Hydrogen bonding switches the spin state of diphenylcarbene from triplet to singlet

Article abstract of DOI:10.1002/anie.201400176

Spin specificity is one of the most important properties of carbenes in their reactions. Alcohols are typically used to probe the reactive spin states of carbenes: O-H insertions are assumed to be characteristic of singlet states, whereas C-H insertions a

Full text of DOI:10.1002/anie.201400176

.
Angewandte
Communications
DOI: 10.1002/anie.201400176
Hydrogen Bonds
Hydrogen Bonding Switches the Spin State of Diphenylcarbene from
Triplet to Singlet**
Paolo Costa and Wolfram Sander*
Dedicated to Professor Wolfgang Kirmse
Abstract: Spin specificity is one of the most important
properties of carbenes in their reactions. Alcohols are typically
used to probe the reactive spin states of carbenes: OÀH
insertions are assumed to be characteristic of singlet states,
whereas CÀH insertions are typical for the triplets. Surpris-
ingly, the experiments presented here suggest that the spin
ground state of diphenylcarbene 1 switches from triplet to
singlet if the carbene is allowed to interact with methanol.
Carbene 1 and methanol form a strongly hydrogen-bonded
singlet ground state complex that was synthesized in low-
temperature matrices and characterized by IR spectroscopy.
This methanol complex is only metastable, and even at 3 K
slowly rearranges to form the product of OÀH insertion
through quantum chemical tunneling. Thus, the ground state
triplet (in the gas phase) carbene 1 forms exclusively the
products expected from a singlet carbene. Whereas the
assumption of spin specific reactions of carbenes is correct,
the spin state itself can be changed by solvent interactions, and
Scheme 1. Mechanism of the reaction of diphenylcarbene 1 with meth-
anol.
therefore widely accepted conclusions drawn from earlier
experiments have to be revisited.
C
arbenes are molecules with divalent carbon atoms that
from the singlet state S-1 to the triplet state T-1, by pico-
[10,14,15]
play key roles as organic reactive intermediates, as ligands in
metal organic chemistry, and as highly potent organocata-
second fluorescence spectroscopy.
The lifetime of S-1 is
limited by ISC to 340 ps in acetonitrile and to only 95 ps in the
less polar isooctane. By assuming that isopropene and
methanol are selective scavengers for T-1 and S-1, respec-
[
1–3]
lysts.
The chemistry of carbenes is controlled by their spin
[
2,4]
state, which is either triplet or singlet,
and this spin-
dependent reactivity of carbenes forms the basis of carbene
chemistry. In particular, reactions of carbenes with alcohols
were extensively studied to develop general rules for the spin-
tively, the singlet triplet gap DG was determined to
ST
À1
À1
À3.36 kcalmol in acetonitrile and to À4.75 kcalmol in
isooctane, thus in both solvents the triplet state is consid-
[
5,6]
[14]
selective chemistry of carbenes:
Triplet carbenes undergo
erably more stable than the singlet state. The smaller S-T
[7]
insertion into CÀH bonds, whereas singlet carbenes insert
gap in acetonitrile compared to isooctane was rationalized by
the better stabilization of the more polar singlet state in the
[
8]
into the OÀH bonds of alcohols.
[16]
Diphenylcarbene 1 is an archetypal triplet-ground-state
solvent of higher polarity.
However, these results were challenged by Griller
arylcarbene, and its reaction with methanol has been exten-
[8–12]
[13]
[11,17]
sively studied by time resolved
and low-temperature
et al.
From the temperature dependence of the quenching
spectroscopy (Scheme 1). Eisenthal et al. were able to directly
of 1 by methanol, activation barriers E were determined to
a
À1
À1
determine k , the rate for the intersystem crossing (ISC)
1.66 kcalmol
in acetonitrile and to 3.61 kcalmol
in
ST
isooctane. These barriers are considerably lower than the
barriers that are estimated for an essentially diffusion-
controlled reaction of 1 with methanol. The conclusion was
that the assumption of a spin-selective reaction of 1 with
methanol—which forms the basis for the determination of
kTS—is flawed, and 1 is able to react directly with methanol
from its triplet state. This implies that the ISC occurs along
the reaction coordinate rather than at a stationary point.
Because of its fundamental importance, we re-investi-
gated the reaction between 1 and methanol under the
conditions of matrix isolation using low-temperature IR and
[
*] P. Costa, Prof. Dr. W. Sander
Lehrstuhl fꢀr Organische Chemie II, Ruhr-Universitꢁt Bochum
44781 Bochum (Germany)
E-mail: wolfram.sander@rub.de
Homepage: http://www.rub.de/aksander
[
**] We thank Prof. W. Kirmse for valuable discussions. This work was
supported by the Cluster of Excellence RESOLV (EXC 1069) funded
by the Deutsche Forschungsgemeinschaft.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201400176.
5
122
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 5122 –5125

Angewandte
Chemie
EPR spectroscopy. Additional EPR experiments were per-
formed in methanol glass.
rapidly evaporates). To our surprise, T-1 rapidly reacted with
methanol if the matrix was allowed to warm from 3 K to 25 K.
The reaction of T-1 could be monitored by both EPR and IR
spectroscopy. The EPR signals of T-1 decrease by more than
40% within 5 min, but no new signals are formed, indicating
that the reaction product is diamagnetic (Figure 1 and
Figures S3 and S4 in the Supporting Information). This is
puzzling, since in methanol glass at the same temperature no
reaction of T-1 is found even after several hours.
Visible-light photolysis (530 nm) of diphenyldiazome-
thane (DPDM) in frozen methanol at 4 K produces diphe-
nylcarbene 1 in high yields. The formation of 1 in its triplet
state can be directly followed by EPR spectroscopy. At
temperatures below 30 K the EPR signals of T-1 are persis-
tent, and even after several hours no decrease in intensity is
observed. Warming the methanol glass above 60 K, however,
results in a slow decrease of the signals of T-1 and concurrent
formation of a triplet radical pair 4. Platz et al. reported that
T-1 reacts in frozen 2-propanol at 77 K preferentially through
CÀH abstraction–recombination to form an alcohol as CÀH
IR spectroscopy allowed us to directly monitor the
reaction of T-1 with methanol and formation of diamagnetic
products (Figure 2). The IR spectra clearly show that the
[13]
insertion product. Analysis of the decay kinetics and the
very large CÀH/D kinetic isotope effect prompted these
authors to claim that H atom tunneling is rate determining for
the CÀH insertion into alcohols. However, in a later publi-
cation it was described that the formal CÀH insertion results
from secondary photochemistry of T-1 rather than from
[
13b]
a thermal reaction.
In our experiments, we find that the
rates for the reaction of T-1 with methanol are getting
exceedingly slow below 40 K (see Table S1). A tunnelling
reaction from the lowest vibrational state is expected to be
independent of temperature, which is in obvious contra-
diction to the observation described here.
To gain more insight into the mechanism of the reaction
between 1 and methanol we performed experiments in argon
matrices doped with 1% methanol. As expected, 530 nm
irradiation of DPDM in these matrices at 3 K produced triplet
carbene T-1 in high yields. The IR and EPR spectra of T-
Figure 2. IR spectra showing the formation of the complex between S-
1
and CH OD. a) T-1 in Ar doped with 1% CH OD at 3 K. b) Differ-
3 3
1
obtained under these conditions are in excellent agreement
ence IR spectrum of the same matrix showing changes after annealing
for 10 minutes at 25 K. Bands pointing downwards assigned to T-
[
13,18,19]
with previously published spectra.
below 20 K the diffusion of species matrix-isolated in argon is
efficiently inhibited, and thus bimolecular reactions cannot
occur. To allow the diffusion of small molecules in solid argon,
it is necessary to anneal the matrix at temperatures between
At temperatures
1
and CH OD are disappearing, and bands pointing upwards assigned
3
[20]
to the complex between S-1 and CH OD are appearing. c) Difference
3
IR spectrum of the same matrix showing changes after 17 h at 3 K,
and bands pointing downwards are assigned to the complex between
S-1 and CH OD. Bands pointing upwards are assigned to [D ]-3. d) IR
3
1
2
5 and 35 K for several minutes (at higher temperatures argon
spectrum of [D ]-3 matrix-isolated in argon at 3 K.
1
reaction of T-1 with methanol exclusively leads to the product
of OÀH insertion, ether 3, whereas the expected product of C-
H insertion, alcohol 5, is not formed. Apparently, the results
of these experiments are not in agreement with the generally
accepted rules of spin-selective reactivity of carbenes. Careful
analysis of the IR spectra revealed that ether 3 is not directly
formed and that a metastable intermediate with a very strong
À1
absorption at 1327.8 cm is the primary product of the
reaction between T-1 and methanol. This labile intermediate
rearranges to the final product 3 even at temperatures as low
as 3 K. Several isotopomers were synthesized by allowing T-
1
to react with CH OH, CH OD, and CD OH (Figure 2 and
3 3 3
1
3
Figures S5–S10), and [ C]-T-1 (carbene center labeled with
Figure 1. EPR spectra showing the reaction of T-1 in an argon matrix
doped with 1% methanol. a) Matrix at 5 K showing the triplet
spectrum of 1 with the zero-field splitting (zfs) parameters
13
À1
C) to react with CH OD. The band at 1327.8 cm shows
3
13
À1
a very strong C isotopic shift of À23.2 cm . While using
À1
À1
CD OH results in no significant isotopic shift of this band,
3
D=0.417 cm , E=0.019 cm . b) After annealing for 5 minutes at
5 K 42% of the signal intensity of T-1 is lost. The signals of the
À1
with CH OD a blue-shift of + 2 cm is observed, suggesting
2
3
that the carbene center directly interacts with the OÀH/D
radicals are formed during the initial photolysis of the precursor and
do not change in intensity during annealing.
group of methanol.
Angew. Chem. Int. Ed. 2014, 53, 5122 –5125
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5123

.
Angewandte
Communications
[
12]
The labile intermediate shows an OÀH stretching vibra-
constant for this proton transfer of only 9 ps in methanol.
À1
tion at 2802 cm (Figure S7), which corresponds to a huge
DFT calculations in the gas phase predict that the reaction of
À1
red-shift of 864 cm if compared to matrix-isolated CH OH.
the S-1–methanol complex to 3 is highly exothermic (À55 kcal
3
À1
À1
The OÀD red-shift is 520 cm (Figure 2). This evidences that
mol , B3LYP/6-311 ++ G(d,p) + ZPE, Figure S12) and the
the intermediate is a very strongly hydrogen-bonded complex
between carbene 1 and methanol. Since this complex is EPR
silent, we conclude that the singlet state S-1 is acting as
hydrogen bond acceptor for methanol rather than the triplet
state T-1. This is confirmed by DFT calculations, which
predict an IR spectrum of the S-1–methanol complex that is in
excellent agreement with the experimental spectrum.
ion pair 6 rather corresponds to a transition state than
a reaction intermediate. In an unpolar argon matrix, we do
not expect the ion pair to be an intermediate. However, in
polar solvents the ion pair should be long-lived enough to be
detected spectroscopically.
The results presented here require to re-evaluate pre-
viously published data on spin selective reactions of carbenes.
Although the reaction of 1 with methanol is indeed spin
specific as outlined in Scheme 1, the basic assumption that
1 and related carbenes have triplet ground states in all
solvents is wrong. The singlet state of a carbene can become
more stable than the triplet state by hydrogen bonding with
The highly polar S-1 state is stabilized by hydrogen-
À1
bonding with methanol by À7.7 kcalmol , whereas T-1 forms
only a weak van der Waals complex with a binding energy of
À1
À1.8 kcalmol (B3LYP/6-311 ++ G(d,p) + ZPE, Figure 3).
Thus, a single molecule of methanol stabilizes S-1 enough to
make it thermodynamically more stable than T-1 even in an
unpolar environment such as solid argon.
a single molecule of alcohol, even if DG is larger than
À5 kcalmol in the gas phase. Since many triplet carbenes
ST
À1
have singlet–triplet gaps in this range, we expect that this type
of singlet-state stabilization is of general importance for
carbene reactions not only with alcohols, but also with other
solvents and reagents that can act as hydrogen bond donors.
This effect is due to specific solvation and goes beyond the
general stabilization of the polar singlet states of carbenes
with respect to the less polar triplet states with increasing
solvent polarity.
Interestingly, while the reaction of T-1 in methanol-doped
argon at 25 K is fast, no reaction is observed in bulk methanol
at the same temperature, and the EPR spectra show very
strong signals of T-1. This could be related to methanol
molecules involved in the hydrogen-bond network in solid
methanol being weaker hydrogen-bond donors than matrix-
isolated methanol molecules. Since for the S-1–methanol
complex in the gas phase the singlet state is calculated to be
À1
only 0.5 kcalmol more stable than the triplet state, in bulk
methanol the order of stability might be reversed. This opens
a whole range of new possibilities for controlling the spin state
of carbenes by rational design of solvent systems, for example,
solvent mixtures with only low concentrations of methanol or
more hydrophobic alcohols as solvents.
The concept presented here of switching the ground state
multiplicity of a molecule by solvent interactions represents
an unique way to control chemical reactivity. This new
principle not only contributes to the understanding of carbene
reactivity in solution, but more importantly, it paves the way
to design new carbene reactions by tailoring the solvent
system.
Figure 3. Relative energies of T-1, S-1, and their most stable complexes
with methanol. Calculations performed at the (U)B3LYP/6-311+ +G-
(
d,p) and (U)B3LYP-D3/6-311+ +G(d,p) (in parenthesis) levels of
theory. The bond angles at the carbene centers and the non-bonding
C···HO distances are shown in green.
To elucidate the reaction mechanism, we measured the
kinetics of the rearrangement of the complex between S-1 and
CH OH, CH OD, and CD OH at 3 and at 12 K. First-order
3
3
3
kinetics was found, and with CH OH and CD OH the rate
3
3
À5
À1
was determined to 6.8 ꢀ 10 Æ 0.2 s , whereas with CH OD
3
À5
a rate of 1.3 ꢀ 10 Æ 0.2 was observed, corresponding to
Received: January 7, 2014
Published online: April 2, 2014
a kinetic isotope effect (KIE) of 5. Between 3 and at 12 K the
rates are independent of temperature, and thus the Arrhenius
activation barrier for the rearrangement is zero. This is strong
evidence for a tunneling mechanism.
Keywords: carbenes · hydrogen bonds · IR spectra ·
.
matrix isolation · singlet–triplet splitting
The most likely mechanism for the formation of ether 3
from the S-1–methanol complex is proton transfer to form the
ion pair 6 followed by combination of the ions to give 3. This
is in accordance with the observation of Kirmse et al. that in
solution diarylcarbenes react with alcohols to give diaryl-
[
1] M. Melaimi, M. Soleilhavoup, G. Bertrand, Angew. Chem. 2010,
22, 8992 – 9032; Angew. Chem. Int. Ed. 2010, 49, 8810 – 8849.
1
[
2] W. Sander, G. Bucher, S. Wierlacher, Chem. Rev. 1993, 93, 1583 –
[
8]
carbenium ions. Kohler and co-workers measured a time
1621.
5
124
www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 5122 –5125

Angewandte
Chemie
[
[
[
[
[
[
[
3] D. Martin, M. Soleilhavoup, G. Bertrand, Chem. Sci. 2011, 2,
89 – 399.
4] W. Kirmse, Carbene Chemistry, Academic Press, New York,
971.
5] W. Kirmse in Advances in carbene chemistry, Vol. 1 (Ed.: U. H.
Brinker), Jai, Greenwich, CT, 1994, pp. 1 – 57.
6] W. Kirmse in Advances in Carbene Chemistry, Vol. 3 (Ed.: U. H.
Brinker), Elsevier Science B.V., Amsterdam, 2001, pp. 1 – 51.
7] L. M. Hadel, M. S. Platz, J. C. Scaiano, J. Am. Chem. Soc. 1984,
[12] J. Peon, D. Polshakov, B. Kohler, J. Am. Chem. Soc. 2002, 124,
3
6428 – 6438.
[13] a) M. S. Platz, V. P. Senthilnathan, B. B. Wright, C. W. Mccurdy,
J. Am. Chem. Soc. 1982, 104, 6494 – 6501; b) E. Leyva, R. L.
Barcus, M. S. Platz, J. Am. Chem. Soc. 1986, 108, 7786-7788.
[14] K. B. Eisenthal, N. J. Turro, E. V. Sitzmann, I. R. Gould, G.
Hefferon, J. Langan, Y. Cha, Tetrahedron 1985, 41, 1543 – 1554.
1
[
15] K. B. Eisenthal, N. J. Turro, M. Aikawa, J. A. Butcher, Jr., C.
DuPuy, G. Hefferon, W. Hetherington, G. M. Korenowski, M. J.
McAuliffe, J. Am. Chem. Soc. 1980, 102, 6563 – 6565.
1
06, 283 – 287.
8] W. Kirmse, J. Kilian, S. Steenken, J. Am. Chem. Soc. 1990, 112,
399 – 6400.
9] E. V. Sitzmann, Y. Wang, K. B. Eisenthal, J. Phys. Chem. 1983,
7, 2283 – 2285.
10] K. B. Eisenthal, R. A. Moss, N. J. Turro, Science 1984, 225, 1439 –
445.
11] D. Griller, A. S. Nazran, J. C. Scaiano, J. Am. Chem. Soc. 1984,
06, 198 – 202.
[
16] J. Wang, J. Kubicki, H. Peng, M. S. Platz, J. Am. Chem. Soc. 2008,
1
30, 6604 – 6609.
17] D. Griller, A. S. Nazran, J. C. Scaiano, Acc. Chem. Res. 1984, 17,
83 – 289.
6
[
2
8
[
[
18] W. W. Sander, J. Org. Chem. 1989, 54, 333 – 339.
19] R. W. Murray, E. Wasserman, W. A. Yager, A. M. Trozzolo, J.
Am. Chem. Soc. 1962, 84, 3213 – 3214.
[
[
1
[
20] A. Mardyukov, R. Crespo-Otero, E. Sanchez-Garcia, W. Sander,
Chem. Eur. J. 2010, 16, 8679 – 8689.
1
Angew. Chem. Int. Ed. 2014, 53, 5122 –5125
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5125