observed over the entire delay range of our instrument (0-1000
ps), consistent with the long-lived BP triplet state (τ ≈ 14 µs
in acetonitrile33).
The ratio of the maximum signal in methanol at 435 nm to
the BP triplet at 50 ps is 1.25 from the data in Figure 9. The
molar absorption coefficient of the benzophenone triplet state
is ꢀBP(525nm) ) 6500 M-1 cm-1, and the triplet quantum yield,
φBP, is equal to unity.33 The molar absorption coefficient of
(28) Lutz, H.; Breheret, E.; Lindqvist, L. J. Phys. Chem. 1973, 77, 1758.
(29) Amand, B.; Bensasson, R. Chem. Phys. Lett. 1975, 34, 44.
(30) Wintgens, V.; Johnston, L. J.; Scaiano, J. C. J. Am. Chem. Soc. 1988, 110,
511.
(32) Tamai, N.; Asahi, T.; Masuhara, H. Chem. Phys. Lett. 1992, 198, 413.
(33) Gramain, J.-C.; Remuson, R. J. Chem. Soc., Faraday Trans. 1 1980, 76,
1800.
(31) Bakac, A.; Burrows, H. D. Appl. Spectrosc. 1997, 51, 1916.
9
6432 J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002

Proton Transfer from an Alcohol to Diphenylcarbene
A R T I C L E S
Ph2CH+ has not been reported in neat methanol due to the
cation’s short lifetime in this nucleophilic solvent. To estimate
this quantity, we used the value measured in acetonitrile,
ꢀ(435nm) ) 44 000 M-1 cm-1 25
. Little is known about solvent
Scheme 2. Deactivation Pathways for Photoexcited
Diphenyldiazomethane
effects on Ph2CH+ absorption; however, similar absorption is
expected in methanol and acetonitrile on account of their similar
dielectric properties. Using these values in eq 1, φ ) 0.18 is
calculated for Ph2CH+ at 18 ps, the time of its maximum
concentration. The uncertainty in the quantum yield is estimated
to be 20-30%.34
of excess vibrational energy. Vibrationally hot molecules display
red-shifted absorption spectra that blue shift as they thermally
equilibrate with the cooler solvent molecules.39,40 Transient
absorption at probe wavelengths on the red edge of a given
absorption band generally show decays, while signals at higher
energy show rises. The decay times observed here (9 ps in
cyclohexane and 21 ps in acetonitrile) are in line with times
observed for other polyatomic molecules. For example, Troe’s
group reported deactivation times of ≈14 ps for azulene in
cyclohexane and acetonitrile.41 Since vibrational cooling usually
results in probe wavelength dependent decay times, the char-
acterization of this process by a single, global fitting parameter
is approximate. We previously reported evidence for vibrational
cooling in 2-naphthyl(carbomethoxy)carbene following UV
photolysis of its diazo precursor.8
Discussion
(1) Formation of Singlet Diphenylcarbene from Diphenyl-
diazomethane. The absorption spectrum of 1Ph2C has not
been reported previously. Dix and Goodman studied two di-
phenylcarbene derivatives, bis(p-chlorophenyl)carbene and bis-
(p-methoxyphenyl)carbene, using 355 nm laser flash photolysis
with 25 ps pulses.4 They observed transient absorption between
400 and 500 nm in heptane, tetrahydrofuran, and acetonitrile,
which they assigned to the singlet carbenes. They failed to detect
transient absorption between 400 and 700 nm upon picosecond
photolysis of Ph2CN2. The absorption spectrum of 1Ph2C (Figure
1) is blue-shifted relative to the spectra reported by Dix and
Goodman. The same spectral blue shift is observed when the
absorption spectrum of Ph2CH+ is compared to those of the
para-substituted cations.25 Turro et al. reported λmax ) 355 nm
for the singlet state of phenylchlorocarbene.35 Absorption near
360 nm appears to be a general characteristic of singlet aryl
carbenes. Ab initio calculations indicate that these transitions
have considerable π-π* character and involve molecular
orbitals that are delocalized over the phenyl rings.36
The subpicosecond decay observed immediately after excita-
tion of Ph2CN2 is the most enigmatic feature of the dynamics.
This spikelike signal was seen in all solvents investigated, and
in our earlier study of 2-naphthyl(carbomethoxy)carbene.8 Since
the signal at all times changes linearly with pump pulse energy,
multiphoton excitation is not responsible. Chergui and co-
workers observed a ≈250 fs decay following UV photolysis of
Ph2CN2 in isooctane at 390 and 780 nm.19 They assigned this
dynamical feature to energy relaxation in the hot carbene by
internal conversion and vibrational relaxation. The spectrum
during this phase of the signal (squares in Figure 4) resembles
a broadened version of the 1Ph2C spectrum shown in Figure 1,
as expected for a vibrationally hot species.42 However, vibra-
tional cooling in the lowest singlet state of 1Ph2C was excluded
for two reasons. First, vibrational cooling of hot molecules
typically occurs on a time scale of order 10 ps.42 Second,
pronounced decays are observed at 320 and 360 nm on the high-
1
As seen in Figure 1, λmax of Ph2C has approximately the
same value in highly polar acetonitrile as in nonpolar cyclo-
hexane. In contrast, the absorption spectrum of the lowest singlet
state of 2-naphthyl(carbomethoxy)carbene depends significantly
on the solvent.36 Ab initio calculations combined with a realistic
solvation model have reproduced both effects.36
The intermediate time constant (τ2 in Table 1) has the smallest
amplitude at each probe wavelength. We assign this weak signal
component to vibrational cooling of the nascent 1Ph2C. Eisenthal
and co-workers observed 3Ph2C* fluorescence following single
photon excitation of Ph2CN2 at 266 nm from which they
concluded that the carbon-nitrogen bond strength is <2 eV.17
A high-level ab initio estimate of the minimum carbon-nitrogen
bond dissociation energy for gas-phase CH2N2 with respect to
singlet products gave a value De ) 38.2 kcal mol-1 (1.7 eV).37
Simon and Peters reported a value ∆rH ) 0.0 ( 1.7 kcal mol-1
for the dissociation of Ph2CN2 to 3Ph2C and molecular nitrogen
in benzene.38 Even though this value is a lower limit, it provides
evidence that the carbon-nitrogen bond of diazo compounds
is indeed weak.
1
energy side of the Ph2C absorption band. In contrast, rising
signals are usually observed at probe wavelengths on the blue
edge of the absorption band of vibrationally hot molecules.40
The short-time dynamics are difficult to assign because so
little is known about the earliest events following electronic
excitation of diazo compounds. Scheme 2 shows some possible
deactivation pathways. The initially prepared excited singlet state
of the diazo (1Ph2CN2*) could intersystem cross to a triplet diazo
state, lose nitrogen, or internally convert to regenerate the
ground-state diazo compound. Since we have ruled out absorp-
tion by 1Ph2C or 3Ph2C for the short-time signal, absorption by
an excited state must be considered. Excited-state absorption
Following photolysis of the weak carbon-nitrogen bond, the
1Ph2C fragment is expected to have at least 2.6 eV (21000 cm-1)
(34) Previous measurements of ꢀ for Ph2CH+ in various solvents are summarized
in ref 25. On the basis of these values, a cautious estimate of the uncertainty
in the molar absorption coefficent of Ph2CH+ is 20-30%. The uncertainty
in the quantum yield will be roughly the same, since the other quantities
required for the calculation are known more precisely.
3
by Ph2C* is unlikely given the nanosecond lifetime of this
state.43,44 As already mentioned, Eisenthal and co-workers
(35) Turro, N. J.; Butcher, J. A., Jr.; Moss, R. A.; Guo, W.; Munjal, R. C.;
Fedorynski, M. J. Am. Chem. Soc. 1980, 102, 7576.
(39) Laermer, F.; Elsaesser, T.; Kaiser, W. Chem. Phys. Lett. 1989, 156, 381.
(40) Miyasaka, H.; Hagihara, M.; Okada, T.; Mataga, N. Chem. Phys. Lett. 1992,
188, 259.
(41) Schwarzer, D.; Troe, J.; Votsmeier, M.; Zerezke, M. J. Chem. Phys. 1996,
105, 3121.
(42) Elsaesser, T.; Kaiser, W. Annu. ReV. Phys. Chem. 1991, 42, 83.
(36) Peon, J.; Kohler, B.; Barone, V.; Hadad, C. M.; Platz, M. S. To be submitted
for publication.
(37) Papakondylis, A.; Mavridis, A. J. Phys. Chem. A 1999, 103, 1255.
(38) Simon, J. D.; Peters, K. S. J. Am. Chem. Soc. 1988, 110, 3336.
9
J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002 6433

A R T I C L E S
Peon et al.
detected a small amount of the triplet-triplet fluorescence at
510 nm from 3Ph2C*, which appeared within their instrumental
resolution.17,43 Since this excited triplet population has a singlet-
steps by first-order kinetics, even though the decay of the
carbenium ion could exhibit more complex, nonclassical kinetics
due to reaction in the geminate ion pair (pathway A′ in Scheme
1). This issue is discussed in more detail later. We define C(t)
to be the ratio of the Ph2CH+ population to the initial population
of 1Ph2C. Since no Ph2CH+ is present at time zero, C(t) is given
by
1
state precursor, it is possible that some or all of the Ph2C is
formed in an electronically excited state, 1Ph2C*. This popula-
tion would then internally convert to 1Ph2C in competition with
3
intersystem crossing to Ph2C*. Internal conversion between
upper singlet states is thought to occur in hundreds of femto-
seconds, consistent with the fast decay of the short-time signal.
However, the strength of the transition necessary to explain the
spikelike feature seems inconsistent with the generally weak
absorption of upper singlet states.
k1
C(t) )
[exp(-k2t) - exp(-k1t)]
(3)
k1 - k2
The maximum value of C(t), Cmax, depends only on the ratio,
r ) k1/k2, of the two rate constants:
In our opinion, the most likely possibility for the short-time
signal is absorption by the initial diazo excited state, 1Ph2CN2*.
In this case, the ultrafast decay rate is the sum of the rates of
all deactivation processes, including N2 extrusion, intersystem
crossing, and internal conversion to the diazo ground state. The
decay time of ≈ 300 fs indicates that carbon-nitrogen bond
scission is not impulsive, suggesting that the excited-state
potential is not purely repulsive. The strong wavelength
dependence of the photodestruction quantum yields of diazo
compounds45 is consistent with excited-state deactivation path-
ways other than N2 loss. In addition, the broad absorption band
of the short-time feature, which extends from 320 to beyond
500 nm (squares in Figure 4), resembles the spectrum of
Ph2CN2•- reported by Chateauneuf.46,47 Excited-state absorption
by singlet states of aromatic compounds frequently resembles
that of the corresponding radical anions due to their similar
electronic configurations.
Cmax ) r1/(1-r)
(4)
For small r (i.e. k1 , k2), there is negligible intermediate
population at all times (Cmax ≈ 0), while for large r, population
accumulates in the intermediate state, before decaying to product
(Cmax ≈ 1). Of course, regardless of the value of r, every 1Ph2C
is eventually protonated, according to eq 2. Equation 4 quantifies
the maximum amount of intermediate that can be detected at
one instant. This quantity is measured in our experiments (see
Figure 9) and will be discussed in more detail later.
Fitting transient signals from intermediates such as those in
Figure 6 to eq 3 is difficult for two reasons. First, it is
challenging to accurately determine values of k1 and k2 by
nonlinear least-squares fitting. In the presence of noise, extract-
ing rate constants from multiexponential decays can be highly
uncertain, particularly when the rates do not differ signifi-
cantly.49 Second, even when the rates can be accurately
determined by fitting, the association of rate constants with
individual reaction steps can be ambiguous.50 The decaying
signal at long times is always due to the step with the lower
rate, but the rate-limiting step can be either the first or second
step in eq 2.51 Our solutions to these difficulties are presented
in the next two subsections.
There are few reports of photophysical measurements for
diazo compounds. Kirmse determined a quantum yield of
photodestruction of 0.78 for diphenyldiazomethane exposed to
quasi-monochromatic light at 294 nm.45 Quantum yields were
approximately 10-fold smaller when the diazo compound was
excited in its long wavelength absorption band.45 The quantum
yields for other diazo compounds were significantly lower. This
suggests that other pathways must compete with N2 extrusion
in the singlet excited state of Ph2CN2. Silva and Thomas reported
a quantum yield of photolysis of ≈ 0.05 for 1-pyrenyldiazo-
methane for excitation between 340 and 380 nm.48 They
observed both fluorescence and ISC in this compound, which
they attributed to strong mixing between electronic states of
pyrene with those of the diazo group. Clearly, an important
question to answer in the future is the quantum yield of N2 loss
(b) Determination of k1 and k2. Attempts at fitting the
signals to a biexponential function of the form A1e-k t + A2e-k t
by nonlinear least squares generally did not produce unique
values of the parameters.52 To obtain reliable estimates, the
natural logarithm of the signal at delay times t greater than a
cutoff time, tcut, was fit to a straight line. The slope of this line
1
2
1
and the primary quantum yield for Ph2C formation.
(49) Istratov, A. A.; Vyvenko, O. F. ReV. Sci. Instrum. 1999, 70, 1233.
(50) Measurement of the rate of disappearance of 1Ph2C (or the rate of formation
of the ether) eliminates this ambiguity. Unfortunately, the UV-absorbing
ether product is not detected in our experiments. The singlet absorption
can be monitored as already shown in Figures 2 and 3. In methanol,
however, transient absorption at this wavelength contains a significant
contribution by Ph2CH+ due to the strength of this transition. This results
in the weak maximum seen at 360 nm in Figure 5. Due to the admixture
of 1Ph2C and Ph2CH+ this signal’s decay is governed by both rate constants,
just like the signal at 435 nm, and assignment is not possible.
(51) The emission from pyrene excimers illustrates the kinetics of an intermediate
in the limit k2 > k1. See: Van Dyke, D. A.; Pryor, B. A.; Smith, P. G.;
Topp, M. R. J. Chem. Educ. 1998, 75, 615. In this case, the growth of the
time-resolved excimer emission is determined not by the time scale of their
formation but by their short emission lifetime.
(2) Ph2CH+ Growth and Decay. (a) Sequential First-Order
Reaction Kinetics. The pronounced rise and decay of the
transient absorption signals in Figure 6 offer a striking example
on the ultrafast time scale of the time-dependent population of
an intermediate in a sequential two-step reaction mechanism:
k1
9
k2
1Ph2C
8 Ph2CH+ 98 ether
(2)
In eq 2, the rate of proton transfer is given by k1, while solvolysis
of the diphenylmethyl cation occurs with rate k2. We treat both
(52) This is a reflection of the flat nature of the ø2 surface with respect to τ1
and τ2. For example, the data from Ph2CH+ in ethanol (Figure 8b) were fit
by holding τ1 constant but allowing the other parameters to be optimized.
When τ1 was varied between 20 and 34 ps, ø2 for the fit varied by less
than 12%. Although the fits all appeared very good by eye, closer inspection
showed that many of them failed to accurately reproduce the long-time
signal decay. Note that independent amplitudes must be used (as opposed
to eq 3) since the signal at 435 nm contains contributions from both 1Ph2C
and Ph2CH+. Since Ph2CH+ absorption dominates, the best-fit values of
A1 and A2 are nearly equal and of opposite sign.
(43) Wang, Y.; Sitzmann, E. V.; Novak, F.; Dupuy, C.; Eisenthal, K. B. J. Am.
Chem. Soc. 1982, 104, 3238.
(44) Horn, K. A.; Allison, B. D. Chem. Phys. Lett. 1985, 116, 114.
(45) Kirmse, W.; Horner, L. Justus Liebigs Ann. Chem. 1959, 625, 34.
(46) Chateauneuf, J. E. J. Phys. Chem. 1990, 94, 7177.
(47) Chateauneuf, J. E. Chem. Phys. Lett. 1992, 188, 439.
(48) Silva, S.; Olea, A. F.; Thomas, J. K. Photochem. Photobiol. 1991, 54, 511.
9
6434 J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002

Proton Transfer from an Alcohol to Diphenylcarbene
A R T I C L E S
Table 2. Time Constantsa for Formation (τ1) and Decay (τ2) of Diphenylmethyl Cation in Neat Alcohols and Protonation Fractions (fp)
∆Amax/∆ACH3OH
fp
c
d
solvent
τ1 /ps
τ2 /ps
τsolvb /ps
Cmax
max
CH3OH
CH3OD
ethanol
1-propanol
2-propanol
9.0 ( 0.5
15 ( 1
16 ( 2
17 ( 2
35 ( 5
30.5 ( 1.5
34.8 ( 2.1
65 ( 6
5.0,e 6.80 ( 0.05f
0.60
0.53
0.63
0.72
0.62
1
0.30
0.30
0.15
0.14
0.09
7.46 ( 0.05f
0.88
0.54
0.54
0.31
16e
26e
117 ( 8
130 ( 15
a See text for detailed description of fitting procedure. Uncertainties are estimated standard deviations from the fits, assuming a standard deviation of
3 × 10-4 for an individual ∆A measurement. b Solvation times determined by dynamic Stokes shift measurements. c Evaluated from eq 4 and r ) τ2/τ1.
d Protonation fraction, see text for details. e From ref 76. f From ref 62.
fixed the value of the slower rate constant, which was labeled
k2. It will be shown in the next section that k2 actually is the
rate constant for the second reaction step in eq 2. Use of the
long-time signal decay to determine the slowest rate constant
of a multiexponential function is known as the peeling method.49
An initial guess for the value of tcut was made by eye from a
log plot of the signal. It was verified in every case that tcut was
always 3-5 times greater than τ1 ) 1/k1. Next, nonlinear least-
squares fitting was used to determine the value of k1, holding
k2 fixed at the value determined in the first step. The value of
k2 was insensitive to the precise value of tcut. The time constants,
τ1,2 ) 1/k1,2, determined by this procedure are listed in Table
2. The fits are shown by the solid curves in Figures 5-7.
Chateauneuf studied Ph2CN2 photolysis on the picosecond
time scale in several alcohols.3 In his experiments, Ph2CH+ was
formed within the duration of the 18 ps, 355 nm pump pulse
used to photoexcite diphenyldiazomethane. The cation popula-
tion decayed in 40, 70, and 85 ps in methanol, ethanol, and
2-propanol, respectively. There is good agreement with our τ2
values in methanol and ethanol. However, the decay time of
Ph2CH+ reported here in 2-propanol (τ ) 130 ps) is considerably
slower than the value reported by Chateauneuf. The reason for
this discrepancy is unclear at present. There are no previous
direct measurements of τ1 to compare our values with. It will
be shown later that the τ1 values are consistent with rate
constants inferred from product studies for the reaction of 1Ph2C
with alcohols.
(c) Protonation More Rapid Than Solvolysis. As already
noted, fitting to the intermediate’s growth and decay does not
by itself determine whether the first or second step of eq 2 has
the faster rate. Decisive evidence that protonation is the faster
step comes from the solvent isotope effect in methanol. The
fitting results in Table 2 indicate that the fast step has an isotope
effect, while the slower step shows a much smaller change of
only 14%, which is close to our experimental uncertainty. An
isotope effect is expected for protonation of singlet diphenyl-
carbene (pathway A in Scheme 1). On the other hand, the
disappearance of Ph2CH+ due to reaction with either an alkoxide
ion (pathway A′) or a neutral alcohol molecule is not expected
to show an isotope effect. Since τ1 shows an isotope effect, we
conclude that it is the time constant of protonation. If ylide
formation were the correct mechanism (pathway B in Scheme
1), then an isotope effect would be expected for the second step
(proton shift shown by pathway B′), but not for the first
(electrophilic attack). In this case, τ1 and τ2 would be associated
with the second and first steps of eq 2, respectively. While an
argument can be made that an intramolecular proton shift should
occur more rapidly than attack by a carbene electrophile on a
neutral alcohol molecule, the results in the ether THF rule out
ylide formation. In this solvent, which lacks an acidic proton,
there was no spectroscopic evidence of an ylide reaction
intermediate. Additionally, the decay of Ph2C was no faster
than is expected for ISC in this solvent.15,53
1
Further evidence that τ1 in Table 2 is the time constant for
proton transfer is provided by the relative signal amplitudes in
CH3OH and CH3OD at 435 nm. From the data in Figure 6, the
ratio of the maximum signal in CH3OD to the maximum signal
in CH3OH is 0.87. This ratio can be independently calculated
from eq 4 and the measured time constants. Using the values
of τ1 and τ2 in Table 2, Cmax(CH3OD)/Cmax(CH3OH) ) 0.88.
On the other hand, interchanging the values of τ1 and τ2 yields
a value of 1.3 for Cmax(CH3OD)/Cmax(CH3OH). Thus, assign-
ment of τ1 to protonation is the only way to reproduce the
observed isotope effect. A tacit assumption in this analysis is
that the molar absorption coefficient of Ph2CH+ in CH3OH is
the same at 435 nm as that of Ph2CD+ in CH3OD. We are
unaware of any data on the effect of deuteration on carbocation
absorption. The absorption spectrum of Ph2CH+ is insensitive
to solvent,26 so a solvent isotope effect on the absorption
spectrum can be ruled out. By the Born-Oppenheimer ap-
proximation, the potential energy surfaces of both cations are
identical. A difference in absorption would require that one or
more vibrational modes sensitive to deuteration be strongly
coupled to the electronic transition. The compound 3-hydroxy-
flavone, which has an exchangeable proton, shows identical
absorption in CH3OH and CH3OD.54 Identical absorption by
the two cations thus appears reasonable.
(3) Decay Rates of 1Ph2CH+ in Neat Alcohols. Protonation
of diphenylcarbene by alcohol creates an alkoxide/Ph2CH+ ion
pair. This reaction is essentially irreversible due to the high
basicity of the carbene. Thus, geminate recombination of the
two partners to reform the alcohol and 1Ph2C is not important.
However, geminate reaction is possible in which Ph2CH+
undergoes nucleophilic attack by the alkoxide ion to yield the
ether product, as shown by pathway A′ in Scheme 1. However,
carbocations are highly electrophilic, and they react rapidly with
any available nucleophile, including a solvent molecule.55 Thus,
the final ether product could also be formed by electrophilic
addition of an alcohol molecule to the carbocation, giving rise
to an oxonium ion, which subsequently deprotonates. On one
hand, the Coulombic attraction of the two ions for each other
could provide a strong driving force for ion pair collapse. On
the other hand, the large number of neutral alcohol molecules
surrounding the carbocation in the neat solvent might favor the
homogeneous reaction. In this section, we discuss our current
understanding of the importance of these two different channels
for the different alcohols studied.
(53) This value was estimated from data in Figure 1 of ref 15.
(54) Strandjord, A. J. G.; Barbara, P. F. J. Phys. Chem. 1985, 89, 2355.
(55) McClelland, R. A. Tetrahedron 1996, 52, 6823.
9
J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002 6435

A R T I C L E S
Peon et al.
In methanol and ethanol, the reaction of Ph2CH+ with a
neutral alcohol molecule appears to dominate ion pair collapse.
Bartl et al. measured the rates of reaction between Ph2CH+ and
various nucleophilic quenchers, including neutral alcohol mol-
ecules, by nanosecond photoheterolysis of Ph2CHCl in aceto-
nitrile.25,26 From these rates and the molarities of the neat
solvents, time constants of 34, 69, and 246 ps are predicted for
methanol, ethanol, and 2-propanol, respectively. Our measured
decay times in methanol and ethanol (τ2 in Table 2) agree very
well with these estimates, and we conclude that solvolysis by a
neutral molecule is faster than addition of the geminate alkoxide
ion to Ph2CH+. In the case of 2-propanol, however, the value
of τ2 in Table 2 is nearly a factor of 2 smaller than predicted
from the bimolecular rate constant measured in acetonitrile by
Bartl et al.
Of course, caution must be used when extrapolating bimo-
lecular rate constants measured in acetonitrile to the neat alcohol
solvents. Kirmse et al. reported pronounced solvent effects on
the reaction rates of Ph2CH+ with a given nucleophile.56 These
authors found that reaction rates were 40-100 times lower in
2,2,2-trifluoroethanol (TFE) than in acetonitrile, while still lower
rates were observed in 3,3,3,3′,3′,3′-hexafluoro-2-propanol
(HFIP). They argued that hydrogen bonding to the nucleophile
hinders its reaction with the carbocation.56 However, lower rates
of reaction with increasing hydrogen bonding would predict
slower rates in the neat solvents, in contrast to our observation
in 2-propanol. In other systems, scavenging rates have been
reported to increase faster than expected with quencher con-
centration. For example, a plot of the decay rate of Ph2CH+
vs alcohol concentration shows upward curvature in 1,2-
dichloroethane57 and in HFIP.56 Kirmse et al. rationalized this
behavior by the ability of a second alcohol molecule to act as
a base, preventing dissociation of an initial alcohol-carbocation
adduct.56 In this case, it is unclear why this mechanism would
accelerate quenching in 2-propanol, but not in ethanol and
methanol. In the latter two solvents the decay rates agree with
those inferred from measurements of the bimolecular rate con-
stants at low alcohol concentration. Furthermore, the quenching
rates for alcohols in acetonitrile were reported to be independent
of alcohol concentration.58
with solvent separated ion pairs are familiar from electron trans-
fer in polar solvents.60 This could retard the geminate reaction,
offsetting the Coulombic attraction of the two oppositely charged
ions. Further study is required to fully understand the competi-
tion between geminate ion pair collapse and reaction with a
neutral alcohol molecule. A similar competition occurs in
excited-state proton-transfer reactions to solvent at low pH.
Complex kinetics can be observed in these systems due to
the diffusion-influenced reaction of geminate partners. These
effects can be modeled with the time-dependent Smoluchowski
equation.61
To end this section, we note that the rate of decay of the
diphenylmethyl cation is just 10% slower in CH3OD than in
CH3OH. Yoshihara and co-workers determined by dynamic
Stokes shift measurements that solvation of photoexcited
coumarin 102 occurs approximately 10% more slowly in deu-
terated methanol.62 This small isotope effect is thus consistent
with rate-limiting solvation and is not believed to indicate proton
transfer.
(4) Intermolecular Proton Transfer from Alcohols to
1Ph2C. The τ1 values in Table 2 show that proton transfer from
oxygen to carbon can take place on an ultrafast time scale.
Intermolecular proton transfer reactions are challenging to study
by time-resolved methods because of the difficulty of initiating
reaction at the same instant of time for an ensemble of donors
and acceptors. Photoacids, molecules that undergo proton-
transfer reactions in an electronically excited state, overcome
this problem and have been extensively studied in recent
years.63-65 The fastest known photoacids transfer a proton to a
water molecule in slightly less than 10 ps. Excited-state proton
transfer to H2O occurs in 8 ( 1 ps at 25 °C for 5-cyano-1-
naphthol63 and in 7.1 ps for 7-hydroxy-4-methylflavylium.66
Proton transfer from excited 8-hydroxypyrene-1,3,6-trisulfonate
to acetate ion, which is present at high concentration in aqueous
solution, occurs in 3 ps.64 1Ph2C, which accepts a proton in
9 ps in neat methanol, belongs to this same class of ultra-
fast intermolecular proton-transfer reactions. Interestingly,
proton transfer to methanol by 5-cyano-1-naphthol is nearly
2 orders of magnitude slower than in water, requiring 390 ps
in CH3OH.63
The anomalous kinetics in 2-propanol are best explained in
our view by reaction between Ph2CH+ and the alkoxide ion
formed by protonation of the carbene. The reactive decay of
this ion pair is apparently faster than nucleophilic attack by
neutral 2-propanol, which is a weaker nucleophile than either
methanol or ethanol. A crude estimate is that Ph2CH+ and the
2-propoxide ion react in 100 ps. The geminate reaction rate is
likely limited by the energetic cost of desolvating two singly
charged ions in order for reaction to take place. If the rate of
protonation of 1Ph2C is controlled by the solvent reorganization
necessary to solvate the incipient alkoxide ion (see next section),
then solvent separated alkoxide/Ph2CH+ ion pairs could be
formed quickly.59 The large reorganization energies associated
The high basicity of singlet carbenes follows from the high
pKa values of their corresponding carbocations. For example,
the stable carbene imidazole-2-ylidene, which deprotonates
weakly acidic hydrocarbons such as indene, is estimated to have
a pKa of 24, or a pKb of -10.67 By comparison, the powerful
photoacid 5-cyano-1-naphthol has an excited-state pKa of -2.8.63
Ab initio calculations predict a lower proton affinity for
imidazol-2-ylidene68 than for diphenylcarbene,9 suggesting that
the latter may be even more basic.
The rate constant for the disappearance of singlet carbenes
longer lived than 1Ph2C have been measured in the presence of
(60) Gould, I. R.; Farid, S. Acc. Chem. Res. 1996, 29, 522.
(61) Agmon, N.; Szabo, A. J. Chem. Phys. 1990, 92, 5270.
(62) Shirota, H.; Pal, H.; Tominaga, K.; Yoshihara, K. J. Phys. Chem. 1996,
100, 14575.
(63) Pines, E.; Pines, D.; Barak, T.; Magnes, B. Z.; Tolbert, L. M.; Haubrich,
J. E. Ber. Bunsen-Ges. Phys. Chem. 1998, 102, 511.
(64) Genosar, L.; Cohen, B.; Huppert, D. J. Phys. Chem. A 2000, 104, 6689.
(65) Cohen, B.; Huppert, D. J. Phys. Chem. A 2001, 105, 2980.
(66) Lima, J. C.; Abreu, I.; Brouillard, R.; Macanita, A. L. Chem. Phys. Lett.
1998, 298, 189.
(56) Kirmse, W.; Guth, M.; Steenken, S. J. Am. Chem. Soc. 1996, 118, 10838.
(57) Sujdak, R. J.; Jones, R. L.; Dorfman, L. M. J. Am. Chem. Soc. 1976, 98,
4875.
(58) As reported in ref 26. The authors did not specify the range of quencher
concentrations used in acetonitrile, but the maximum concentration is
presumably considerably below 1 M.
(59) Alkoxide ions have high mobilities due to the proton relay mechanism
(Grotthus mechanism). Slower escape by the bulkier alkoxide ion derived
from 2-propanol could be another reason that ion pair recombination is
more important in this solvent than in the shorter alcohols.
(67) Alder, R. W.; Allen, P. R.; Williams, S. J. J. Chem. Soc., Chem. Commun.
1995, 1267.
(68) Dixon, D. A.; Arduengo, A. J. J. Phys. Chem. 1991, 95, 4180.
9
6436 J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002

Proton Transfer from an Alcohol to Diphenylcarbene
A R T I C L E S
proton donors in acetonitrile.5,56 The observed rates were
somewhat smaller than the diffusion limit, and the isotope effects
were close to unity. It was argued that the lack of an isotope
effect was due to a nonlinear transition-state geometry, or a rate-
limiting preassociation step. In contrast to the results of Steenken
and co-workers, we observe a kinetic isotope effect of 1.7 on
protonation. This value is close to values observed for photo-
acids.63
The 9.0 ps time constant for protonation in methanol
corresponds to a bimolecular rate of reaction of 4.5 × 109 M-1
s-1. This rate is near the diffusion limit for methanol, but since
photolysis occurs in the neat solvent the nascent carbene is
surrounded by a number of potential proton donors, and no
diffusion is necessary. In that sense, these experiments are the
proton-transfer analogue of ultrafast intermolecular electron
transfer where the solvent acts as electron donor.12,69,70 Electron
donation from a solvent can occur in <100 fs, but here proton
donation requires significantly longer times. This indicates that
a barrier exists to reactionsa barrier that we suggest arises from
the solvent reaction coordinate.
(13.1 M) gives a rate of 2.2 × 109 M-1 s-1. This bimolecular
rate is a factor of 2.0 slower than our measured rate in methanol.
Eisenthal and co-workers also varied the concentration of
methanol between 0 and 10 M in acetonitrile containing 0.1 M
isoprene and measured the ratio of singlet to triplet products.
From this independent measurement they derived a rate constant
of 7.4 × 109 M-1 s-1 for the reaction of 1Ph2C with methanol,
corresponding to a lifetime of 5.5 ps in neat methanol. This is
in reasonable agreement with our value of 9.0 ps.
These time constants must be compared cautiously with ours
due to the variation in solvent. As seen with the carbocations,
bimolecular reaction rates between singlet carbenes and alcohols
increase with increasing alcohol concentration.5,56,79 Kirmse et
al. argued that the increased rates reflect the kinetic advantage
of a second alcohol molecule which can stabilize the incipient
alkoxide ion by hydrogen bonding.56 Steenken observed in-
creased bimolecular rates at increased alcohol concentration for
carbenes reacting with weakly acidic alcohols.5 This was
attributed to more than one proton donor molecule participating
in the protonation. These ideas are consistent with a rate-limiting
role for the protic solvent.
The values of τ1 support the notion that protonation of
1Ph2C is limited by the time for the solvent to reorganize. The
protonation of 1Ph2C by an alcohol molecule creates an ion pair
from two neutral species. It is therefore reasonable to expect
that solvent fluctuations will be strongly coupled to the reaction
coordinate, particularly since the solvent plays a dual role as
proton donor and solvating medium. Many workers have
emphasized the connection between solvation dynamics and
proton transfer.71-74 Solvent reorientation may be necessary
before proton transfer can proceed.75 Hynes and co-workers have
suggested that solvent fluctuations are the reaction coordinate
for nonadiabatic and adiabatic proton transfer.74 Pines et al. have
emphasized that exothermic intermolecular proton-transfer reac-
tions are controlled by solvent fluctuations.63 The protonation
times reported here correlate reasonably well with solvation
times of the various alcohols determined by Maroncelli and co-
workers (see Table 2).76 The correlation is not perfect, since
the isotope effect of 1.7 is considerably larger than that observed
for solvation in deuterated methanol. For intermolecular proton
transfer to occur, it is believed that a hydrogen bond must first
be formed between donor and acceptor.77 In this case, the solvent
must first form a hydrogen bond with the incipient carbene. In
neat methanol, molecular dynamics simulations suggest that
hydrogen bond lifetimes are ≈5 ps.78 It is possible that this time
increases in the deuterated solvent.
(5) Quantum Yield of Protonation. According to the
actinometric results in Figure 9, the quantum yield of Ph2CH+,
φp, reaches a maximum value of 0.18, 18 ps after the pump
pulse. Since φp first increases, and then decreases, it is more
useful to measure the fraction of all photoexcited diazo
molecules that are ultimately protonated. This quantity, which
we term the protonation fraction, fp, is given by φp/Cmax. Cmax
is defined in eq 4 and is calculated from the time constants
τ1 and τ2 determined by fitting. The value for methanol,
fp(CH3OH), is equal to 0.30. If the relative signal amplitudes
in Figure 8 are correct, then the protonation fraction in the other
alcohols is given by
∆Amax
0.18
Cmax
fp )
‚
(5)
CH3OH
∆A
max
where ∆Amax is the maximum absorbance change due to Ph2CH+
in each alcohol. This assumes that the value of ꢀ at 435 nm
is similar for Ph2CH+ in all alcohols. Values of fp are
summarized in Table 2.
The protonation fraction decreases steadily on going from
methanol to 2-propanol. In other words, fewer total carbocations
are formed than are predicted from the rate constants of eq 2.
This could be due to a decreased branching ratio for protonation
in the less acidic alcohols. Belt et al. previously used nanosecond
laser flash photolysis to estimate quantum yields of protonation
for a series of diarylcarbenes in 1:1 TFE/acetonitrile.6 They
observed greater yields for longer-lived carbocations. Interest-
ingly, the quantum yield for Ph2CH+ in this solvent mixture
was just 0.007. The quantum yields reported by Belt et al. cannot
be primary yields since they were measured with nanosecond
pulses. It is important to ask how many carbocations still remain
several nanoseconds after the laser pulse. Even if the rate of
reaction between TFE (a weak nucleophile) and Ph2CH+ is low,
the rate of the geminate reaction between the carbocation and
the TFE anion could be much higher. We speculate that the
low yields reported by Belt et al. are due to the fact that
relatively few carbocations can escape these initial ion pairs.
The bimolecular rate compares well with the value of 5.6 ×
109 M-1 s-1 estimated by Eisenthal et al. for the reaction of
1Ph2C with methanol in N2 saturated acetonitrile.18 These authors
measured a 1.6-fold slower rate of reaction for 2-propanol. The
value of 35 ps from Table 2 and the molarity of neat 2-propanol
(69) Yoshihara, K. AdV. Chem. Phys. 1999, 107 (Pt. 2), 371.
(70) Xu, Q.-H.; Scholes, G. D.; Yang, M.; Fleming, G. R. J. Phys. Chem. 1999,
103, 10348.
(71) Marcus, R. A. Farday Discuss. Chem. Soc. 1982, 74, 7.
(72) Moog, R. S.; Maroncelli, M. J. Phys. Chem. 1991, 95, 10359.
(73) Mente, S.; Maroncelli, M. J. Phys. Chem. A 1998, 102, 3860.
(74) Ando, K.; Hynes, J. T. Acid-Base Proton Transfer and Ion Pair Formation
in Solution; Advances in Chemical Physics; Wiley: New York, 1999; p
381.
(75) Kyrychenko, A.; Herbich, J.; Izydorzak, M.; Gil, M.; Dobkowski, J.; Wu,
F. Y.; Thummel, R. P.; Waluk, J. Isr. J. Chem. 1999, 39, 309.
(76) Horng, M. L.; Gardecki, J. A.; Papazyan, A.; Maroncelli, M. J. Phys. Chem.
1995, 99, 17311.
(77) Eigen, M. Angew. Chem., Int. Ed. Engl. 1964, 3, 1.
(78) Matsumoto, M.; Gubbins, K. E. J. Chem. Phys. 1990, 93, 1981.
(79) Griller, D.; Liu, M. T. H.; Scaiano, J. C. J. Am. Chem. Soc. 1982, 104,
5549.
9
J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002 6437

A R T I C L E S
Peon et al.
It is not clear at this stage whether other reaction pathways
are needed to explain the O-H insertion reaction. While the
decreasing protonation fractions observed in the longer alcohols
could be explained by increased branching into a concerted
reaction pathway,6 it could also be explained by less efficient
formation of 1Ph2C in these solvents. For example, if every third
diazo excited state would form singlet carbene, then the total
amount of Ph2CH+ observed in methanol would be explained,
assuming that protonation is the only decay channel. This is
quite a bit below the photodestruction yield of 0.78 reported
by Kirmse,45 but, as discussed earlier, the initial quantum yield
of 1Ph2C is unknown. Although Kirmse argued that the
photodecomposition quantum yield of Ph2CN2 is independent
of solvent, 80 his experiments measured only the net destruction
rate of the diazo compound and do not provide any insight into
how the carbene population might partition into different spin
states. It is conceivable that partitioning of the initially excited
diazo singlet state (1Ph2CN2* in Scheme 2) into 1Ph2C and other
states such as 3Ph2C or 3Ph2C* could depend on solvent.
Eisenthal showed that some amount of 3Ph2C* is formed in less
than 20 ps after excitation at 266 nm.17 Some evidence for lower
primary quantum yields of 1Ph2C in the longer alcohols comes
from a comparison of the signals at 360 nm. When transients
at 360 nm (data not shown) are plotted with the t ) 0 features
scaled to have the same amplitude,81 then the initial signal level
at the conclusion of the spike decreases from methanol to
2-propanol. This suggests that less 1Ph2C is formed immediately
after loss of N2. Further studies are necessary before it can be
1
decided whether every Ph2C formed is ultimately protonated
or whether reaction channels exist7 that do not require proto-
nation.
Summary and Conclusions
The results presented here indicate that carbocations are short-
1
lived intermediates when Ph2C reacts in neat alcohols. Rates
of protonation were measured directly and shown to correlate
with solvation times in alcohols from methanol to 2-propanol.
The rate-limiting step in the overall O-H insertion reaction is
not protonation, but the decay of Ph2CH+ by reaction with a
nucleophile. In methanol and ethanol, ion pair collapse does
not appear to compete with reaction of Ph2CH+ with a neutral
solvent molecule. In 2-propanol, however, the decay rate of the
cation is faster than expected from the bimolecular reaction rate
measured for 2-propanol at low concentration in acetonitrile.
This is suggested to arise from geminate reaction in the initial
ion pair. The ability to generate powerful bases on an ultrafast
time scale opens up new possibilities for studying adiabatic
proton-transfer reactions in polar solvents. For those carbenes
with singlet ground states, which are also strong bases, it should
be possible to study intermolecular proton-transfer reactions for
which all participants are in their electronic ground states. This
will simplify theoretical study of this important class of
reactions.
Acknowledgment. We thank Professors Matthew S. Platz and
Christopher M. Hadad of the Ohio State University for helpful
comments.
(80) Kirmse, W. Justus Liebigs Ann. Chem. 1963, 666, 9.
(81) The CH3OH/CH3OD data were recorded in back-to-back scans under
identical experimental conditions. The spikes had the same amplitude,
indicating no solvent isotope effect on this feature.
JA017485R
9
6438 J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002