Scope and mechanism in palladium-catalyzed isomerizations of highly substituted allylic, homoallylic, and alkenyl alcohols

Article abstract of DOI:10.1021/ja508736u

Herein we report the palladium-catalyzed isomerization of highly substituted allylic alcohols and alkenyl alcohols by means of a single catalytic system. The operationally simple reaction protocol is applicable to a broad range of substrates and displays a wide functional group tolerance, and the products are usually isolated in high chemical yield. Experimental and computational mechanistic investigations provide complementary and converging evidence for a chain-walking process consisting of repeated migratory insertion/β-H elimination sequences. Interestingly, the catalyst does not dissociate from the substrate in the isomerization of allylic alcohols, whereas it disengages during the isomerization of alkenyl alcohols when additional substituents are present on the alkyl chain.

Full text of DOI:10.1021/ja508736u

Article
pubs.acs.org/JACS
Scope and Mechanism in Palladium-Catalyzed Isomerizations of
Highly Substituted Allylic, Homoallylic, and Alkenyl Alcohols
Evgeny Larionov,† Luqing Lin,† Laure Guenee,‡ and Clement Mazet*,†
́
́
́
†Department of Organic Chemistry, University of Geneva, 30 quai Ernest Ansermet, 1211 Geneva, Switzerland
‡Laboratory of Crystallography, University of Geneva, 24 quai Ernest Ansermet, 1211 Geneva, Switzerland
S
* Supporting Information
ABSTRACT: Herein we report the palladium-catalyzed
isomerization of highly substituted allylic alcohols and alkenyl
alcohols by means of a single catalytic system. The operationally
simple reaction protocol is applicable to a broad range of
substrates and displays a wide functional group tolerance, and
the products are usually isolated in high chemical yield.
Experimental and computational mechanistic investigations
provide complementary and converging evidence for a chain-
walking process consisting of repeated migratory insertion/β-H
elimination sequences. Interestingly, the catalyst does not
dissociate from the substrate in the isomerization of allylic
alcohols, whereas it disengages during the isomerization of
alkenyl alcohols when additional substituents are present on the alkyl chain.
but unfortunately mask the numerous challenges that still
remain in the field of transition-metal-catalyzed isomerizations
of analogous allylic systems.
INTRODUCTION
■
Driven by the contemporary needs of developing sustainable
processes, the efficiency of a chemical reaction is no longer a
simple measure of its yield, but fundamental principles such as
atom economy, step economy, and redox economy have
become primary concerns.1,2 In order to readily access
elaborated structures, the notion of rapid increase in molecular
complexity has been paralleling these concepts, leading to the
design of novel catalytic and selective reactions.3 Consequently,
cascade, tandem, and isomerization reactions have attracted
renewed interest because they reduce oxidation/reduction
sequences, protection/deprotection events, and functional
group manipulations while streamlining access to more
complex target molecules.
Metal-catalyzed directed isomerizations of functionalized
olefins are particularly attractive because, upon migration of a
CC bond, the substrates undergo refunctionalization in an
overall redox-neutral operation with no chemical waste
generated.4,5 Among some of the most relevant examples, the
Rh-catalyzed enantioselective isomerization of allylic amines to
chiral enamines constitutes a notorious illustration of the
potential of a nearly perfect isomerization.6 Over 30 years ago,
joint efforts from academia and industry resulted in the
development of an extraordinarily active and enantioselective
rhodium catalyst for the quantitative isomerization of N,N-
dialkylgeranylamine derivatives with excellent regioselectivity
and perfect enantioselectivity. The use of extremely low
loadings of a recyclable catalyst under mild reaction conditions
enabled its implementation as the key step in the industrial
production of menthol, the world’s best-selling aroma
ingredient. These achievements are undoubtedly formidable
Even though some useful progress has been made, the
seemingly related isomerization of allylic alcohols has not
reached the same level of refinement (Figure 1A).5 Recent
advances include the discovery of Ru, Rh, and Ir catalysts for
the isomerization of primary and secondary allylic alcohols
under relatively mild reaction conditions. Concomitantly,
efforts have also been directed toward the development of
enantioselective variants of these isomerizations, mostly with
primary allylic alcohols. Notable examples from the Fu group
(Rh),7 our group (Ir),8 and more recently Okhuma and co-
workers (Ru)9 are worth mentioning. Examples of asymmetric
isomerization of secondary allylic alcohols are rare.10
Unfortunately, all of these systemswhether enantioselective
or notare usually very substrate-specific, and to date there is
no general catalyst for the isomerization of both primary and
secondary allylic alcohols. It is usually admitted that a direct
correlation exists between the degree of substitution of the C
C bond and the difficulty associated with the isomerization of
allylic alcohols.5 Substrates with more-substituted double bonds
are less reactive, as schematically shown in Figure 1B. This
trend is explained by the fact that it is difficult for transition-
metal complexes to bind and react with heavily substituted
olefins, and not surprisingly, isomerizations of allylic alcohols
with a tetrasubstituted double bond are very rare.8a,f Therefore,
it is widely appreciated that further developments will benefit
Received: August 25, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja508736u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 1. (A) Challenges in allylic and alkenyl alcohol isomerizations. (B) Reactivity as a function of olefin substitution pattern. It should be noted
that no catalyst is currently effective for a breadth of both primary and secondary allylic alcohols. (C) Bridging the gap between “biased” allylic
alcohols and remote homo-, bis-homo-, and alkenyl alcohols.
from the discovery of a more general catalyst that would
indifferently isomerize primary and secondary allylic alcohols
independently of the extent of substitution of the olefinic
moiety.
selectivity for the more hindered position enables the formation
of quaternary stereocenters with high enantioselectivity.
Despite these formidable achievements, the oxidative con-
ditions employed and the strong influence of both coupling
partners on the regioselectivity of the initial migratory insertion
are current limitations of the approach.
Interestingly, none of the Ru, Rh, or Ir catalysts mentioned
above is able to isomerize homoallylic alcohols, bis-homoallylic
alcohols, or more remote alkenyl alcohols. This clearly indicates
that the hydroxyl directing group plays a crucial role in
imparting catalytic activity, independently of the mechanistic
subtleties that distinguish these systems. In 2007, Grotjhan and
co-workers reported a bifunctional Ru catalyst for the long-
range migration of unhindered double bonds (up to 30
units).11 The substrate scope included several examples of
primary and secondary alkenyl alcohols with a terminal CC
bond that were converted into the corresponding carbonyl
derivatives. Of note, when geraniol was subjected to the
optimized reaction conditions, the homoallylic alcohol (E)-iso-
geraniol rather than citronellal was generated in 61% yield. To
our knowledge, catalysts that can perform the isomerization of
both allylic and alkenyl alcohols are scarce; they are limited to
unhindered terminal or disubstituted double bonds, and the
yields of the products are usually low or moderate.5−10,12
The propensity of Pd catalysts to migrate CC bonds over
several positions by the so-called zipper or chain-walking
process was first exploited in the context of olefin polymer-
ization.13 Recently, several groups have capitalized on this
ability in more synthetic-oriented settings. Relevant examples
include the catalytic cycloisomerization of various 1,n-dienes by
Kochi and co-workers.14 Their approach allows the formation
of five-membered rings by facile C−C bond formation upon
isomerization of a double bond and intramolecular interception
of a cyclohexene moiety. However, it is applicable only to
substrates with a terminal olefin, which facilitates the first
insertion of the putative palladium hydride intermediate.
Gooßen and co-workers have developed an isomerizing olefin
metathesis for the conversion of fatty acids (i.e., substrates with
an internal disubstituted double bond) into dicarboxylic acids
of adjustable chain lengths.15 The system relies on the
successful combination of a Pd isomerization catalyst and the
Grubbs ruthenium metathesis catalyst. In a remarkable series of
papers, Sigman and co-workers have described highly
enantioselective redox-relay Heck arylations of acyclic alkenyl
alcohols.16 Using either aryl diazonium salts (Heck−Matsuda)
or boronic acid derivatives (oxidative Heck), the methods
provide access to remotely functionalized arylated carbonyls.
When trisubstituted olefins are employed, the preferred site
As part of our program directed toward the discovery of well-
defined transition-metal hydrides for isomerization reactions,8,9
we recently disclosed a novel [Pd−H] catalyst for the selective
rearrangement of di- and trisubstituted epoxides into aldehydes
and ketones.17 Computational and experimental studies
pointed to an unusual addition/elimination mechanism
reminiscent of the typical sequence usually proposed for alkene
migrations.18 This finding and the ability of Pd catalysts to
sustain olefin migration along extended carbon chains (vide
supra) motivated us to identify a well-defined catalyst that
would isomerize both primary and secondary allylic alcohols as
well as alkenyl alcohols independently of the olefinic
substitution pattern. If successful, such a system would bridge
the gap between directed and nondirected isomerizations and
certainly complement existing synthetic alternatives (Figure
1C).
In this article, we describe our progress toward achieving this
goal, specifically a functional-group-tolerant catalytic system
that can isomerize a variety of substrates. Furthermore, our
study provides detailed mechanistic information obtained by a
combined experimental and theoretical approach.
RESULTS AND DISCUSSION
■
Reaction Optimization. The system initially reported for
the isomerization of epoxides employed the well-defined
dinuclear palladium complex [(dippp)3(Pd−H)2)](OTf)2
1
(dippp = 1,3-bis(diisopropylphosphino)propane).17 For our
initial investigations, we chose to explore the isomerization of
four distinct geometrically pure substrates: the trisubstituted
primary allylic alcohol (E)-2a, the disubstituted secondary
allylic alcohol (E)-2b, and the di- and trisubstituted homoallylic
alcohols (E)-4b and (E)-4c, respectively. The reactions were
conducted under conditions similar to those employed for the
isomerization of epoxides (5 mol % [Pd], 85 °C, 18 h), but
tetrahydrofuran was replaced by 1,2-dichloroethane (DCE), a
less-coordinating solvent. Productive isomerizations of sub-
strates (E)-2b, (E)-4b, and (E)-4c were observed, and the
desired carbonyl compounds 3b, 5b, and 5c were isolated in
62%, 66%, and 40% yield, respectively. The remainder was
essentially the unreacted starting material in all cases. In
B
dx.doi.org/10.1021/ja508736u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
contrast, isomerization of (E)-2a led to a mixture of aldehyde
3a, (Z)-2a, and the corresponding homoallylic alcohol with a
terminal olefin, 4a (Scheme 1). The disparate results obtained
Scheme 2. Syntheses of Complexes 7, 9, 11, and 13 Along
with the X-ray Structures of 11 and 13
Scheme 1. Isomerization of Representative Substrates by
Complex 1
a
1
Conversion determined by H NMR analysis of the crude reaction
b
mixture. Yield after chromatography.
phenyl)borate) and 50 mol % cyclohexene in DCE at 85 °C for
18 h (Table 1). The two complexes supported by bisphosphine
ligands resulted in complete conversion of (E)-2a into 3a
(entries 1 and 4), while the catalysts supported by the (P,N)
for the isomerization of (E)-2a and (E)-4c already suggest the
existence of (subtle) mechanistic differences for the isomer-
ization of highly substituted allylic and homoallylic alcohols.
To circumvent this selectivity issue and in order to identify a
more general catalyst, we set out to explore the influence of the
supporting ligand on the outcome of these reactions.
Unfortunately, attempts to synthesize and isolate [Pd−H]
analogues of 1 with other chelating ligands (such as 6, 8, or 10)
were not successful. Inspired by precedents in olefin polymer-
ization13 and by the work of Kochi and co-workers,14 we turned
our attention to the synthesis of neutral palladium complexes of
general formula [LnPd(Me)(Cl)] where Ln is a chelating
bidentate ligand. Admittedly, these species provide in situ
access to the corresponding cationic hydride [Ln(Pd−H)]+ by
halide abstraction in the presence of a sacrificial olefin additive.
The three bidentate ligands dippp (6), 2,2′-bipyridine (bipy, 8),
and 2-(2-(diphenylphosphino)phenyl)-4,5-dihydrooxazole
(phox, 10) were selected because of their distinct donor
abilities. The ligand 1,2-bis(dicyclohexylphosphino)ethane
(dcpe, 12) was chosen because of its smaller bite angle with
respect to 6, a parameter that we anticipated might be
important for tuning the hydricity of the palladium
intermediate.19 The corresponding complexes 7, 9, 11, and
13 were prepared in good yields (67−86%; Scheme 2). Of
note, complex 11 was obtained as a single isomer, as was
Table 1. Survey of Catalysts
a
b
c
entry
substrate
catalyst
cons. (%)
conv. (%)
1
2
(E)-2a
(E)-2a
(E)-2a
(E)-2a
(E)-4c
(E)-4c
(E)-4c
(E)-4c
(E)-2b
(E)-2b
(E)-4b
(E)-4b
7
9
>99
>99
>99
>99
65
96 (44)
d
<5
d
3
11
13
7
<5
4
>99 (68)
e
5
16
f
6
9
58
<5
g
7
11
13
7
51
<5
h
8
93
24
9
80
65 (30)
97 (92)
71 (50)
83 (70)
10
11
12
13
7
99
93
13
92
a
b
1
apparent from H, 13C{1H}, and 31P{1H} NMR spectroscopy.
Reactions were performed on a 0.25 mmol scale. Consumption of
1
starting material as determined by H NMR spectroscopy using an
internal standard. Conversion to product as determined by H NMR
spectroscopy using an internal standard. Values in parentheses are
yields after chromatography. Complex reaction mixture and catalyst
decomposition. Along with a 9% yield of (Z)-4c and unidentified
The methyl group was proposed to be located trans to the N-
donor atom on the basis of the characteristic coupling constant
3JPH = 3.3 Hz. This was later confirmed by X-ray analysis of a
single crystal of 11. A crystal structure was also obtained for the
air-stable complex 13.
The performances of these four palladium complexes were
first evaluated in the isomerizations of (E)-2a and (E)-4c using
5.5 mol % NaBArF (BArF = tetrakis(3,5-bis(trifluoromethyl)-
c
1
d
e
f
oligomeric products. Along with an 8% yield of (Z)-4c, an 8% yield of
the terminal bis-homoallylic alcohol, and unidentified oligomeric
g
h
materials. Along with a 15% yield of (Z)-4c. Along with unidentified
oligomeric materials.
C
dx.doi.org/10.1021/ja508736u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
and (N,N) ligands led to intractable mixtures that did not
contain any traces of the product (entries 2 and 3). For
homoallylic alcohol (E)-4c, the catalytic activities of 7 and 13
were substantially reduced, and 5c was formed in 16% and 24%
yield, respectively (entries 5 and 8). These reactions were
accompanied by the formation of (Z)-4c and unidentified
oligomeric materials.20 No productive isomerization was
detected with 9 and 11 (entries 6 and 7). In evaluating 7
and 13 further, the latter turned out to offer the best
performances, as isomerization of secondary alcohol (E)-2b
and homoallylic alcohol (E)-4b afforded the corresponding
carbonyl derivatives 3b and 5b in 92% and 70% yield,
respectively (entries 10 and 12). These results were not
improved by additional optimization of the reaction conditions
(see the Supporting Information (SI)).
Scheme 3. Isomerization of Allylic Alcohols with 13
From these studies, it appears that complexes 1 and 7, which
are expected to generate a common palladium hydride
intermediate, give contrasting results. The former performs
best for homoallylic alcohols but poorly for allylic alcohols,
while the latter follows the opposite trend. Complexes 9 and 11
are poor catalysts even for the more reactive substrates. This is
especially surprising for 9, as most Pd-catalyzed long-range
olefin migrations reported in the literature rely on complexes
supported by (N,N) ligands.13,14,16 Finally, the straightforward
availability of 13 along with the well-balanced reactivity pattern
obtained for substrates (E)-2a/2b and (E)-4b/4c prompted us
to carry out further investigations with this air-stable complex.
Reaction Scope. A collection of primary and secondary
allylic alcohols with a representative set of olefin substitution
patterns and functional groups were synthesized and submitted
to the reaction conditions (Scheme 3). Primary and secondary
cinnamyl alcohols were efficiently converted to the correspond-
ing aldehydes and ketones in moderate to excellent yields (3b−
j, 62−92% yield). Allylic alcohols with a 2,3,3- or a 2,2,3-
trisubstituted olefinic pattern were isomerized with similar
efficiencies (3k−p, 51−94% yield; 3q and 3r, 90−92% yield).
While steric hindrance did not affect the reaction (R1 = o-
ClC6H4, Cy; R2 = i-Pr, Cy; R4 = i-Pr), sensitive functional
groups and heterocycles such as fluoride, chloride, bromide,
methoxy, nitro, hydroxyl, and N-methylpyrrole were all very
well tolerated. The latter example is interesting because hydride
insertion occurs at a bisbenzylic position. Isomerization of a
primary allylic alcohol with a tetrasubstituted olefin delivered
aldehyde 3s in 70% yield as a 1:1 mixture of cis and trans
isomers. In contrast, isomerization of 2t, a secondary allylic
alcohol with a tetrasubstituted CC bond, led to the
formation of 4H-chromene 14 in 69% yield (Figure 2).
The exact same catalytic system was next evaluated in the
isomerization of 16 different primary and secondary homo-
allylic alcohols 4a−p (Scheme 4). Interestingly, 3-phenyl-3-
butenol, a substrate with a terminal olefin, was converted into
5a in 52% yield. This is noteworthy, as one may expect
unproductive insertion of Pd at the least encumbered olefinic
position to occur preferentially. Homoallylic alcohols with a
disubstituted olefin were isomerized efficiently (5b, 70% yield;
5d, 58% yield). Whereas 5c was obtained in only 29% yield,20
4e, 4o, and 4p reacted with increased efficiency (isolated yields:
5e, 51%; 5o, 87%; 5p, 94%). Although we had initial concerns
regarding the site-selective insertion of Pd for substrates with a
3,3,4-trisubstituted olefinic pattern (4f−n), the result obtained
for 4a prompted us to explore their reactivity. Gratifyingly,
productive isomerization to the desired carbonyl derivatives was
systematically observed. Homoallylic alcohols with an aromatic
a
Reactions were performed on a 0.25 mmol scale. The substrates were
E-configured, unless otherwise noted. Yields after chromatography are
shown. On a 0.5 mmol scale. With 10 mol % cyclohexene.
b
c
Figure 2. Tentative isomerization of 2t, a tetrasubstituted secondary
allylic alcohol.
substituent at both C3 and C4 delivered the products in
moderate yields (5f−5j, 37−55% yield), while substrates with
an aryl group at C4 and an alkyl substituent at C3 proved
particularly suitable (5l−n, 80−89% yield). Isomerization of
alkenyl alcohols with a more remote trisubstituted double bond
was also successfully achieved. Thus, 5q was isolated in 75%
yield from the corresponding bis-homoallylic alcohol. Interest-
ingly, while rac-citronellol was isomerized to rac-3,7-dimethy-
loctanal (rac-5r) in only 38% yield, the corresponding
secondary alcohol delivered rac-5s in 71% yield. In both
cases, the remainder consisted of unreacted starting material.
These results are remarkable because the presence of an
additional alkyl substituent along the alkyl chain does not
interrupt the catalysis. These observations are in line with
recent findings by Sigman and co-workers.16d Finally, isomer-
ization of a terminal (4t) or disubstituted (4u) CC bond
over 11 positions afforded 5t and 5u in 92% and 79% yield,
respectively.
D
dx.doi.org/10.1021/ja508736u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Scheme 4. Isomerization of Homoallylic and Alkenyl Alcohols with 13
a
Reactions were performed on a 0.25 mmol scale. Substrates were E-configured, unless otherwise noted. Yields after chromatography are shown.
Without cyclohexene. 48 h. On a 0.75 mmol scale. 4 h, 93% purity. On a 0.1 mmol scale.
b
c
d
e
f
Figure 3. (a−c) Labeling experiments: (a) with 2m and in situ generation of [Pd−D] using various Pd loadings; (b) with 2a and in situ generation
of [Pd−D]; (c) with 2u-d2 and in situ generation of [Pd−H]. (d) Crossover experiment. (e) Mechanistic rationale.
Mechanistic Experiments. In order to obtain preliminary
insights into the reaction mechanism, a series of labeling
experiments was carried out (Figure 3). First, reactions with
substrate 2m were conducted using different loadings of Pd in
the presence of cyclohexene-d10 to generate a [Pd−D] catalyst
in situ.21 After full consumption of 2m, 2H NMR analysis of the
crude reaction mixture showed that deuterium was incorpo-
rated at C1 and C3 exclusively. No incorporation at C2 was
detected. Quantitative H NMR analysis indicated that the
1
extent of deuterium incorporation at C3 matched the initial
loading of 13, clearly establishing that the first insertion of
palladium deuteride into the CC bond is irreversible.22 In
contrast, deuterium scrambling at C1 is consistent with
postreaction reversible insertion of palladium deuteride across
E
dx.doi.org/10.1021/ja508736u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
the CO bond of the aldehyde (vide infra). This leads to
equilibration of the [Pd−H] and [Pd−D] intermediates and
accounts for the higher extent of D incorporation at C1 with
respect to the initial loading of 13 (Figure 3a). When 2a was
reacted with 18 mol % 13 and a 10-fold excess of cyclohexene-
d10, no significant incorporation of deuterium on the methyl
group was detected, suggesting that Pd inserts exclusively at C2
even for sterically less-demanding allylic alcohols (Figure 3b).
Labeling experiments employing dideuterated substrate 2u-d2
were carried out using the standard procedure for the in situ
generation of a [Pd−H] catalyst (Figure 3c). After completion,
1H and 2H NMR analyses of the crude reaction mixture showed
deuterium scrambling at C1, C2, and C3. Although in apparent
contradiction with the experiments of Figure 3a (no D
incorporation at C2), this result can be explained by reversible
insertion of the [Pd−D] species into the transient enol
intermediate B. The subsequent insertion product C undergoes
irreversible β-H elimination to give aldehyde 3u. Postreaction
reinsertion of the [Pd−H] occurs preferentially by formation of
the Pd−alkoxy intermediate D rather than the α-hydroxy Pd−
alkyl complex C. Competing β-H/β-D elimination and
statistical formation of [Pd−H]/[Pd−D] species accounts for
later scrambling of deuterium at C3 and C1. Finally, a crossover
experiment was conducted with an equimolar mixture of
dideuterated substrate 2u-d2 and nondeuterated alcohol 2o
for the development of chiral catalysts for imparting stereo-
control on pre-existing stereocenters in racemic mixtures.
Enantioselective Catalysis. To evaluate the possibility of
developing an enantioselective variant of these isomerization
reactions, complex 15 was prepared using (R,R)-i-Pr-
DUPHOS, a ligand that we anticipated may share electronic
and steric similarities with dcpe (Scheme 6). When the reaction
a
Scheme 6. Enantioselective Isomerizations
a
Reactions were performed on a 0.25 mmol scale. Isolated yields and
ee’s determined by chiral HPLC are shown.
was conducted at room temperature, encouraging preliminary
results were obtained for the isomerization of allylic alcohol
(E)-2n and its homoallylic congener (E)-4v. The correspond-
ing aldehydes (S)-3n and (S)-5v were isolated in 65% yield
with 73% ee and in 36% yield with 62% ee, respectively.
Furthermore, the reaction was found to be stereospecific
because isomerization of (Z)-2n delivered (R)-3n in 50% yield
with 48% ee.
Computational Investigations. To further investigate the
reaction mechanism, we performed a series of density
functional theory (DFT) calculations using 1,2-bis(di-tert-
butylphosphino)ethane (dtbpe) and 1,3-bis(di-tert-
butylphosphino)propane (dtbpp) as model ligands. Replace-
ment of the cyclohexyl and isopropyl substituents in dcpe and
dippp by tert-butyl groups reduces the number of conformers
and has only a minor effect on the reaction profile, as was
examined for the most favorable reaction pathways. This is
further supported by the similar reactivities observed with
[(dtbpe)Pd(Me)(Cl)] and 13 for selected substrates (see the
SI for details).27 The model palladium hydride complex 16 was
used as a starting point of our investigations because its
formation under the experimental conditions employed above
is well-documented.13,14
2
(Figure 3d). H NMR analysis of the crude mixture showed
significant incorporation of deuterium at C3 but no measurable
incorporation at C2 in aldehyde 3o-dn. This implies that the
[Pd−H] species decoordinates from the substrate only after
complete conversion to the carbonyl product but not at the
enol stage. The lower deuterium incorporation at C3 for 3u-dn
compared with 3o-dn indicates a higher rate of isomerization of
substrate 2u-d2 (see the SI). Finally, additional control
experiments also indicated that no exchange involving the
hydroxyl proton occurs during or after catalysis (see the SI).23
The potential decoordination of the catalyst during isomer-
ization is a fundamental aspect of chain-walking processes.
Isomerization of (S)-citronellol (4r) and its secondary alcohol
analogue (4S)-4s were conducted using our optimized protocol
(Scheme 5). The corresponding carbonyl derivatives 5r and 5s
a
Scheme 5. Isomerization of Citronellol Derivatives
Isomerization of Allylic Alcohol (E)-2a Catalyzed by
[(dtbpe)Pd−H]+. A reasonable reaction profile was calculated,
and the most relevant intermediates and transition states are
depicted in Figure 4. Coordination of allylic alcohol (E)-2a to
16 is an endergonic process that leads to the formation of
INT1 (ΔGrxn = 14.5 kcal/mol). Subsequent migratory insertion
of [Pd−H] across the double bond of the allylic alcohol occurs
with a relatively low activation barrier of 1.8 kcal/mol (INT1
→ TS1 → INT2). The resulting secondary alkyl-Pd
intermediate INT2 is characterized by a stabilizing agostic
interaction with C3−Ha (i.e., formerly the hydride ligand).28
This C−H bond is slightly elongated (1.18 Å), and the Pd−
C2−C3−H dihedral angle is 7.1° (Figure 5).
a
Reactions were performed on a 0.25 mmol scale. Isolated yields and
ee’s determined by chiral HPLC or GC are shown.
were obtained in 38% and 64% yield, respectively, but in
racemic form in both cases. These results indicate that the
catalyst disengages from the substrate during the isomerization
process. This is in contrast with previous observations. Using
[Fe(CO)5] as an alkene-zipper catalyst, Stockis and co-workers
found that partial epimerization of 4r and complete
racemization of 4s occur under thermal reaction conditions.24
More recently, Sigman and co-workers convincingly established
that during redox-relay Heck reactions with alkenyl alcohols,
the putative [(N,N)Pd−H]+ intermediate does not dissociate
from the substrate upon iterative migration along the alkyl
chain.16 We believe that this result may provide an opportunity
Notably, intermediate INT2 contains three β-hydrogens
susceptible to β-hydride elimination: Ha, Hb, and Hc (Figure 5).
Direct β-H elimination of Hb or Hc is not achievable because of
the long Pd···H distances (3.54 and 3.37 Å, respectively) and
F
dx.doi.org/10.1021/ja508736u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 4. Computed reaction profile for the productive isomerization of (E)-2a to 3a. The most relevant intermediates and transition states are
represented. DFT method: PCMDCE-B3LYP/[6-311+G(d,p); LANL2DZ]//M06L/[6-31G(d) on C, H, O, P; LANL2DZ on Pd]. ΔG values are
given in kcal/mol.25,26
the large Pd−C2−C1−H dihedral angles (Pd−C2−C1−Hb =
71.2°; Pd−C2−C1−Hc = 47.5°). Rotation around the Pd−C2
bond is required in order to bring Pd−C2−C1−Hb or Pd−
C2−C1−Hc into a single plane. The former rotation via TS2 is
energetically more favorable than the latter via TS3
(ΔΔG⧧TS2−TS3 = 5.3 kcal/mol). The resulting intermediate
INT3.E is again stabilized by an agostic interaction (Pd−Hb =
1.92 Å; Pd−C2−C1−H = 24.2°). The Pd−C2 rotation has an
activation barrier of 9.5 kcal/mol and compares well with the
value computed by Wang, Wang, and co-workers for the related
redox-relay Heck reaction (ΔG⧧ = 11.2 kcal/mol).29 As a direct
consequence of a switch from one agostic interaction to
another, both the C−Ha and C−Hb bonds are slightly
elongated in TS2 (1.12 and 1.14 Å, respectively) (Figure 5).
Next, complex INT3.E undergoes β-Hb elimination via TS4
to generate INT4.E, which is characterized by a π-bound (E)-
enol. This process has a low activation barrier (ΔG⧧ = 2.1 kcal/
mol). Dissociation of the (E)-enol from INT4.E is a highly
exergonic process (ΔGrxn = −9.3 kcal/mol). Decoordination of
the (E)-enol from the palladium center in the sequence INT4.E
→ [16 + (E)-enol] → INT5.E is in apparent contradiction with
the results of the labeling experiments (Figure 3). Nonetheless,
we also located reasonable isomeric structures of INT5.E in
which the Pd remains on the same face of the olefin that are
G
dx.doi.org/10.1021/ja508736u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
transition state of the bimolecular process lies far higher in
energy than other stationary points of the reaction profile (ΔG⧧
= 41.6 kcal/mol; see the SI). The palladium-catalyzed
tautomerization pathway we computed compares more
favorably (Figure 4, bottom profile). Migratory insertion of
[Pd−H] across the CC bond of the (E)-enol via the
sequence INT5.E → TS6 → INT6.E is a kinetically favorable
exergonic process (ΔG⧧ = 0.6 kcal/mol, ΔGrxn = −7.9 kcal/
mol) in which intermediate INT6.E is stabilized by an agostic
interaction (Pd−H = 2.02 Å; Pd−C1−C2−H = 21.4°).
Rotation around the Pd−C1 bond leads to the more stable
coordinatively unsaturated palladium complex INT7.E (Pd···O
= 2.26 Å, Ptrans−Pd−C1 = 162.8°), which undergoes
tautomerization to generate the product-bound palladium
hydride complex INT8. Unfortunately, we were not able to
locate a reasonable transition state for this elementary step.16c,27
Final decoordination of [Pd−H] liberates aldehyde 3a, which is
11.1 kcal/mol more stable than the initial allylic alcohol (E)-2a.
This high exergonicity constitutes the driving force for the
isomerization process.
The postreaction [Pd−H] insertion across the CO bond
of aldehyde 3a from INT8 was also computed. It proceeds via
TS8 and generates INT9. The low activation barrier of this
process (ΔG⧧ = 3.1 kcal/mol) is in full agreement with the
labeling experiments of Figure 3, where scrambling of the
deuterium label at C1 position was observed.
In summary, rotation around the Pd−C2 bond is turnover-
determining in the isomerization of allylic alcohol (E)-2a
catalyzed by [(dtbpe)Pd−H]+ (via TS2). The activation free
energy of 20.0 kcal/mol is also in good agreement with the
experimental reaction conditions (typically, T = 85 °C).
Isomerization of Allylic Alcohol (Z)-2a Catalyzed by
[(dtbpe)Pd−H]+. To understand the influence of the substrate
geometry on the isomerization of allylic alcohols, we computed
the reaction profile using (Z)-2a as the substrate (Figure 6).
Migratory insertion of [Pd−H] into the CC bond has a
similar activation barrier as in the case of (E)-2a (16.1 vs 16.3
kcal/mol, respectively). Rotation around the Pd−C2 bond
proceeds with a much lower activation barrier than for the
Figure 5. Most relevant intermediates and transition states in the
isomerization of (E)-2a.
accessible from INT4.E without decoordination of the
substrate (see the SI for details). Alternatively, dissociation of
the enol could occur within the solvent cage and would still be
consistent with the labeling studies.
The subsequent uncatalyzed tautomerization ((E)-enol →
3a), although thermodynamically favorable (ΔGrxn = −11.0
kcal/mol), is kinetically not feasible because the six-membered
Figure 6. Computed reaction profile for the isomerization of (Z)-2a to 3a. The most relevant intermediates and transition states are depicted. ΔG
values are given in kcal/mol.
H
dx.doi.org/10.1021/ja508736u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
isomerization of (E)-2a (4.4 vs 9.5 kcal/mol, respectively). A
comparative analysis of the two transition states reveals a strong
steric interaction between the tert- butyl substituent of the
ligand and the phenyl group of the substrate (C−H distance =
2.60 Å) in TS2, whereas the orientation of the smaller methyl
group of the substrate toward the ligand backbone has a less
destabilizing influence in TS10 (C−H distance = 2.65 Å)
(Figure 7).
labeling studies, directly convert to INT5.E without substrate
dissociation (ΔGrxn = 0.6 kcal/mol) (vide supra).
In summary, isomerization of (Z)-2a has a lower activation
barrier than isomerization of (E)-2a (16.4 vs 20.0 kcal/mol,
respectively). This is in perfect agreement with the result of
monitoring experiments where (Z)-2a was found to react
slightly faster than (E)-2a (see the SI). Whereas rotation
around the Pd−C2 bond is rate-determining in the isomer-
ization of (E)-2a, β-H elimination via TS11 has the highest
energy in the isomerization of (Z)-2a.
Competing Pathways in the Isomerization of Allylic
Alcohol (E)-2a. We experimentally found that isomerization
of (E)-2a to aldehyde 3a was potentially accompanied by the
formation of the terminal homoallylic alcohol 4a and the
isomeric allylic alcohol (Z)-2a (Scheme 1 and Table 1).
Because the reaction is stereospecific and in view of the future
development of a highly enantioselective version of this
process, we sought to gain insight into the factors influencing
the site selectivity in the migratory insertion of [Pd−H] across
the CC bond (Figure 8).
Coordination of (E)-2a to 16 leading to INT14 is slightly
more demanding than the binding leading to its regioisomer
INT1 (18.0 vs 14.5 kcal/mol, respectively). Migratory insertion
into the double bond via TS12 requires a very low activation
energy (ΔG⧧ = 0.6 kcal/mol), although the overall activation
barrier for this sequence (ΔG⧧ = 18.6 kcal/mol; Figure 8) is
higher than the corresponding barrier for the regioisomeric
pathway via TS1 (ΔG⧧ = 16.3 kcal/mol; Figure 4). This
difference is attributable to the increased steric interaction
between the C3 substituents and the ligand backbone, as
reflected by the longer Pd−C3 distance in TS12 (2.36 Å;
Figure 9) compared with the Pd−C2 distance in TS1 (2.21 Å;
Figure 5).
Figure 7. Structural comparison of the TSs for the Pd−C rotations in
the isomerizations of (E)-2a (TS2) and (Z)-2a (TS10).
Subsequent β-H elimination via TS11, which is a facile
process (activation barrier ΔG⧧ = 1.3 kcal/mol), generates the
π-bound (E)-enol cationic complex INT13. As discussed above,
this can either decoordinate in an exergonic process ([16 +
(E)-enol]; ΔGrxn = −11.7 kcal/mol) or, in agreement with the
Figure 8. Computed reaction profile for the competing isomerization processes: (E)-2a/(Z)-2a, (E)-2a/4a, and (Z)-2a/4a. The most relevant
intermediates and transition states are depicted. ΔG values are given in kcal/mol.
I
dx.doi.org/10.1021/ja508736u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
kcal/mol and produces complex INT19 with the Z-configured
allylic alcohol. Decoordination of the substrate is exergonic, and
there is no significant thermodynamic preference for one allylic
alcohol isomer over the other. Overall, the isomerization of
allylic alcohol (E)-2a to either 4a or (Z)-2a presents a lower
activation barrier than the productive isomerization leading to
aldehyde 3a shown in Figure 4 (18.6 vs 20.0 kcal/mol,
respectively). These computational results are not fully
consistent with the labeling experiments conducted with (E)-
2a, where no D incorporation on the methyl substituent was
detected (Figure 3b). Although the activation energy of the
competing processes might be underestimated, the drastically
different result obtained in the isomerization of (E)-2a with 1
suggests that subtle variations in either the ligand design or the
reaction conditions may dramatically impact the outcome of
the isomerization reaction. Similarly, substrates with bulkier
substituents at C3 may favor productive isomerization over
other competing processes. Importantly, the thermodynamic
sink constituted by the productive isomerization of (E)-2a to
3a acts as a strong driving force for the whole process.
Isomerization of Homoallylic Alcohols (E)-4c and (E)-4l
Catalyzed by [(dtbpe)Pd−H]+. Because of the strong influence
of the olefinic substitution pattern on the isomerization of
homoallylic alcohols, the reaction profiles for the isomerization
of homoallylic alcohols (E)-4c and (E)-4l catalyzed by
[(dtbpe)Pd−H]+ were computed. The reaction profile for the
isomerization of (E)-4c is similar to that for the isomerization
of (E)-2a, and the most significant portion is displayed in
Figure 10 (see the SI for the complete reaction profile). While
coordination of homoallylic alcohol (E)-4c to [Pd−H] is
slightly more favorable than coordination of allylic alcohol (E)-
2a (13.4 vs 14.5 kcal/mol), the subsequent migratory insertion
across the CC bond in (E)-4c proceeds with a higher
Figure 9. Most relevant intermediates and transition states in the
competing isomerization processes.
Rotation around the Pd−C3 single bond in INT15 via TS13
proceeds with a low barrier (ΔG⧧ = 5.6 kcal/mol) and
generates intermediate INT16, which is characterized by an
agostic interaction between the Pd atom and one C−Hd bond
of the methyl group (blue pathway). Subsequent β-H
elimination via TS15 (ΔG⧧ = 6.5 kcal/mol) affords INT17
and is followed by dissociation of the terminal homoallylic
alcohol 4a. The overall process (E)-2a → 4a is slightly
endergonic (ΔGrxn = 2.6 kcal/mol) (Figure 8).
Alternatively, rotation around C3−C2 in intermediate
INT15 (INT15 → TS14 → INT18; ΔG⧧ = 4.0 kcal/mol)
poises one of the two diastereotopic hydrogen atoms (He) in a
favorable position for subsequent β-H elimination (green
pathway). This elementary step has an activation barrier of 6.3
Figure 10. Comparative and partial representation of the computed reaction profiles for the isomerizations of (E)-2a, (E)-4c, and (E)-4l with the
model catalysts 16 and 17. ΔG values are given in kcal/mol.
J
dx.doi.org/10.1021/ja508736u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 11. Comparative structural analysis of the turnover-determining transition state (TS2) in the isomerizations of (E)-2a, (E)-4c, and (E)-4l
with model catalysts 16 and 17. The color code is the same as the one used in Figure 10.
activation barrier (18.1 vs 16.3 kcal/mol, respectively). For the
isomerization of homoallylic alcohol (E)-4c, INT2, TS2, and
INT3.E are higher in energy than the corresponding stationary
points for the isomerization of allylic alcohol (E)-2a. More
importantly, the transition state corresponding to the rotation
around the newly formed Pd−C bond is rate-determining in
both cases (compare the blue and black lines in Figure 10: 24.2
vs 20.0 kcal/mol for TS2). Inspection of the respective
molecular structures of TS2 reveals that the Pd−Ha distance is
shorter than the Pd−Hb distance for the isomerization of (E)-
4c (2.29 vs 2.39 Å; Figure 11), whereas Pd−Ha is longer than
Pd−Hb for the isomerization of (E)-2a (2.44 vs 2.29 Å; Figure
5). These computational results are in line with the
experimental data, where homoallylic alcohols with this olefinic
substitution pattern displayed a lower reactivity compared with
the analogous allylic alcohols (Schemes 3 and 4).
The overall reaction profile for the isomerization of (E)-4l is
similar to that of (E)-4c. The activation free energies of the
turnover-determining transition state TS2 are very close (24.5
vs 24.2 kcal/mol, respectively; compare the blue and purple
lines in Figure 10). The structural parameters of TS2 are also
very similar for the two alcohols (Figure 11). At first sight, the
computational results may not properly reflect the experimental
data, as 3,3,4-trisubstituted homoallylic alcohols delivered the
carbonyl derivatives in higher yields than their 3,4,4-
trisubstituted analogues. Nonetheless, consumption of the
starting material proceeded with similar efficiencies in these
two cases, but the isomerization of 3,4,4-trisubstituted
homoallylic alcohols was accompanied by the formation of
additional side products.20
and the blue line with the green line in Figure 10; ΔΔGrxn = 4.1
kcal/mol for (E)-2a and 3.5 kcal/mol for (E)-4c). This is a
direct consequence of the wider bite angle of dtbpp with
respect to dtbpe (105.9° in 17 vs 92.3° in 16, respectively)
which imposes a stronger steric interaction between the P
substituents and the substrate substituents.19,30 The free
energies of the turnover-determining TS2 are also higher
with the dtbpp ligand (ΔG⧧ = 26.5 kcal/mol for (E)-4c and
ΔG⧧ = 27.2 kcal/mol for (E)-2a). Interestingly, we found large
variations in the bite angles of the chelating bisphosphine ligand
in 17 and TS2 for dtbpp (17 → TS2: 105.9° → 97.6° for (E)-
2a). In contrast, dtbpe adopts a more rigid conformation (16
→ TS2: 92.3° → 89.7° for (E)-2a). These observations may
account for the difference in reactivity between the catalysts
supported by dcpe and dippp ligands. Overall, these computa-
tional results are also in agreement with the lower reactivity of
(E)-4c in the isomerization reaction catalyzed by complex 1 or
7. Similarly, the isomerization of (E)-2a catalyzed by [(dtbpe)-
Pd−H]+ is more favorable than that catalyzed by [(dtbpp)Pd−
H]+ (ΔΔG⧧ = 7.2 kcal/mol). This again reflects the
experimental results obtained for this allylic alcohol (Scheme
1 and Table 1).
As a final complementary note, we never observed
coordination of the hydroxyl functionality to the Pd atom
throughout the computed reaction profiles of the productive
isomerization of either the allylic alcohols or the homoallylic
alcohols investigated.
CONCLUSION
■
We have discovered a readily accessible palladium catalyst that
isomerizes a variety of allylic, homoallylic, and alkenyl primary
and secondary alcohols into the corresponding carbonyl
derivatives. To the best of our knowledge, this is the first
example of an isomerization catalyst that bridges the gap
between allylic and alkenyl alcohols as well as between primary
and secondary substrates. The system operates under relatively
mild conditions, tolerates a number of potentially reactive
functional groups, andmore importantlyis virtually in-
sensitive to the number, position, and nature of the substituents
of the olefinic moiety. The preliminary results obtained for the
enantioselective isomerization of a handful of substrates augur
well for the development of a more general asymmetric version
of these isomerization reactions.
Isomerizations of (E)-2a and (E)-4c Catalyzed by
[(dtbpp)Pd−H]+. To get a better understanding of the influence
of the ligand structure on the isomerization of allylic and
homoallylic alcohols, the reaction profiles for the isomerizations
of the structurally related allylic and homoallylic alcohols (E)-
2a and (E)-4c were also computed using [(dtbpp)Pd−H]+
(17) as a model for catalyst 1 or 7. For the sake of clarity, the
results for the [Pd−H] insertion/rotation sequence are also
summarized in Figure 10 (green and red lines; see the SI for the
complete profiles).
The coordination of either substrate to the coordinatively
unsaturated complex [(P,P)Pd−H]+ (i.e., the formation of
INT1) is clearly a less favorable process when dtbpp is
employed as a ligand (compare the black line with the red line
K
dx.doi.org/10.1021/ja508736u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
chai, J.; Dumas, A. M.; Bode, J. W. Angew. Chem., Int. Ed. 2012, 51,
10954.
(4) (a) Trost, B. M.; Toste, F. D.; Pinkerton, A. B. Chem. Rev. 2001,
101, 2067. (b) Yue, C. J.; Liu, Y.; He, R. J. Mol. Catal. A: Chem. 2006,
259, 17.
The results of a combined experimental and theoretical
mechanistic study are in support of a chain-walking process that
operates by iterative migratory insertion/β-H elimination
sequences for both allylic and homoallylic alcohols. The
irreversible nature of the first migratory insertion across the
CC bond constitutes one of the key features of the proposed
mechanism. The rotation around the newly formed Pd−C
bond was identified as the turnover-determining step for the
productive isomerization of E-configured substrates. This
process has markedly lower activation energy for the isomer-
ization of (Z)-allylic alcohols. A postreaction insertion of [Pd−
H] across the CO bond of the carbonyl products has been
identified experimentally and reproduced in silico. Interestingly,
we found experimentally that the catalyst does not dissociate
from the substrate in the isomerization of allylic alcohols,
whereas it disengages during the isomerization of alkenyl
alcohols when an additional substituent is present on the alkyl
chain. Finally, the effect of the bite angle of the chelating
bisphosphine ligand on the catalyst activity has been well-
reproduced computationally. This subtle structural change may
be important for the development of improved generation of
achiral and chiral catalysts.
(5) (a) van der Drift, R. C.; Bouwman, E.; Drent, E. J. Organomet.
Chem. 2002, 650, 1. (b) Uma, R.; Crev
́
isy, C.; Gree
́
, R. Chem. Rev.
2003, 103, 27. (c) Kuznik, N.; Krompiec, S. Coord. Chem. Rev. 2007,
́
251, 222. (d) Krompiec, S.; Krompiec, M.; Penczek, R.; Ignasiak, H.
Coord. Chem. Rev. 2008, 252, 1819. (e) Aubert, C.; Fensterbank, L.;
Garcia, P.; Malacria, M.; Simonneau, A. Chem. Rev. 2011, 111, 1954.
(f) Mantilli, L.; Mazet, C. Chem. Lett. 2011, 40, 341. (g) Ahlsten, N.;
Bartoszewicz, A.; Martin-Matute, B. Dalton Trans. 2012, 41, 1660.
(6) Seminal contribution: (a) Kumobayashi, H.; Akutagawa, S.;
Otsuka, S. J. Am. Chem. Soc. 1978, 100, 3949. For recent reviews, see:
(b) Akutagawa, S. In Comprehensive Asymmetric Catalysis; Jacobsen, E.
N., Pfaltz, A., Yamamoto, H., Eds.; Springer: Berlin, 1999; Vol. 2,
Chapter 23, pp 813−832. (c) Akutagawa, S. In Comprehensive
Asymmetric Catalysis; Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.;
Springer: Berlin, 1999; Vol. 3, Chapter 41.1, pp 1461−1472. (d) Fu,
G. C. In Modern Rhodium-Catalyzed Organic Reactions; Evans, P. A.,
Ed.; Wiley-VCH: Weinheim, Germany, 2005; Chapter 4, pp 79−91.
(7) (a) Tanaka, K.; Qiao, S.; Tobisu, M.; Lo, M. M.-C.; Fu, G. C. J.
Am. Chem. Soc. 2000, 122, 9870. (b) Tanaka, K.; Fu, G. C. J. Org.
Chem. 2001, 66, 8177. For a related contribution, see: (c) Chapuis,
C.; Barthe, M.; de Saint Laumer, J.-Y. Helv. Chim. Acta 2001, 84, 230.
(8) (a) Mantilli, L.; Mazet, C. Tetrahedron Lett. 2009, 50, 4141.
ASSOCIATED CONTENT
* Supporting Information
■
S
(b) Mantilli, L.; Ger
Chem., Int. Ed. 2009, 48, 5413. (c) Mantilli, L.; Mazet, C. Chem.
Commun. 2010, 46, 445. (d) Mantilli, L.; Gerard, D.; Torche, S.;
Besnard, C.; Mazet, C. Chem.Eur. J. 2010, 16, 12736. (e) Mantilli,
L.; Gerard, D.; Besnard, C.; Mazet, C. Eur. J. Inorg. Chem. 2012, 3320.
́
ard, D.; Torche, S.; Besnard, C.; Mazet, C. Angew.
Experimental procedures, spectral data and crystallographic
data (CIF) for complexes 11 (CCDC 1020394) and 13
(CCDC 1020782), computational data, and full ref 24. This
material is available free of charge via the Internet at http://
pubs.acs.org.
́
́
(f) Li, H.; Mazet, C. Org. Lett. 2013, 15, 6170. For other work using
iridium, see: (g) Li, J.-Q.; Peters, B.; Andersson, P. G. Chem.Eur. J.
2011, 17, 11143.
AUTHOR INFORMATION
Corresponding Author
clement.mazet@unige.ch
■
(9) (a) Arai, N.; Sato, K.; Azuma, K.; Ohkuma, T. Angew. Chem., Int.
Ed. 2013, 52, 7500. For related contributions, see: (b) Sun, Y.;
LeBlond, C.; Wang, J.; Blackmond, D. G.; Lauirada, J.; Sowa, J. R., Jr. J.
Am. Chem. Soc. 1995, 117, 12647. (c) Bizet, V.; Pannecoucke, X.;
Renaud, J.-L.; Cahard, D. Angew. Chem., Int. Ed. 2012, 51, 6467.
(d) Wu, R.; Beauchamps, M. G.; Laquidara, J. M.; Sowa, J. R., Jr.
Angew. Chem., Int. Ed. 2012, 51, 2106.
(10) (a) Kitamura, M.; Manabe, K.; Noyori, R. Tetrahedron Lett.
1987, 28, 4719. (b) Hiroya, K.; Kurihara, Y.; Ogasawara, K. Angew.
Chem., Int. Ed. Engl. 1995, 34, 2287. (c) Ito, M.; Kitahara, S.; Ikariya,
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by the University of Geneva and
Roche. We thank Prof. Tomasz A. Wesolowski and Dr. Amalia
Poblador-Bahamonde (University of Geneva) for offering us
́
́
T. J. Am. Chem. Soc. 2005, 127, 6172. (d) Fernandez-Zumel, M. A.;
Lastra-Barreira, B.; Scheele, M.; Díez, J.; Crochet, P.; Gimeno, J.
Dalton Trans. 2010, 39, 7780.
́
access to their computational facilities. We thank Dr. Celine
Besnard (University of Geneva) for assistance with X-ray
diffraction measurements and crystal structure resolution. All of
the molecular structures were generated using CYLview.31 One
referee is acknowledged for suggesting important control
labeling experiments.
(11) (a) Grotjahn, D. B.; Larsen, C. R.; Gustafson, J. L.; Nair, R.;
Sharma, A. J. Am. Chem. Soc. 2007, 129, 9592. For related studies from
the same group, see: (b) Larsen, C. R.; Grotjahn, D. B. J. Am. Chem.
Soc. 2012, 134, 10357. (c) Erdogan, G.; Grotjahn, D. B. Org. Lett.
2014, 16, 2818. For a related study, see: (d) Ishibashi, K.; Takahashi,
M.; Yokota, Y.; Oshima, K.; Matsubara, S. Chem. Lett. 2005, 34, 664.
(12) For a general review of palladium hydrides, see: (a) Grushin, V.
V. Chem. Rev. 1996, 96, 2011. For important contributions on Pd-
catalyzed isomerization processes, see: (b) Ionescu, A.; Ruppel, M.;
Wendt, O. F. J. Organomet. Chem. 2006, 691, 3806. (c) Coquerel, Y.;
REFERENCES
■
(1) (a) Anastas, P. T.; Warner, J. C. Green Chemistry: Theory and
Practice; Oxford University Press: New York, 1998. (b) Sheldon, R. A.;
Arends, I.; Hanefeld, U. Green Chemistry and Catalysis; Wiley-VCH:
Weinheim, Germany, 1997. (c) Anastas, P.; Eghbali, N. Chem. Soc. Rev.
2010, 39, 301.
́
Bremond, P.; Rodriguez, J. J. Organomet. Chem. 2007, 692, 4805.
(d) Lim, H. J.; Smith, C. R.; RajanBabu, T. V. J. Org. Chem. 2009, 74,
4565. (e) Gauthier, D.; Lindhardt, A. T.; Olsen, E. P. K.; Overgaard, J.;
Skrydstrup, T. J. Am. Chem. Soc. 2010, 132, 7998. (f) Sabitha, G.;
Nayak, S.; Bhikshapathi, M.; Yadav, J. S. Org. Lett. 2011, 13, 382. For a
related account, see: (g) Widenhoefer, R. A. Acc. Chem. Res. 2002, 35,
905.
(13) (a) Johnson, L. K.; Killian, C. M.; Brookhart, M. J. Am. Chem.
Soc. 1995, 117, 6414. (b) Johnson, L. K.; Mecking, S.; Brookhart, M. J.
Am. Chem. Soc. 1996, 118, 267. (c) Guan, Z.; Cotts, P. M.; McCord, E.
F.; McLain, S. J. Science 1999, 283, 2059. (d) Berkefeld, A.; Mecking,
(2) (a) Trost, B. M. Science 1991, 254, 1471. (b) Trost, B. M. Angew.
Chem., Int. Ed. Engl. 1995, 34, 259. (c) Wender, P. A.; Verma, V. A.;
Paxton, T. J.; Pillow, T. H. Acc. Chem. Res. 2008, 41, 40. (d) Wender,
P. A.; Miller, B. L. Nature 2009, 460, 197. (e) Burns, N. Z.; Baran, P.
S.; Hoffmann, R. W. Angew. Chem., Int. Ed. 2009, 48, 2854.
(3) (a) Comprehensive Asymmetric Catalysis; Jacobsen, E. N., Pfaltz,
A., Yamamoto, H., Eds.; Springer: Berlin, 1999; Vols. 1−3. (b) Walsh,
P. J.; Kozlowski, M. C. Fundamentals of Asymmetric Catalysis;
University Science Books: Sausalito, CA, 2009. (c) Hoveyda, A. H.;
Evans, D. A.; Fu, G. C. Chem. Rev. 1993, 93, 1307. (d) Mahatthanan-
L
dx.doi.org/10.1021/ja508736u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
S. Angew. Chem., Int. Ed. 2006, 45, 6044. (e) Okada, T.; Park, S.;
Takeuchi, D.; Osakada, K. Angew. Chem., Int. Ed. 2007, 46, 6141.
(f) Okada, T.; Takeuchi, D.; Shishido, A.; Ikeda, T.; Osakada, K. J. Am.
Chem. Soc. 2009, 131, 10852. (g) Guan, Z. Chem.Asian J. 2010, 5,
1058. (h) Takeuchi, D. J. Am. Chem. Soc. 2011, 133, 11106. (i) Okada,
T.; Park, S.; Takeuchi, D.; Osakada, K. Angew. Chem., Int. Ed. 2007, 46,
6141. For noncatalytic long-range isomerization, see: (j) Chinkov, N.;
Levin, A.; Marek, I. Angew. Chem., Int. Ed. 2006, 45, 465. (k) Chinkov,
N.; Levin, A.; Marek, I. Synlett 2006, 501.
(31) Legault, C. Y. CYLview, version 1.0.561b; Universite
Sherbrooke: Sherbrooke, QC, 2014; http://www.cylview.org
́
de
(14) Kochi, T.; Hamasaki, T.; Aoyama, Y.; Kawasaki, J.; Kakiuchi, F.
J. Am. Chem. Soc. 2012, 134, 16544.
(15) Ohlmann, D. M.; Tschauder, N.; Stockis, J.-P.; Gooßen, K.;
Dierker, M.; Gooßen, L. J. J. Am. Chem. Soc. 2012, 134, 13716.
(16) (a) Werner, E. W.; Mei, T.-S.; Burckle, A. J.; Sigman, M. S.
Science 2012, 338, 1455. (b) Mei, T.-S.; Werner, E. W.; Burckle, A. J.;
Sigman, M. S. J. Am. Chem. Soc. 2013, 135, 6830. (c) Xu, L.; Hilton, M.
J.; Zhang, X.; Norrby, P.-O.; Wu, Y.-D.; Sigman, M. S.; Wiest, O. J. Am.
Chem. Soc. 2014, 136, 1960. (d) Mai, T.-S.; Patel, H. H.; Sigman, M. S.
Nature 2014, 508, 340.
(17) Vyas, D. J.; Larionov, E.; Besnard, C.; Guen
́ ́
ee, L.; Mazet, C. J.
Am. Chem. Soc. 2013, 135, 6177.
(18) Larionov, E.; Li, H.; Mazet, C. Chem. Commun. 2014, 50, 9816.
(19) (a) Raebiger, J. W.; Miedaner, A.; Curtis, C. J.; Miller, S. M.;
Anderson, O. P.; DuBois, D. L. J. Am. Chem. Soc. 2004, 126, 5502.
(b) Morris, R. H. J. Am. Chem. Soc. 2014, 136, 1948. (c) Bullock, R.
M.; Appel, A. M.; Helm, M. L. Chem. Commun. 2014, 50, 3125.
(20) The main component was identified as a dimerization product
reminiscent of those observed by Ryu and co-workers: Doi, T.;
Fukuyama, T.; Minamino, S.; Husson, G.; Ryu, I. Chem. Commun.
2006, 1875. See the Supporting Information for additional details.
(21) Ishibashi, K.; Matsubara, S. Chem. Lett. 2007, 36, 724.
(22) The slight variations between the extent of D incorporation and
the initial catalyst loading are within the experimental margin of error
(ca. 1%).
(23) The imperfect balance of D incorporation at C2 in the
experiments of Figure 3c,d is not due to competing tautomerization
reactions involving the starting allylic alcohol, enol intermediates, and/
or the final aldehyde. Indeed, mixing 3u-dn with 2o under the reaction
conditions for catalysis did not change the extent of D incorporation at
C2. Similar results were obtained in the absence of the Pd catalyst. See
the SI for details.
(24) Weissberger, E.; Stockis, A.; Carr, D. D.; Giebfried, J. Bull. Soc.
Chim. Belg. 1980, 89, 281.
(25) Frisch, M. J.; et al. Gaussian 09, revision A.02; Gaussian, Inc.:
Wallingford, CT, 2009.
(26) (a) For the LANL2 effective core potential, see: Hay, P. J.;
Wadt, W. R. J. Chem. Phys. 1985, 82, 270. (b) For the PCM solvation
model, see: Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev. 2005,
105, 2999.
(27) To further validate the use of dtbpe as a model for dcpe, this
noncommercial ligand was synthesized according to the following
literature procedure: Scherer, W.; Herz, V.; Bruck, A.; Hauf, C.;
̈
Reiner, F.; Altmannshofer, S.; Leusser, D.; Stalke, D. Angew. Chem., Int.
Ed. 2011, 50, 2845. The corresponding complex [(dtbpe)Pd(Me)
(Cl)] was evaluated for the isomerization of representative model
substrates (2a, 2b, 2h, 2m, and 4c) and provided yields similar to
those obtained with 13 under the prototypical reaction conditions for
isomerization (62%, 89%, 79%, 58%, and 44% respectively). See the SI
for details.
(28) (a) Scherer, W.; McGrady, G. S. Angew. Chem., Int. Ed. 2004, 43,
1782. (b) Brookhart, M.; Green, M. L. H.; Parkin, G. Proc. Natl. Acad.
Sci. U.S.A. 2007, 104, 6908.
(29) Dang, Y.; Qu, S.; Wang, Z.-X.; Wang, X. J. Am. Chem. Soc. 2014,
136, 986.
(30) (a) Dierkes, P.; van Leeuwen, P. W. N. M. J. Chem. Soc., Dalton
Trans. 1999, 1519. (b) van Leeuwen, P. W. N. M.; Kamer, P. C. J.;
Reek, J. N. H.; Dierkes, P. Chem. Rev. 2000, 100, 2741. (c) Kamer, P.
C. J.; van Leeuwen, P. W. N. M.; Reek, J. N. H. Acc. Chem. Res. 2001,
34, 895.
M
dx.doi.org/10.1021/ja508736u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX