Mechanistic insights into the phosphine-free RuCp*-diamine-catalyzed hydrogenation of aryl ketones: Experimental and theoretical evidence for an alcohol-mediated dihydrogen activation

Article abstract of DOI:10.1021/ja051920q

The commercially available chiral diamine quincorine-amine, originally derived from quinine, was found to be a highly active catalyst for ruthenium-catalyzed hydrogenation of ketones. The complex formed between the quincorine-amine, containing both a prim

Full text of DOI:10.1021/ja051920q

Published on Web 10/08/2005
Mechanistic Insights into the Phosphine-Free
RuCp*-Diamine-Catalyzed Hydrogenation of Aryl Ketones:
Experimental and Theoretical Evidence for an
Alcohol-Mediated Dihydrogen Activation
Christian Hedberg,† Klas Ka¨llstro¨m,† Per I. Arvidsson,† Peter Brandt*,‡ and
Pher G. Andersson*,†
Contribution from Organic Chemistry, Department of Chemistry, Uppsala UniVersity, Box 599,
BMC S-751 24 Uppsala, Sweden, and Department of Chemistry, BioVitrum AB,
112 76 Stockholm, Sweden
Received March 25, 2005; E-mail: Pher.Andersson@kemi.uu.se; peter.brandt@biovitrum.com
Abstract: The commercially available chiral diamine quincorine-amine, originally derived from quinine, was
found to be a highly active catalyst for ruthenium-catalyzed hydrogenation of ketones. The complex formed
between the quincorine-amine, containing both a primary and a quinuclidine amino function, and RuCp*Cl
catalyzes the hydrogenation of aromatic and aliphatic ketones in up to 90% ee approximately 24 times
faster than previously reported Ru-diamine complexes. The pseudo-enantiomer of the quincorine-amine,
i.e., quincoridine-amine, also showed high activity; however, the enantioselectivities obtained with this catalyst
were lower. The reason for the lower, but opposite stereoselectivity seen with the quincoridine-amine, as
compared to the quincorine-amine, was rationalized by a kinetic and computational study of the mechanism
of the reaction. The theoretical calculations also revealed a significantly lower activation barrier for the
alcohol-mediated split of dihydrogen, as compared to the nonalcohol-mediated process, a finding of utmost
implication also for the diphosphine/diamine-mediated enantioselective hydrogenation of ketones.
Introduction
arene or Rh/Ir-cyclopentadiene complexes. The most efficient
catalysts for hydrogenation of functionalized ketones with H2
are based on Rh and Ru diphosphine catalysts.4 The method of
choice for reducing unfunctionalized ketones, on the other hand,
is Noyori’s newly developed Ru-diphosphine/diamine catalyst
system.5
Catalytic asymmetric synthesis is, according to a newly
published report,1 the most important synthetic method for the
production of chiral products. Definitely, catalytic enantiose-
lective hydrogenation is the reaction most widely used for this
purpose, both industrially and in the laboratory.2 This interest
has made homogeneous hydrogenation of functionalized olefins
the best-studied enantioselective catalytic reaction.
Of comparable practical interest to the enantioselective
reduction of CdC bonds is the hydrogenation of CdO and Cd
N bonds. Two methods have been developed for homogeneous
enantioselective hydrogenation of ketones. These are transfer
hydrogenation using organic sources of dihydrogen and the
direct use of molecular hydrogen. In the transfer hydrogenation
reaction the proton is transferred from 2-propanol or formic acid
to the ketone through an iridium, ruthenium, or rhodium
catalyst.3 The most efficient ligands used for this reaction are
based on â-amino alcohols or â-diamines together with Ru-
Although both Ru-diphospine/diamine complexes 1 (e.g., P-P
) (S)-BINAP, N-N ) (S,S)-1,2-diphenylethylenediamine) and
the widely used Ru-arene/diamine complex RuCl(TsDPEN)-
(3) (a) Zassinovich, G.; Mestroni, G.; Gladiali, S. Chem. ReV. 1992, 92, 1051.
(b) Hashiguchi, S.; Fujii, A.; Noyori, R. J. Synth. Org. Chem. (Jpn) 1996,
54, 818. (c) Noyori R.; Hashiguchi, S. Acc. Chem. Res. 1997, 30, 97. (d)
Everaere, K.; Mortreux, A.; Carpentier, J.-F. AdV. Synth. Catal. 2003, 345,
67. (e) Palmer, M. J.; Wills, M. Tetrahedron: Asymmetry 1999, 10, 2045.
(f) Nordin, S. F. M.; Roth, P.; Tarnai, T.; Alonso, D. A.; Brandt, P.;
Andersson, P. G. Chem. Eur. J. 2001, 7, 1431.
† Uppsala University.
‡ Biovitrum AB.
(1) Rouhi, A. M. Chem., Eng. News 2004, 82, 47.
(4) For a recent review see: Noyori, R.; Kitamura, M.; Ohkuma, T. Proc.
Natl. Acad. Sci. U.S.A. 2004, 101, 5356.
(2) For reviews, see: (a) Ohkuma, T.; Noyori, R. In ComprehensiVe Asymmetric
Catalysis; Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.; Springer: Berlin,
1999; Vol. 1, pp 199-246. (b) Blaser, H.-U.; Springer, F. In ComprehensiVe
Asymmetric Catalysis; Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.;
Springer: Berlin, 1999; Vol. 1, pp 247-265. (c) Noyori, R. In Asymmetric
Catalysis in Organic Synthesis; Wiley: New York, 1994; pp 16-94. (d)
Noyori, R.; Ohkuma, Angew. Chem., Int. Ed. 2001, 40, 40. (e) T. Blaser,
H.-U.; Malan, C.; Pugin, B.; Spindler, F.; Steiner, H.; Studer, M. AdV. Synth.
Catal. 2003, 345, 103.
(5) (a) Ohkuma, T.; Ooka, H.; Hashiguchi, S.; Ikariya, T.; Noyori, R. J. Am.
Chem. Soc. 1995, 117, 2675. (b) Ohkuma, T.; Ooka, H.; Ikariya, T.; Noyori,
R. J. Am. Chem. Soc. 1995, 117, 10417. (c) Doucet, H.; Ohkuma, T.;
Murata, K.; Yokozawa, T.; Kozawa, M.; Katayama, E.; England, A. F.;
Ikariya, T.; Noyori, R. Angew. Chem., Int. Ed. 1998, 37, 1703. (d) Ohkuma,
T.; Ishii, D.; Takeno, H.; Noyori, R. J. Am. Chem. Soc. 2000, 122, 6510.
(e) Ohkuma, T.; Koizumo, M.; Muniz, K.; Hilt, G.; Kabuto, C.; Noyori,
R. J. Am. Chem. Soc. 2002, 124, 6508.
9
10.1021/ja051920q CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 15083-15090
15083

A R T I C L E S
Hedberg et al.
Scheme 1. Cinchona Alkaloid-Derived Diamines 4 and 5 Used as Ligands for Cp*Ru-Mediated Asymmetric Hydrogenation
(η6-arene) (TsDPEN ) N-(p-toluenesulfonyl)-(S,S)-1,2-diphe-
nylethylenediamine) 2 contain a characteristic NH functionality,
the catalytic activity of these two catalyst systems differs
tremendously. The Ru-arene/Tsdiamine complex, typically used
for transfer hydrogenation, hardly reacts with H2,6 while the
Ru-diphospine/diamine complex does so easily.
Recently, Ikariya et al. demonstrated that RuCp*-1,2-diamine
complexes of general type 3 (Cp* ) pentamethylcyclopenta-
dienyl, η5-C5-(CH3)5), which are isoelectronic to the transfer
hydrogenation catalyst 2, were active catalysts for the hydro-
genation of ketones with H2 as hydrogen source.7 This finding
is one of the few known example of a homogeneous hydrogena-
tion catalyst that is capable of activating molecular hydrogen
without having at least one phosphine ligand around the metal
center.8 This study also concluded that the most active catalyst
was obtained with a diamine having one tertiary and one primary
amino function.
On the basis of these findings, we figured that an even more
potent catalyst could be obtained by increasing the Lewis
basicity of the tertiary nitrogen center of the ligand.9 One
possibility to accomplish this would be to incorporate a
quinuclidine function in the ligand. Gratifyingly, two pseudo-
enantiomeric 1,2-diamines containing a quinuclidine function
and a primary amino function are commercially available. The
quincorine-amine (QCI-amine) 4 and quincoridine-amine (QCD-
amine) 5 are derived from the Cinchona alkaloids quinine and
quinidine, respectively), Scheme 1.10
to good enantioselectivities for most substrates investigated. A
kinetic and computational study was also undertaken. In addition
to providing a working selectivity model, this study revealed a
significant lowering of the activation energy for the dihydrogen
split when a coordinating alcohol molecule mediated this
process. Additional calculations showed that the same mecha-
nism is the lowest energy pathway also in the Noyori type of
diphosphine/diamine system.
Computational Details
General Procedures. Geometries of all substrates were calculated
using the Jaguar program14 using the B3LYP hybrid density functional,15
together with the LACVP** and LACV3p+** basis sets. Normal-mode
analysis revealed one imaginary frequency for each transition state.
LACVP in Jaguar defines a combination of the LANL2DZ basis set16
for ruthenium and the 6-31G basis set for other atoms. LACVP implies
the use of an effective core potential for 28 core electrons of ruthenium
and a (5s,6p,4d) primitive basis contracted to [3s,3p,2d]. Final energies
were retrieved from single-point calculations at B3LYP/LACV3p+**.
LACV3p+** differs from LACVP** by using the 6-311+G** basis
set in place of 6-31G**.
Model Systems. Several model systems have been used to cover
different aspects of the reaction (Figure 1). To start with, in the
These â-diamines, containing four stereogenic centers each,
including a fixed stereogenic (S)-configured N-bridgehead, have
previously been used as ligands in Ir-catalyzed asymmetric
transfer hydrogenation of ketones11 and as chiral acylation
catalysts.12 Other ligands based on the quincorine and quinco-
ridine scaffold, inlcuding P,N-ligands, have also been de-
scribed.13 However, to the best of our knowledge, no QCI/QCD-
based ligands have previously been used in asymmetric
hydrogenation reactions.
In this report, we demonstrate how the catalytic activity of
the RuCp*-1,2-diamine complex, used for hydrogenation of
ketones with H2, may be substantially enhanced by the quinu-
clidine-based ligands 4 and 5. These ligands also furnish fair
Figure 1. Catalytic systems used in the computational study.
investigation of the activation and splitting of H2, Cp* was replaced
by Cp. Also, the vinyl group in the ligands 4 and 5 was replaced by a
hydrogen atom. In the study of regioselectivity in the addition of H2 to
Ru and in the study of the enantioselectivity in ketone reduction, the
real systems were used.
(6) Haack, K.; Hashigushi, J. S.; Fujii, A.; Ikaraya, T.; Noyori, R. Angew.
Chem., Int. Ed. Engl. 1997, 36, 285.
(7) Ito, M.; Hirakawa, M.; Murata, K.; Ikaraya, T. Organometallics 2001, 20,
379.
(8) (a) Blum, Y.; Czarkie, D.; Rahamim, Y.; Shvo, Y. Organometallics 1985,
4, 1459. (b) Vicente, C.; Shulpin, G. B.; Moreno, B.; Sabo-Etienne, S.;
Chaudret, B. J. Mol. Catal. A: Chem. 1995, 98, L5.
(9) Ohwada, T.; Hirao, H.; Ogawa, A. J. Org. Chem. 2004, 69, 7486.
(10) Hoffmann, H. M. R.; Frackenpohl, J. E. J. Org. Chem. 2004 (21), 4293.
(11) Hartikka, A.; Modin, S. A.; Andersson, P. G.; Arvidsson, P. I. Org. Biomol.
Chem. 2003, 1, 2522.
(12) Kuendig, E. P.; Lomberget, T.; Bragg, R.; Poulard, C.; Bernardinelli, G.
Chem. Commun. 2004, 1548.
(13) Saluzzo, C.; Breuzard, J.; Pellet-Rostaing, S.; Vallet, M.; Le Guyader, F.;
Lemaire, M. J. Organomet. Chem. 2002, 643, 98.
Results and Discussion
Mechanism. A number of efficient Ru catalysts have been
developed for the asymmetric reduction of ketones into optically
(14) Jaguar 4.2; Schro¨dinger, Inc.: Portland, OR, 1991-2000.
(15) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (b) Lee, C.; Yang, W.;
Parr, R. G. Phys. ReV. B 1988, 37, 785.
(16) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299.
9
15084 J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005

Alcohol-Mediated Dihydrogen Activation
A R T I C L E S
Figure 2. Potential energy surface at the B3LYP/LACV3P+**//B3LYP/LACVP** level including zero-point corrections for addition of H2 to the 16-
electron Ru complex and subsequent heterolytic cleavage in the diamine C1 model system. Left: Nonalcohol-mediated pathway. Right: alcohol-mediated
pathway.
active alcohols, either by hydrogenation or by transfer hydro-
genation.17 It has been postulated that both these reactions, when
the ligand contains at least one ruthenium-coordinated sp3 NH
donor atom, proceed via a concerted mechanism where a Ru-
hydride and a NH proton are transferred simultaneously from
the catalyst to the CdO bond.18 This mechanism is very likely
the mechanism of the hydrogenation of ketones by 3, and we
assume the same mechanism is valid for catalysts C4 and C5.
Initially, we looked into the steps where dihydrogen is activated
by ruthenium. This part of the reaction is of particular interest
considering the low number of phosphine-free catalysts able to
use dihydrogen gas as a source of hydrogen atoms. In addition,
some reported calculations have shown relatively high activation
energies (up to 25.2 kcal/mol) for similar heterolytic splittings
of H2.18,19 A computed activation energy for a diamine/
diphosphine complex has been reported as 13.4 kcal/mol,
whereas the experimental activation enthalpy for this reaction
in benzene was reported to be only 7.6-8.6 kcal/mol.18c
Ikaraya and co-workers have proposed a mechanism of
heterolytic splitting of H2 in which a solvent 2-propanol
molecule participates in the transition state.7 This mechanism
has recently received support in the literature.20 Morris and co-
workers reported18c that the hydrogenation of acetophenone
catalyzed by a diamine diphosphine complex was independent
of the ketone and alcohol concentrations. However, the same
group has also reported that the reactions with related bis-
(phosphine)21 and tetradentate20b systems are much faster in
2-propanol than in benzene. This effect was presumed to be
caused by the higher dielectric constant of 2-propanol compared
to benzene and by hydrogen bonds from the solvent to the
polarized transition state.
In this part of the study, we have investigated dihydrogen
coordination and cleavage both for the diamine system used
here and for the Noyori type of diamine/diphosphine system,
with and without participation of solvent alcohol. The reaction
paths studied are depicted in Figures 2 and 3, and the results
are summarized in Tables 1 and 2.
The results in Table 1 indicate similar potential energy
surfaces for the systems studied, except for C7, which has a
higher affinity for dihydrogen. The rate-determining step for
the formation of ruthenium hydride complexes is the heterolytic
cleavage of η2-coordinated dihydrogen. The activation energy
for this step is around 17 kcal/mol. Experimentally, the ruthen-
ium hydride complexes have been postulated to be formed revers
ibly for C1.7 This reversible formation of ruthenium hydrides
was used to rationalize a H/D scrambling observed in the study.
We find formation of ruthenium hydride complexes exothermic
by about -5 kcal/mol, and thus, the barrier for the reversible
reaction, i.e., the decomposition of the ruthenium hydride to
form H2 from P/C1, will be substantial (21.8 kcal/mol).
(17) For a recent review see: Clapham, S.; Hadzovic, A.; Morris, R. H. Coord.
Chem. ReV. 2004, 248, 2201.
(18) (a) Yamakawa, M.; Ito, H.; Noyori, R. J. Am. Chem. Soc. 2000, 122, 1466.
(b) Noyori, R.; Yamakawa, M.; Hashigushi, S. J. Org. Chem. 2001, 66,
7931. (c) Abdur-Rashid, K.; Clapham, S. E.; Hadzovic, A.; Harvey, J. N.;
Lough, A. J.; Morris, R. H. J. Am. Chem. Soc. 2002, 124, 15104.
(19) (a) Crabtree, R. H.; Siegbahn, P. E. M.; Eisenstein, O.; Rheingold, A. L.;
Koetzle, T. Acc. Chem. Res. 1996, 29, 348. (b) Shubina, E. S.; Belkova,
N. V.; Epstein, L. M. J. Organomet. Chem. 1997, 536, 17. (c) Crabtree, R.
H. Science 1998, 282, 2000. (d) Guari, Y.; Ayllon, J. A.; Sabo-Etienne,
S.; Chaudret, B.; Hessen, B. Inorg. Chem. 1998, 37, 640. (e) Grundemann,
S.; Ulrich, S.; Limbach, H.-H.; Golubev, N. S.; Denisov, G. S.; Epstein, L.
M.; Sabo-Etienne, S.; Chaudret, B. Inorg. Chem. 1999, 38, 2550. (f) Fryzuk,
M. D.; MacNeil, P. A.; Rettig, S. J. J. Am. Chem. Soc. 1987, 109, 2803.
(20) (a) Sandoval, C. A.; Ohkuma, T.; Mun˜iz, K.; Noyori, R. J. Am. Chem.
Soc. 2003, 125, 13490. (b) Rautenstrauch, V.; Hoang-Cong, X.; Churland,
R.; Abdur-Rashid, K.; Morris, R. H. Chem. Eur. J. 2003, 9, 4954. (c) Casey,
C. P.; Johnson, J. B.; Singer, S. W.; Cui, Q. J. Am. Chem. Soc. 2005, 127,
3100.
(21) Abbel, R.; Abdur-Rashid, K.; Faatz, M.; Hadzovic, A.; Lough, A. J.; Morris
R. H. J. Am. Chem. Soc. 2005, 127, 1870.
9
J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005 15085

A R T I C L E S
Hedberg et al.
Figure 3. Comparison of reaction pathways for the solvent-nonassisted and -assisted heterolytic dihydrogen cleavage in diamine/diphospine model system
C7 (potential energy surface at the B3LYP/LACV3P+**//B3LYP/LACVP** level including zero-point corrections). Left: Nonalcohol-mediated pathway.
Right: alcohol-mediated pathway.
Table 1. Calculated Energies (kcal/mol) for Dihydrogen
reference. Hydrogen bonding between the ruthenium complex
Coordination and Cleavage without Participation of Solvent
and alcohol will substantially increase the reversibility of the
reaction with an activation barrier for the formation of a
hydrogen-hydrogen bond around 18 kcal/mol. However, con-
sidering the activation free energy for the formation of IR′OH
from PR′OH with C1 (24.5 kcal/mol), the mechanism proposed
by Ikaraya and co-workers for hydrogen/deuterium exchange
could not be corroborated.
Morris and co-workers have reported a calculated activation
energy for the heterolytic cleavage of H2 by diamine/diphosphine
system C7 of 13.4 kcal/mol. Our results indicate a considerable
decreased activation energy with an active participation of the
solvent, from 10.9 to 5.3 kcal/mol for the conversion of I/C7
to P/C7. The importance of an alcoholic solvent for the diamine/
diphosphine system was also observed in Noyori’s initial
report,5a and later Casey found a similar role of the alcoholic
solvent for the Shvo catalyst.20c For the diamine system C1,
the same activation energy decreased from 6.2 to 4.4 kcal/mol.
Here, the accelerating effect could instead be attributed to a
stabilization of the dihydrogen complex.
As mentioned above, Ikaraya and co-workers have concluded
from a series of experiments using deuterated i-PrOH and D2
that the addition of dihydrogen over the Ru-N bond is
reversible.7 Thus, the release of i-PrOH and the subsequent
coordination/reduction of substrate was concluded to be slower
than re-formation of the 16-electron complex. According to our
calculations, the activation energy for reduction of acetophenone
by PR′OH/C1 is 14.2 kcal/mol (19.1 kcal/mol in free energy)
and the reverse reaction, i.e., formation of a H-H bond, has an
activation energy of 18.3 kcal/mol (24.5 kcal/mol in free
energy).
To summarize the calculations, we propose a reaction
mechanism starting from R. This species coordinates a solvent
alcohol molecule, and this in turn coordinates the dihydrogen,
a step greatly facilitated by the hydrogen bond to the ruthenium-
coordinated amide nitrogen. The last step is a heterolytic splitting
of H2 mediated by the alcohol hydroxyl group. Thereby, the
Alcohola
model system
species
C1b
C1
C1c
C2
C3
C3
′
C7
R
0
0
0
0
0
0
0
TS1
I
TS2
P
12.3
12.8
18.2
-5.2
12.4
11.0
17.2
-4.6
17.5
13.3
20.3
-4.3
11.2
10.0
15.9
-6.1
11.5
10.9
18.6
-4.3
12.0
11.2
17.6
-4.2
n.d.
6.9
17.8
0.4
a Energies at B3LYP/LACV3P+**//B3LYP/LACVP** including zero-
point corrections. b Energies at B3LYP/LACVP** including zero-point
corrections. c Free energies at B3LYP/LACV3P+**//B3LYP/LACVP**.
Table 2. Calculated Energies (kcal/mol) for Dihydrogen
Coordination and Cleavage with Active Participation of Solvent
Alcohola
model system
c,e
species
C1b,d
C1d
C1f
C2d
C3d
C3c,d
C5e
C5
′
C7d
R
0
0
0
0
0
0
0
0
0
RR′OH
TS1R′OH -0.2
IR′OH
TS2R′OH
PR′OH
-7.8 -4.8 3.2 -5.2 -5.1 -4.5 -7.8 -12.3 -2.4
2.9 14.4
0.1 10.3 -0.9
4.6 19.5 3.6
2.1
3.2
0.0
4.5
4.3
0.4 -6.5 -7.3 -3.4
5.8 -3.2 2.5 1.9
n.d.
n.d n.d.
-1.8
0.8
-17.6 -13.7 -5.0 -14.0 -13.8 -13.1 -19.2 -16.9 -9.8
a Energies at B3LYP/LACV3P+**//B3LYP/LACVP** including zero-
point corrections. b Energies at B3LYP/LACVP** including zero-point
corrections. c Addition of H2 cis to the methyleneamine group. d R′OH )
MeOH. e R′OH ) i-PrOH. f Free energies at B3LYP/LACV3P+**//B3LYP/
LACVP**.
Table 2 shows the results from the calculations on the alcohol-
mediated activation and cleavage of H2. The calculated transi-
tion state structures for complexes C7 and C5 are shown in
Figure 4. The rate-determining step is again found to be the
cleavage of the η2-coordinated dihydrogen, but now, the
activation energies are substantially decreased. In particular, the
alcohol stabilizes the coordination of H2 and the subsequent
change in coordination geometry around the metal center. As a
consequence, the activation energy is decreased by ca. 7-10
kcal/mol using the hydrogen-bonded alcohol complex as a
9
15086 J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005

Alcohol-Mediated Dihydrogen Activation
A R T I C L E S
Figure 4. Transition state structures for the alcohol-mediated heterolytic cleavage of coordinated dihydrogen for complexes C7 and C5.
Scheme 2. Preparation of RuCp*-Diamine Complex C5
Table 3. Rates of the Reduction of Acetophenone Using
Complexes C5, C8, and C9 at Different Concentrationsa
1
complex
conc (mM)
rate (µ )
mol s-
C5
C5
C5
C8
C9
4.75
1.9
0.95
4.75
4.75
2.10
0.64
0.17
0.086
0.39
catalyst is ready to replace the coordinated solvent by a substrate
molecule and perform the hydrogenation of the carbonyl. We
assume breaking and making of hydrogen bonds to solvent and
substrate is a fast and solvent-assisted process in 2-propanol.
In absolute numbers, the free energy of activation for the
heterolytic clevage of H2 by C1 is slightly lower with an active
participation of alcohol. This is largely due to the entropic effect
on the free energy caused by solvent coordination. This effect
is likely overestimated as the reaction is run in 2-propanol as
solvent and not in the gas phase. The reduction of ketones by
the ruthenium hydride species was found to be slightly lower
in free energy than the heterolytic clevage of H2 (19.1 compared
to 19.5 kcal/mol); however, the low solubility of hydrogen in
2-propanol will make the heterolytic cleavage of H2 rate
determining. Thus, the overall reaction is expected to be zero-
order in substrate, and R will be the resting state of the catalyst.
Kinetics. According to our calculations and proposed reaction
mechanism, the reaction catalyzed by C5 would be first order
with respect to catalyst and hydrogen pressure. This mechanism
is in conflict with a reversible formation of a ruthenium hydride
complex. To investigate this hypothesis, we prepared complex
C5 by reacting quincoridine 5 with [RuCp*Cl]4 in 2-propanol
in the presence of potassium tert-butoxide for 30 min, Scheme
2, and utilized this complex for the reduction of acetophenone.
As shown in Table 3 and Figure 5a, the reaction showed the
expected first-order dependence on both catalyst concentration
and hydrogen pressure.We also confirmed the expected zero-
order dependence in substrate, Figure 5b.
a The substrate concentration was 0.475 M in each case. Reactions were
performed at 25 °C at 25 bar pressure of H2.
of the quinuclidine nitrogen present in C5 should yield a more
potent catalyst than C8 and C9. The importance of an alcoholic
solvent for this reaction was supported by performing the
hydrogenation in a nonprotic solvent (THF), where the hydro-
genation was found to be substantially slower, as was also
observed in Ikariya’s initial report.7 However, it should be noted
that the solubility of H2 varies with solvents. While this
quinuclidine-based diamine system is the most reactive phos-
phine-free ruthenium catalyst reported to date, it is still less
efficient than Noyori type of catalysts for the hydrogenation of
prochiral ketones.5
Selectivity. Encouraged by the high reaction rates obtained
for the quincoridine-Ru complex C5, we next wanted to examine
the enantioselectivity of this catalyst in the reduction of prochiral
ketones. The results of these reactions are shown in Table 4.
Interestingly, and in sharp contrast with many other asym-
metric hydrogenation catalysts, aryl ketones having bulky alkyl
groups reacted with higher enantioselectivity than those having
smaller substituents. While acetophenone (Table 4, entry 1)
was hydrogenated to (S)-phenylethanol in 75% ee, the sterically
more demanding 2,2-dimethylpropiophenone and valerophenone
gave the corresponding alcohols in 90% ee (Table 4, entries 3
and 5).
In an earlier study using Ru(p-cymene)(2-azanorbornyl-3-
methanols) as catalysts, we found that ∆∆Eq(R-S)Exp was
negatively correlated with the width of the alkyl substituent of
the substrate.22 The parameter used to describe this property
was the STERIMOL B1 parameter. In the present study, no
To compare the reactivity between catalysts, the same reaction
was also performed with previously reported7 complexes C8
and C9. Table 3 shows that catalyst C5 is approximately 24
times faster than complex C8, in agreement with the postulation
made at the beginning of this study that the increased basicity
(22) Brandt, P.; Roth, P.; Andersson, P. G. J. Org. Chem. 2004, 69, 4885.
9
J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005 15087

A R T I C L E S
Hedberg et al.
Figure 5. (a) Observed rate of the reaction vs hydrogen pressure (slope ) 0.098 µmol bar-1 s-1, intercept ) -0.3 µmol s-1, R2 ) 0.995) for the C5-
catalyzed reduction of acetophenone. The hydrogenations were performed at 25 °C as 0.475 M solutions of substrate and 4.75 mM solutions of catalyst. The
negative intercept may well be explained by a different rate-determining step at low pressure. (b) Plot of initial rates as a function of substrate concentration;
the slope shows reaction order zero with respect to acetophenone.
Table 4. Asymmetric Hydrogenation of Aryl Alkyl Ketones
Catalyzed by C5a
abs
configc
entry
Ar
R
conv (%)
ee (%)b
time (min)
1
2
3
4
5
6
7
8
C6H5
C6H5
C6H5
C6H5
C6H5
o-CH3C6H4
m-CH3C6H4
p-CH3C6H4
CH3
+99
99
99
99
99
+99
+99
99
+99
99
75
74
90
81
90
82
71
76
79
86
62
54
44
50
81
11
30
30
30
30
15
15
11
11
14
10
11
40
42
30
(S)
(S)
(S)
(S)
(S)
(S)
(S)
(S)
(S)
(S)
(S)
(S)
(S)
(S)
(S)
i-C3H7
t-C4H9
n-C4H9
i-C4H9
CH3
Figure 6. Results of the hydrogenation of aryl ketones. Labels refer to the
p-substituent on the aryl ring. ln(er) vs σp (er ) enantiomeric ratio) (slope
) -1.26, intercept ) 1.85, R2 ) 0.993).
CH3
CH3
9
p-CH3OC6H4 CH3
o-CH3OC6H4 CH3
10
11
12
13
14
15
p-CF3C6H4
p-Cl-C6H4
3,4,5-FH2C6
2,6-FH2C6
1-naphthyl
CH3
CH3
CH3
CH3
CH3
99
99
99
99
+99
a All reactions were complete within 42 min. Experimental conditions:
Substrate-to-catalyst ratio 100:1, 25 bar H2, 0.475 M in i-PrOH according
to the general procedure in the Experimental Section. b Enantiomeric excess
determined by chiral GC or HPLC. c Absolute configuration correlated from
previous reported GC or HPLC elution order.
Figure 7. Coordination and alcohol-assisted cleavage of dihydrogen to
16e Ru complex R/C5.
such straightforward trends could be observed. This is indicative
of a more complex interaction at the interface between the
substrate and the catalyst. Despite the reversed geometrical
preference in the favored TS (see also the selectivity model
presented below), where the aryl group of the substrate is facing
the alkyl part of the ligand, there is an electronic effect of similar
size to that for regular Ru-(amino alcohol)(arene) catalysts.22
That is, the more electron rich the substrate, the better the
selectivity. This correlation is depicted in Figure 6. The observed
trend implies either of two alternatives, namely, that there is a
repulsive electrostatic component between the aryl of the
substrate and the formally negatively charged Cp* ligand of
the catalyst. Or, there is an attractive electrostatic interaction
between the alkyl part of the ligand and the aryl of the substrate.
This observation could potentially be used to find substrates
electrostatically more suitable for this class of catalysts.
The diastereomeric diamine ligand 4, derived from quincorine,
had reactivity similar to that of C5 and gave products of opposite
configuration but in a markedly less selective manner (99%
conversion, ee ) 41% (R)-phenylethanol).
Selectivity Model. To see how well the postulated mechanism
correlates with the experimental results, we compared the selec-
tivities obtained for various ketones with the catalyst structures
derived from the DFT calculations. This analysis is greatly
facilitated by the presence of the η5-bonded Cp* ligand, since
it occupies three coordination sites on the octahedral Ru and
thus reduces the number of possible diastereomeric complexes.
In the initial step, the alcohol-mediated addition of dihydrogen
can take place from two sides of the 16e complex R/C5, leading
to two diastereomeric complexes. Calculations favor formation
of complex Pi-PrOH/C5, which has the lowest activation energy
for the heterolytic, alcohol-mediated cleavage of the coordinated
dihydrogen (8.8 kcal/mol for Pi-PrOH/C5 compared to 11.6 kcal/
mol for Pi-PrOH/C5′, Figure 7).
The reduction step is found to involve a concerted, but
asynchronous, addition of a hydride to the carbonyl carbon
followed by the transfer of a proton from nitrogen to oxygen.
The calculated and experimentally observed selectivities for the
9
15088 J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005

Alcohol-Mediated Dihydrogen Activation
A R T I C L E S
Table 5. Comparison between Experimental and Calculated Enantioselectivities (energies in kcal/mol) for the Diamine-Ru-Cp*-Catalyzed
Hydrogenation of Acetophenone and 2,2-Dimethylpropiophenone
b
c
entry
catalyst
substrate
∆∆Eq(R
−
S)a
ee, % (R or S)Exp
∆∆Eq(R
−S)Exp
calcd
1
2
3
4
C5
C8
C4
C5
PhCOMe
PhCOMe
PhCOMe
PhCOtBu
1.9
-2.0
-1.1
5.6
75 (S)
72 (R)
41 (R)
90 (S)
1.15
-1.07
-0.52
1.74
a Refers to the absolute configuration of the alcohol product. b All calculations were performed on the full structures. Energies reported are from the
B3LYP/LACV3P+**//B3LYP/LACVP** including zero-point correction. c Calculated from the observed ee.
Figure 8. Interactions between the substrate, ligand, and the Cp* ring in
the C5-catalyzed addition of hydride to the re-face of the ketone.
Figure 10. Interactions between acetophenone, the quincorine-derived
ligand 4, and the Cp* ring in the addition to the si-face of the ketone. Values
given in parentheses refer to the reduction of acetophenone using C5.
phenyl ring of the substrate (Figure 10). This interaction leads
to destabilization of the TS giving si-face addition, and the
calculated energy difference between the two diastereomeric TSs
is now only 1.2 kcal/mol (compared to 2.0 kcal/mol for the
quincoridine complex).
Finally, the increased enantioselectivity observed for the bulky
2,2-dimethylpropiophenone is also in agreement with the DFT
Figure 9. Interactions between the substrate, ligand, and the Cp* ring in
model, which predicts the si-addition to be favored by 5.5 kcal/
the C5-catalyzed addition of hydride to the si-face of the ketone.
mol. This can be attributed to (i) a reasonable fit of the t-Bu
group into the pocket formed between the H-Ru-N-H moiety
and the Cp* ligand (Figure 8) and (ii) the collision between the
t-Bu group and the quinuclidine in the TS, giving si-face addition
(Figure 9).
enantioselective reduction step for different catalysts and
substrates are summarized in Table 5.
The experimental and theoretically predicted selectivities
agree well for C5, showing that the model may be used to
provide both quantitative and qualitative information concerning
the stereochemical outcome of the reaction. The addition to the
Conclusions
re-face of the ketone (Figure 8) is favored by 2.2 kcal/mol,
probably as this will avoid a steric clash between the phenyl
group and the Cp* ligand, being present in the corresponding
addition to the si-face, Figure 9.
As seen in Table 5, the model also correctly predicts the si-
face addition by C8 to be favored over re-face addition, which
is in agreement with the results published earlier by Ikariya.7
The lower selectivities obtained for complex C4 originate
from a collision between the vinyl group of the ligand and the
In conclusion, we have shown that the complex formed
between [RuCp*Cl]4 and the commercially available chiral
diamine quincorine-amine, originally derived from quinine, is
a highly reactive catalyst for the phosphine-free enantioselective
hydrogenation of aryl ketones. A detailed mechanistic investiga-
tion revealed that the quinuclidine-based ligand is approximately
24 times more reactive than previously described catalysts, and
the enantioselectivities obtained are modest to good (up to 90%
ee). The straightforward methodology and the availability of
9
J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005 15089

A R T I C L E S
Hedberg et al.
all reagents imply that this method will find broad practical use.
Interestingly, the diastereomer of quincorine-amine, i.e., quin-
coridine-amine, also showed high activity; however, the enan-
tioselectivities obtained with this catalyst were lower. This
finding represents a rare exception where these diastereoisomers
do not function as pseudo-enantiomeric reagents for an asym-
metric reaction. The reason for the lower, but opposite stereo-
selectivity seen with the quincoridine-amine, as compared to
the quincorine-amine, could be rationalized through a compu-
tational study of the mechanism of the reaction. In addition to
providing a working selectivity model, these calculations also
revealed that the activation energy for the dihydrogen split
involved is significantly lowered when a solvent alcohol
molecule mediates the process. This finding is of utmost
importance, as it allows a rational design of a second-generation
phosphine-free catalysts; moreover, it provides an explanation
for the reported discrepancy between experimental and theoreti-
cal activation energies reported for Noyori’s diphosphine/
diamine-mediated enantioselective hydrogenation system, which
was computationally shown to involve the same alcohol-
mediated activation.
Acknowledgment. Financial support from Vetenskapsrådet
(The Swedish Research Council), SSF (The Swedish Foundation
for Strategic Research), and Solvias AG is gratefully acknowl-
edged. We also thank Dr. C. von Riesen and Buchler GmbH-
Chininfabrik Braunschweig for a generous gift of QCI- and
QCD-amine.
Supporting Information Available: General experimental
methods, details for the chiral separation methods, and coor-
dinates for all calculated structures. This material is available
free of charge via the Internet at http://pubs.acs.org.
JA051920Q
9
15090 J. AM. CHEM. SOC. VOL. 127, NO. 43, 2005