Scope and mechanistic analysis of the enantioselective synthesis of allenes by rhodium-catalyzed tandem ylide formation/[2,3]-sigmatropic rearrangement between donor/acceptor carbenoids and propargylic alcohols

Article abstract of DOI:10.1021/ja3061529

Rhodium-catalyzed reactions of tertiary propargylic alcohols with methyl aryl- and styryldiazoacetates result in tandem reactions, consisting of oxonium ylide formation followed by [2,3]-sigmatropic rearrangement. This process competes favorably with the standard O-H insertion reaction of carbenoids. The resulting allenes are produced with high enantioselectivity (88-98% ee) when the reaction is catalyzed by the dirhodium tetraprolinate complex, Rh ₂(S-DOSP)₄. Kinetic resolution is possible when racemic tertiary propargylic alcohols are used as substrates. Under the kinetic resolution conditions, the allenes are formed with good diastereoselectivity and enantioselectivity (up to 6.1:1 dr, 88-93% ee), and the unreacted alcohols are enantioenriched to 65-95% ee. Computational studies reveal that the high asymmetric induction is obtained via an organized transition state involving a two-point attachment: ylide formation between the alcohol oxygen and the carbenoid and hydrogen bonding of the alcohol to a carboxylate ligand. The 2,3-sigmatropic rearrangement proceeds through initial cleavage of the O-H bond to generate an intermediate with close-lying open-shell singlet, triplet, and closed-shell singlet electronic states. This intermediate would have significant diradical character, which is consistent with the observation that the 2,3-sigmatropic rearrangement is favored with donor/acceptor carbenoids and more highly functionalized propargylic alcohols.

Full text of DOI:10.1021/ja3061529

Article
pubs.acs.org/JACS
Scope and Mechanistic Analysis of the Enantioselective Synthesis of
Allenes by Rhodium-Catalyzed Tandem Ylide Formation/[2,3]-
Sigmatropic Rearrangement between Donor/Acceptor Carbenoids
and Propargylic Alcohols
Zhanjie Li,† Vyacheslav Boyarskikh,†,‡ Jørn H. Hansen,† Jochen Autschbach,§ Djamaladdin G. Musaev,‡
and Huw M. L. Davies*,†
†Department of Chemistry, Emory University, 1515 Dickey Drive, Atlanta, Georgia 30322, United States
‡Cherry L. Emerson Center for Scientific Computation, Emory University, 1521 Dickey Drive, Atlanta, Georgia 30322, United States
§Department of Chemistry, University at Buffalo, The State University of New York, Buffalo, New York 14260, United States
S
* Supporting Information
ABSTRACT: Rhodium-catalyzed reactions of tertiary propargylic
alcohols with methyl aryl- and styryldiazoacetates result in tandem
reactions, consisting of oxonium ylide formation followed by [2,3]-
sigmatropic rearrangement. This process competes favorably with the
standard O−H insertion reaction of carbenoids. The resulting allenes
are produced with high enantioselectivity (88−98% ee) when the
reaction is catalyzed by the dirhodium tetraprolinate complex, Rh2(S-DOSP)4. Kinetic resolution is possible when racemic
tertiary propargylic alcohols are used as substrates. Under the kinetic resolution conditions, the allenes are formed with good
diastereoselectivity and enantioselectivity (up to 6.1:1 dr, 88−93% ee), and the unreacted alcohols are enantioenriched to 65−
95% ee. Computational studies reveal that the high asymmetric induction is obtained via an organized transition state involving a
two-point attachment: ylide formation between the alcohol oxygen and the carbenoid and hydrogen bonding of the alcohol to a
carboxylate ligand. The 2,3-sigmatropic rearrangement proceeds through initial cleavage of the O−H bond to generate an
intermediate with close-lying open-shell singlet, triplet, and closed-shell singlet electronic states. This intermediate would have
significant diradical character, which is consistent with the observation that the 2,3-sigmatropic rearrangement is favored with
donor/acceptor carbenoids and more highly functionalized propargylic alcohols.
Scheme 1. Proposed Tandem Oxonium Ylide Formation/
[2,3]-Sigmatropic Rearrangement
INTRODUCTION
■
Metal−carbenoid insertions into X−H bonds (X = C, O, N,
etc.) have been extensively studied over the last few decades.1
In particular, metal−carbenoid insertions into O−H bonds
have received considerable attention as an effective method for
the synthesis of α-alkoxy and α-hydroxy carbonyl compounds,
which are important motifs in natural products and
pharmaceutical targets.2 These O−H insertion reactions are
believed to proceed, mechanistically, via formation of an
oxonium ylide followed by a proton transfer. Although some
chiral copper catalysts result in O−H insertion with high levels
of asymmetric induction, in general no asymmetric induction is
observed in rhodium-catalyzed reactions.3,4
report the results of this study, which includes both
experimental results to define the scope of the transformations
and computational studies to explain why this reaction is
favored over O−H insertions for reactions with donor/acceptor
carbenoids and highly functionalized propargylic alcohols.
Recently, we discovered that the rhodium(II)-catalyzed
reactions of donor/acceptor carbenoids with highly substituted
allyl alcohols do not lead to O−H insertion products.5,6
Instead, a tandem oxonium ylide formation/[2,3]-sigmatropic
rearrangement process occurs. The discovery of this new
unexpected reaction pathway between carbenoids and alcohols
has prompted us to explore the reactions of donor/acceptor
carbenoids with propargylic alcohols. A similar tandem
oxonium ylide formation/[2,3]-sigmatropic rearrangement
would generate α-hydroxy allenes (Scheme 1).7 Herein, we
RESULTS AND DISCUSSION
■
Wood and co-workers have previously explored the reactions of
carbenoids with propargylic alcohols using α-diazoketones as
the carbenoid precursors.8 The reaction of diazoketone 1 with
propargyl alcohol generates an alkoxy enol intermediate 2,
which undergoes either a [3,3]-sigmatropic rearrangement to
Received: June 24, 2012
Published: August 27, 2012
© 2012 American Chemical Society
15497
dx.doi.org/10.1021/ja3061529 | J. Am. Chem. Soc. 2012, 134, 15497−15504

Journal of the American Chemical Society
Article
give α-hydroxy allene 3 when the electron-rich catalyst
Rh2(cap)4 is used or a [2,3]-sigmatropic rearrangement to
give the isomeric α-hydroxy allene 4 when the electron-
deficient catalyst Rh2(tfa)4 is used (Scheme 2). However, as the
seen when an even more highly substituted propargylic alcohol,
2-methyl-3-hexyn-2-ol (6c), was used as the substrate (entry 3).
The [2,3]-sigmatropic rearrangement product 8c was cleanly
formed in 61% isolated yield with 79% ee. An increase in
asymmetric induction occurred on lowering the reaction
temperature to 0 °C, and, under these conditions, compound
8c was cleanly formed in 85% isolated yield with 85% ee (entry
4).
Scheme 2. Reaction of Diazoketones with Propargylic
Alcohols
The competition between allene formation and O−H
insertion is highly dependent on the nature of the carbenoids.
The reaction of 6c with ethyl diazoacetate (5b) gave only the
O−H insertion product 7d (entry 5). The reaction of 6c with
methyl diazomalonate (5c) gave a 7:1 mixture of [2,3]-
sigmatropic rearrangement product 8e and the O−H insertion
product 7e, with 8e isolated in 59% yield (entry 6). The most
impressive result was obtained from the reaction of 6c with
styryldiazoacetate 5d. This reaction cleanly formed the [2,3]-
sigmatropic rearrangement product 8f, which was isolated in
74% yield with 96% ee (entry 7). These results indicate that
donor/acceptor carbenoids are the best-suited carbenoids for
the tandem oxonium ylide formation/[2,3]-sigmatropic rear-
rangement of propargylic alcohols under these conditions.
The reactions of 6c can be extended to a series of
styryldiazoacetates 9 (Table 2). Styryldiazoacetates with
electron-withdrawing groups such as Br−, CF3, and Cl− on
the aryl group were tolerated in the reaction, and [2,3]-
sigmatropic rearrangement products 10a−e were formed with
good yield and very high levels of enantioselectivity (entries 1−
4, 77−85% yield, 85−97% ee). However, the reaction with
styryldiazoacetate 9e, containing electron-donating groups on
the aryl ring, gave low yield of 10e (34%, entry 5), although the
enantioselectivity was still high (92% ee). In all of these cases,
the O−H insertion product was not observed (ratio of [2,3]-
sigmatropic rearrangement/O−H insertion: >20:1).
Wood protocol generates 4 via the intermediacy of the planar
enol 2, this mechanism would be expected to preclude the
transfer of any asymmetric induction generated during ylide
formation into the final product 4. Indeed, when we replicated
this reaction using our standard chiral catalyst, Rh2(S-DOSP)4,
we obtained the α-hydroxy allene 4 in modest yield with no
asymmetric induction.
We hypothesized that enol formation would be suppressed if
a diazoester was used instead of a diazoketone as the carbenoid
precursor. Therefore, we examined the Rh2(S-DOSP)4-
catalyzed reactions of diazoacetates 5a−d with a series of
functionalized propargylic alcohols 6 (Table 1). The reaction of
phenyldiazoacetate 5a with propargyl alcohol (6a) gave none of
the allene product (entry 1). Instead, the O−H insertion
product 7a was isolated in 50% yield, and as is typical of
rhodium-catalyzed O−H insertions, 7a was formed with no
asymmetric induction. We knew from our studies with allyl
alcohols that the [2,3] sigmatropic rearrangement of the
oxonium ylide is favored when the allyl group is more highly
substituted.5 The reaction of 5a with the tertiary propargylic
alcohol 6b did result in the formation of the [2,3]-sigmatropic
rearrangement product 8b in 42% yield and 27% ee (entry 2).
The racemic O−H insertion byproduct 7b was still formed,
albeit in diminished yield (12%). Further improvement was
The reaction was applied to a range of propargylic alcohols
11 as illustrated in Table 3. The styryldiazoacetate 5d was used
as the reference system. The desired [2,3]-sigmatropic
rearrangement products were obtained with uniformly excellent
levels of enantioselectivity. In all cases, the O−H insertion
product was not observed. Alkyl groups (linear or cyclic, entries
1−6), TBS protected alcohols (entries 7−8), and substituents
containing phenyl groups (entries 9−12) were all compatible
with this reaction. This suggests that the allene formation is a
highly favorable process as many of these substrates have
Table 1. Optimization of the Asymmetric Tandem Ylide Formation/[2,3]-Sigmatropic Rearrangement
entry
diazo compound
propargyl alcohols
temp, °C
product (s)
yield of 7, %
ee of 7, %
yield of 8, %
ee of 8, %
1
2
3
4
5
6
7
5a
5a
5a
5a
5b
5c
5d
6a
6b
6c
3c
3c
3c
3c
23
23
23
0
7a
50
12
0
0
<5
42
61
85
<5
59
74
7b, 8b
8c
27
79
85
<5
8c
<5
0
7d
27
0
0
7e, 8e
8f
∼10
<5
0
96
15498
dx.doi.org/10.1021/ja3061529 | J. Am. Chem. Soc. 2012, 134, 15497−15504

Journal of the American Chemical Society
Article
Table 2. Reaction of Styryldiazoacetate 9a−e with 2-Methyl-
Table 3. Reaction of Styryldiazoacetate 5d with Alcohols
11a−n
a
a
3-hexyn-2-ol (6c)
a
Standard reaction conditions: 9 (1.0 mmol, 2.0 equiv) in pentane (9
mL) with the minimum amount of toluene required for solubilization
was added to a solution of 6c (0.5 mmol, 1.0 equiv) and Rh2(S-
DOSP)4 (0.005 mmol, 1 mol %) in pentane (1 mL) over 1.5 h at 0 °C.
b
1H NMR of the crude reaction mixtures revealed that the [2,3]-
sigmatropic rearrangement/O−H insertion ratio in each case was
c
d
>20:1. Isolated yield of 10a−e. Determined by chiral HPLC.
potentially active sites for C−H functionalization or cyclo-
propanation. Alcohols with very bulky R groups, such as t-butyl
and trimethylsilyl, however, gave only moderate yields of
product (entries 13,14). The absolute configuration of 12e was
determined by X-ray crystallography (see the Supporting
Information), and the absolute configuration of the other
products was assigned by analogy.
The reaction could also be extended to propargylic alcohols,
13a−c, containing cyclic subunits as shown in Scheme 3. In
each case, the allenes 14a−c were cleanly formed in 69−85%
yield with 88−95% ee.
a
b
The reaction conditions described in Table 2 were used. 1H NMR of
the crude reaction mixtures revealed that the [2,3]-sigmatropic
rearrangement/O−H insertion ratio in each case was >20:1. Isolated
yield of 12a−n. Determined by chiral HPLC.
c
d
Scheme 3. Formation of Cyclic Allenes 14
The reactions described so far generate allenes with a single
stereogenic center. To challenge the reaction further, substrates
were examined in which two new stereogenic centers would be
generated (Table 4). It was envisioned that the size difference
between methyl and the second alkyl group, c-hexyl, i-Pr, or t-
Bu, could lead to kinetic resolution of chiral racemic propargylic
alcohols. Several examples of kinetic resolution are known in
carbenoid chemistry,9 but none of these examples involve the
reaction of carbenoids with alcohols. The reaction of racemic
alcohol 15a with styryldiazoacetate 5d gave moderate
diastereocontrol, and the allenic alcohols 16a and 17a were
formed in 6.1:1 dr and in a combined yield of 49% (entry 1).10
The major diastereomer 16a was formed in 88% ee. Under
these reaction conditions, 15a was recovered in 35% yield and
was found to be enriched to 95% ee, confirming that we were
indeed observing a kinetic resolution of the starting material.
Similar results were obtained for the i-Pr and t-Bu derivatives,
15b and 15c (entries 2 and 3).
as the catalyst (see the Supporting Information for details). The
results of the reactions of styryldiazoacetate 5d with
enantiomeric enriched 15a are summarized in Table 5. We
observed distinct matched/mismatched reactions. For example,
the reaction of (S)-15a with Rh2(S-DOSP)4 or (R)-15a with
Rh2(R-DOSP)4 resulted in the formation of 16a in high yield
with extremely high diastereo- and enantioselectivity (>20:1 dr,
>99% ee, Table 5, entries 1 and 3). In the mismatch situation,
(S)-15a with Rh2(R-DOSP)4 or (R)-15a with Rh2(S-DOSP)4,
much inferior results were obtained (Table 5, entries 2 and 4).
The diastereoselectivity was low (∼2:1 dr), favoring 17a.
Although compound 17a was formed in 96−97% ee, the
To further understand the stereochemical outcome with
chiral alcohols, we carried out the reaction with enantiomeri-
cally enriched alcohols, (R) and (S)-15a, which were obtained
by conducting the reaction of racemic 15a with 5d on a larger
scale (5 mmol) with either Rh2(S-DOSP)4 or Rh2(R-DOSP)4
15499
dx.doi.org/10.1021/ja3061529 | J. Am. Chem. Soc. 2012, 134, 15497−15504

Journal of the American Chemical Society
Article
a
Table 4. Reaction of Styryldiazoacetate 5d with Racemic Alcohols 15
b
c
d
c
d
entry alcohol
R
16:17 ratio
yield of 16 + 17, %
ee of 16, % (configuration)
yield of recovered 15, %
ee of 15, % (configuration)
1
2
3
15a
15b
15c
c-hex
i-Pr
6.1:1
3.3:1
6.7:1
49
47
50
88 (2R, 4S)
93 (2R, 4S)
35
36
95 (R)
81 (R)
t-Bu
93 (2R, 4S)
42
77(R)
a
b
c
1
Reactions were performed using the standard conditions described in Table 2. Determined by H NMR of the crude reaction mixture. Isolated
d
yield. Determined by chiral HPLC.
Table 5. Reaction of Styryldiazoacetate 5d with (R)- and (S)-15a
b
e
entry
alcohol
Rh(II) catalyst
16a:17a
yield, %
ee of 16a, %
ee of 17a, %
c
1
2
3
4
(S)-15a (95% ee)
(S)-15a (95% ee)
(R)-15a (96% ee)
(R)-15a (96% ee)
Rh2(S-DOSP)4
Rh2(R-DOSP)4
Rh2(R-DOSP)4
Rh2(S-DOSP)4
>20:1
70
>99
51
d
f
1:1.7
>20:1
1:2.1
42
−97
c
f
79
>−99
d
f
39
−59
96
a
b
1
Reactions were performed with Rh2(S- or R-DOSP)4 under the standard reactions conditions described in Table 2. Determined by H NMR of
c
d
e
f
the crude reaction mixture. Isolated yield of 16a. Combined yield of (16a + 17a). Determined by chiral HPLC. “Negative” value signifies the
opposite enantiomeric series.
formation of the minor diastereomer was less enantioselective
(51−59% ee). Compound 16a from the reaction of (S)-15a
and Rh2(S-DOSP)4 (entry 1, >99% ee) was recrystallized from
cold hexanes, and the resulting crystals were analyzed by X-ray
crystallography. The configuration of the tertiary alcohol in 16a
was determined by X-ray crystallography and agrees with the
determined absolute configuration of 12e. The configuration of
the allene component of 16 is (S) (see the Supporting
Information). The configurations of 16b and 16c were assigned
by analogy to 16a. A similar kinetic resolution was conducted
with the racemic alcohol 18 (Scheme 4). The recovered alcohol
The stereoselective conversion of chiral allenic alcohol into
2,5-dihydrofuran has been extensively studied,12,13 and also
successfully applied to the total synthesis of complex natural
products such as amphidinolide X and Y,14 and boivinnianin
B.15 The highly substituted allenic alcohol 16a can also be easily
transformed into various 2,5-dihydrofuran derivatives with
excellent chirality transfer (Scheme 5). Treatment of 16a with
Scheme 5. Cyclization of Allenes to Dihydrofurans
Scheme 4. Kinetic Resolution of 18
AgNO3 and CaCO3 provided dihydrofuran 20, while treatment
with NBS provided the bromodihydrofuran 21 with two
quaternary stereogenic centers at the 2,5-positions of the
dihydrofuran. Both products were formed in good yield and
>99% ee.
MECHANISTIC DFT STUDIES
■
The allene formation is believed to occur via ylide formation,
followed by a [2,3]-sigmatropic rearrangement of the rhodium-
associated ylide. Normally, racemic O−H insertion occurs
readily in the presence of alcohols with most rhodium
carbenoids, and, indeed, this is observed with simple
propargylic alcohols in this study. Computational studies on
the O−H insertion mechanism of rhodium carbenoids with
water have been reported,4,16 and the reaction preferentially
goes through a pathway involving a free enol, thereby losing
from the Rh2(S-DOSP)4-catalyzed reaction was assigned as
(R)-18 from comparison of its optical rotation with literature
values (recovered (R)-18: 96% ee, [α]20D: +1.94° (c 6.03,
Et2O); lit. +1.44° (c 6.03, Et2O)).11 On the assumption that the
same sense of kinetic resolution would be observed with the
propargylic alcohols 15 and 18, the recovered alcohols 15 from
the Rh2(S-DOSP)4-catalyzed reactions were assigned as (R)
configurations.
15500
dx.doi.org/10.1021/ja3061529 | J. Am. Chem. Soc. 2012, 134, 15497−15504

Journal of the American Chemical Society
Article
any asymmetric induction that may be generated during ylide
formation.4 Considering the high levels of asymmetric
induction observed in the tandem oxonium ylide formation/
[2,3]-sigmatropic rearrangement, a more detailed mechanistic
investigation of the [2,3]-sigmatropic rearrangement was
warranted. These studies could shed light on the following
vital questions: (1) What are the factors that govern
partitioning between O−H insertion and the [2,3]-sigmatropic
rearrangement? (2) Why is [2,3]-sigmatropic rearrangement so
strongly favored with donor/acceptor carbenoids and more
highly functionalized propargylic alcohols? (3) Why is ylide
formation so highly enantioselective? (4) What is the
mechanism of the chirality transfer during the [2,3]-sigmatropic
rearrangement?
For these purposes, we conducted detailed DFT calculations
on ylide formation followed by either the O−H insertion or the
[2,3]-sigmatropic rearrangement pathways.17 The study was
conducted on an unsubstituted vinylcarbenoid reacting with
both the primary propargylic alcohol 6a, which favors O−H
insertion, and the highly substituted propargylic alcohol 21a,
which favors the [2,3]-sigmatropic rearrangement.
vinylcarbenoid VC-I, formed through nitrogen extrusion
reaction between Rh2(O2CH)4 and methyl 2-diazobutenoate,
is illustrated in Figure 1. The reaction starts with the formation
of a prereaction complex PC-I, where propargylic alcohol 21a
and vinylcarbenoid VC-I are bound via two weak interactions:
the (O2···H1) hydrogen bonding with d(O2···H1) = 2.316 Å
and (O1···C1) bonding with d(O1···C1) = 2.884 Å. The weak
(O1···C1) bonding in PC-I did not impact the hybridization
state of the reactive sp2-hybridized carbenoid carbon (impr.
angle = 5°). From PC-I, the reaction proceeds via the transition
state TS-I, where the (O2···H1) and (O1···C1) bonding became
stronger (with d(O2···H1) = 1.917 Å) and d(O1···C1) = 1.966
Å), and the pyramidalization of C1 is more enhanced (impr.
angle = 22°). This two-point interaction, that is, the (O2···H1)
and (O1···C1) bonding, nascent in prereaction complex PC-I
and developed via transition state TS-I, defines the orientation
of the allylic alcohol as it approaches the carbenoid and is likely
to be a crucial factor that impacts the high asymmetric
induction in this chemistry. A similar type of two-point
interaction, involving hydrogen bonding to the carboxylate
ligands, has been observed in computational studies on
rhodium-catalyzed cyclopropenation of internal alkynes.19
The calculated relative energy of TS-I is ΔH/ΔG(ΔGsol) =
−5.8/10.2(12.0) kcal/mol. Intrinsic reaction coordinate (IRC)
calculations confirm that TS-I connects PC-I with ylide YL-I
intermediate: the calculated energy of singlet ground state of
ylide intermediate YL-I is −12.8/2.7(3.6) kcal/mol relative to
reactants. The ylide YL-I has a reinforced hydrogen
(d(O2···H1) = 1.705 Å) and C−O bonding (d(O1···C1) =
1.483 Å). The carbenoid center of YL-I has strongly
pyramidalized (impr. angle = 43°). One should mention that
the ground electronic states of PC-I, TS-I, and YL-I are the
closed-shell singlet states. Their triplet states are significantly
higher in energy (see below).
The calculated17,18 energies presented below were referenced
to the reactants, that is, VC-I plus 21a (6a), and presented as
ΔH/ΔG(ΔGsol), where ΔH and ΔG are the gas-phase enthalpy
and Gibbs free energy, respectively. ΔGsol is calculated as ΔGs +
[ΔG − ΔE], where ΔGs is the PCM calculated free energy in
solution, and ΔE is the gas-phase total energy.
Reaction of VC-I with 21a. A summary of the calculated
reaction pathways between propargylic alcohol 21a and
Figure 1. Mechanism of the reaction of vinylcarbene VC-I with alcohol 21a.
15501
dx.doi.org/10.1021/ja3061529 | J. Am. Chem. Soc. 2012, 134, 15497−15504

Journal of the American Chemical Society
Article
After formation of the YL-I intermediate, the reaction may
proceed via either of two distinct pathways: (1) the [2,3]-
sigmatropic rearrangement or (2) proton transfer to the
carbonyl group (O1−H1 insertion). The diazo compound and
the propargylic alcohol are the control elements for which
pathway will occur with donor/acceptor carbenoids and highly
substituted propargylic alcohols strongly favoring the [2,3]-
sigmatropic rearrangement. The [2,3]-sigmatropic rearrange-
ment, as shown in the literature, may proceed via a concerted
pathway or in a stepwise fashion, involving either homolytic or
heterolytic bond cleavage/recombination mechanisms.20 Cal-
culations show that singlet YL-I undergoes heterolytic C2−O1
bond cleavage with almost no energy barrier at the singlet
transition state TS-II-A: the energy of the closed-shell singlet
TS-II-A is calculated to be −13.2/2.6(3.2) kcal/mol relative to
singlet reactants. IRC calculations confirm that singlet TS-II-A
connects singlet YL-I with the closed-shell singlet intermediate
I-I, which is −13.9/−7.4(−2.8) kcal/mol lower in energy than
reactants.
system, the transition state TS-II-A has no significant radical
character.
The performed Mulliken spin density analyses are consistent
with the above-presented findings. Indeed, in the triplet state of
YL-I, two α-spins are located, mostly on the Rh1 [1.15 |e|], Rh2
[0.49 |e|], and C6 [0.20 |e|] centers. The spin density
distribution in the triplet state transition state TS-II-A is very
similar to that for the prereaction complex YL-I: Rh1 [1.17 |e|],
Rh2 [0.43 |e|], and C6 [0.22 |e|]. However, in case of
intermediate I-I, the largest portion (∼1.15 e) of the two α-
spins is located on substrate 21a: in triplet I-I spin density is
distributed as Rh2 [0.29 |e|], C1[0.29 |e|], C2[0.65 |e|], C4[0.51 |
e|], and C6[0.47 |e|]. Thus, the electronic structure of transition
state TS-II-A is very much “reactant-like”. However, the
progression of the reaction via YL-I → TS-II-A → I-I is
accompanied by significant spin buildup at the reactive (at the
next stage of the reaction) carbon centers (C1, C2, and C4) that
facilitates the allene formation (C4−C1 formation) via a radical
coupling mechanism. It is expected that having alkyl
substituents on the propargylic alcohol and electron-donating
groups on the carbenoid fragment will further promote spin
buildup on the reactive carbon centers and stabilize transition
state TS-II-A associated with the [2,3]-sigmatropic rearrange-
ment by making its electronic structure more “product-like”.
This conclusion is in excellent agreement with our experimental
findings presented in Table 1. Neither the reaction of ethyl
diazoacetate with a trialkylated propargylic alcohol (entry 5)
nor the reaction of methyl phenyldiazoacetate with the
unsubstituted propargylic alcohol (entry 1) give any of the
[2,3]-sigmatropic rearrangement products because only one
radical center in the diradical intermediate would be stabilized.
In contrast, when a donor/acceptor carbenoid is reacted with a
highly substituted propargylic alcohol, as in entries 3 and 7,
exclusive formation of the 2,3-sigmatropic rearrangement
products is observed. In these cases, both of the radical centers
are stabilized. These conclusions also help rationalize the
similar product distributions trends seen in the reaction of allyl
alcohols with donor/acceptor carbenoids.5a
One should note that we were not able to calculate the exact
location of the closed-shell singlet and triplet states of the allene
formation TS-II-B on the potential energy surface of the
reaction. Instead, we estimated the upper limit of energy
barriers associated with these transition states by scanning PESs
of the reactions: they are ΔEtot = −14.7 and −6.7 kcal/mol,
respectively. Both states of TS-II-B are lower in total energy
(ΔEtot) than TS-II-A by −0.1 and −6.0 kcal/mol, respectively,
but higher than I-I by 1.7 and 9.4 kcal/mol, respectively. IRC
calculations indicate that the transition state TS-II-B connects
I-I with the allene product P-I as a complex with
Rh2(OCOH)4.
Thus, the above presented findings indicate that after
formation of YL-I, which requires a 3.3/10.2(12.0) kcal/mol
barrier, the overall [2,3]-sigmatropic rearrangement occurs with
no significant energy barrier. The mechanism predicts that Re-
face attack on the carbenoid would lead to the R-configuration
of the quaternary stereocenter in P-I, and this prediction is
consistent with experimental observations for reactions with
Rh2(S-DOSP)4.1b
The homolytic C2−O1 bond cleavage of YL-I proceeds via a
triplet TS-II-A transition state, which is calculated to be 1.5/
14.1(14.1) kcal/mol higher in energy than singlet state
reactants, that is, VC-I + 21a. These values are over 10 kcal/
mol higher than those for the singlet state TS-II-A. IRC
calculations confirm that triplet TS-II-A connects the triplet
YL-I (which is calculated to be 13.6/9.0(8.9) kcal/mol higher
than its singlet ground state) with the triplet intermediate I-I.
The energy barrier at triplet TS-II-A, calculated from the triplet
YL-I, is found to be only 0.7/2.4(1.6) kcal/mol. Unlike the
prereaction complex YL-I and transition state TS-II-A,
intermediate I-I has close-lying singlet and triplet states:
closed-shell singlet, open-shell singlet, and triplet states with
energies of −13.9/−7.4(−2.8), −15.8/−3.3(−2.3), and
−14.7/-4.5(−2.7) kcal/mol, respectively.
The data discussed in the preceding paragraph suggest that
the reaction involves the closed-shell singlet states of YL-I and
TS-II-A, because of calculated large singlet−triplet energy gaps
in these systems. However, the electronic structure of the
intermediate I-I is less obvious, because its closed-shell singlet,
open-shell singlet, and triplet states are close in energy. Thus,
one should apply DFT-based methods to I-I with caution. In
this case, the multideterminant approaches, such as CASSCF
and MRD-CI, are required to identify the degree of radical
character in the total wave function of I-I. However, such
methods have a significantly higher computational cost than
DFT, and their use is beyond the scope of the present work.
Thus, we conclude that the reaction YL-I → TS-II-A → I-I
starts from the closed-shell singlet prereaction complex YL-I,
proceeds via closed-shell singlet (“reactant-like”) transition
state TS-II-A, and leads to intermediate I-I with several lower-
lying electronic states. In other words, this reaction proceeds via
heterolytic C2−O1 bond cleavage mechanism, but leads to
product with a significant diradical (i.e., homolytic C2−O1 bond
cleavage) character. To estimate the region of the potential
energy surface (PES) where the singlet and triplet states may
cross, we did scan it for all three states (closed-shell singlet,
open-shell singlet, and triplet states) starting from the
intermediate I-I by fixing of C2−O1 (reaction coordinate
leading back to the starting material) bond distances but
optimizing all other parameters. These calculations clearly
indicate that closed-shell and open-shell states cross in the
vicinity of intermediate I-I. Thus, in this simplified model
The alternative pathway from YL-I is proton transfer (to the
carbonyl group) that leads to enol P-II and proceeds via
transition state, TS-III. Later, enol P-II could tautomerize to
the formal O−H insertion product. All reactants, transition
states, and products of this reaction have a well-defined closed-
15502
dx.doi.org/10.1021/ja3061529 | J. Am. Chem. Soc. 2012, 134, 15497−15504

Journal of the American Chemical Society
Article
Figure 2. Comparison of the important steps of the O−H insertion and [2,3]-sigmatropic rearrangement pathways for alcohols 6a and 21a,
respectively. Schemes are scaled to ΔGsolv. All energetics presented in this figure are for the singlet states of the intermediates, transition states, and
products.
shell singlet state; that is, they have no radical character. One
should mention that IRC calculations confirm that TS-III
connects P-II with YL-I. Formation of the enol explains why
O−H insertion products are racemic in Rh(II)-catalyzed
processes. The involvement of enol intermediates in the
rhodium-catalyzed O−H insertion is in agreement with
previous theoretical calculations on O−H insertion.4
The formation of allene P-I versus O−H insertion is
controlled by the relative energetics of TS-II-A and TS-III,
respectively. As seen in Figure 2 (left), the calculated
insignificant energy barrier for [2,3]-sigmatropic rearrangement
for the ylide derived from 21a makes it favorable over the O−H
insertion, which requires a +4.9/4.6(5.0) kcal/mol energy
barrier.
used as substrates. A predictive model was developed on the
basis of a detailed computational study that explains the factors
that control the partitioning between the [2,3]-sigmatropic
rearrangement and O−H insertion. The model also rationalizes
the diastereoselectivity observed in these reactions. The most
distinctive aspects of the model include a two-point binding
during ylide formation and the diradical character of the [2,3]-
sigmatropic rearrangement.
ASSOCIATED CONTENT
* Supporting Information
■
S
Synthetic details, computational details, and spectral data. This
material is available free of charge via the Internet at http://
pubs.acs.org.
We also calculated reaction of VC-I with the nonsubstituted
propargylic alcohol 6a (Figure 2, right). The overall reaction
pathways for 6a are similar to 21a (see the Supporting
Information). The reaction of alcohol 6a and VC-I forms PC-I′
that rearranges to ylide YL-I′ via the transition state TS-I′. The
required energy barrier for this process is 4.1/8.2(8.5) kcal/
mol. The intermediate YL-I′ undergoes either O1−H1 insertion
via transition state TS-III′ to give enol product P-II′, or [2,3]-
sigmatropic pathway via the transition state TS-II-A′ to give
intermediate I-I′. The associated barriers for O1−H1 insertion
and [2,3]-sigmatropic rearrangement are calculated to be 2.7/
4.5(4.6) and 2.8/5.6(5.3) kcal/mol, respectively. Comparison
of these barriers for 6a with those reported above for 21a shows
that the lack of two methyl groups did not introduce a
significant change in the free energy barrier of the O1−H1
insertion, whereas it significantly increased the energy barrier
for the C2−O1 bond cleavage step of [2,3]-sigmatropic
rearrangement. Intermediate I-I′ possessed an ambiguous
electronic structure similar to that of I-I; however, this
intermediate became less important because the barriers at
the transition state TS-II-A′ leading to this intermediate cannot
compete with that at TS-III′ leading to enol product P-II′.
AUTHOR INFORMATION
Corresponding Author
hmdavie@emory.edu
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This material is based on work supported by the National
Institutes of Health (GM099142). We thank Dr. Ken
Hardcastle for the X-ray crystallographic structural determi-
nation. We gratefully acknowledge NSF-MRI-R2 grant (CHE-
0958205) and the use of the Cherry Emerson Center for
Scientific Computation. We also acknowledge the University at
Buffalo’s Center for Computational Research for technical
support.
REFERENCES
■
(1) (a) Doyle, M. P.; McKervey, M. A.; Ye, T. Modern Catalytic
Methods for Organic Synthesis with Diazo Compounds: From Cyclo-
propanes to Ylide; Wiley: New York, 1998. (b) Davies, H. M. L.;
Beckwith, R. E. J. Chem. Rev. 2003, 103, 2861−2904. (c) Moody, C. J.
Angew. Chem., Int. Ed. 2007, 46, 9148−9150. (d) Zhang, Z.; Wang, J.
Tetrahedron 2008, 64, 6577−6605. (e) Davies, H. M. L.; Manning, J.
R. Nature 2008, 451, 417−424. (f) Doyle, M. P.; Duffy, R.; Ratnikov,
M.; Zhou, L. Chem. Rev. 2010, 110, 704−724.
CONCLUSIONS
■
In summary, we have discovered that the reaction of carbenoids
with propargylic alcohols can lead to a tandem oxonium ylide
formation/[2,3]-sigmatropic rearrangement, instead of O−H
insertion. The reaction is favored when donor/acceptor
carbenoids and highly functionalized propargylic alcohols are
(2) (a) Miller, D. J.; Moody, C. J. Tetrahedron 1995, 51, 10811−
10843. (b) Zhu, S. F.; Zhou, Q. L. Acc. Chem. Res. 2012, ASAP (DOI:
10.1021/ar300051u).
15503
dx.doi.org/10.1021/ja3061529 | J. Am. Chem. Soc. 2012, 134, 15497−15504

Journal of the American Chemical Society
Article
(3) (a) Maier, T. C.; Fu, G. C. J. Am. Chem. Soc. 2006, 128, 4594−
4595. (b) Chen, C.; Zhu, S. F.; Liu, B.; Wang, L.-X.; Zhou, Q.-L. J. Am.
Chem. Soc. 2007, 129, 12616−12617. (c) Zhu, S.-F.; Chen, C.; Zhou,
Q.-L. Angew. Chem., Int. Ed. 2008, 47, 932−934. (d) Zhu, S.-F.; Song,
X.-G.; Li, Y.; Cai, Y.; Zhou, Q.-L. J. Am. Chem. Soc. 2010, 132, 16374−
16376.
(4) For a computational study regarding the difference between
copper(I) and dirhodium(II) complexes in catalytic asymmetric OH
insertion reactions, see: Liang, Y.; Zhou, H.; Yu, Z.-X. J. Am. Chem. Soc.
2009, 131, 17783−17785.
(14) (a) Lepage, O.; Kattnig, E.; Furstner, A. J. Am. Chem. Soc. 2004,
126, 15970−15971. (b) Furstner, A.; Kattnig, E.; Lepage, O. J. Am.
Chem. Soc. 2006, 128, 9194−9204.
(15) Miura, T.; Shimada, M.; de Mendoza, P.; Deutsch, C.; Krause,
N.; Murakami, M. J. Org. Chem. 2009, 74, 6050−6054.
(16) Liu, Z.; Liu, J. Cent. Eur. J. Chem. 2009, 8, 223−228.
(17) Calculations were performed using the Gaussian 09 software
package: Frisch, M. J.; et al. Gaussian 09, revision A0.2; Gaussian Inc.:
Wallingford, CT, 2009. See the Supporting Information for complete
reference.
(18) The calculations were conducted at the M06L/{(SDD
+4f)Rh+[6-311+G(d,p)]} level of theory. The optimized structures
of the reactants, intermediates, transition states, and products of this
reaction were calculated in the gas phase without any symmetry
constrains. Solvent effects (n-pentane with ε = 1.8371 was used as a
solvent) were incorporated by performing single-point energy
calculations on the gas-phase-optimized geometries using the PCM
polarizable conductor calculation model. For the M06L method, see:
Zhao, Y.; Truhlar, D. G. J. Chem. Phys. 2006, 125, 194101−18. For
SDD basis sets and associated effective core potential, see: (a) Kaupp,
M.; Schleyer, P. v. R.; Stoll, H.; Preuss, H. J. Chem. Phys. 1991, 94,
1360. (b) Bergner, A.; M., D.; Kuechle, W.; Stoll, H.; Preuss, H. Mol.
Phys. 1993, 80, 1431. (c) Dolg, M.; S., H.; Preuss, H.; Pitzer, R. M. J.
Phys. Chem. 1993, 97. See also: Hansen, J.; Autschbach, J.; Davies, H.
M. L. J. Org. Chem. 2009, 74, 6555−6563. For PCM method, see:
(d) Tomasi, J.; Persico, M. Chem. Rev. 1994, 94, 2027−2094.
(e) Cammi, R.; Tomasi, J. J. Comput. Chem. 1995, 16, 1449−1458.
(19) (a) Briones, J. F.; Hansen, J. H.; Hardcastle, K.; Autschbach, J.;
Davies, H. M. L. J. Am. Chem. Soc. 2010, 132, 17211−17215. (b) Goto,
T.; Takeda, K.; Shimada, N.; Nambu, H.; Anada, M.; Shiro, M.; Ando,
K.; Hashimoto, S. Angew. Chem., Int. Ed. 2011, 50, 6803−6808.
(20) Sigmatropic rearrangements are well known to display both
concerted and diradical character in computational studies. For a
discussion of this topic, see: Bachrach, S. M. Computational Organic
Chemistry; John Wiley & Sons, Inc.: Hoboken, NJ, 2007.
(5) (a) Li, Z.; Davies, H. M. L. J. Am. Chem. Soc. 2010, 132, 396−
401. (b) Parr, B. T.; Li, Z.; Davies, H. M. L. Chem. Sci. 2011, 2, 2378−
2382. (c) Li, Z.; Parr, B.; Davies, H. M. L. J. Am. Chem. Soc. 2012, 134,
10942−10946.
(6) For representative examples of donor/acceptor carbenoids and
their rich chemistry, see refs 1b, 1e, and: (a) Davies, H. M. L.; Denton,
J. R. Chem. Soc. Rev. 2009, 38, 3061−3071. (b) Morton, D.; Davies, H.
M. L. Chem. Soc. Rev. 2011, 40, 1857−1869.
(7) For reviews on the asymmetric synthesis of allenes, see:
(a) Miesch, M. Synthesis 2004, 746−752. (b) Ogasawara, M.
Tetrahedron: Asymmetry 2009, 20, 259−271. (c) Yu, S.; Ma, S.
Angew. Chem., Int. Ed. 2012, 51, 3074−3112.
(8) Moniz, G. A.; Wood, J. L. J. Am. Chem. Soc. 2001, 123, 5095−
5097.
(9) For examples of matched/mismatched reactions in the rhodium
carbenoid field, see: (a) Martin, S. F.; Spaller, M. R.; Liras, S.;
Hartmann, B. J. Am. Chem. Soc. 1994, 116, 4493−4494. (b) Doyle, M.
P.; Dyatkin, A. B.; Kalinin, A. V.; Ruppar, D. A.; Martin, S. F.; Spaller,
M. R.; Liras, S. J. Am. Chem. Soc. 1995, 117, 11021−11022. (c) Doyle,
M. P.; Kalinin, A. V.; Ene, D. G. J. Am. Chem. Soc. 1996, 118, 8837−
8846. (d) Davies, H. M. L.; Venkataramani, C.; Hansen, T.; Hopper,
D. W. J. Am. Chem. Soc. 2003, 125, 6462−6468. (e) Davies, H. M. L.;
Walji, A. M. Angew. Chem., Int. Ed. 2005, 44, 1733−1735. (f) Nadeau,
E.; Ventura, D. L.; Brekan, J. A.; Davies, H. M. L. J. Org. Chem. 2010,
75, 1927−1939. (g) Lian, Y.; Miller, L. C.; Born, S.; Sarpong, R.;
Davies, H. M. L. J. Am. Chem. Soc. 2010, 132, 12422−12425.
(10) The absolute configuration of the allene moiety in compound
17a−c was assigned on the assumption that the catalyst has a
controlling effect on the tertiary alcohol stereogenic center in these
compounds. For a relevant discussion, see ref 5a.
(11) Damm, L. G.; Hartshorn, M. P.; Vaughan, J. Aust. J. Chem. 1976,
29, 1017−1021.
(12) For representative examples of the enantioselective synthesis of
α-allenic alcohols containing a secondary alcohol stereogenic center,
see: (a) Corey, E. J.; Yu, C.-M.; Lee, D.-H. J. Am. Chem. Soc. 1990,
112, 878−879. (b) Brown, H. C.; Khire, U. R.; Narla, G. J. Org. Chem.
1995, 60, 8130−8131. (c) Marshall, J. A.; Adams, N. D. J. Org. Chem.
1997, 62, 8976−8977. (d) Yu, C.-M.; Yoon, S.-K.; Baek, K.; Lee, J.-Y.
Angew. Chem., Int. Ed. 1998, 37, 2392−2395. (e) Furstner, A.; Mendez,
M. Angew. Chem., Int. Ed. 2003, 42, 5355−5357. (f) Xu, D.; Li, Z.; Ma,
S. Tetrahedron: Asymmetry 2003, 14, 3657−3666. (g) Inoue, M.;
Nakada, M. Angew. Chem., Int. Ed. 2006, 45, 252−255. (h) Xia, G.;
Yamamoto, H. J. Am. Chem. Soc. 2007, 129, 496−497. (i) Miura, T.;
Shimada, M.; Ku, S.-Y.; Tamai, T.; Murakami, M. Angew. Chem., Int.
Ed. 2007, 46, 7101−7103. (j) Li, J.; Kong, W.; Fu, C.; Ma, S. J. Org.
Chem. 2009, 74, 5104−5106.
(13) (a) Marshall, J. A.; Pinney, K. G. J. Org. Chem. 1993, 58, 7180−
7184. (b) Marshall, J. A.; Bartley, G. S. J. Org. Chem. 1994, 59, 7169−
7171. (c) Krause, N.; Laux, M.; Hoffmann-Roder, A. Tetrahedron Lett.
2000, 41, 9613−9616. (d) Hoffmann-Roder, A.; Krause, N. Org. Lett.
2001, 3, 2537−2538. (e) Krause, N.; Hoffmann-Roder, A.; Canisius, J.
Synthesis 2002, 1759−1774. (f) Deutsch, C.; Gockel, B.; Hoffmann-
Roder, A.; Krause, N. Synlett 2007, 1790−1794. (g) Poonoth, M.;
Krause, N. Adv. Synth. Catal. 2009, 351, 117−122. (h) Li, J.; Kong, W.;
Yu, Y.; Fu, C.; Ma, S. J. Org. Chem. 2009, 74, 8733−8738. (i) Alcaide,
B.; Almendros, P.; Martinez del Campo, T.; Quiros, M. T. Chem.-Eur.
J. 2009, 15, 3344−3346.
15504
dx.doi.org/10.1021/ja3061529 | J. Am. Chem. Soc. 2012, 134, 15497−15504