Mechanistic insights into iridium-catalyzed asymmetric hydrogenation of dienes

Article abstract of DOI:10.1002/chem.200500762

Hydrogenation of 2,3-diphenylbutadiene (1) with the chiral carbene-oxazoline-iridium complex C has been studied by means of a combined experimental and computational approach. A detailed kinetic profile of the reaction was obtained with respect to consumption of the substrate and formation of the intermediate half-reduction products, 2,3-diphenylbut-1-ene (2) and the final product, 2,3-diphenylbutane (3). The data generated from these analyses, and from NMR experiments, revealed several facets of the reaction. After a brief induction period (presumably involving reduction of the cyclooctadiene ligand on C), the diene concentration declines in a zero-order process primarily to give mono ene intermediates. When all the diene is consumed, the reaction accelerates and compound 3 begins to accumulate. Interestingly, the prevalent enantiomer of the monoene intermediate 2 is converted mostly to meso-3 so the enantioselectivity of the reaction appears to reverse. The reaction seems to be first-order with respect to the catalyst when the catalyst concentration is less than 0.0075 M; diffusion of hydrogen across the gas-liquid interface complicates the analysis at higher catalyst concentra tions. Similarly, these diffusion effects complicated measurements of reaction rate versus applied pressure of dihydrogen; other factors like stir speed and flask geometry come into play under some, but not all, the conditions examined. Density functional theory (DFT) calculations, using the PBE method, were used to probe the reaction. These studies indicate a transoid-η ⁴-diene-di-hydride complex forms in the first stages of the catalytic cycle. Further reaction requires dissociation of one alkene ligand to give a η⁴-diene-dihydride-dihydrogen intermediate. A catalytic cycle that features Ir³⁺/Ir⁵⁺ seems to be involved thereafter.

Full text of DOI:10.1002/chem.200500762

FULL PAPER
DOI: 10.1002/chem.200500762
Mechanistic Insights into Iridium-Catalyzed Asymmetric Hydrogenation of
Dienes
Xiuhua Cui, Yubo Fan, Michael B. Hall,* and Kevin Burgess*[a]
Abstract: Hydrogenation of 2,3-diphe-
nylbutadiene (1) with the chiral car-
bene–oxazoline–iridium complex C has
been studied by means of a combined
experimental and computational ap-
proach. A detailed kinetic profile of
the reaction was obtained with respect
to consumption of the substrate and
formation of the intermediate half-re-
duction products, 2,3-diphenylbut-1-
ene (2) and the final product, 2,3-di-
phenylbutane (3). The data generated
from these analyses, and from NMR
experiments, revealed several facets of
the reaction. After a brief induction
period (presumably involving reduction
of the cyclooctadiene ligand on C), the
diene concentration declines in a zero-
order process primarily to give mono-
ene intermediates. When all the diene
is consumed, the reaction accelerates
and compound 3 begins to accumulate.
Interestingly, the prevalent enantiomer
of the monoene intermediate 2 is con-
verted mostly to meso-3 so the enantio-
selectivity of the reaction appears to
reverse. The reaction seems to be first-
order with respect to the catalyst when
the catalyst concentration is less than
0.0075m; diffusion of hydrogen across
the gas–liquid interface complicates the
analysis at higher catalyst concentra-
tions. Similarly, these diffusion effects
complicated measurements of reaction
rate versus applied pressure of dihydro-
gen; other factors like stir speed and
flask geometry come into play under
some, but not all, the conditions exam-
ined. Density functional theory (DFT)
calculations, using the PBE method,
were used to probe the reaction. These
studies indicate a transoid-h4-diene–di-
hydride complex forms in the first
stages of the catalytic cycle. Further re-
action requires dissociation of one
alkene ligand to give a h2-diene–dihy-
dride–dihydrogen intermediate. A cata-
lytic cycle that features Ir3+/Ir5+ seems
to be involved thereafter.
Keywords: asymmetric catalysis
·
carbene ligands · density functional
calculations
iridium
·
hydrogenation
·
Introduction
ties and conversions for a broad range of substrate types.[1–3]
However, the subsequent lack of studies indicate others
have been reluctant to develop and apply these catalysts,
probably because they are not particularly accessible, tend
to require 5 mol% loadings, use high hydrogen pressures,
and are very air-sensitive. The data obtained by using zirco-
nium-based systems to reduce tetrasubstituted arylalkenes is
still the best reported to date,[3] but subsequent research on
hydrogenations of tri- and disubstituted alkenes has focused
on iridium-based catalysts. In many cases, these are easier to
prepare, moderately air-stable, require only ambient or
slightly elevated hydrogen pressures, and are useful at lower
catalyst loadings. Thus, using Crabtree’s catalyst A as a con-
ceptual template,[4] Pfaltz and co-workers pioneered the ap-
plication of ligands with N,P-coordinating groups[5–7] in this
area, particularly in phosphine oxazoline complexes like
B.[6,8–14] Others,[14–16] including our group,[17] subsequently de-
signed and tested similar complexes. More recently, studies
from our laboratories demonstrated that N-heterocyclic car-
bene–oxazoline ligands like C also could be used to reduce
From a synthetic perspective, asymmetric hydrogenation of
largely unfunctionalized, aryl-substituted monoenes is a ma-
turing field. About a decade ago two catalyst types had
emerged that inspired a flurry of interest in this area. One
of these was titanium/zirconium catalysts from the Buchwald
group, which were shown to give excellent enantioselectivi-
[a] X. Cui, Y. Fan, M. B. Hall, Prof. K. Burgess
Department of Chemistry, Texas A & M University
P.O. Box 30012, College Station, Texas 77842 (USA)
Fax : (+1)979-845-8839
E-mail: burgess@tamu.edu
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author. The Supporting
Information contains the synthetic procedures for 2, 4, and 5 provid-
ing authentic samples for assignments of absolute configurations; de-
tails of the kinetic, NMR, and catalyst screening experiments; and co-
ordinates from the DFT calculations for P–W.
Chem. Eur. J. 2005, 11, 6859 – 6868
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
6859

and ruthenium–BINAP catalysts. Besides these examples,
Burk and co-workers had explored rhodium-mediated hy-
drogenation of the substrates F (and isomers);[22] only the
a,b-double bond was reduced giving the products G, so in a
sense these are archetypical reductions of dehydroamino
acids, that is, of well-studied, appropriately functionalized,
monoenes. We have not been able to find any other exam-
ples of asymmetric reductions of dienes prior to our work.
As an initial study, we chose 2,3-diphenylbutadiene (1) as
a substrate. It was predictable that stereocontrol for reduc-
tion of this particular diene would be difficult because each
tri- and disubstituted arylalkenes with good conversions and
enantioselectivities.[18,19] The outcome of all these studies is
that some iridium-based catalysts are known to mediate re-
duction of a variety of common, unfunctionalized, trisubsti-
tuted arylalkenes, with high conversions and enantioface se-
lectivities. Further modifications to the ligand designs are
only likely to give incremental improvements for iridium-
catalyzed hydrogenations of these particular substrates.
Despite the interest in iridium-catalyzed hydrogenations
of arylalkenes, there are some important unmet goals in the
area. Two of these specifically relate to the research de-
scribed here: 1) reductions of largely unfunctionalized al-
kenes and 2) elucidation of mechanistic features of the reac-
tion. All the work on iridium-mediated hydrogenations with
“Crabtree-like” catalysts has focused on aryl-substituted
monoenes, and the mechanism(s) of these reactions remain
largely unexplored.
We have identified a logical progression in the field to be
asymmetric hydrogenations of dienes (and polyenes). These
tend to be more difficult substrates to explore than mono-
enes, because both relative and absolute stereochemistries
must be controlled if two or more chiral centers are formed.
However, the rewards are greater if this can be achieved.
This is because the chirons that could be generated are
more complex, less accessible through other routes, but still
potentially useful in stereoselective syntheses.
alkene unit in it is 1,1-disubstituted, and reduction of this
type of alkene tends to be difficult to achieve with high face
selectivities. In any case, the products, isomers of 2,3-diphe-
nylbutane, have no evident immediate synthetic applica-
tions. Nevertheless, we regarded this as a useful starting
point for our investigations, because so many unknowns sur-
rounded the hydrogenation of even this simple diene. Those
unknowns can be highlighted by the following questions.
Could the reaction be manipulated to give the optically
active product 3 with high enantiomeric excess? If so, what
is the diastereoselectivity of the process? What are the rela-
tive rates of hydrogenation of the two double bonds, and do
the monoenes 2 accumulate in the reaction? Related to this,
what are the kinetic factors that control the overall stereose-
lectivity of the reaction? Is the reaction complicated by
double bond migration to give the tetrasubstituted alkene
4? Can any inferences be drawn from these studies regard-
ing the mechanism of iridium-mediated hydrogenations of
aryl-substituted monoenes? In the event, all these issues can
be addressed, at least to some extent. Our investigations of
2,3-diphenylbutadiene (1), which we initially anticipated
would be relatively superficial,[23] evolved into an in-depth
and revealing study.
Surprisingly, the literature reveals little prior art with re-
spect to asymmetric hydrogenations of dienes of any kind.
Reduction of the diacid D had been explored, with moder-
ate success,[20] while conversions and enantioselectivities for
the di(phosphine oxide) E were found to be considerably
better.[21] Both these studies featured functionalized dienes
Results and Discussion
The strategy in this project was as follows. Stereochemical
assignments were made for the products and intermediates
of the reduction process, then the reaction was followed by
means of gas chromatography (GC) to develop a kinetic
profile of the relative rates of conversion. These observa-
tions were supported by a competition experiment to reveal
the origin of a small amount of meso-3 that formed in the
early stage of the reaction, and by NMR analyses of cata-
lyst/hydrogen, catalyst/substrate mixtures. A series of experi-
ments was then performed using different stirring speeds to
explore if hydrogen diffusion effects can alter the outcome
of the reaction, then dependencies on catalyst concentration
6860
www.chemeurj.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6859 – 6868

Asymmetric Catalysis
FULL PAPER
and hydrogen pressure were investigated. All these experi-
ments were performed with a carbene–oxazoline catalyst
that had previously proved to be a preferred one for asym-
metric hydrogenations of aryl-substituted monoenes. The
last phase of the study was to screen a small library of alter-
native carbene–oxazoline catalysts to probe for increased
stereoselectivities.
In parallel with the experiments described above, an ex-
tensive series of density functional theory (DFT) calcula-
tions were performed to simulate a preferred reaction mech-
anism. Basically, relative energies of a series of diastereo-
meric intermediates were calculated to formulate a model of
the preferred reaction path and exclude relatively high-
energy routes to the products.
Kinetic profile of the reaction under typical conditions: Two
approaches were used to establish the absolute stereoche-
mistries of compounds 2 and 3. First, an authentic sample of
2 was prepared by a Wittig reaction on the corresponding
ketone, which was in turn made using an asymmetric synthe-
sis.[24] Second, two literature reports describe close and sig-
nificant optical rotations for compound 3,[25,26] hence it was
possible to determine the stereochemistry of the material
formed in the test reaction by means of polarimetry and cor-
relation. These two approaches were complementary and
led to the same conclusions.
Figure 1 gives the reaction conditions used to develop a
kinetic profile of the hydrogenation process, the relative
rates determined from GC analyses, and the absolute stereo-
chemistries of the compounds involved. This experiment
was repeated several times and gave consistent results.
After an induction period of about 4 min, the reaction
proceeds at a steady rate indicative of zero-order consump-
tion of diene. In the first 475 min of the hydrogenation, until
all the diene 1 is consumed, the monoenes 2 and ent-2 are
the major products with the former prevailing, but only in
an enantiomeric excess of approximately 12%. It is the
events after the diene is consumed that largely govern the
enantiomeric excess of the product 3 formed at the end of
the reaction. Two matched/mismatched pairs should be con-
sidered at the critical point when all the diene is consumed.
The catalyst interacts constructively with substrate 2 to give
meso-3, and ent-3 is only formed slowly in that reaction.
Conversely, the favored product from ent-2 is the optically
active product 3, and meso-3 only forms from this substrate
at about half the rate. The reaction generates an enantio-
meric excess of 3, because the major alkene 2 formed when
the diene is consumed is rapidly drained off to meso-3; si-
multaneously, the minor enantiomer of the alkene inter-
mediate, ent-2, preferentially gives 3 that emerges as the
preferred enantiomer from the reaction.
Figure 1. a) The test reaction; b) hydrogenation of 2,3-diphenylbutadiene
1, with rates in units of 10À4 molLÀ1 minÀ1; and c) concentrations of the
reaction components as the reaction proceeds.
the second phase of the reaction; and 3) the enantiomeric
excess of 2 formed in the first phase of the reaction is not
constant. These observations are now discussed in more
depth.
Competition experiment to probe dissociation of the sub-
strate from the metal: Formation of meso-3 in the first
phase of the reaction could occur through two consecutive
addition reactions of H2 to the diene without dissociation of
the substrate from the metal. Alternatively, partial hydroge-
nation of 1 could occur, followed by dissociation to free 2
and ent-2 then recombination of these with the metal to give
meso-3. These two possibilities are expressed in Figure 2a. A
competition experiment was devised to distinguish between
these two pathways. This was not straightforward, because,
by definition, the diene has to be present in the first phase
of the reaction. Consequently, conditions were designed, as
Several features of the data shown in Figure 1 are inter-
esting. These are 1) a significant concentration of meso-3,
but not ent-3, is formed in the first phase of the reaction; 2)
small quantities of the double-bond migration product 4 are
formed in the first phase of the reaction, but its Z isomer is
not formed, and the concentration of 4 remains constant in
Chem. Eur. J. 2005, 11, 6859 – 6868
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6861

K. Burgess, M. B. Hall et al.
Figure 3. a) Concentration of alkene 4 under the conditions shown in Fig-
ure 1a; b) enantiomeric excess of 2 throughout the reaction, shown in
green, ee%=(2Àent-2)/(2+ent-2)×100; and the enantiomeric excess of 2
adjusted for production of 4 from only ent-2, shown in pink, ee%=
(2À(ent-2+4))/(2+(ent-2+4))×100 as indicated in part c).
Figure 2. a) Simple representation of dissociative and nondissociative
routes to meso-3 in the first phase of the reaction and b) a competition
experiment showing that the nondissociative mechanism is favored, but
both are operative.
shown in Figure 2b, whereby 2 was generated, then an
excess of the tolyl derivative 5 was added. Hydrogenation of
this mixture gave a disproportionately large amount of
meso-3. This was not due to faster reaction of rac-2 than
rac-5, because control experiments showed the opposite;
rac-5 under our standard reaction conditions was hydrogen-
ated slightly faster than rac-2. Thus the overall conclusion is
that both the nondissociative and the dissociative pathways
are operative for the formation of meso-3 in the first phase
of the reaction, but the nondissociative mechanism is domi-
nant.
The other two unusual features of the kinetic data, that is,
in the first phase of the reaction the small quantities of the
double-bond migration product 4 formed and the variation
of the enantiomeric excess of 2 seem to be related. Evidence
for this assertion is as follows. The concentration of the tet-
rasubstituted alkene 4 varied as shown in Figure 3a. The en-
antiomeric excess of the monoene 2 varied as shown in Fig-
ure 3b (green line) during the same time period. We were
curious about this unusual variation in enantiomeric excess
of 2 and realized that if it were assumed that the tetrasubsti-
tuted alkene 4 arose solely from ent-2 then the ee of 2 cor-
rected for this is almost constant in the first phase of the re-
action (Figure 3b, pink line). We conclude that under these
particular conditions 4 may be formed preferentially from
ent-2.
NMR experiments: NMR data were collected in two solvent
systems: CDCl3 and CD3C6H5. Deuterochloroform was
chosen because it is a common solvent for these hydrogena-
tion reactions, (though it was not the one used in the kinetic
experiments), and because no solubility issues arose. Deuter-
ated toluene was used to correspond with the medium used
for the kinetic experiments, but in this case material tended
to separate from the solution as an oil (vide infra). In both
solvents, treatment of catalyst C with 25 equivalents of
6862
www.chemeurj.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6859 – 6868

Asymmetric Catalysis
FULL PAPER
diene 1 for 12 h at 258C gave no significant change in the
1H NMR spectrum indicating that hydrogen is required to
displace the cyclooctadiene (COD) ligand.
duction products 2 were observed to co-exist in the same so-
lution. These data indicate that the reaction proceeds some-
what differently under these two sets of conditions. It is pos-
sible that solvent coordination effects of toluene are suffi-
cient to stabilize intermediates in the process, whereas de-
composition events under similar conditions in CDCl3 were
more prevalent.
For [D8]toluene as solvent, the hydrogenation of diene 1
(1 atm H2, 20 min, diene present from t=0) gave proton
NMR spectra that were different to those described above.
Just as in the kinetic experiments, only significant quantities
of the monohydrogenated intermediates 2 were observed.
Resonances associated with the ligand and hydridic region
were complicated and difficult to interpret.
When catalyst C in CDCl3 was treated with hydrogen
under atmospheric pressure in the absence of diene, the
1H NMR spectrum of the solution changed significantly over
a period of approximately 40 min. During this time the reso-
nances attributed to the COD ligand disappeared and a
peak corresponding to cyclooctane appeared, but the reso-
nances due to the catalyst were indicative of formation of a
mixture of products. No significant hydrogenation occurred
when diene was added to this solution. We infer from these
experiments that the catalyst is rapidly deactivated in essen-
tially non-coordinating solvents in the absence of alkene
substrate. It seems likely that degradation to oligometallic
clusters occurred under these conditions, similar to Crab-
tree’s catalysts[4,27,28] and their chiral phosphine–oxazoline
analogues.[29] However, at the present time we have no evi-
dence to support this for the particular case of catalyst C;
our efforts to isolate and crystallize these species using simi-
lar conditions have been unsuccessful.
When the hydrogenation reaction of diene 1 was per-
formed with diene present from t=0 in CDCl3, the proton
NMR spectra were different to those described above. They
were complicated to interpret, except the hydride region
showed only one resonance (d=À15.4 ppm); this signal was
not observed in the reaction of catalysts with hydrogen
alone. Thus the overall conclusion from these experiments is
that both diene (or alkene) and hydrogen are required to
form the active catalyst in these reactions.
Hydrogen diffusion effects: Until this point, the emphasis of
these studies had been on the relative rates of reaction
under a given set of conditions. The next stage was to con-
sider relationships of the reaction rate to catalyst and hydro-
gen concentration. However, work by Blackmond and
others has highlighted situations in which applied hydrogen
pressure and concentration of hydrogen in solution are not
always related in a simple way for catalytic hydrogenation
reactions.[31,32] Pfalz et al. have also noted that H2 diffusion
effects might be important for catalysts formed from their
phosphine–oxazolines for cases in which the substrate reacts
relatively fast.[9,33] Further, there is evidence that, unlike
some rhodium-catalyzed hydrogenation reactions in which
oxidative addition of hydrogen tends to be rate-limiting, the
corresponding reactions of cationic Ir+ complexes tend to
be fast even at À808C.[34,35] This is to be expected due to the
In [D8]tolune, the catalyst behaved differently. After
treatment with 1 atm of hydrogen in the absence of diene
for 1 h, the COD ligand is reduced to free cyclooctane, and
À
greater thermodynamic stability of Ir H bonds relative to
À
Rh H bonds. Thus, there is a possibility that diffusion of hy-
1
an oily material separated from the solution. The H NMR
drogen is rate-limiting. The likelihood that hydrogen diffu-
sion across a gas–liquid interface becomes rate-limiting in-
creases for substrates that are intrinsically fast to react, es-
pecially at increased catalyst concentrations and relatively
low hydrogen pressures.
spectrum of the complex remaining in solution gave two
coupled hydridic resonances at d=À16.00 and À15.97 ppm
(J=5.7 Hz). After prolonged treatment with hydrogen (e.g.,
5 h), even more material separated out of the solution, and
the resonances of the dissolved material diminished. We
infer from this experiment, that an iridium dihydride com-
Diene 1 is less reactive than most monoenes that we have
tested in the presence of catalyst C. For instance, the diene
requires reaction times of the order of 10 h, whereas (E)-
1,2-diphenylpropene under similar conditions, but with less
catalyst, required reaction times of only 2 h for complete
conversion. This factor implies that hydrogen diffusion rates
might not be so important for diene 1 as they are for mono-
enes.
Experiments to test if the diffusion of hydrogen is rate-
limiting or not can feature experiments with different cata-
lyst loadings, vessels with different gas–liquid contact sur-
face areas, and/or experiments with different stir speeds. It
was not practical to vary the catalyst loadings in these ex-
periments, because the reaction becomes inconveniently
slow when significantly less than 1 mol% of catalyst is used,
and competitive deactivation of the catalyst after prolonged
reaction times also becomes an issue. Conversely, experi-
ments with high catalyst loadings are not truly representa-
tive of the conditions used, and would consume large
À
plex with two inequivalent Ir H moieties was formed. This
is consistent with observations from Pfaltz et al. on their
iridium phosphine–oxazoline complexes treated with dihy-
drogen in THF.[30] Unlike those phosphine–oxazoline com-
plexes, derivatives of catalyst C do not have 31P nuclei, so
further structural assignments were relatively difficult.
When diene 1 was added to a solution of catalyst C that had
been pretreated with H2 (1 atm, 5 h), hydrogenation did
occur, that is, contrary to what had been observed for the
CDCl3 solution. However, the product distribution was not
the same as that observed in the kinetic experiments (com-
pare with Figure 1). In the experiments outlined in Figure 1,
only a small amount of completely reduced product, 2,3-di-
phenylbutane (3), was observed until all the starting materi-
al was consumed. However, for the experiment here, in
which the catalyst was pre-treated with H2, significant
amounts of diene 1, product 3, and the intermediate half-re-
Chem. Eur. J. 2005, 11, 6859 – 6868
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6863

K. Burgess, M. B. Hall et al.
Table 2. Control experiments for H2 diffusion effects.[a]
amounts of the valuable catalyst C. For these reasons, we
elected to used stir speeds as a prime method for determin-
ing the importance of hydrogen diffusion effects on the re-
duction of diene 1. Different vessel geometries were used in
some cases to accentuate the effects.
Entry
Stir speed
[rpm]
Conv (1)
[%]
yield (3)
[%]
ee (3)
dr (3)
[%]
[%]
1
2
500
100
100
95
95
93
93
70/30
71/69
>1100[b]
[a] Data shown are the average of two experiments. [b] Maximum stir
speed of magnetic stirrer was used.
Table 1 shows rates of consumption of diene as a function
of various catalyst concentrations and stir speeds in the
same reaction vessel. Each value is based on 5–6 data
(just as in the kinetic experiments), and the stir speeds were
varied between two extremes. Table 2, entries 1–2 show that
the product distribution and stereoselectivities did not vary
with stir speed under these conditions, confirming that H2
diffusion was not determinant in the stereoselectivities
under these conditions. The product distribution and stereo-
selectivities are very close to those observed in the kinetic
analyses (Figure 1).
Slight differences in the data obtained from the experi-
ments outlined in Figure 1 and those in Table 2 may indicate
small variations in the relative rates would be observed
under conditions under which mass-transfer effects were a
distant possibility. However, the influence of mass transfer
under the conditions shown in Figure 1 is largely inconse-
quential to the overall conclusions regarding the origin of
the optical purity of the product.
Table 1. Variance in rate of consumption of diene with catalyst concen-
tration and stir speed.
Entry
Cat. conc.
[molLÀ1
stir speed
[rpm]
Àd[diene]/dt
]
[10À3 molLÀ1 minÀ1
]
H2 diffusion effects unimportant
1
2
3
4
5
0.0075
0.0075
0.0075
0.0075
0.0075
300
500
700
900
1100
1.9
2.0
1.9
2.0
2.0
borderline region
6
7
8
0.01
0.01
0.01
500
700
1100
2.2
2.6
2.6
Dependence on catalyst concentration: Rates of consump-
tion of the diene 1 were measured at different catalyst con-
centrations under the conditions indicated in Figure 4a. The
same reaction vessel and stir speed were used throughout,
so the influence of diffusion rates was constant. The rate of
reaction was linearly related to the catalyst concentration:
the reaction rate is first-order with respect to the catalyst
under these conditions.
H2 diffusion effects important
9
10
11
0.015
0.015
0.015
300
700
1100
2.5
3.4
4.2
points, and nearly all the experiments were performed twice.
The data show that at the catalyst concentration 0.0075m
there is no evidence that stir speed effects the measured
rate of consumption of the diene (entries 1–5). However,
when the catalyst concentration was increased 0.015m then
H2 diffusion effects were clearly significant (entries 9–11).
The kinetic studies performed early in this work featured
catalyst concentrations of 0.010m. This was a deliberate
choice. We desired to keep the reactions rates high enough
that the kinetics could be measured conveniently, without
encountering excessive complications from H2 diffusion ef-
fects. The data shown in entries 6–8 of Table 1 indicate that
this is a borderline region (the vessel used here was the
same as for the kinetic experiments). Consequently, a series
of control experiments were undertaken to check that H2
diffusion effects do not significantly affect the outcome of
the reaction under these types of conditions. These are now
described below.
Dependence on hydrogen pressure: The experiments with
variation of stirring speeds proved that diffusion rates influ-
ence the enantioselective hydrogenation of diene 1. Further,
we have already reported that enantiomeric excesses in hy-
drogenations of some, but not all, aryl-substituted monoenes
can vary with hydrogen pressure.[19] This observation is indi-
cative of a change in the relative rates in a given pathway,
or perhaps even a more fundamental mechanistic shift. Such
mechanistic variations include transitions between situations
in which the diffusion of hydrogen is and is not a controlling
factor. Significantly, one of the aryl-substituted monoenes
for which ee and H2 pressure were related was the 1,1-disub-
stituted substrate H, a compound that is structurally similar
to diene 1. These considerations indicate that only limited
inferences can be drawn from the effects of pressure on the
rate of the hydrogenation reaction. Nevertheless variance of
enantioselectivity with pressure was measured for diene 1
(Figure 4b).
The first set of experiments used a vessel with a larger
gas–liquid interface than the one used in the kinetic experi-
ments, that is, a geometry that does not favor rate-limiting
H2 diffusion. The catalyst concentration was kept at 0.010m
The data shown in Figure 4b for the elevated pressures
are the composite sets of experiments (5–6 data points each)
repeated four times. They show that the rate of consumption
of diene under these conditions appears to be linearly relat-
6864
www.chemeurj.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6859 – 6868

Asymmetric Catalysis
FULL PAPER
are relevant here. Our confidence in that modeling study
was reinforced by the good correspondence between calcu-
lated and experimental values of enantiomeric excesses.
Our studies on the asymmetric hydrogenation of arylal-
kenes with catalyst C indicated the catalytic cycles favored
the key steps shown in Figure 5. After the COD ligand is re-
duced, the iridium(iii)–dihydride–dihydrogen complex I is
formed. The key steps are rupture of the dihydrogen ligands
À
À
and formation of a C H bond and a C Ir bond via the tran-
sition state J to give the Ir5+ complex K, and the reductive
elimination process from K to M via L. These steps are cal-
culated to have similar energetics, so either could be turn-
over-limiting, but only the second one is irreversible. Thus,
the cycle features transitions between Ir3+ and Ir5+ oxida-
tion states but not Ir+.
Another important aspect of the original modeling study
was as follows. The relative energies of the intermediates I,
K, and M in the arylalkene hydrogenation corresponded to
the relative energies of the transition states J and L without
crossover. This result is also significant, because the inter-
mediates are easier to model than the transition states. Con-
sequently, the relative energies of only the intermediates
were modeled in the work described here.
Before the DFT calculations outlined above, two of us
had hypothesized that the hydrogenation of 2,3-diphenylbu-
tadiene (1) perhaps involved Ir+/Ir3+ cycling via an h4-diene
complex like N in the first phase of the reaction, because
the diene precludes formation of a dihydride–dihydrogen
complex, then, after all the diene is consumed, an Ir3+/Ir5+
pathway dominates.[23] The previous set of DFT calculations
(for aryl-substituted alkenes) and the ones performed for
this study now allow us to revise this hypothesis.
Modeling studies indicate that the diene coordinated in a
transoid configuration as in O (not cisoid as in N) is prefer-
red (Figure 5b). Optimized geometries of the intermediates
reveal the reason for this: the transoid form has the phenyl
group comfortably resting behind the adamantyl group of
the ligand (Figure 5c), whereas the cisoid isomer incurs a se-
rious repulsive interaction between an isopropyl group on
the ligand and one of the phenyl substituents on the diene
(Figure 5d). We also observed that one of the two p-coordi-
nated alkene bonds in the transoid form is twisted away
from the metal. This seems likely to be the alkene that dis-
sociates to accommodate a hydrogen for reduction to occur
through an Ir3+/Ir5+ pathway. Two possibilities for such an
alkene dissociation/H2 addition are depicted in Figure 6a.
The h4-diene complexes (P and R) are significantly more
stable than the corresponding h2-forms (Q and S). This ob-
servation accounts for the fact that hydrogenation of the
diene in the first phase of the reaction is slower than reduc-
tion of the monoenes in the second one. It also explains why
the catalyst preferentially coordinates and reduces diene
even when monoene is present in the reaction mixture. It is
unnecessary to calculate all the intermediates in the reac-
tion, and there are so many possible conformational states
beyond this stage that it is not practical to simulate them all
in any case. However, none of the trial calculations that
Figure 4. a) Variation of reaction rate with catalyst concentration. Each
data point is the average of two sets of experiments (each of 5–6 data
points), values obtained from each set of experiments were identical to
within 0.1 molLÀ1. b) Variation of the rate of the reaction shown with ap-
plied hydrogen pressure. Each data point for the elevated pressures is the
average of four sets of experiments (each of 5–6 data points); error bars
shown indicate one standard deviation.
ed to the applied pressure. However, as indicated above,
these observations do not prove first-order rate dependence
on hydrogen in the catalytic cycle.
Modeling of the reaction: We recently communicated a de-
tailed DFT computational study of arylalkene hydrogena-
tion mediated by our catalyst C.[36] Both the approach to
that modeling study and the conclusions that were reached
Chem. Eur. J. 2005, 11, 6859 – 6868
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6865

K. Burgess, M. B. Hall et al.
Figure 5. a) The mechanism of hydrogenation of selected arylalkenes by catalyst C as modeled using DFT calculations; b) cisoid (N) and transoid (O) h4-
diene complexes; c) calculated structure of the transoid intermediate and d) that of the disfavored cisoid intermediate.
were performed indicated that Ir+/Ir3+ cycling occurs, so
that part of our original hypothesis now seems unlikely. Fig-
ure 6a illustrates the calculated relative energies for the two
diastereomeric transoid h4-diene complexes, and the corre-
sponding two diastereomeric h2-forms after complexation of
H2. The most stable dihydride dihydrogen complex would
lead to the alkene 2 that was actually observed in the reac-
tion (Figure 6b). The relative free energies of the complexed
monoene intermediates were also calculated. These relative
energies are in accord with the measured reaction rates
from the monoene intermediates, that is, formation of meso-
3 is faster than ent-3 from 2, and 3 forms faster then meso-3
from ent-2. This supports our assumption that the relative
energies of these intermediates correspond to those of the
transition states (not calculated) without crossover.
using catalyst C was somewhat depressed (75 vs 87% ee at
1 atm). There are indications in the data collected that a su-
perior catalyst for hydrogenation of substrate 1 could be
identified. For instance, the catalysts represented in en-
tries 2, 3, and 5 gave higher enantioselectivities and larger
3:meso-3 ratios than catalyst C under the same conditions.
These leads were not pursued further, since the objective of
these studies was not to maximize the optical purity of sub-
strate 3. Nevertheless, the data indicate that catalyst C is not
necessarily the best possible one for hydrogenations of
dienes, even though it was conspicuously good for hydroge-
nations of aryl-substituted monoenes. This information may
be valuable in studies that are ongoing in these laboratories
to investigate hydrogenations of dienes and polyenes that
give more synthetically useful products.
Screening other catalysts: A small library of alternative cat-
alysts was screened for hydrogenation of the diene 1
(Table 3). To accelerate the process, these tests were per-
formed by using 10 atm H2 pressure. Under these conditions,
the enantiomeric excess of the product 3 that was obtained
Conclusion
Assimilation of the experimental and theoretical data accu-
mulated in this study provided a reasonably lucid view of
6866
www.chemeurj.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6859 – 6868

Asymmetric Catalysis
FULL PAPER
Figure 6. a) Calculated relative free energies (kcalmolÀ1, P as origin) of the four probable intermediates formed in the hydrogenation of diene 1; b) cor-
relation of the h2-forms with the monoenes 2; and c) relative free energies (V as origin) of the complexed monoene intermediates and the reduced prod-
ucts that would arise from these.
Table 3. Screening of alternative catalysts for hydrogenation of diene 1.
The hydrogenation of 2,3-di-
phenylbutadiene (1) with cata-
lyst C is initially zero-order in
diene and probably first-order
in catalyst; however, transfer of
hydrogen across the gas–liquid
interface becomes an issue at
Entry
R1
R2[a]
Conv
[%]
products [%]
ee [%]
(R,R)-3
3:meso-3
high catalyst concentrations and
under conditions that favor rel-
atively slow diffusion of hydro-
gen into the solution. DFT cal-
2
4
3
1
2
3
4
5
6
7
8
9
2,6-iPr2C6H3
2,6-Et2C6H3
2,4,6-Me3C6H2
3,5-tBu2-4-MeOC6H2
2,5-tBu2C6H2
2,6-iPr2C6H3
2,6-iPr2C6H3
Ph2CH
1-Ad
1-Ad
1-Ad
1-Ad
1-Ad
3,5-tBu2C6H3
tBu
tBu
1-Ad
1-Ad
>99
>99
>99
>99
>99
99
13
15
1.0
0.5
0
0
0
0
0
0.4
3
4
0.4
0.2
15
5
4
1
5
8
2
0.1
0.2
0
85
95
96
99
95
90
8
11
0.7
0.2
75
79
79
86
91
-68
28:72
36:64
41:59
23:77
31:69
8:92
culations indicate that an Ir3+
/
Ir5+ pathway prevails via trans-
oid h4-diene complexes. These
must dissociate one alkene
group to give an h2-complex
before the hydrogenation can
proceed. These studies do not
provide all the details of the re-
action mechanism. For instance,
[b]
[b]
–
–
–
–
–
–
–
–
[b]
[b]
[b]
[b]
Ph2CH
tBu
[b]
[b]
10
[a]1-Ad=1-adamantyl. [b] Not determined due to small quantities involved.
the course of events that lead to products in the hydrogena-
tion of substrate 1. They also indicate how the mechanism
of the reaction can vary, or appear to vary, if some particular
experimental parameters are changed. This combined practi-
cal and computational approach also point to some key fea-
tures that are likely to be general to hydrogenations of aryl-
alkene substrates by iridium catalysts.
it is quite likely that h3-allyl complexes feature in the mech-
anism of this reaction, and neither the kinetic or DFT ex-
periments can support or refute this idea. Nevertheless, an
outline of the mechanistic picture for these reactions is now
clear.
Chem. Eur. J. 2005, 11, 6859 – 6868
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6867

K. Burgess, M. B. Hall et al.
[18] M. T. Powell, D.-R. Hou, M. C. Perry, X. Cui, K. Burgess, J. Am.
Chem. Soc. 2001, 123, 8878–8879.
Computational Methods
[19] M. C. Perry, X. Cui, M. T. Powell, D.-R. Hou, J. H. Reibenspies, K.
Burgess, J. Am. Chem. Soc. 2003, 125, 113–123; X. Cui, K. Burgess,
Chem. Rev., in press.
[20] H. Muramatsu, H. Kawano, Y. Ishii, M. Saburi, Y. Uchida, J. Chem.
Soc. Chem. Commun. 1989, 769–770.
[21] V. Beghetto, U. Matteoli, A. Serivanti, Chem. Commun. 2000, 155–
156.
[22] M. J. Burk, K. M. Bedingfield, W. F. Kiesman, J. G. Allen, Tetrahe-
dron Lett. 1999, 40, 3093–3096.
Briefly, with respect to the computational approach, we showed that
there was good agreement with data generated using the B3LYP and
PBE methods for arylalkene hydrogenation mediated by catalyst C.[36]
That was extremely important because a comprehensive set of B3LYP
calculations for the complete structures of the catalyst C and of the
alkene substrate would have been far too computationally expensive
(even using our state-of-the-art facilities), but they are practical using the
PBE method.
Methods used in the computational experiments described here build
upon the previous approach described above. Thus all the calculations
were carried out using the Gaussian 03[37] implementation of PBE[38] den-
sity functional theories, which for these systems produces results similar
to B3LYP.[39,40] The basis sets used were LANL2DZ with ECP for Ir[41,42]
and D95v for all other elements.[43] This switch from B3LYP to PBE was
made because density-fitting functions[44,45] can be used with PBE and
make these calculations much faster; hence the problem was computa-
tionally accessible. Throughout, intermediates were modeled but not the
transition states. All structures were fully optimized, and analytical fre-
quency calculations were performed at the same theoretical level on each
structure to ensure a minimum was achieved. Zero-point energies and
thermodynamic functions were computed for 298.15 K and 1 atm. Only
the cation of the complex was modeled.
[23] X. Cui, K. Burgess, J. Am. Chem. Soc. 2003, 125, 14212–14213.
[24] O. Roy, A. Riahi, F. Henin, J. Muzart, Eur. J. Org. Chem. 2002,
3986–3994.
[25] N. D. Berova, B. J. Kurtev, Tetrahedron 1969, 25, 2301–2311.
[26] H. H. Richmond, E. J. Underhill, A. G. Brook, G. F. Wright, J. Am.
Chem. Soc. 1947, 69, 937–939.
[27] H.-H. Wang, A. L. Casalnuovo, B. J. Johnson, A. M. Mueting, L. H.
Pignolet, Inorg. Chem. 1988, 27, 325–331.
[28] H.-H. Wang, L. H. Pignolet, Inorg. Chem. 1980, 19, 1470–1480.
[29] S. P. Smidt, A. Pfaltz, Organometallics 2003, 22, 1000–1009.
[30] C. Mazet, S. P. Smidt, M. Meuwly, A. Pfaltz, J. Am. Chem. Soc.
2004, 126, 14176–14181.
[31] Y. Sun, R. N. Landau, J. Wang, C. LeBlond, D. G. Blackmond, J.
Am. Chem. Soc. 1996, 118, 1348–1353.
[32] Y. Sun, J. Wang, C. LeBlond, R. A. Reamer, J. Laquidara, J. R.
Sowa, Jr., D. G. Blackmond, J. Organomet. Chem. 1997, 548, 65–72.
[33] S. P. Smidt, N. Zimmermann, M. Studer, A. Pfaltz, Chem. Eur. J.
2004, 10, 4685–4693.
[34] B. F. M. Kimmich, E. Somsook, C. R. Landis, J. Am. Chem. Soc.
1998, 120, 10115–10125.
[35] R. H. Crabtree, H. Felkin, G. E. Morris, J. Chem. Soc. Chem.
Commun. 1976, 716–717.
[36] Y. Fan, X. Cui, K. Burgess, M. B. Hall, J. Am. Chem. Soc. 2004, 126,
16688–16689.
[37] Gaussian 03, Revision B.05, M. J. Frisch, G. W. Trucks, H. B. Schle-
gel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, Montgomery J. A.
Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar,
J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega,
G. A. Petersson, H. Nakatsuki, M. Hada, M. Ehara, K. Toyota, R.
Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao,
H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross,
V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann,
O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y.
Ayala, K. Morokuma, G. A. voth, P. Salvador, J. J. Dannenberg,
V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C. Strain, O.
Farkas, D. K. Malick, A. D. Rabuck, B. B. Stefano, G. Liu, A. Lia-
shenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith,
Al-M. A. Laham, C. Y. Peng, A. Nanyakkara, M. Challacombe,
P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonzalez, J. A.
Pople, Gaussian, Inc.: Willingford, CT, 2004.
Acknowledgements
Financial support for this work was provided by The National Science
Foundation (CHE-0456449, 98-00184, DMR 02-16275, and 05-18074) and
The Robert Welch Foundation (A1121 and A0648). We thank Johnson
Matthey plc for help with some precious metals, Dr Shane Tichy and the
TAMU/LBMS-Applications Laboratory for MS support, and NMR Lab-
oratory at Texas A&M University, supported by a grant from the Nation-
al Science Foundation (DBI-9970232) and the Texas A&M University
System.
[1] R. B. Grossman, R. A. Doyle, S. L. Buchwald, Organometallics 1991,
10, 1501–1505.
[2] R. D. Broene, S. L. Buchwald, J. Am. Chem. Soc. 1993, 115, 12569–
12570.
[3] M. V. Troutman, D. H. Appella, S. L. Buchwald, J. Am. Chem. Soc.
1999, 121, 4916–4917.
[4] R. Crabtree, Acc. Chem. Res. 1979, 12, 331–337.
[5] A. Pfaltz, Synlett 1999, 835–842.
[6] T. Bunlaksananusorn, K. Polborn, P. Knochel, Angew. Chem. 2003,
115, 4071–4073; Angew. Chem. Int. Ed. 2003, 42, 3941–3943.
[7] G. Helmchen, A. Pfaltz, Acc. Chem. Res. 2000, 33, 336–345.
[8] A. Lightfoot, P. Schnider, A. Pfaltz, Angew. Chem. 1998, 110, 3047–
3050; Angew. Chem. Int. Ed. 1998, 37, 2897–2899.
[38] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77,
3865–3868.
[39] A. D. Becke, J. Chem. Phys. 1993, 98, 5648–5652.
[40] C. Lee, W. Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785–789.
[41] P. J. Hay, W. R. Wadt, J. Chem. Phys. 1985, 82, 270–283.
[42] W. R. Wadt, P. J. Hay, J. Chem. Phys. 1985, 82, 299–310.
[43] T. H. Dunning, P. J. Hay in Modern Theoretical Chemistry, Vol. 3
(Ed.: H. F. Schaefer III), New York, 1976.
[9] D. G. Blackmond, A. Lightfoot, A. Pfaltz, T. Rosner, P. Schnider, N.
Zimmermann, Chirality 2000, 12, 442–449.
[10] J. Blankenstein, A. Pfaltz, Angew. Chem. 2001, 113, 4577–4579;
Angew. Chem. Int. Ed. 2001, 40, 4445–4447.
[11] D. Drago, P. S. Pregosin, A. Pfaltz, Chem. Commun. 2002, 286–287.
[12] F. Menges, A. Pfaltz, Adv. Synth. Catal. 2002, 344, 40–44.
[13] F. Menges, M. Neuburger, A. Pfaltz, Org. Lett. 2002, 4, 4713–4716.
[14] P. G. Cozzi, F. Menges, S. Kaiser, Synlett 2003, 833–836.
[15] G. Xu, S. Gilbertson, Tetrahedron Lett. 2003, 44, 953–955.
[16] W. Tang, W. Wang, X. Zhang, Angew. Chem. 2003, 115, 973–976;
Angew. Chem. Int. Ed. 2003, 42, 943–946.
[44] B. I. Dunlap, J. Mol. Struct. 2000, 529, 37–40.
[45] B. I. Dunlap, J. Chem. Phys. 1983, 78, 3140–3142.
[17] D.-R. Hou, J. H. Reibenspies, T. J. Colacot, K. Burgess, Chem. Eur.
J. 2001, 7, 5391–5400.
Received: July 2, 2005
Published online: September 15, 2005
6868
www.chemeurj.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6859 – 6868