Cyclopropylhydroxycarbene

Article abstract of DOI:10.1021/ja204507j

Cyclopropylhydroxycarbene was generated by high-vacuum flash pyrolysis of cyclopropylglyoxylic acid at 960 °C. The pyrolysis products were matrix-isolated in solid Ar at 11 K and characterized by means of IR spectroscopy. Upon photolysis, the carbene undergoes ring expansion, thereby paralleling the reactivity of other known cyclopropylcarbenes. The ring expansion product, cyclobut-1-en-1-ol, was characterized for the first time. Matrix-isolated cyclopropylhydroxycarbene undergoes [1,2]H-tunneling through a barrier of approximately 30 kcal·mol^-1, yielding cyclopropylcarboxaldehyde. The cyclopropyl moiety acts as a π-donor and increases the half-life by almost a factor of 10 compared to parent hydroxymethylene, resulting in a temperature-independent half-life of τ = 17.8 h at both 11 and 20 K. Hence, cyclopropylhydroxycarbene is the first hydroxycarbene that differs from other members of its family by a significantly prolonged half-life. As expected, the O-deuterated analogue does not show tunneling. Our findings are rationalized by accurate CCSD(T)/cc-pVnZ (n = D, T)//M06-2X/6-311++G(d,p) computations. The half-life of cyclopropylhydroxycarbene was verified by tunneling computations employing the Wentzel-Kramers-Brillouin formalism. By comparison with other experimentally known hydroxycarbenes, we determine the electronic donor capabilities of the carbenes' substituents to be a dominant factor governing their half-lives.

Full text of DOI:10.1021/ja204507j

ARTICLE
pubs.acs.org/JACS
Cyclopropylhydroxycarbene
David Ley, Dennis Gerbig, J. Philipp Wagner, Hans P. Reisenauer, and Peter R. Schreiner*
Justus-Liebig University, Heinrich-Buff-Ring 58, 35392 Giessen, Germany
S Supporting Information
b
ABSTRACT: Cyclopropylhydroxycarbene was generated by high-va-
cuum flash pyrolysis of cyclopropylglyoxylic acid at 960 °C. The pyrolysis
products were matrix-isolated in solid Ar at 11 K and characterized by
means of IR spectroscopy. Upon photolysis, the carbene undergoes ring
expansion, thereby paralleling the reactivity of other known cyclopro-
pylcarbenes. The ring expansion product, cyclobut-1-en-1-ol, was char-
acterized for the first time. Matrix-isolated cyclopropylhydroxycarbene
undergoes [1,2]H-tunneling through a barrier of approximately 30
kcal molꢀ1, yielding cyclopropylcarboxaldehyde. The cyclopropyl moi-
3
ety acts as a π-donor and increases the half-life by almost a factor of 10
compared to parent hydroxymethylene, resulting in a temperature-independent half-life of τ = 17.8 h at both 11 and 20 K. Hence,
cyclopropylhydroxycarbene is the first hydroxycarbene that differs from other members of its family by a significantly prolonged
half-life. As expected, the O-deuterated analogue does not show tunneling. Our findings are rationalized by accurate CCSD(T)/cc-
pVnZ (n = D, T)//M06-2X/6-311++G(d,p) computations. The half-life of cyclopropylhydroxycarbene was verified by tunneling
computations employing the WentzelꢀKramersꢀBrillouin formalism. By comparison with other experimentally known hydro-
xycarbenes, we determine the electronic donor capabilities of the carbenes’ substituents to be a dominant factor governing their
half-lives.
Scheme 1. Common Reactivity Patterns of Known Hydroxy-
and Cyclopropylcarbenes
’ INTRODUCTION
In the chemistry of highly reactive alkylcarbenes, cyclopro-
pylcarbenes are unique because the strained, three-membered
ring is an excellent electron donor.1,2 The electron deficiency of
the carbene center leads to ring expansion reactions to the corre-
sponding cyclobutenes, with the driving force arising from the
addition of a double bond and from providing all carbon atoms
with the desired electron octet. This ring expansion has been
observed for cyclopropylmethylene3ꢀ7 (3a) and its derivatives
cyclopropylmethylcarbene8,9 (3b) and cyclopropylchloro- and
cyclopropylfluorocarbene (3c,d; see Scheme 1),10ꢀ15 whose ring
expansion reactions yield the cyclobutenes 4aꢀd.11,16ꢀ18 The
driving force for this ring expansion facilitates the preparation of
extremely strained cyclobutene derivatives.19ꢀ21
The carbene family has recently been extended by the class
of hydroxycarbenes (1), for which a remarkable hydrogen tunnel-
ing mechanism in noble gas matrices was revealed, even at
temperatures as low as 11 K.22 The parent hydroxycarbene,
hydroxymethylene (1a), decays through [1,2]H-tunneling with
a measured half-life of τ = 2.0 h to give formaldehyde (2a). By
analogy, phenylhydroxycarbene (1b)23 and methylhydroxycar-
bene (1c)24 give benzaldehyde (2b) and acetaldehyde (1c),
respectively, with similar half-lives. However, in contrast to
1aꢀc, the reactivities of dihydroxycarbene (1d) and methoxyhy-
droxycarbene (1e) are strikingly different.25 Whereas all experi-
mentally known hydroxycarbenes without heteroatoms other
than that of the single hydroxy group undergo H-tunneling,
hydroxycarbenes with a second heteroatom are persistent under
cryogenic conditions (Scheme 1). This is most likely due to the
increased overall positive mesomeric effect on the carbene center
by two donor substituents. In the course of our ongoing
investigations concerning the nature of H-tunneling and the
factors that govern it, we sought to prepare a hydroxycarbene that
constitutes an intermediate between the tunneling and nontun-
neling species known so far, that is, one with a considerably larger
half-life than 1aꢀc. Thus, we considered attaching to the
hydroxycarbene center a π-donating group stronger than phenyl
but weaker than oxygen. The unique electronic properties
exhibited by the Walsh orbitals of the cyclopropyl moiety make
Received: May 29, 2011
Published: July 27, 2011
r
2011 American Chemical Society
13614
dx.doi.org/10.1021/ja204507j J. Am. Chem. Soc. 2011, 133, 13614–13621
|

Journal of the American Chemical Society
ARTICLE
it a most suitable candidate for our efforts to increase the stability
of 1 toward tunneling decay without entirely shutting down this
pathway. Therefore, we sought to prepare and matrix-isolate
cyclopropylhydroxycarbene (3e; Scheme 2). Regarding a possi-
ble ring expansion reaction of 3e like in carbenes 3aꢀd,22,23 we
were curious about the reactivity of 3e.
were carried out at 600, 750, and 960 °C. In contrast to our
previous work on other hydroxycarbenes, the matrix-isolated
pyrolysis mixture from 6 contained only small amounts of the
respective aldehyde, in this case cyclopropylcarboxaldehyde (7).
The major product was identified as methyl vinyl ketone (10),
which is most likely the last link in a chain of subsequent
rearrangements involving ring expansion of 3e to cyclobut-1-
en-1-ol (8), electrocyclic ring opening, and eventual ketoꢀenol
tautomerism of the intermediate 2-hydroxybuta-1,3-diene (9) to
yield 10 (Scheme 2). However, the pyrolysis mixture mainly
consisted of 6, 7, 10, and 3e in an estimated ratio of approxi-
mately 10:5:8:1 (based on relative IR band intensities). Carbene
3e was sufficiently present in the matrix and could unequivocally
be identified by means of IR spectroscopy (Figure 1). No traces
of cyclobutanone, the keto tautomer of 8, could be found.
All characteristic IR absorptions (νobs) could be matched
with the computed unscaled harmonic vibrational frequencies
(ωtheor) of the energetically most favorable conformation 3etout
with the OH group pointing away from the cyclopropyl group
and an s-trans-conformation of the HOCC moiety (Figure 1).
The vibrational frequencies (Table 1a) were evaluated at the
CCSD(T)/cc-pVTZ (frozen core, fc) (see Computational
Details) level of theory.
’ RESULTS AND DISCUSSION
Preparation and Characterization of Cyclopropylhydroxy-
carbene 3e. Following our generic route for the preparation
of hydroxycarbenes,22ꢀ25 we obtained 3e through thermal
decarboxylation of cyclopropylglyoxylic acid (6), which is readily
accessible through alkaline potassium permanganate oxidation of
cyclopropylmethylketone (5).26 Precursor 6 was evaporated
from a small storage vial into the pyrolysis zone of the matrix
apparatus (see Experimental Section for details); the pyrolyses
Scheme 2. Generation of 3e by Thermal Carbon Dioxide
Extrusion from Precursor 6, as Well as Thermal and Photo-
chemical Rearrangement Reactions
The UV absorption maximum of 3etout, computed with time-
dependent density functional theory (TD-DFT) at the B3LYP/
6-311++G(d,p) level, was determined as λmax = 373 nm (1A0 f
1A00, open-shell). Accordingly, irradiation of the matrix at 366 nm
resulted in complete disappearance of the respective signals. New
signals appeared, though, which were in accordance with the
computed spectrum of the primary ring expansion product 8; a
detailed discussion of 8, which we characterize here for the first
Figure 1. IR spectrum of pyrolysis mixture from 6 (top) and corresponding difference spectrum after 5 min of irradiation at 366 nm (middle);
disappearing signals point downward and are matched with the unscaled computed spectrum of 3etout (bottom) [CCSD(T)/cc-pVTZ]; new signals are
due to the formation of 8. (Inset) Structure of 3etout optimized at CCSD(T)/cc-pVTZ. This pyrolysis was carried out at 960 °C, which is the highest
possible temperature for our experimental setup. Lower temperatures resulted in a larger amount of remaining precursor 6 and thus a lower yield of 3e.
13615
dx.doi.org/10.1021/ja204507j |J. Am. Chem. Soc. 2011, 133, 13614–13621

Journal of the American Chemical Society
ARTICLE
Table 1. Unscaled Computeda Harmonic Vibrational
Frequencies ωtheor of 3etout and d-3etout
spectrum before and after irradiation of the matrix, depicted in
Figure 1, originate from the interconversion of different rotamers
of 10 and 6 (see next section on conformational analysis).
All matrix experiments with O-deuterated acid d-6 were
conducted accordingly, yielding the respective deuterated spe-
cies d-3e (Table 1b) and d-10 (through d-7, d-8, and d-9) as well
as d-8 (Table 2b) upon irradiation of d-3e at 366 nm. Bands of
deuterated but-1-en-1-one and deuterated hydroxyacetylene
were found at 2098 and 2198 cmꢀ1, respectively.
approx
νobs
sym (cmꢀ1
ωtheor
Iobs (cmꢀ1
Itheor
description
)
)
(km molꢀ1
)
3
(a) IR Bands of 3etout
OH str
a0 3567
a0 1363
a0 1317
a0 1217
a0 1112
1110
w
s
3796
1410
1360
1268
1132
122
45
CCH def + COH def
COH def + CCH def
CO str
s
70
Conformational Analysis of Cyclopropylhydroxycarbene
3e. Two precursor rotamers are capable of extruding carbon
dioxide, namely, 6tin (methyne H and carbonyl O are syn) and
6tout (methyne H and carbonyl O are anti), which lead to
carbenes 3etout and 3ecout via TS1 and TS3, respectively,
with a very similar barrier for decarboxylation (Scheme 3). The
m
s
130
114
COH def + CCH def
ring def
a0
917
807
739
m
m
s
952
834
772
125
24
ring def + OH oop def a00
ring def + OH oop def a00
two rotamers of 6 are separated by 1.4 kcal molꢀ1 in favor of
3
51
6tout and are connected by transition state TS2 that lies +7.4
kcal molꢀ1 above 6out [CCSD(T)/cc-pVDZ//M06-2X/6-311+
(b) IR Bands of d-3e
3
OD str
CCH def
CO str
a0 2637
m
s
2764
1396
1268
1202
944
71
92
+G(d,p)].
a0 1352
a0 1231
a0 1158
a0
While pyrolysis at 960 °C allows the equilibration of both
carbene conformers, 3etout is thermodynamically favored; con-
former 3ecout could not be detected in the matrix. Whereas the
rotation around the central carbonꢀcarbon bond is almost
barrierless for related structure 1c, it is more hindered in 3e
due to the overlap between the empty p-orbital on the carbene
center and the occupied Walsh-type orbital of the cyclopropyl
moiety (Figure 2).
s
123
34
ring def
w
ring def
134
34
COD def + CH2 oop def a0
a CCSD(T)/cc-pVTZ.
824
s
840
The degree of Walsh orbital interaction is evident in the
conformational analysis of the rotation around the exocyclic
carbonꢀcarbon bond (rotation of OH) in 3e, devised from a
relaxed scan of the potential energy surface at the M06-2X/6-311
++G(d,p) level (Scheme 4): By forcing the COH moiety into the
molecular plane, the central p-orbital becomes perpendicular to
the Walsh-type orbital, resulting in zero overlap and thus leading
to an energetically unfavorable arrangement. This conformation
Table 2. Unscaled Computeda Harmonic Vibrational
Frequencies ωtheor of 8 and d-8
νobs
(cmꢀ1
ωtheor
(cmꢀ1
Itheor
normal mode
sym
)
Iobs
)
(km molꢀ1
)
3
(a) IR Bands of 8
OH str
CdC str
COH def + CCH2 str a0
COH def + CH2 wag a0
COH def + CH2 wag a0
a0
a0
3618
1644
1388
1220
1114
737
s
3823
1693
1432
1246
1134
745
51
180
17
is responsible for the sizable rotational barrier of 10.1 kcal
3
molꢀ1. An even higher barrier was found earlier for parent 3a.6
During rotation of the hydroxy group in 3e, two minima are
s
w
m
w
s
passed, namely, 3etout and 3ecout, which are only 1.8 kcal molꢀ1
62
3
apart in favor of the 3etout conformer. A similar scan of 1b and 1c
122
34
CCH oop def
a00
reveals in both cases a lower barrier of CꢀC rotation compared
to 3e (0.9 and 6.9 kcal molꢀ1, respectively), which shows that
3
(b) IR Bands of d-8
the stabilizing effect of the interaction of the empty p-orbital with
a Walsh-type orbital of the cyclopropyl group is indeed much
larger than in 1b or 1c.
OD str
a0
a0
2670
1646
w
s
2782
1691
1370
1009
745
34
171
82
CdC str
COD def + CCH2 str a0
For structural and reactivity comparisons, both parent systems
of 3e have to be considered, namely, cyclopropylcarbenes and
hydroxycarbenes (Scheme 5, bottom). While the exocyclic
carbonꢀcarbon bond in 3etout is long compared to other
members of the cyclopropylcarbene family, it is short in compar-
ison to the other known hydroxycarbenes (1b,c), demonstrating
the strong interaction of the occupied Walsh-type orbital with the
carbene’s empty p-orbital. The empty p-orbital of the carbene
center is partly saturated by substituents with strong electron-
donor capabilities, enforcing the carbene’s singlet character
COD def + ring def
a0
a0
68
CCH oop def
a CCSD(T)/cc-pVTZ.
740, 737
s
35
time, is given below. In the initial spectrum of the pyrolysis
mixture, a strong signal was found in the ketene region at
2099 cmꢀ1, which vanished upon irradiation at 366 nm. We
assigned this band to but-1-en-1-one, which is likely to result
thermally from 3e in the pyrolysis tube and for which a TD-DFT
computation at B3LYP/6-311++G(d,p) reveals a UV transition
[ΔEST = 35.4 kcal molꢀ1 at CCSD(T)/cc-pVTZ//M06-2X/
3
6-311++G(d,p)]. Hence, the electron-donor strength of a sub-
stituent can be associated with the magnitude of the carbene’s
singletꢀtriplet gap ΔEST. This value can be used as a compara-
tive measure for a substituent’s electron-donor strength toward
the empty p-orbital of the carbene center. Scheme 5 correlates
the substituent’s electron-donor strength with the length of the
1
at 353 nm (1A00 f A0, open-shell). In addition, irradiation
entailed a new signal at 2199 cmꢀ1, which matches the strongest
band of matrix-isolated (Ar) hydroxyacetylene,27 a fragmentation
product of 3e (see Supporting Information for further dis-
cussion). Other changes in signal intensities in the difference
13616
dx.doi.org/10.1021/ja204507j |J. Am. Chem. Soc. 2011, 133, 13614–13621

Journal of the American Chemical Society
ARTICLE
Scheme 3. Two Conformers of Precursor 6 and Their Reactions Leading to 3ea
a After decarboxylation, a weakly bound complex of the carbene and CO2 (3etout CO2 or 3ecout CO2) can computationally be located but is not
3
3
persistent at the pyrolysis temperatures and immediately dissociates into the carbene and carbon dioxide. Regular type = CCSD(T)/cc-pVDZ//
M06-2X/6-311++G(d,p); in brackets = M06-2X/6-311++G(d,p).
Scheme 4. Energy Profiles for Rotation around Central
CarbonꢀCarbon Bonds in 1b, 1c, and 3ea
Figure 2. (Left) HOMOꢀ5 orbital arising from electron donation
from the oxygen p-orbital to the empty p-orbital of the carbene
center. (Right) Interaction of the Walsh orbital with the empty
p-orbital of the central carbon (middle lobe) in the HOMOꢀ1
natural bond orbital [M06-2X/6-311++G(d,p)]. The displayed in-
teractions stabilize the carbene center through n f p overlap,
enhancing the singlet character of the carbene.
distal carbonꢀcarbon bond (top) and the adjacent carbonꢀ
carbon bond (bottom): For members of the cyclopropylcarbene
family, stronger electron-donating substituents increase this
bond length and thus weaken both the adjacent as well as the
distal carbonꢀcarbon bond.28 In hydroxycarbenes, stronger
electron-donating substituents decrease the carbonꢀcarbon
bond length adjacent to the carbene center and thereby strength-
en the bond. Being a member of both families, 3e is located at the
intersection of both lines. Only 1c (in parentheses) slightly
deviates from the observed trends.
Identification and Characterization of Cyclobut-1-en-1-ol
8. By comparison with the computed frequencies, the main
photolysis product of 3e could be identified as the cyclobutenol
8, with the OH proton pointing toward the double bond
(Figure 3). The Cs symmetrical molecule features a rhomboidal
shape as the cyclobutene scaffold is strongly distorted by the OH
group: The endocyclic double bond is the shortest bond in the
ring (1.348 Å) facing the longest endocyclic single bond (1.573
Å), but the two other opposing single bonds also have signifi-
cantly different lengths (1.506 versus 1.572 Å). The length of the
corresponding double bond in vinyl alcohol is 1.337 Å compared
to 1.348 Å in 8. It should be noted that the cyclobutene moiety in
8 exhibits an even higher ring strain than the cyclopropyl moiety
in 3e.29 The amount of strain energy strongly favors the ring-
opening reaction in the hot pyrolysis zone so that intermediate 8
cannot be trapped directly in the pyrolysis mixture, but it can be
generated from 3e through irradiation. Yet 8 is stable toward
a M06-2X/6-311++G(d,p), not corrected for zero-point vibrational
energies.
irradiation with wavelengths of 290 nm e λ e 800 nm. In
accordance, the computed UV spectrum [B3LYP/6-311++G-
(d,p)] of 8 shows no absorptions above 300 nm. The observed
vibrational frequencies (νobs) of 8 are listed in Table 2 and are
matched with the computed harmonics (ωtheor).
Expansion and Rearrangement Reactions in the Pyrolysis
Zone. All energies given in this paragraph have been computed at
the CCSD(T)/cc-pVTZ//M06-2X/6-311++G(d,p) + ZPVE
level unless stated otherwise. As found for other cyclopropylcar-
benes, the ring expansion of 3e via TS4 is favored over all other
reactions during the passage of the pyrolysis zone, and the
interaction of electron-rich cyclopropyl orbitals with empty
p-orbital at the carbene center generally promotes the C-shift
over the H-shift in cyclopropylcarbenes.11,16 The expansion takes
place in the pyrolysis zone as the temperature is high enough to
overcome the barrier of approximately +21 kcal molꢀ1. The
3
energetically less favorable conformer 3ecout could in principle
also undergo ring expansion, yet with a huge barrier of almost
+52 kcal molꢀ1. The reaction sequence starting from ring
3
expansion product 8 to 10 proceeds exothermically with about
ꢀ22 kcal molꢀ1, and the energy arising from the transformation
3
of 3e to 8 (approximately ꢀ36 kcal molꢀ1) likely facilitates
3
the electrocyclic ring opening of 8 to 9 via TS5, which is mainly
13617
dx.doi.org/10.1021/ja204507j |J. Am. Chem. Soc. 2011, 133, 13614–13621

Journal of the American Chemical Society
ARTICLE
driven by the decrease in ring strain.29 The subsequent [1,3]H-
shift in 9 finally gives 10 via TS6 (Scheme 6). The yields of these
reactions cannot be estimated reliably, owing to the unknown
absorption coefficients of the products. No traces of cyclopro-
pylidenemethanol (12), the product of a thermal [1,2]H-shift via
TS7 from the R-carbon of the cyclopropyl moiety to the carbene
center, were found, due to the high reaction barrier of approxi-
Scheme 5. Correlation of (top) Distal CarbonꢀCarbon
Bond Lengths in Some Cyclopropylcarbenes and (bottom)
Adjacent CarbonꢀCarbon Bond Lengths in Hydroxy- and
mately +36 kcal molꢀ1. A CꢀH insertion in 3e via TS8 would
a
3
Cyclopropylcarbenes with ΔEST
lead to highly strained and thus energetically unfavorable bicyclo-
[1.1.0]butan-2-ol (13) and is furthermore hindered by a barrier
of approximately +35 kcal molꢀ1. In accordance with these
3
computational results, 13 was not observed experimentally.
[1,2]H-Tunneling in Cyclopropylhydroxycarbene 3e. The
structure of 3e sits in a deep energy minimum on the potential
energy hypersurface, and the surrounding enthalpic barriers are
too high to be overcome at cryogenic temperatures. In this
regard, 3e should be persistent in the matrix. However, even in
the dark, the signals of matrix-isolated 3e slowly vanish over the
course of several hours to yield cyclopropylcarboxaldehyde (7)
by transfer of the hydroxy H to the central carbon atom
(Scheme 7). The increase of 7 could unequivocally be detected
by means of IR spectroscopy (see Supporting Information). As
for other hydroxycarbenes, the temperature-independent decay
is likely to be due to a [1,2]H-tunneling mechanism and follows
first-order kinetics. The half-life of 3e was determined experi-
mentally to be τ = 17.8 h at both 11 and 20 K, which is the longest
half-life known to date for all hydroxycarbenes undergoing
[1,2]H-tunneling.22,23 The temperature independence of the
first-order decay constant for the generation of 7 from 3e and
the unreactivity of d-3e are in accordance with a quantum
mechanical tunneling mechanism. Employing the same theore-
tical WentzelꢀKramersꢀBrillouin (WKB) approach30 as for
1aꢀc (see Computational Details), the half-life of 3e was
computed to be τ = 16.6 h, in excellent agreement with
experiment. In contrast, the O-deuterated species (d-3e) persis-
tent under the same conditions as the tunneling mechanism is
suppressed by the higher mass of deuterium, resulting in a
computed half-life of approximately 105 years.
a ΔEST = singletꢀtriplet gap of carbenes. All structures were computed
at the M06-2X/6-311++G(d,p) level.
The tunneling mechanism favors the reaction of 3e toward
more stable product 7, which is, however, kinetically disfavored.
The H-tunneling in 3e effectively reverses the expected reactivity
(i.e., no reaction at cryogenic temperatures) under the conditions
applied. In order to induce ring expansion of matrix-isolated 3e to
yield 8, irradiation is necessary (see above). Since even at
cryogenic temperatures the kinetically disfavored pathway is
populated, the classical reactivity of the system is inverted by
Figure 3. Structures of vinyl alcohol (left) and cyclobut-1-en-1-ol 8
(right), optimized at the CCSD(T)/cc-pVTZ level of theory.
Scheme 6. Potential Energy Surface of 3e Featuring H- and C-Migrationsa
a Products of the reactions in gray are not observed experimentally. Computations are based on the most stable conformers of all species. Boldface
type = CCSD(T)/cc-pVTZ//M06-2X/6-311++G(d,p) + ZPVE; lightface type = M06-2X/6-311++G(d,p) + ZPVE.
13618
dx.doi.org/10.1021/ja204507j |J. Am. Chem. Soc. 2011, 133, 13614–13621

Journal of the American Chemical Society
ARTICLE
Scheme 7. Details of the Potential Energy Hypersurface of 3e
Featuring both Kinetic and Tunneling Reaction Pathsa
Scheme 9. Correlation of Hydroxycarbene Half-Life τ and
Electron-Donating Capability of the Substituenta with an
Isodesmic Relationship
a Boldface type = CCSD(T)/cc-pVTZ//M06-2X/6-311++G(d,p) + ZPVE;
lightfacce type = M06-2X/6-311++G(d,p) + ZPVE.
a Substituent = H, Me, Ph, cyc-C3H5, OH, OMe, or NH2.
is accompanied by polarization of the σ-bond due to a noticeable
difference in electronegativity, the interaction of the hydroxy,
methoxy, and amino groups with the empty p-orbital over-
compensates bond polarization and renders 1dꢀf stable toward
H-tunneling. Parent 1a is the only carbene that has no carbon
attached to its carbene center and is therefore expected to deviate
from the above-mentioned relationship.
Scheme 8. Correlation of Hydroxycarbene Half-Life τ and
Relative Electron-Donating Capability of Substituenta
Isodesmic Evaluation of Hydroxycarbene Relative Stabi-
lities with Respect to Tunneling. In order to further investigate
the relationship between half-life and substituent electron-donat-
ing capability, we devised an isodesmic equation that describes
these relative stabilities (Scheme 9):
a Substituent = H, Me, Ph, cyc-C3H5, HO, MeO, or H2N.
the tunneling mechanism, as observed for the related 1b.24 That
is, tunneling control directs the reaction path away expected from
classic kinetic control.
The reaction enthalpies ΔrH0 at 0 K of these hypothetical
reactions are computed at the CCSD(T)/cc-pVDZ//M06-2X/
6-311++G(d,p) and CCSD(T)/cc-pVDZ//B3LYP/6-311+
+G(d,p) levels (see Supporting Information for the latter) for
the most stable conformers of all species involved; ΔrH0°K corre-
sponds to the gain in energy by exchanging carbene 1c for 1b,
1dꢀf, and 3e. The relative isodesmic stabilization energies
become increasingly negative in the sequence of increasing half-
life (Scheme 9). Although evaluation of the isodesmic equation
only assesses the carbenes’ relative thermodynamic stabilities,
the latter depend on the electron-donor strength of the substit-
uents. Thus, the carbenes’ half-lives can indirectly be corre-
lated with electron-donating capability of the substituents:
More enthalpically stable hydroxycarbenes, which are indeed
those with stronger electron-donating substituents, are those that
have longer half-lives under cryogenic conditions.
Correlation of Substituent’s Electron-Donor Capabilities
with Hydroxycarbene Half-Lives. The half-life of τ = 17.8 h
suggests that the electronic effect of the cyclopropyl moiety on
the carbene center is just enough to keep the H-tunneling decay
at an observable rate, whereas heteroatomic π-donors, that is,
ꢀOH or ꢀOCH3, completely suppress [1,2]H-tunneling in the
hydroxycarbenes 1d and 1e. The large singletꢀtriplet energy gap
ΔEST also reflects the stabilizing effect exhibited by the cyclo-
propyl moiety (Schemes 5 and 7): The first triplet state of 3e, for
3
which the energetically most favorable conformation is 3etin,
with the OH group pointing toward the cyclopropyl group and
an s-trans-conformation of the HOCC moiety, lies about +35
kcal molꢀ1 higher in energy than the singlet state. The increase
3
in half-life with an additional electron-donating substituent
attached to the carbene’s central carbon atom is also evident
for the other members of the hydroxycarbene family, namely
1aꢀc and aminohydroxycarbene 1f. By qualitatively correlating a
hydroxycarbene’s half-life and the relative electron donor
strength of its substituent, the currently known hydroxycarbenes
can be arranged in a sequence of ascending stability (Scheme 8).
In 1c, the methyl group stabilizes the carbene through
hyperconjugation. In 1b and 3e, the substituents’ connecting
carbons are (effectively) sp2-hybridized and exhibit a slight ꢀI
effect, which is overcompensated by the +M effect of the phenyl
and cyclopropyl moieties, respectively. This influence is stronger
in 3e, resulting in an increased half-life. For stronger π-donors
like oxygen and even more so nitrogen, [1,2]H-tunneling cannot
be observed experimentally. Although the heteroatom +M effect
’ CONCLUSION AND OUTLOOK
The ambivalent reactivity of hitherto unknown cyclopropyl-
hydroxycarbene 3e renders it both a typical cyclopropylcarbene
and a distinctive hydroxycarbene that undergoes H-tunneling.
The ring expansion of 3e is similar to the reactions of carbenes
3aꢀd and yields hitherto unknown cyclobutenol 8, which under-
goes further thermal reactions. In solid Ar at cryogenic tempera-
tures, the characteristic [1,2]H-migration in 3e occurs by quan-
tum mechanical tunneling, as observed for 1aꢀc. The lifetime of
3e under inert matrix conditions is increased by the interaction of
the carbene’s empty p-orbital and the Walsh-type orbitals of the
cyclopropyl donor, demonstrating the significance of π-effects
on the H-tunneling phenomenon at work in hydroxycarbenes.
13619
dx.doi.org/10.1021/ja204507j |J. Am. Chem. Soc. 2011, 133, 13614–13621

Journal of the American Chemical Society
ARTICLE
Fine-tuning of the substituent’s electronic nature will be the
key step to fully understand tunneling control at work in
hydroxycarbenes and related systems.
the quartz pyrolysis tube. At a distance of approximately 50 mm, all
pyrolysis products were co-condensed with a large excess of argon
(typically 60ꢀ120 mbar from a 2000 mL storage bulb) on the surface of
the matrix window at 11 K (20 K). Several experiments with pyrolysis
temperatures ranging from 600 to 960 °C were performed in order to
determine the optimal pyrolysis conditions. A high-pressure mercury
lamp (HBO 200, Osram) with a monochromator (Bausch & Lomb) was
used for irradiation.
’ COMPUTATIONAL DETAILS
The potential energy surface (PES) surrounding 3e was established
by employing density functional theory (DFT) at the M06-2X/6-311+
+G(d,p)31,32 and B3LYP/6-311++G(d,p)33ꢀ35 levels. The computed
structures were then augmented with high-level coupled-cluster single
and double excitation (with triple excitations) [CCSD(T)]36ꢀ39 en-
ergies (using the frozen-core approximation throughout), utilizing
the ORCA40,41 program package with the Dunning-type correlation-
consistent basis sets cc-pVDZ and cc-pVTZ.42 Geometry optimizations
and frequency computations at the coupled-cluster level were carried
out with the CFOUR program package, which features analytic first
and second derivatives.43 All computed harmonic vibrational frequen-
cies presented in this work are unscaled. In general, the cc-pVTZ basis
was used whenever computationally feasible. Otherwise, the less de-
manding cc-pVDZ basis set was employed. For the WKB evaluation of
the half-life τ, the intrinsic reaction path44 (IRP, starting from TS9) and
the zero-point vibrational energy corrections (ZPVE) of the projected
frequencies along the path were computed at the M06-2X/6-311++G(d,p)
level using the Hessian-based predictor corrector algorithm45 as
implemented in Gaussian09.46 Among many tested DFT functionals,
the M06-2X functional was found to give half-lives closest to the
experimental ones in theoretical tunneling examinations of the other
known hydroxycarbenes.22,23 A triple-ζ basis set is generally necessary to
ensure an accurate IRP generation, at which, averaged over the half-lives
of all currently known hydroxycarbenes, the 6-311++G(d,p) Pople basis
has been found to be slightly superior to the cc-pVTZ Dunning basis.
The tunneling probabilities were evaluated by computing one-dimen-
sional barrier action integrals along the IRP and invoking the WKB
relation. The attempt energy of the tunneling particle was set equal to
the zero-point energy of the frequency corresponding to the reaction
coordinate ξ. Algebraic equations were solved with the Mathematica
program package.47
Preparation of Cyclopropylglyoxylic Acid (6).26 To a me-
chanically stirred solution of 5.05 g (60 mmol) of cyclopropyl methyl
ketone (5) in 30 mL of aqueous NaOH at 0 °C in a 250 mL round-
bottom flask was added dropwise, over 3 h, a solution of 18.96 g (120
mmol) of KMnO4 in 480 mL of H2O. The reaction mixture was stirred
overnight and allowed to warm up to 25 °C. After MnO2 was filtered off,
the volume of the clear, colorless filtrate was reduced in vacuo to
approximately 100 mL. Then 7.05 g of aqueous HCl (37%) was added,
and the reaction mixture was extracted three times with approximately
20 mL of CH2Cl2 each time. The combined organic layers were dried
over Na2SO4 and the solvent was removed. Due to partial decarboxyla-
tion and reoxidation of the generated aldehyde, the crude mixture
contained small amounts of cyclopropylcarboxylic acid and was hence
fractionally distilled in vacuo. The O-deuterated precursor (d-6) was
generated by repeated dissolution of 6 in D2O, followed by evaporation
of the solvent. Matrix experiments with d-6 were conducted accordingly.
Colorless liquid, 1.8 g (16 mmol, 27% yield). 1H NMR (400 MHz,
CDCl3) δ = 1.23ꢀ1.31 (4H, m), 2.84ꢀ2.90 (1H, m), 10.27 (1H, s); 13
C
NMR (100 MHz, CDCl3) δ = 15.51 (CH2), 17.08 (CH), 160.77
(COCOOH), 195.35 (COCOOH); UV/vis (EtOH) λmax (ε) = 330 nm
(29 molꢀ1 dm3 cmꢀ1).
3
3
’ ASSOCIATED CONTENT
S
Supporting Information.
Twenty-four figures, nine
b
schemes, and nine tables showing full-matrix IR spectra of
pyrolyses of 6 and d-6 and reference spectra of 7 and 10, both
before and after irradiation at 366 nm; discussion of fragmenta-
tion products hydroxyacetylene and but-1-en-one; alternative
version of Figure 1 featuring B3LYP/6-311++G(d,p) frequen-
cies with anharmonic corrections; B3LYP/6-311++G(d,p) struc-
tures and corresponding CCSD(T) energies; and full references
for electronic structure codes. This material is available free of
charge via the Internet at http://pubs.acs.org.
Despite the good results for 3e, though, the one-dimensional WKB
treatment of the [1,2]H-tunneling reaction has a serious caveat:
Whereas the DFT-computed half-life of 3e is in good agreement with the
experimental results, a tunneling computation including CCSD(T)/cc-
pVTZ energies on all points of the DFT-generated IRP greatly under-
estimates the lifetime of 3e, mainly due to a significantly reduced
reaction barrier at the coupled cluster level. Yet CCSD(T) results of
other hydroxycarbenes are in very good agreement with their experi-
mental half-lives even on DFT-generated paths. We are currently
examining this apparent dichotomy, which is most probably due to
the relatively small cc-pVTZ basis set used for the evaluation of single
point energies with respect to this particular computational problem.
’ AUTHOR INFORMATION
Corresponding Author
prs@org.chemie.uni-giessen.de
’ ACKNOWLEDGMENT
’ EXPERIMENTAL SECTION
We gratefully acknowledge funding by the Deutsche For-
schungsgemeinschaft (Grant Schr 597/13-1). D.L. thanks the
Fonds der Chemischen Industrie for a scholarship, and D.G.
thanks the Justus-Liebig University for a Graduiertenstipendium.
We thank Wesley D. Allen and Attila G. Csꢀaszꢀar for helpful
discussions.
Matrix Apparatus Design. For the matrix isolation studies, we
used an APD Cryogenics HC-2 cryostat with a closed-cycle refrigerator
system, equipped with an inner CsI window for IR measurements.
Spectra were recorded with a Bruker IFS 55 FT-IR spectrometer with a
spectral range of 4500ꢀ300 cmꢀ1 and a resolution of 0.7 cmꢀ1. For the
combination of high-vacuum flash pyrolysis with matrix isolation, we
employed a small, home-built, water-cooled oven, which was directly
connected to the vacuum shroud of the cryostat. The pyrolysis zone
consisted of an empty quartz tube with an inner diameter of 8 mm, which
was resistively heated over a length of 50 mm by a coaxial wire. The
temperature was monitored with a NiCrꢀNi thermocouple. Cyclopro-
pylglyoxylic acid (6) was evaporated at ꢀ20 °C from a storage bulb into
’ REFERENCES
(1) de Meijere, A. Angew. Chem., Int. Ed. 1979, 18, 809–826.
(2) de Meijere, A.; Kozhushkov, S. I. Chem. Rev. 2000, 100, 93–142.
(3) Hoffmann, R.; Zeiss, G. D.; Van Dine, G. W. J. Am. Chem. Soc.
1968, 90, 1485–1499.
(4) Schoeller, W. W. J. Org. Chem. 1980, 45, 2161–2165.
13620
dx.doi.org/10.1021/ja204507j |J. Am. Chem. Soc. 2011, 133, 13614–13621

Journal of the American Chemical Society
ARTICLE
(5) Wang, B.; Deng, C. Tetrahedron 1988, 44, 7355–7362.
(6) Shevlin, P. B.; McKee, M. L. J. Am. Chem. Soc. 1989, 111,
519–524.
(44) Fukui, K. J. Phys. Chem. 1970, 74, 4161–4163.
(45) Hratchian, H. P.; Schlegel, H. B. J. Chem. Phys. 2004, 120,
9918–9924.
(7) Chou, J. H.; McKee, M. L.; De Felippis, J.; Squillacote, M.;
Shevlin, P. B. J. Org. Chem. 1990, 55, 3291–3295.
(8) Modarelli, D. A.; Platz, M. S.; Sheridan, R. S.; Ammann, J. R.
J. Am. Chem. Soc. 1993, 115, 10440–10441.
(46) Gaussian09, Revision B.02; Gaussian, Inc., Wallingford CT,
2009 (see Supporting Information for full reference).
(47) Mathematica, Version 7.0.1;WolframResearchInc., ChampaignIL,
2008.
(9) Thamattoor, D. M.; Jones, M.; Pan, W.; Shevlin, P. B. Tetra-
hedron Lett. 1996, 37, 8333–8336.
(10) Moss, R. A.; Fantina, M. E. J. Am. Chem. Soc. 1978, 100, 6788–
6790.
(11) Liu, M. T. H.; Bonneau, R. J. Phys. Chem. 1989, 93, 7298–7300.
(12) Ho, G. J.; Krogh-Jespersen, K.; Moss, R. A.; Shen, S.; Sheridan,
R. S.; Subramanian, R. J. Am. Chem. Soc. 1989, 111, 6875–6877.
(13) Bonneau, R.; Liu, M. T. H.; Rayez, M. T. J. Am. Chem. Soc. 1989,
111, 5973–5974.
(14) Moss, R. A.; Ho, G. J.; Shen, S.; Krogh-Jespersen, K. J. Am.
Chem. Soc. 1990, 112, 1638–1640.
(15) Moss, R. A.; Liu, W.; Krogh-Jespersen, K. J. Phys. Chem. 1993,
97, 13413–13418.
(16) Moss, R. A.; Ho, G. J.; Liu, W. J. Am. Chem. Soc. 1992, 114,
959–963.
(17) Moss, R. A.; Jang, E. G.; Fan, H.; Wlostowski, M.; Krogh-Jespersen,
K. J. Phys. Org. Chem. 1992, 5, 104–107.
(18) Ammann, J. R.; Subramanian, R.; Sheridan, R. S. J. Am. Chem.
Soc. 1992, 114, 7592–7594.
(19) Wiberg, K. B.; Burgmaier, G. J.; Warner, P. J. Am. Chem. Soc.
1971, 93, 246–247.
(20) de Meijere, A.; Wenck, H.; Kopf, J. Tetrahedron 1988, 44,
2427–2438.
(21) de Meijere, A.; von Seebach, M.; Z€ollner, S.; Kozhushkov, S. I.;
Belov, V. N.; Boese, R.; Haumann, T.; Benet-Buchholz, J.; Yufit, D. S.;
Howard, J. A. K. Chem.—Eur. J. 2001, 7, 4021–4034.
(22) Schreiner, P. R.; Reisenauer, H. P.; Pickard, F. C., IV; Simmonett,
A. C.; Allen, W. D.; Mꢀatyus, E.; Csꢀaszꢀar, A. G. Nature 2008, 453,
906–909.
(23) Gerbig, D.; Reisenauer, H. P.; Wu, C.-H.; Ley, D.; Allen, W. D.;
Schreiner, P. R. J. Am. Chem. Soc. 2010, 132, 7273–7275.
(24) Schreiner, P. R.; Reisenauer, H. P.; Ley, D.; Gerbig, D.; Wu, C.-H.;
Allen, W. D. Science 2011, 332, 1300–1303.
(25) Schreiner, P. R.; Reisenauer, H. P. Angew. Chem., Int. Ed. 2008,
47, 7071–7074.
(26) Larionov, O. V.; de Meijere, A. Adv. Synth. Catal. 2006, 348,
1071–1078.
(27) Hochstrasser, R.; Wirz, J. Angew. Chem. 1989, 101, 183–185.
(28) G€unther, H. Tetrahedron Lett. 1970, 11, 5173–5176.
(29) Wiberg, K. B. Angew. Chem., Int. Ed. 1986, 25, 312–322.
(30) Razavy, M. Quantum Theory of Tunneling; World Scientific:
Singapore, 2003.
(31) Zhao, Y.; Truhlar, D. G. Acc. Chem. Res. 2008, 41, 157–167.
(32) Zhao, Y.; Truhlar, D. G. Theo. Chem. Acc. 2008, 120, 215–241.
(33) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785–789.
(34) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652.
(35) Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J.
J. Phys. Chem. 1994, 98, 11623–11627.
ꢁ
(36) Cíꢁzek, J. J. Chem. Phys. 1966, 45, 4256–4266.
(37) Raghavachari, K.; Trucks, G. W.; Pople, J. A.; Head-Gordon, M.
Chem. Phys. Lett. 1989, 157, 479–483.
(38) Bartlett, R. J.; Watts, J. D.; Kucharski, S. A.; Noga, J. Chem. Phys.
Lett. 1990, 165, 513–522.
(39) Stanton, J. F. Chem. Phys. Lett. 1997, 281, 130–134.
(40) Wennmohs, F.; Neese, F. Chem. Phys. 2008, 343, 217–230.
(41) Neese, F.; Hansen, A.; Wennmohs, F.; Grimme, S. Acc. Chem.
Res. 2009, 42, 641–648.
(42) Dunning, J. T. H. J. Chem. Phys. 1989, 90, 1007–1023.
(43) CFOUR, Coupled-Cluster Techniques for Computational
Chemistry; quantum chemical program package, http://www.cfour.de
(see Supporting Information for full reference).
13621
dx.doi.org/10.1021/ja204507j |J. Am. Chem. Soc. 2011, 133, 13614–13621