Asymmetrie transfer hydrogenation of ketones catalyzed by amino acid derived rhodium complexes: on the origin of enantioselectivity and enantioswitchability

Article abstract of DOI:10.1002/chem.200900548

Amino acid based thioamides, hydroxamic acids, and hydrazides have been evaluated as ligands in the rhodium-catalyzed asymmetric transfer hydrogenation of ketones in 2propanol. Catalysts containing thioamide ligands derived from L-valine were found to selectively generate the product with an R configuration (95 % ee), whereas the corresponding Lvaline-based hydroxamic acids or hydrazides facilitated the formation of the (S)-alcohols (97 and 91% ee, respectively). The catalytic reduction was examined by performing a structureactivity correlation investigation with differently functionalized or substituted ligands and the results obtained indicate that the major difference between the thioamide and hydroxamic acid based catalysts is the coordination mode of the ligands. Kinetic experiments were performed and the rate constants for the reduction reactions were determined by using rhodiumarene catalysts derived from amino acid thioamide and hydroxamic acid ligands. The data obtained show that the thioamide-based catalyst systems dem-onstrate a pseudo-first-order dependence on the substrate, whereas pseudozero-order dependence was observed for the hydroxamic acid containing catalysts. Furthermore, the kinetic experiments revealed that the rate-limiting steps of the two catalytic systems differ. From the data obtained in the structure-activity correlation investigation and along with the kinetic investigation it was concluded that the enantioswitchable nature of the catalysts studied originates from different ligand coordination, which affects the ratelimiting step of the catalytic reduction reaction.

Full text of DOI:10.1002/chem.200900548

FULL PAPER
DOI: 10.1002/chem.200900548
Asymmetric Transfer Hydrogenation of Ketones Catalyzed by Amino Acid
Derived Rhodium Complexes: On the Origin of Enantioselectivity and
Enantioswitchability
Katrin Ahlford,[a] Jesper Ekstrçm,[a] Alexey B. Zaitsev,[a] Per Ryberg,[b]
Lars Eriksson,[c] and Hans Adolfsson*[a]
Abstract: Amino acid based thio-
ACHTUNGTRENNUNGamides, hydroxamic acids, and hydra-
zides have been evaluated as ligands in
the rhodium-catalyzed asymmetric
transfer hydrogenation of ketones in 2-
propanol. Catalysts containing thio-
ligands and the results obtained indi-
cate that the major difference between
the thioamide and hydroxamic acid
based catalysts is the coordination
mode of the ligands. Kinetic experi-
ments were performed and the rate
constants for the reduction reactions
were determined by using rhodium–
arene catalysts derived from amino
acid thioamide and hydroxamic acid li-
gands. The data obtained show that the
thioamide-based catalyst systems dem-
onstrate a pseudo-first-order depend-
ence on the substrate, whereas pseudo-
zero-order dependence was observed
for the hydroxamic acid containing cat-
alysts. Furthermore, the kinetic experi-
ments revealed that the rate-limiting
steps of the two catalytic systems
differ. From the data obtained in the
structure–activity correlation investiga-
tion and along with the kinetic investi-
gation it was concluded that the enan-
tioswitchable nature of the catalysts
studied originates from different ligand
coordination, which affects the rate-
limiting step of the catalytic reduction
reaction.
ACHTUNGTRENNUNGamide ligands derived from l-valine
were found to selectively generate the
product with an R configuration (95%
ee), whereas the corresponding l-
valine-based hydroxamic acids or hy-
drazides facilitated the formation of
the (S)-alcohols (97 and 91% ee, re-
spectively). The catalytic reduction was
examined by performing a structure–
activity correlation investigation with
differently functionalized or substituted
Keywords: amino acids · asymmet-
ric catalysis · kinetics · reaction
mechanisms · rhodium
Introduction
numerous synthetic procedures leading to compounds with
interesting biologically activity. One particularly attractive,
mild, and highly selective method for ketone reductions is
the asymmetric transfer hydrogenation (ATH) protocol.[1,2]
In the ATH reaction, a prochiral ketone is selectively re-
duced by transfer of a hydride and a proton from a suitable
hydride donor, most often mediated by a chiral transition-
metal complex. When developing a catalytic system for the
ATH reaction there are three major aspects to consider:
1) the choice of hydride donor, 2) the transition-metal
source, and 3) the chiral ligand, which will enable the pro-
cess to proceed with good stereoselectivity. There are in
principle two different types of donors frequently used in
the ATH reaction. These can be classified as either reversi-
ble or irreversible donors, with 2-propanol being the classic
example of a reversible donor and formic acid a good exam-
ple of an irreversible donors. The use of formic acid, alkali
metal formate, or the triethylamine/formic acid (2:5) azeo-
trope (TEAF) is considered more advantageous because the
The asymmetric reduction of ketones to yield enantiomeri-
cally enriched secondary alcohols is a key transformation in
[a] K. Ahlford, Dr. J. Ekstrçm, Dr. A. B. Zaitsev, Prof. Dr. H. Adolfsson
Department of Organic Chemistry, Stockholm University
The Arrhenius Laboratory, 10691 Stockholm (Sweden)
Fax : (+46)8-154908
E-mail: hansa@organ.su.se
[b] Dr. P. Ryberg
Global Process R&D, AstraZeneca, 15185 Sçdertꢀlje (Sweden)
[c] Dr. L. Eriksson
Department of Physical Chemistry, Inorganic Chemistry and
Structural Chemistry
The Arrhenius Laboratory, 10691 Stockholm (Sweden)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200900548. Experimental proce-
dures for ligand and complex formation, catalytic and kinetic experi-
ments, and ligand characterization data.
Chem. Eur. J. 2009, 15, 11197 – 11209
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
11197

resulting carbon dioxide formed after the hydride/proton
transfer is thermodynamically much more stable, which
makes the reaction almost completely irreversible. However,
the downside is that only a narrow range of catalysts that
tolerate the use of formic acid as the hydride donor are
available. In contrast, ATH reactions performed under re-
versible conditions in secondary alcohols like 2-propanol
can be catalyzed by a wider range of transition-metal com-
plexes.[3] For an ATH process to favorably give the desired
product under reversible conditions, the substrate needs to
have an oxidation potential that differs from that of the hy-
dride donor.[4] Even so, the donor needs to be present in
large excess because the erosion of product enantioselectivi-
ty is often observed even if the equilibrium is in favor of the
substrate. There are several metal sources available that
have the ability to mediate the hydride transfer from the
donor to the substrate. Even if main-group metals like alu-
minium have historically been used in the transfer hydroge-
nation reaction,[5–7] todayꢂs catalysts of choice are transition-
metal complexes predominantly of ruthenium, rhodium, and
iridium.[1] In addition, a few highly interesting reports de-
scribing the use of iron com-
These catalytic systems behave in accordance with amino
alcohol systems and only show activity with alcohols as hy-
dride donors. When we investigated the catalytic activity
and selectivity of Ru/Rh complexes composed of the differ-
ent ligands 2–4 in the ATH reactions of acetophenone, we
obtained a most interesting result. By using ligands made
from natural (S)-amino acids, we found that catalysts de-
rived from pseudo-dipeptides (2) and hydroxamic acids (4)
selectively catalyzed the formation of the (S)-alcohol,[11a–f,h]
whereas complexes derived from the corresponding thio-
AHCTUNGTREGaNNUN mide ligands (3) gave the product with the R configura-
tion.[11g] Hence, the same amino acid scaffold can evidently
be used for the construction of catalysts that selectively
favor the formation of either of the product enantiomers,
simply by changing the nature of the amino acid C terminus.
Herein we present results that serve to further our under-
standing of the reasons behind the observed enantioswitch-
AHCTUNGTRENNUNG
Results and Discussion
plexes have recently been pub-
lished.[8] The most well-known
In our previous studies on the use of pseudo-dipeptides (2)
as ligands for ruthenium and rhodium in the ATH reaction
we were neither able to isolate nor obtain sufficient spectro-
scopic data to be able to conclude the structure of the pre-
and successful transition-metal
complex used to catalyze the
ATH reaction is unquestionably
the Noyori catalyst (1), which is
ACHTUNGTRENcNUNG atalyst or the active catalyst. Nevertheless, by using chemi-
formed
from
[{RuCl
(p-
cal methods we found that the nature of the ligandsꢂ func-
tional groups was strongly correlated to the activity and se-
lectivity of the catalysts.[11b] In fact, most changes made to
the structure of 2 led to a significant decrease in the catalyt-
ic activity. We found that it is essential that the N terminus
of the ligand is “protected” as a base-stable carbamate be-
cause catalysts formed with ligands containing either carbox-
amides, sulfonamides, or simply unprotected amines gave no
conversion to the product. The peptide bond, referred to as
the central amide, that is formed upon coupling of the 1,2-
amino alcohol to the amino acid proved to be essential for
catalytic activity. If the central amide was formed from a
secondary amine, hence creating a tertiary amide, the cata-
lytic activity was lost. The alcohol functionality at the C ter-
minus of the pseudo-dipeptide was equally important for
the activity of the catalyst, and when this group was convert-
ed into the corresponding methyl ether, no product forma-
tion was detected in the catalytic reduction. In addition, op-
timization studies of the catalytic reduction of acetophenone
cymene)}2] and the monotosy-
lated 1,2-diphenyl-1,2-diamino-
ethane (TsDPEN).[1h,9]
This particular complex shows high catalytic activity with
both categories of hydride donors, and prochiral ketones as
well as imines have successfully been reduced with high
enantioselectivity in TEAF. In addition to 1, ruthenium
complexes of vicinal amino alcohols show high activity and
often good-to-excellent selectivity in the reduction of ke-
tones. However, these reactions only work properly with al-
cohols as the hydride donor.[10]
We and others have demonstrated that a-amino acids can
serve as efficient ligand precursors to ATH catalysts.[11,12]
The modular nature of a-amino acids make them highly at-
tractive as ligand building blocks and a multitude of ligands
with different structural features can easily be prepared and
evaluated. We recently reported that ruthenium and rhodi-
um complexes of N-carbamate-protected amino acid deriva-
tives, pseudo-dipeptides (2), thioamides (3), and hydroxamic
acids (4), successfully catalyze the ATH reactions of aryl
alkyl ketones with high activity and excellent enantioselec-
tivity.[11]
in 2-propanol using 2 and [{RuCl2ACHTUNRTGNEUNG(p-cymene)}2] showed that
good activity was obtained with a base/ligand ratio of >3:1.
The common feature of the working ligands is the presence
of three acidic sites, which could explain the necessity of
using a minimum of 3 equiv of sodium hydroxide or 2-prop-
oxide. In most cases the pseudo-dipeptides studied con-
tained two stereogenic centers and from the reaction out-
come we found a good correlation between the stereocenter
of the amino acid part of the ligand and the absolute config-
uration of the product obtained in the ATH reaction. Evi-
11198
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 11197 – 11209

Asymmetric Transfer Hydrogenation of Ketones
FULL PAPER
Table 1. Rhodium-catalyzed ATH reaction of acetophenone using thio-
dently the chiral induction originated primarily from the
amino acid and the stereocenter present in the amino alco-
hol part played a secondary role. Based on these data we
concluded that the metal most likely interacts with all of the
available functional groups in the pseudo-dipeptide ligand,
however, the exact structure could not be elucidated. The
problems encountered when trying to isolate a metal com-
plex containing ligand 2 led to further structural variations.
We argued that higher complex stability could be obtained
by converting the central amide of the ligand into its corre-
sponding thioamide, thereby increasing the acidity of this
functionality.[13] To our surprise we found that this modifica-
tion led to a ligand class with significantly different proper-
ties. In the ruthenium-catalyzed ATH reaction of acetophe-
none using a thioamide derived from a natural amino acid
pseudo-dipeptide, we unexpectedly obtained the product
with the opposite configuration.[11g] After catalyst optimiza-
tion we discovered that the use of rhodium was superior to
ruthenium and that the alcohol moiety of the ligand was re-
dundant. Thus, the most efficient and selective catalyst was
obtained from ligand 3a (R1 =tBu, R2 =iPr, R3 =(S)-
CHPhCH3) and [{RhCl2Cp*}2].[14] The observed switch in
enantioselectivity on going from N-Boc-protected pseudo-
dipeptides to thioamides was rather intriguing and we there-
fore decided to further elaborate the ligand structure to clar-
ify the origin of this effect. The increased acidity of the thio-
amide could of course be one of the reasons for the switch
in enantioselectivity. A similar outcome of the ATH reac-
tion was therefore anticipated for an amino acid with the C
terminus transformed into a hydroxamic acid. The ATH re-
actions of ketones with rhodium complexes of the hydroxa-
mic acid ligands 4 derived from natural amino acids resulted
in the selective formation of the (S)-alcohol, in agreement
with the use of pseudo-dipeptide ligands. Clearly the ob-
served enantioswitch is not connected to the acidity of the
amide functionality, but possibly rather an effect of different
modes of ligand coordination. To obtain data on the ligand
mode of coordination, and thereby an insight into the origin
of the enantioswitch observed, a catalyst structure–activity
correlation investigation was performed. Hence structural
variations were made to the amino acid derived thioamides
and hydroxamic acids and the ligands were evaluated in the
ATH reaction by using acetophenone as the model sub-
strate.
AHCTUNGTRENNUNG
amide ligands.[a]
Entry
Ligand
Conversion [%][b]
Enantioselectivity [%][c]
1
3a
3a
3b
3b
3c
3c
3d
3d
3e
3e
3 f
3 f
3g
3g
3h
3i
91
88
72
87
64
67
31
43
20
24
1
86 (R)
95 (R)
88 (R)
89 (R)
85 (R)
86 (R)
4 (R)
9 (R)
62 (R)
63 (R)
–
2[d]
3
4[d]
5
6[d]
7
8[d]
9
10[d]
11
12[d]
13[e]
14[d,e]
15
2
–
82
65
–
2
–
47 (R)
49 (R)
–
16[f]
17
–
–
3j
[a] Reaction conditions: Acetophenone (1 equiv, 0.2m in 2-propanol),
[{RhCl2Cp*}2] (0.25 mol%), ligand (0.55 mol%), and 2-PrONa
(5 mol%), 2 h, room temperature. [b] The conversion was determined by
GLC analysis. [c] The enantiomeric excess was determined by GLC anal-
ysis (CP Chirasil DEXCB). [d] LiCl (5 mol%) was added to the reaction
mixture. [e] Reaction time: 15 min. [f] [{RuCl
was used as the metal precursor.
2ACHTNUGTREN(UNGN p-cymene)}2] (0.5 mol%)
estingly, when the diastereomeric ligand 3b was used in the
catalytic reduction, the same product isomer was obtained
with good ee values (entries 3 and 4). Intrigued by these re-
sults we prepared compounds 3c and 3d and evaluated their
performance in the ATH reaction. From the results present-
ed in Table 1 (entries 5–8) it is clear that the presence of the
isopropyl group in the amino acid is essential for the enan-
tioselectivity and that the stereocenter in the amine part of
the ligand is of less importance. We then turned our atten-
tion towards the N terminus of the ligand. Replacing the
carbamate by an acetamide (3e) resulted in lower activity
and selectivity of the reduction reaction (entries 9 and 10),
whereas the use of ligand 3 f practically gave no conversion
(entries 11 and 12). When ligand 3g was employed in the
ATH reaction of acetophenone we achieved good activity,
albeit with significantly lower enantioselectivity (entry 13).
Interestingly, when the reaction was performed in the pres-
ence of LiCl, no significant improvement in the enantiose-
Thioamide ligands derived from a-amino acids: In our pre-
vious study on the development of a-amino acid thioamide
ligands for the ATH reaction of prochiral ketones, ligand
optimization revealed that valine-derived 3a in combination
with [{RhCl2Cp*}2] gave the best catalytic activity and selec-
tivity.[11g] Thus, the use of catalyst Rh–3a for the reduction
of acetophenone resulted in high conversion and good enan-
tioselectivity of the (R)-alcohol (Table 1, entry 1). When the
reaction was performed with LiCl as an additive, an even
higher ee was obtained (entry 2). Ligands based on other
amino acids as well as on different amines were also investi-
gated, but in all cases inferior results were obtained. Inter-
Chem. Eur. J. 2009, 15, 11197 – 11209
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11199

H. Adolfsson et al.
Table 2. Rhodium-catalyzed ATH reaction of acetophenone using hy-
droxamic acid ligands.[a]
lectivity was observed (entry 14). The catalyst derived from
ligand 3h, which lacks the N terminus, was completely inac-
tive. When the structure of the C terminus of the ligand was
varied, we found that the catalyst formed with thioamide 3i,
derived from the corresponding tertiary amide, was essen-
tially inactive (entry 16). The same result was obtained with
the methylated thioamide 3j (entry 17). Evidently, in agree-
ment with our previous observations using pseudo-dipeptide
ligands, it is of the utmost importance for the catalytic activ-
ity that the amide in the ligand can be deprotonated. To
conclude, the structural features of the ligands that lead to
active and stereoselective thioamide-based rhodium cata-
lysts appear to be 1) steric bulk at the a position of the
amino acid, 2) a carbamate protecting group at the N termi-
nus, and 3) a removable proton at the C terminus.
Entry
Ligand
Conversion [%][b]
Enantioselectivity [%][c]
1
4a
4a
4b
4b
4c
4c
4d
4d
4e
4e
4 f
4 f
4g
4g
4h
4i
89
82
43
62
45
63
43
47
76
80
2
8
1
2
4
–
2
16
10
8
87 (S)
97 (S)
89 (S)
95 (S)
86 (S)
92 (S)
33 (S)
34 (S)
90 (S)
92 (S)
–
Hydroxamic acid ligands derived from a-amino acids: The
ATH reactions of acetophenone using rhodium catalysts
based on the N-Boc-protected amino acid derived hydrox-
2[d]
3
4[d]
5
ACHTUNGTRENNUNGamic acid ligands 4a–c resulted in good conversions and
6[d]
7
good-to-excellent enantioselectivity of 1-phenylethanol
(Table 2, entries 1–6). In agreement with the reaction out-
comes using pseudo-dipeptide catalysts, and in contrast to
the corresponding thioamide-based catalysts presented
above, we obtained the S-configured secondary alcohol as
the major enantiomer. To further our understanding of the
role and importance of the different functionalities present
in the hydroxamic acids regarding catalytic activity and
mode of coordination we decided to examine a number of
structurally related ligands. The combination of
[{RhCl2Cp*}2] and ligand 4d, in which the Boc group is re-
moved, resulted in decreased catalytic activity and selectivi-
ty (entries 7 and 8). The catalytic reduction of acetophenone
using ligand 4e containing a Cbz group at the N terminus
gave a result similar to the corresponding Boc-protected
ligand (entries 9 and 10), whereas catalysts derived from
ligand 4 f or 4g showed poor activity and selectivity (en-
tries 11–14). Interestingly, the sulfonamide-containing ligand
4 f apparently induces the formation of the (R)-alcohol,
even if the selectivity is poor. As in the case of the corre-
sponding thioamide ligand, the use of compound 4h, which
completely lacks coordinating ability with anything but the
hydroxamic acid, gave practically no product formation
(entry 15). From the above results it is evident that the opti-
mal amino acid derived hydroxamic acid ligand contains a
carbamate protecting group at the N terminus because li-
gands with other functionalities gave no or poor perfor-
mance. Next we focused on the catalytic activity of ligands
with structural variation around the hydroxamic acid func-
tionality. We prepared compounds 4i and 4j from N-Boc-
valine and the corresponding monomethylhydroxylamines.
Employing these and the commercially available 4k as li-
gands in the rhodium-catalyzed ATH reaction of acetophe-
none resulted in overall poor activity. However, the enantio-
selectivity was significantly higher using compound 4j than
with the other two methylated hydroxamic acids (entries 16–
21). These results indicate that the possibility of removing a
8[d]
9
10[d]
11
12[d]
13
53 (R)
–
14[d]
15
–
–
–
–
16
17[d]
18
4i
4j
4j
4k
4k
70 (S)
36 (S)
5 (S)
18 (S)
19[d]
20
21[d]
11
[a] Reaction conditions: Acetophenone (1 equiv, 0.2m in 2-propanol),
[{RhCl2Cp*}2] (0.25 mol%), ligand (0.55 mol%), and 2-PrONa
(5 mol%), 2 h, room temperature. [b] The conversion was determined by
GLC analysis. [c] The enantiomeric excess was determined by GLC anal-
ysis (CP Chirasil DEXCB). [d] LiCl (5 mol%) was added to the reaction
mixture.
proton from the nitrogen atom in the hydroxamic acid leads
to a more selective catalyst.
Hydrazide ligands derived from N-Boc-protected a-amino
acids: As demonstrated by the results presented above using
amino acid thioamide or hydroxamic acid ligands in the rho-
dium-catalyzed ATH reactions of ketones, the observed
enantioswitch appears to be decoupled from the acidity of
the functionality present at the ligand C terminus. This is
further stressed by the fact that using N-Boc-l-valine as the
ligand with [{RuCl2ACTHNUTRGNE(UNG p-cymene)}2] for the reduction of aceto-
phenone resulted in the formation of the S enantiomer, al-
though in poor conversion and enantioselectivity.[10g] Never-
theless, we decided to investigate the catalytic activity and
selectivity by using an additional C terminus functionality.
Amino acid hydrazides 5a–d were efficiently prepared from
N-Boc-protected alanine or valine and the corresponding N-
11200
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 11197 – 11209

Asymmetric Transfer Hydrogenation of Ketones
FULL PAPER
substituted hydrazines. Screening these compounds as li-
gands in the rhodium-catalyzed ATH reaction of acetophe-
none resulted in general in modest selectivity and conver-
sions of up to 56% after 2 h. However, the catalyst formed
with 5b gave rather good enantioselectivity (Table 3,
entry 4). Interestingly, in contrast to the pseudo-dipeptide
thioamide or hydroxamic acid systems, the rhodium catalysts
derived from hydrazide ligands showed catalytic activity
even after a prolonged reaction time. Performing the rhodi-
um-catalyzed reduction of acetophenone using ligand 5b re-
sulted in 67% conversion with maintained enantioselectivity
(91%) after 4.5 h.
Table 3. Rhodium-catalyzed ATH reaction of acetophenone using hydra-
zide ligands.[a]
Figure 1. Possible modes of coordination of amino acid derived thio-
AHCTUNGTERGaNNUN mides and hydroxamic acids.
Entry
Ligand
Conversion [%][b]
Enantioselectivity [%][c]
1
5a
5a
5b
5b
5c
5c
5d
5d
43
9
30
56
20
7
80 (S)
49 (S)
81 (S)
91 (S)
84 (S)
50 (S)
28 (S)
11 (S)
2[d]
3
tivity was obtained by using the rhodium complex of ligand
3a. Structural variations, including the removal of the N-
Boc protecting group, changing it to either a carboxamide
or a sulfonamide, or the use of a ligand completely lacking
the N terminus, led invariably to less active and selective
catalysts. Furthermore, S-alkylation of the thioamide or the
use of a ligand containing a tertiary amide was equally inef-
fective. These observations indicate that the successful
ligand coordinates to the rhodium center through the carba-
mate and the deprotonated thioamide. Hence, the most
likely mode of coordination of the amino acid thioamide li-
gands is shown in structure I or II in Figure 1, although the
soft nature of sulfur atom favors coordination mode I. The
formation of complexes of the type III and IV cannot be
completely excluded, although the larger seven-membered
chelates formed in these cases should be thermodynamically
less favored. The corresponding structure–activity investiga-
tion performed with the amino acid hydroxamic acids indi-
cates that this class of ligand interacts with the metal center
by chelation with the carbamate and most likely the nitro-
gen atom of the hydroxamic acid (i.e., coordination modes
VI or VIII in Figure 1). For the same reason as given above
for complexes III and IV, the interaction between the ligand
and the metal according to VIII should be less energetically
favored.
4[d]
5
6[d]
7
9
5
8[d]
[a] Reaction conditions: acetophenone (1 equiv, 0.2m in 2-propanol),
[{RhCl2Cp*}2] (0.25 mol%), ligand (0.55 mol%), and 2-PrONa
(5 mol%), 2 h, room temperature. [b] The conversion was determined by
GLC analysis. [c] The enantiomeric excess was determined by GLC anal-
ysis (CP Chirasil DEXCB). [d] 5 mol% of LiCl was added to the reaction
mixture.
Catalyst structure: A comparison of the possible theoretical
coordination modes of amino acid derived thioamides and
hydroxamic acids revealed that these ligands can interact
with the metal center in a number of different ways
(Figure 1). As with the pseudo-dipeptides, the ATH reac-
tions with ligands 3 and 4 require a certain amount of base
to work properly (see below). We can therefore predict that
the thioamide and the hydroxamic acid ligands will be ion-
ized upon metal coordination. Hence, the possible modes of
coordination of the thioamides can be visualized to be I–V
in Figure 1 and for the hydroxamic acids to be VI–X.[15]
From the above presented structure–activity correlations
using differently functionalized thioamides and hydroxamic
acids, the following conclusions can be drawn regarding an
individual ligandꢂs mode of coordination. For the amino
acid thioamides, the best catalytic activity and enantioselec-
The obvious approach for obtaining further structural
knowledge of these complexes would be the use of spectro-
scopic or crystallographic methods. Therefore, rhodium–
Chem. Eur. J. 2009, 15, 11197 – 11209
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11201

H. Adolfsson et al.
arene complexes of the thioamide and hydroxamic acid li-
gands were synthesized. In separate reaction vessels, ligands
ent-3a and ent-4a were treated with an excess of sodium hy-
dride in dry dichloromethane, after which 1 equiv of
[{RhCl2Cp*}2] was added. The resulting solutions were
slowly evaporated to yield microcrystalline materials that
could be isolated. In the case of the thioamide ligand, the
crystals obtained were submitted to X-ray diffraction analy-
sis, which yielded the structure presented in Figure 2.[16,17]
finding indicates that the catalytically active rhodium com-
plexes formed with hydroxamic acid ligands have rather
weak ligand–metal interactions. To rule out the possibility
that dimeric catalysts are operating in the reduction, we per-
formed a nonlinear-effect study. The ATH reaction was exe-
cuted with varying degrees of enantiomeric excess of cata-
lyst Rh–4a by using the corresponding ligand ent-4a and the
results obtained showed that the product ee was directly
proportional to the stereochemical purity of the catalyst.
The same analysis was performed by using catalyst Rh–3a
and ent-3a, which displayed identical behavior. The linear
dependence of the systems indicates that monomeric com-
plexes are operating and hence the coordination models de-
picted in Figure 1 can be used to decipher the correct cata-
lyst structure.
Reaction conditions: The standard reaction set-up for the
ATH reactions presented above involves 0.25 mol% of the
rhodium precursor, 0.55 mol% of the ligand, and 5 mol% of
sodium isopropoxide. We had previously noticed that the
amount of base is crucial for the catalytic activity of the sys-
tems under study. For instance, in the pseudo-dipeptide cata-
lyst system a minimum base-to-catalyst ratio of 3:1 was re-
quired for the reduction to proceed.[11b] A lower amount of
base resulted in no reaction, however, higher base ratios
were tolerated and good conversion and excellent enantiose-
lectivity were obtained even with 20 mol% of sodium iso-
propoxide (base/catalyst, 20:1). For the thioamide ligand
system we found that the optimal amount of sodium iso-
propoxide was 10 mol% (10:1 ratio) and the highest conver-
sions were achieved when the base was added subsequent to
all other components, including the substrate.[11g] At lower
base-to-catalyst ratios, the reaction rate was substantially de-
creased. The amino acid based hydroxamic acid ligand
system was investigated in more detail and we found that
the best base-to-metal ratio appeared to be 7:1. As can be
seen in Figure 3, the use of less sodium isopropoxide result-
Figure 2. Molecular structure of the complex formed between ligand ent-
3a and [{RhCl2Cp*}2].
The structure of the isolated complex shows that the
ligand coordinates to the metal in a bidentate fashion, and
much to our surprise, with the nitrogen and the sulfur atoms
of the thioamide group (i.e., coordination mode V in
Figure 1). Furthermore, when the isolated complex was em-
ployed as a precatalyst in the ATH reaction of acetophe-
none using the standard conditions (Table 1), the corre-
sponding secondary alcohol was formed in moderate conver-
sion (77% after 2 h), but with exceptionally poor enantiose-
lectivity (17% in favor of the S isomer). Clearly the isolated
complex is different from the active catalyst formed in situ
between ligand 3a and [{RhCl2Cp*}2]. Because the isolated
rhodium compound is the result of complex formation oc-
curring under thermodynamic conditions, this particular
structure does not necessarily mirror the actual structure of
the active catalyst or the precatalyst operating in the ATH
reaction. In fact, upon closer examination of the reaction
profile obtained by using the thioamide-derived catalyst, de-
viation from the expected exponential behavior was ob-
served (see below). Unfortunately, all attempts to isolate
and characterize the corresponding complex of the hydroxa-
mic acid ligand 4a failed. In fact, the microcrystalline mate-
rial that was isolated from the complex formation reaction
turned out to be the [{RhCl2Cp*}2] starting complex. This
Figure 3. Effect of the sodium isopropoxide-to-catalyst ratio on the ATH
reduction of acetophenone using [{RhCl2Cp*}2] and ligand 4a. The con-
version was monitored after 2 h.
11202
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 11197 – 11209

Asymmetric Transfer Hydrogenation of Ketones
FULL PAPER
ed in lower conversion, and a higher ratio did not result in a
better reaction outcome. The enantioselectivity was not af-
fected by the amount of base and in all the reactions in
which we observed product formation, the ee of 1-phenyle-
thanol was above 85%.
ride. However, for practical reasons (i.e., poor solubility of
the lithium base in 2-propanol) the latter experimental set-
up is preferable. These observations indicate that the alkali
ion also plays an intimate part in these reactions (see Mech-
anistic Considerations below).
The observation that ruthenium-catalyzed transfer hydro-
genation processes in 2-propanol are accelerated in the pres-
ence of base was made by Chowdhury and Bꢀckvall[18] in
1991, and most ATH protocols developed since then have
shown similar behavior. Depending on the metal precursor
used, the added base plays different roles.
When Ru–arene or Rh–Cp* catalyst precursors are em-
ployed, the dominant role of the base is to assist in the com-
plex formation, often by ligand deprotonation. In all of the
ligand systems described above there are at least two acidic
sites that could be deprotonated by the base. The fact that
higher sodium isopropoxide concentrations resulted in
better and faster conversions indicates that the base is in-
volved in additional processes.[18,19]
Initial rate kinetics: To obtain further information on the
different catalytic systems, additional experiments were per-
formed. From the initial rate experiments, the effect of the
different reaction components could be examined. Initially
ketone concentrations were varied in the reactions catalyzed
by either Rh–3a (thioamide) or Rh–4a (hydroxamic acid),
whereas the other components were kept constant. The for-
mation of 1-phenylethanol as a function of time was plotted
for a series of different acetophenone concentrations using
the following conditions: [Rh–3a] or [Rh–4a]=0.001012m,
[NaOPri]=0.009773m,
and
[acetophenone]=0.009773–
0.3909m in 2-propanol at a constant volume of 25.58 mL
(Figure 4). It was found that the two catalytic systems be-
Effect of additives: In the asymmetric transfer hydrogena-
tion reactions using ruthenium catalysts formed with
pseudo-dipeptide ligands, we observed that the addition of
lithium salts to the catalytic reaction had a beneficial effect
on the product formation and the enantioselectivity.[11f]
Alkali ions were found to be crucial for the catalytic turn-
over and reactions performed with additives such as crown
ethers, which efficiently complex such metals, resulted in a
significant drop in activity and selectivity. Moreover, DFT
calculations indicated that the alkali ion is involved in the
hydride transfer step by coordinating to the substrate and
part of the catalyst. This coordination results in a highly or-
ganized transition state. In line with the experimental find-
ings, we found that the presence of the smaller lithium ion
resulted in the tightest TS, which naturally resulted in the
highest selectivity. When we examined the influence of lithi-
um additives on the corresponding rhodium pseudo-dipep-
tide catalytic system we found, to our surprise, that no such
enhancing effect was obtained. Instead of an increase in the
enantioselectivity, we achieved slightly lower ee values. As
seen in Tables 1–3, the addition of lithium chloride had a
beneficial effect on the product enantioselectivity in all of
the catalyst systems studied. The most dramatic effects, an
ee increase of around 10%, were obtained with the best li-
gands in each class (i.e., 3a, 4a, and 5b). To rule out the
possibility that the observed increase in ee was a result of
added chloride, we performed the ATH reaction with hy-
droxamic acid ligand 4a and [{RhCl2Cp*}2] in the presence
of a silver salt. The addition of silver triflate did not affect
the enantioselectivity of the reaction because the product
was obtained in 91% ee after 2 h (no LiCl present). Further-
more, we found that the addition of 10 mol% 15-crown-5 to
the standard reaction set-up (with sodium isopropoxide as
base) resulted in a decrease in conversion (45%) and selec-
tivity (76% ee). The reaction can be performed with lithium
isopropoxide as base, which results in the same outcome as
the use of a mixture of the sodium base and lithium chlo-
Figure 4. Initial reaction rates as a function of acetophenone concentra-
~
&
tion: ) Rh–3a and ) Rh–4a.
haved rather differently. In the case of the Rh–3a-catalyzed
reduction, the reaction rate was linearly correlated with the
ketone concentration and only at a high acetophenone con-
centration (0.39m) was the substrate dependence found to
be of decreasing importance. Conversely, when Rh–4a was
employed as the catalyst, the initial rates were essentially
equal regardless of the ketone concentration and only at
low initial concentrations of acetophenone (<0.1m) did the
reaction proceed at a lower rate. This behavior indicates
pseudo-first-order reaction conditions in ketone for the thio-
amide-derived catalyst system. The Rh–4a-catalyzed reac-
tion, on the other hand, demonstrates a pseudo-zero-order
dependence on acetophenone.
Next, the effect of different 2-propanol concentrations on
the initial rates of the ATH reaction of acetophenone was
examined. By using the same reaction conditions as in the
experiments presented above, but with a constant acetophe-
none concentration of 0.1955m and a variation of the con-
centration of 2-propanol from 3–12.78m with THF as co-sol-
vent, the plots presented in Figure 5 were obtained. The re-
sults show that a first-order dependence on 2-propanol was
found for both catalytic systems. However, at high 2-propa-
Chem. Eur. J. 2009, 15, 11197 – 11209
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11203

H. Adolfsson et al.
rateð2Þ ¼k2½acetophenoneꢁ½Rh ꢀ Hꢁ
ꢀkꢀ2½1 ꢀ phenylethanolꢁ½Rhꢁ
ð2Þ
Because the total catalyst concentration remains constant
throughout the reaction, the concentration of the individual
ꢀ
rhodium species [Rh] and [Rh H] can be expressed by
Equation (3).
½Rhꢁtot ¼ ½Rhꢁ þ ½Rh ꢀ Hꢁ
ð3Þ
Combining the above equations gives the overall rate ex-
pression [Eq. (4)] for the reaction.
Figure 5. Initial reaction rates as a function of 2-propanol concentration:
~
&
) Rh–3a and ) Rh–4a.
ꢀ
ꢁ
k1k2½2 ꢀ propanolꢁ½acetophenoneꢁ ꢀ kꢀ1kꢀ2½acetoneꢁ½1 ꢀ phenylethanolꢁ
ð4Þ
rate ¼ ½Rhꢁtot
k1½2 ꢀ propanolꢁ þ k2½acetophenoneꢁ þ kꢀ1½acetoneꢁ þ kꢀ2½1 ꢀ phenylethanolꢁ
nol concentrations ([2-
propanol]=8–10m),
a
The magnitude of the individual rate constants associated
saturation effect was noticed because the rates did not in-
crease further.
with the different steps in the ATH process can be obtained
from analyzing the results of reactions performed under dif-
ferent conditions (see the Supporting Information for fur-
ther details).[20] The values of the individual rate constants
obtained according to the initial rate analysis for the differ-
ent catalytic systems, employing catalysts Rh–3a and Rh–
4a, are presented in Table 4. An alternative to the use of
the initial rate method is to study the complete reaction of
the individual experiments and to model the obtained pro-
files for the determination of the individual rate constants
(k1, kꢀ1, k2, and kꢀ2) in Equation (4). The experimental data
for the reactions catalyzed by the hydroxamic acid derived
complex (Rh–4a) were modeled by using the DynaFit soft-
ware[21] in which a total of 257 data points were simultane-
ously fitted to give estimated values of k1, kꢀ1, k2, and kꢀ2
(Table 4 and Figure 6). It was, however, not possible to
obtain reliable modeled rate constants for the reaction cata-
lyzed by the thioamide-derived complex Rh–3a. This finding
can either be explained by the use of an incorrect mecha-
nism in the simulation or by possible catalyst deactivation
during the course of the reaction. To gain further under-
standing about the nature of the reaction catalyzed by com-
plex Rh–3a, an experiment was performed in which the ace-
tophenone was added to the mixture of all the other compo-
nents with a delay of 1 h. The reaction profile and hence the
rate of the reaction was found to be considerably different
to those obtained under the standard conditions (Figure 7).
From the curves and the initial rates obtained (4.3 mmminꢀ1
under standard conditions and 2.9 mmminꢀ1 with substrate
addition after 1 h), it can be concluded that the catalyst is
clearly affected over time. The latter finding suggests that
the use of initial rate kinetics for studies on the Rh–3a-cata-
lyzed system is preferable.
At a first glance, the above results indicate that the two
catalytic systems behave rather differently and that the indi-
vidual processes might have different rate-determining
steps. The direct correlation between acetophenone concen-
tration and the initial rate suggests that ketone reduction is
rate-limiting in the ATH reaction catalyzed by Rh–3a. In
contrast, the pseudo-zero-order dependence on acetophe-
none concentration found for the reduction process employ-
ing Rh–4a suggests that rhodium hydride formation could
be the rate-determining step. For a better description of the
behavior of the two different catalytic systems, a more in
depth analysis of the rate constants associated with the indi-
vidual reaction steps was undertaken.
Determination of the rate constants: From the initial rate
experiments it is possible to calculate the rate constants as-
sociated with the ATH reaction. The ATH reaction se-
quence with either of the two different catalysts Rh–3a and
Rh–4a can be divided into two consecutive steps, the forma-
tion of the rhodium hydride followed by the reduction of
the substrate (Scheme 1, steps 1 and 2, respectively).
The rate expressions for the individual steps can be de-
scribed by Equations (1) and (2).
rateð1Þ ¼ k1½2 ꢀ propanolꢁ½Rhꢁ ꢀ kꢀ1½acetoneꢁ½Rh ꢀ Hꢁ
ð1Þ
A comparison of the rate constants for the reaction cata-
lyzed by complex Rh–4a (Table 4) shows that the calculated
initial rates are in good agreement with the modeled ones
representing the complete reaction. The magnitudes of k1,
kꢀ1, k2, and kꢀ2 are the same in both cases. Analysis of the
magnitudes of the individual rate constants presented in
Table 4 shows that the two catalyst systems operate rather
Scheme 1. Schematic representation of the individual reaction steps and
the overall process for the rhodium-catalyzed ATH of acetophenone.
11204
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 11197 – 11209

Asymmetric Transfer Hydrogenation of Ketones
FULL PAPER
Table 4. Kinetic rate constants and calculated initial rates for the ATH reaction of
acetophenone in 2-propanol catalyzed by [Rh–3a] and [Rh–4a].
hydroxamic acid containing cat-
alyst is associated with the ini-
tial step. Furthermore, with the
rate constants in hand it is pos-
sible to calculate the initial
rates associated with the for-
ward reactions of the individual
reaction steps 1 and 2 presented
in Scheme 1. At very low ace-
tone and 1-phenylethanol con-
centrations, the initial rates for
k1
kꢀ1
k2
kꢀ2
Initial rate of
Initial rate of
step 2[a]
[mꢀ1 minꢀ1
]
[mꢀ1 minꢀ1
]
[mꢀ1 minꢀ1
]
[mꢀ1 minꢀ1 step 1[a]
]
[mmminꢀ1
]
[mmminꢀ1
ACHTUNGTERNNUNG ]
initial rate data for 1.12
Rh–3a
initial rate data for 0.198
Rh–4a
100
31.3
137
2.69
3.01
14.5
2.55
3.67
6.18
69.4
27.1
12.3
modeled data for
Rh–4a
0.284ꢂ0.0098 26.0ꢂ2.8
62.2ꢂ6.0 5.37ꢂ0.45
[a] The initial rates were calculated by using Equations (5) and (6). Concentrations used in the calculations
were [Rh]=0.001012m, [2-propanol]=12.78m, and [acetophenone]=0.1955m.
steps 1 and
2 are given by
Equations (5) and (6).
rateð1Þ ¼ k1½2-propanolꢁ½Rhꢁ
ð5Þ
ð6Þ
rateð2Þ ¼ k2½acetophenoneꢁ½Rhꢁ
Comparison of the initial rates obtained for the individual
steps provides valuable information on the rate-determining
step associated with the different processes. Hence, if
k2[acetophenone]!k1[2-propanol], step 2 becomes rate-lim-
iting. On the other hand, if k2[acetophenone]@k1[2-propa-
nol], the hydride-generating step becomes rate-determining.
For the ATH reaction of acetophenone catalyzed by Rh–
4a, the calculated rate for the hydride formation (step 1)
was found to be 3.67 mmminꢀ1 at [acetophenone]=0.1955m,
which is slightly lower than the hydride transfer to the sub-
strate (step 2), for which the rate is 12.3 mmminꢀ1. These
values suggest that, at an early stage of the reaction, the hy-
dride formation (step 1) is rate-limiting. Conversely, in the
reaction catalyzed by complex Rh–3a, the rate for hydride
formation is substantially higher 14.5 mmminꢀ1. Moreover,
the lower value for hydride transfer to the ketone
(6.18 mmminꢀ1) indicates that the rate-determining step in
the Rh–3a-catalyzed reaction is the second step. It can thus
be concluded that the rate-determining step was different
for the two catalyst systems, with the hydride formation
being rate-limiting in the reaction catalyzed by the hydroxa-
mic acid derived catalyst (step 1) and the hydride transfer to
the substrate being rate-determining in the reaction cata-
lyzed by the thioamides-derived catalyst (step 2).
Figure 6. Modeled reaction profiles for the ATH reaction of acetophe-
none catalyzed by Rh–4a. The points represent experimental data and
lines represent the modeled profiles.
Hammett plots: Competitive rate experiments with differ-
ently substituted acetophenones and with the parent ketone
were performed. Plotting the logarithm of the relative rates
against the Hammett s values for the different substrates re-
vealed, not surprisingly, that both catalyst systems display
positive 1 values (Figure 8). The 1 values suggest that the
reaction catalyzed by Rh–3a (1=2.66, whereas 1=1.99 for
Rh–4a) is slightly more sensitive to substrate substituent ef-
fects, a result that reflects the outcome of the initial rate
study. The better linear fit should also be a consequence
thereof.
Figure 7. Reaction profiles as a function of time using catalyst Rh–3a:
~
*
) standard experiment and ) substrate addition after 1 h.
differently. In the reaction catalyzed by the thioamide-con-
taining catalyst Rh–3a, kꢀ1 >k2, which suggests that the
transition state for the initial hydride-forming step is lower
in energy than the second product-generating step. Con-
versely, in the reaction catalyzed by Rh–4a, kꢀ1 is less than
k2. Hence, the energetically highest activation barrier for the
Chem. Eur. J. 2009, 15, 11197 – 11209
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11205

H. Adolfsson et al.
Figure 9. Structural resemblance between the TsDPEN and the amino
acid derived ligands.
over, the stereochemical outcomes of the ATH reductions
suggest that these ligands coordinate the metal in a biden-
tate fashion and, most importantly, with the amino acid part
of the chelate ring.
Figure 8. Hammett plot showing the effect of substituents on the transfer
~
&
hydrogenation using different catalyst systems: ) Rh–3a and ) Rh–4a.
The limited amount of structural information available on
the catalytically active Ru/Rh complexes with either the
pseudo-dipeptide or the amino acid based thioamide, hy-
droxamic acid, or hydrazide ligand systems drastically com-
plicates the prediction of ligand coordination as well as the
mechanistic analysis. The major difference between the
ligand systems examined appears to be their mode of coor-
dination. The isolated and characterized thioamide complex
shown in Figure 2 displays catalytic properties that signifi-
cantly deviate from what we obtain under standard reaction
conditions. Therefore we can conclude that this complex is
not the active catalyst operating in the reaction. Instead, the
thioamides presumably interact with the metal through the
carbamate nitrogen (or oxygen) and the thioamide sulfur
atoms (II or IV of Figure 1). The hydroxamic acids, on the
other hand, most probably coordinate the metal through the
carbamate nitrogen (or oxygen) and the hydroxamic acid ni-
trogen atoms according to VI or VIII of Figure 1. The high
enantioselectivity of either the R or the S isomer of the
product alcohol obtained by using these complexes implies
that two stereogenically different hydride complexes are re-
sponsible for product formation. Moreover, the difference in
the rate-determining step demonstrated for the two catalytic
systems further stresses a mechanistic divergence. As depict-
ed in Scheme 2a, we suggest that the thioamide ligand 3a
interacts with the rhodium precursor to form a complex,
which, upon reaction with 2-propanol, predominantly gives
the (S)-ligand,(S)-rhodium (SL,SRh) diastereomer. In agree-
ment with the results obtained with catalyst 1, the subse-
quent reaction with the substrate selectively gives the (R)-
alcohol. The stereochemically opposite result obtained by
using the rhodium catalyst derived from the hydroxamic
acid 4a suggests that the reduction of the ketone is mediat-
Mechanistic considerations: The mechanism of the asym-
metric transfer hydrogenation using half-sandwich rutheni-
um or rhodium complexes with, in particular, the TsDPEN
ligand has been extensively studied (i.e., the ATH reaction
catalyzed by complex 1).[22,23] It is generally accepted that
the reaction proceeds by an outer-sphere mechanism in
which the ketone substrate interacts with the bifunctional
mono-hydride catalyst and simultaneously receives a hy-
dride from the metal and a proton from the ligand. Howev-
er, recent modeling studies of the transfer hydrogenation in
solution indicate that solvent molecules play an important
role in mediating proton transfers between the catalyst and
the substrate.[24] The high enantioselectivity in the reduction
of aryl alkyl ketones arises from a stabilizing electrostatic in-
teraction (CH–p) between the arene ligand of the catalyst
and the aryl ring of the substrate. Kinetic studies using the
Noyori catalyst 1 in 2-propanol at high ketone concentration
have revealed that the rate-determining step is the hydride
formation/regeneration rather than the ketone reduction.[25]
Because the metal in the catalyst complex becomes a stereo-
genic center upon hydride formation, two different diaste-
reomeric complexes could theoretically be created. The ex-
cellent enantioselectivity obtained by using the Noyori cata-
lyst implies that only one of the two possible isomers is
formed. The amino acid based thioamide, hydroxamic acid,
and hydrazide ligand systems show a large degree of struc-
tural resemblance with the TsDPEN ligand used in catalyst
1. As depicted in Figure 9, all the ligand systems contain a
basic and an acidic site, which are important for complex
formation as well as catalytic activity. The acidic site in the
TsDPEN ligand is evidently the sulfonamide and the corre-
sponding site in the amino acid ligands is the C terminus. In
the Noyori catalyst the amine acts as the basic site, whereas
the corresponding site in the amino acid ligands is less obvi-
ous. From the structure–activity investigation presented
above in which different functional groups were introduced
at the ligand N terminus, we found that carbamates gave the
best catalytic activity and selectivity. Therefore we can con-
clude that even if the carbamate as such is a poor base, the
deprotonated form will be significantly more basic. More-
ed by the (S)-ligand,(R)-rhodium (SL,RRh
)
complex
(Scheme 2b). The pseudo-zero-order dependence on ketone
concentration using Rh–4a suggests that the rate-limiting
step is the metal hydride formation rather than the substrate
reduction. Hence, once the (SL,RRh)-Rh–4a hydride is
formed, a fast reaction with the substrate will lead to the
formation of the S isomer of the product. The underlying
reason why the rhodium complexes of ligands 3a and 4a
11206
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 11197 – 11209

Asymmetric Transfer Hydrogenation of Ketones
FULL PAPER
was found to be crucial for the
catalytic activity. Kinetic studies
were carried out to examine the
reaction order and we found
that the hydroxamic acid based
catalyst system Rh–4a showed
pseudo-zero-order dependence
on the substrate. Furthermore,
the rate constants for the reac-
tions were determined and
from the overall kinetic data we
concluded that the rate-limiting
step for the reaction with the
Rh–4a catalyst is the rhodium
hydride formation. In contrast,
Scheme 2. Suggested rhodium hydride complexes formed with a) thioamide ligand 3a and b) hydroxamic acid
ligand 4a.
the thioACHTNUTRGNEUNGamide-derived catalyst
system Rh–3a demonstrated
pseudo-first-order kinetics with
will generate two different metal hydride diastereomers,
even if they are based on the same amino acid scaffold,
could be explained by secondary steric and/or electronic in-
teractions between the substituents on the thioamide and
the hydroxamic acid and the Cp* ligand. In addition, it is
possible that the Rh–3a complex can undergo hydride for-
mation with 2-propanol to yield both the (SL,SRh) and the
(SL,RRh) hydride intermediates. Because the rate-determin-
ing step is the delivery of a hydride to the substrate, these
two complexes can be in equilibrium with each other and
the lowest barrier towards product formation is via the
(SL,SRh) intermediate (cf. Curtin–Hammett kinetics). The
enhancement of activity and enantioselectivity observed
upon addition of lithium chloride is difficult to explain, how-
ever, we assume that the Lewis acidic metal can interact
with the substrate and the oxygen atom of the carbamate
carbonyl in the hydride transfer step and thereby facilitate a
highly organized transition state.
respect to ketone concentration. This observation, along
with other kinetic data, is in accordance with the results re-
ported for the Noyori catalyst system and implies that there
is an interaction between the substrate and the catalyst in
the rate-determining step. The absence of nonlinear behav-
ior suggests that monomeric metal complexes are operating
as catalysts in the ketone reductions. Thus, when all the re-
sults are summarized, we can conclude that the ligands most
likely coordinate the metal in a bidentate fashion, with the
amino acid part of the chelate ring. The origin of the ambiv-
alent selectivity found by using the amino acid derived
ligand classes can be traced to their different modes of coor-
dination, along with differences in the reaction rates for
product formation. The thioamides and hydroxamic acids
share a common coordination site, namely, the carbamate
nitrogen or possibly the oxygen atom. However, a difference
is found at the “acidic” amino acid C terminus, the depro-
tonated thioamide ligand coordinating the metal through
the sulfur atom and the corresponding hydroxamic acid
ligand binding through the deprotonated hydroxy-amide ni-
ꢀ
Conclusion
trogen. The N S coordination thus obtained with the thioa-
mides results in a complex that favors ketone reduction by
the (SL,SRh) rhodium hydride. The transfer of a proton and a
hydride from (SL,SRh)-Rh–3a to the Si face of the substrate
facilitates the formation of the observed (R)-alcohol. The
corresponding hydroxamic acid ligand, which coordinates
The rhodium-catalyzed asymmetric transfer hydrogenation
of acetophenone using amino acid derived ligands contain-
ing thioamide and hydroxamic acid functionalities have
been investigated to further our understanding of the ob-
served enantioswitchable nature of these catalytic systems.
The reduction of acetophenone using thioamide-containing
ligands from naturally occurring amino acids selectively
gave the (R)-alcohol (95% ee), whereas the use of the cor-
responding hydroxamic acid ligand system gave the S prod-
uct with excellent enantioselectivity (97% ee). The struc-
ture–activity investigation performed on the thioamide and
hydroxamic acid ligands revealed that subtle variations of
the ligand structure caused dramatic effects on the reaction
outcome. Replacing the N terminus of the ligands with
groups other than carbamates resulted in a loss of catalytic
activity and in most cases a significant decrease in the selec-
tivity. Furthermore, an “acidic” proton at the C terminus
ꢀ
the metal in an N N fashion, should thereby favor the for-
mation of the diastereomeric hydride isomer (SL,RRh), which
upon interaction with the substrate will form the opposite
product enantiomer. In addition, we investigated the use of
an additional ligand class derived from carbamate-protected
amino acids, namely hydrazides. We found that the in situ
formed rhodium complex based on the hydrazide ligand
formed from N-Boc-valine and 2,4-dinitrophenylhydrazine
catalyzed the reduction of acetophenone to give a moderate
conversion and good enantioselectivity (91%). Catalysts
based on the hydrazide ligands behaved in accordance with
the hydroxamic acids and should coordinate the metal in a
similar fashion.
Chem. Eur. J. 2009, 15, 11197 – 11209
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11207

H. Adolfsson et al.
[10] a) D. A. Alonso, D. Guijarro, P. Pinho, O. Temme, P. G. Andersson,
J. Org. Chem. 1998, 63, 2749; b) D. A. Alonso, S. J. Nordin, P. Roth,
T. Tarnai, P. G. Andersson, M. Thommen, U. Pittelkow, J. Org.
Chem. 2000, 65, 3116; c) S. J. Nordin, P. Roth, T. Tarnai, D. A.
Alonso, P. Brandt, P. G. Andersson, Chem. Eur. J. 2001, 7, 1431;
d) M. J. Palmer, T. Walsgrove, M. Wills, J. Org. Chem. 1997, 62,
5226; e) M. Wills, M. Gamble, M. J. Palmer, A. R. C. Smith, J. R.
Studley, J. A. Kenny, J. Mol. Catal. A 1999, 146, 139; f) M. J. Palmer,
J. A. Kenny, T. Walsgrove, A. M. Kawamoto, M. Wills, J. Chem. Soc.
Perkin Trans. 1 2002, 416; g) J. Takehara, S. Hashiguchi, A. Fujii, S.
Inoue, T. Ikariya, R. Noyori, J. Chem. Soc. Chem. Commun. 1996,
233.
[11] a) I. M. Pastor, P. Vꢀstilꢀ, H. Adolfsson, Chem. Commun. 2002,
2046; b) I. M. Pastor, P. Vꢀstilꢀ, H. Adolfsson, Chem. Eur. J. 2003, 9,
4031; c) A. Bøgevig, I. M. Pastor, H. Adolfsson, Chem. Eur. J. 2004,
10, 294; d) P. Vꢀstilꢀ, J. Wettergren, H. Adolfsson, Chem. Commun.
2005, 4039; e) J. Wettergren, A. Bøgevig, M. Portier, H. Adolfsson,
Adv. Synth. Catal. 2006, 348, 1277; f) P. Vꢀstilꢀ, A. B. Zaitsev, J. Wet-
tergren, T. Privalov, H. Adolfsson, Chem. Eur. J. 2006, 12, 3218;
g) A. B. Zaitsev, H. Adolfsson, Org. Lett. 2006, 8, 5129; h) K. Ahl-
ford, A. B. Zaitsev, J. Ekstrçm, H. Adolfsson, Synlett 2007, 2541; i) J.
Wettergren, A. B. Zaitsev, H. Adolfsson, Adv. Synth. Catal. 2007,
349, 2556; j) L. Zani, L. Eriksson, H. Adolfsson, Eur. J. Org. Chem.
2008, 4655.
[12] a) T. Ohta, S.-I. Nakahara, Y. Shigemura, K. Hattori, I. Furukawa,
Chem. Lett. 1998, 491; b) D. Carmona, F. J. Lahoz, R. Atencio, L. A.
Oro, M. P. Lamata, F. Viguri, E. San Josꢃ, C. Vega, J. Reyes, F. Joꢄ,
ꢅ. Kathꢄ, Chem. Eur. J. 1999, 5, 1544; c) ꢅ. Kathꢄ, D. Carmona, F.
Viguri, C. D. Remacha, J. Kovꢆcs, F. Joꢄ, L. A. Oro, J. Organomet.
Chem. 2000, 593, 299; d) T. Ohta, S.-I. Nakahara, Y. Shigemura, K.
Hattori, I. Furukawa, Appl. Organomet. Chem. 2001, 15, 699;
e) J. W. Faller, A. R. Lavoie, Organometallics 2001, 20, 5245; f) H. Y.
Rhyoo, Y. A. Yoon, H. J. Park, Y. K. Chung, Tetrahedron Lett. 2001,
42, 5045; g) H. Y. Rhyoo, H.-J. Park, Y. K. Chung, Chem. Commun.
2001, 2064; h) D. Carmona, M. P. Lamata, F. Viguri, I. Dobrinovich,
F. J. Lahoz, L. A. Oro, Adv. Synth. Catal. 2002, 344, 499; i) D. Car-
mona, J. Ferrer, E. Lalaguna, M. Lorenzo, F. J. Lahoz, S. Elipe,
L. A. Oro, Eur. J. Inorg. Chem. 2002, 259; j) H. Y. Rhyoo, H.-J.
Park, W. H. Suh, Y. K. Chung, Tetrahedron Lett. 2002, 43, 269; k) P.
Pelagatti, M. Carcelli, F. Calbiani, C. Cassi, L. Elviri, C. Pelizzi, U.
Rizzotti, D. Rogolino, Organometallics 2005, 24, 5836; l) W. Hoffmu-
eller, H. Dialer, W. Beck, Z. Naturforsch. B 2005, 60, 1278; m) P. Pe-
lagatti, A. Bacchi, F. Calbiani, M. Carcelli, L. Elviri, C. Pelizzi, D.
Rogolino, J. Organomet. Chem. 2005, 690, 4602.
Acknowledgements
The Swedish Research Council, the Knut and Alice Wallenberg Founda-
tion, the Carl Trygger Foundation and the Wenner-Gren Foundation are
gratefully acknowledged for financial support and postdoctoral fellowship
(A.B.Z.).
[1] a) G. Zassinovich, G. Mestroni, S. Gladiali, Chem. Rev. 1992, 92,
1051; b) T. Ohkuma, R. Noyori in Comprehensive Asymmetric Catal-
ysis I-III, Vol. 1 (Eds.: E. N. Jacobsen, A. Pfaltz, H. Yamamoto),
Springer, Berlin, 1999, p. 199; c) H.-U. Blaser, C. Malan, B. Pugin, F.
Spindler, H. Steiner, M. Studer, Adv. Synth. Catal. 2003, 345, 103;
d) K. Everaere, A. Mortreux, J.-F. Carpentier, Adv. Synth. Catal.
2003, 345, 67; e) S. Gladiali, E. Alberico in Transition Metals for Or-
ganic Synthesis, Vol. 2 (Eds.: M. Beller, C. Bolm), Wiley-VCH,
Weinheim, 2004, p. 145; f) A. Zanotti-Gerosa, W. Herns, M.
Groarke, F. Hancock, Platinum Met. Rev. 2005, 49, 158; g) T. Ikar-
iya, K. Murata, R. Noyori, Org. Biomol. Chem. 2006, 4, 393; h) R.
Noyori, S. Hashiguchi, Acc. Chem. Res. 1997, 30, 97; i) C. Bianchini,
L. Glendenning, Chemtracts 1997, 10, 333; j) M. J. Palmer, M. Wills,
Tetrahedron: Asymmetry 1999, 10, 2045; k) T. Ohkuma, R. Noyori,
Compr. Asymmetric Catal. Suppl. 2004, 1, 43; l) M. Kitamura, R.
Noyori in Ruthenium in Organic Synthesis (Ed.: S.-I. Murahashi),
Wiley-VCH, Weinheim, 2004, p. 3; m) S. Gladiali, E. Alberico,
Chem. Soc. Rev. 2006, 35, 226; n) T. Ikariya, K. Murata, R. Noyori,
Org. Biomol. Chem. 2006, 4, 393; o) X. Wu, J. Xiao, Chem.
Commun. 2007, 2449; p) S. Gladiali, R. Taras in Modern Reduction
Methods (Eds.: P. G. Andersson, I. Munslow), Wiley-VCH, Wein-
heim, 2008, p. 135; q) C. Wang, X. Wu, J. Xiao, Chem. Asian J. 2008,
3, 1750.
[2] For recent selected examples, see: a) A. Schlatter, M. K. Kundu, W.-
D. Woggon, Angew. Chem. 2004, 116, 6899; Angew. Chem. Int. Ed.
2004, 43, 6731; b) D. Xue, Y.-C. Chen, X. Cui, Q.-W. Wang, J. Zhu,
J.-G. Deng, J. Org. Chem. 2005, 70, 3584; c) X. Wu, X. Li, F. King, J.
Xiao, Angew. Chem. 2005, 117, 3473; Angew. Chem. Int. Ed. 2005,
44, 3407; d) A. M. Hayes, D. J. Morris, G. J. Clarkson, M. Wills, J.
Am. Chem. Soc. 2005, 127, 7318; e) W. Baratta, G. Chelucci, S.
Gladiali, K. Siega, M. Toniutti, M. Zanette, E. Zangrando, P. Rigo,
Angew. Chem. 2005, 117, 6370; Angew. Chem. Int. Ed. 2005, 44,
6214; f) S. Enthaler, R. Jackstell, B. Hagemann, K. Junge, G. Erre,
M. Beller, J. Organomet. Chem. 2006, 691, 4652; g) M. T. Reetz, X.
Li, J. Am. Chem. Soc. 2006, 128, 1044.
[3] The use of primary diols as hydride donors, which result in irreversi-
ble reaction conditions, has recently been reported, see: H. C.
Maytum, B. Tavassoli, J. M. J. Williams, Org. Lett. 2007, 9, 4387.
[4] H. Adkins, R. M. Elofson, A. G. Rossow, C. C. Robinson, J. Am.
Chem. Soc. 1949, 71, 3622.
[5] a) H. Meerwein, R. Schmidt, Justus Liebigs Ann. Chem. 1925, 444,
221; b) M. Verley, Bull. Soc. Chim. Fr. 1925, 37, 871; c) W. Ponndorf,
Angew. Chem. 1926, 38, 138; d) C. F. de Graauw, J. A. Peters, H. van
Bekkum, J. Huskens, Synthesis 1994, 1007; e) K. Nishide, M. Node,
Chirality 2002, 14, 759.
[6] For recent examples of aluminium-mediated MPVO reactions, see:
a) Y.-C. Liu, B.-T. Ko, B.-H. Huang, C.-C. Lin, Organometallics
2002, 21, 2066; b) T. Ooi, T. Miura, Y. Itagaki, H. Ichikawa, K. Mar-
uoka, Synthesis 2002, 279; c) T. Ooi, H. Ichikawa, K. Maruoka,
Angew. Chem. 2001, 113, 3722; Angew. Chem. Int. Ed. 2001, 40,
3610.
[7] For a recent example of an efficient lithium-mediated MPVO pro-
cess, see: J. Ekstrçm, J. Wettergren, H. Adolfsson, Adv. Synth. Catal.
2007, 349, 1609.
[13] For a comprehensive overview of acidity constants, please see the
following website: http://www.chem.wisc.edu/areas/reich/pkatable/
index.htm.
[14] a) D. G. Blackmond, M. Ropic, M. Stefinovic, Org. Process Res.
Dev. 2006, 10, 457; b) D. S. Matharu, D. J. Morris, A. M. Kawamoto,
G. J. Clarkson, M. Wills, Org. Lett. 2005, 7, 5489; c) X. Wu, D. Vinci,
T. Ikariya, J. Xiao, Chem. Commun. 2005, 4447.
[15] a) E. Farkas, ꢇ. A. Enyedy, G. Micera, E. Garribba, Polyhedron
2000, 19, 1727; b) F. Hintermaier, S. Helding, L. B. Volodarsky, K.
Sꢈnkel, K. Polborn, W. Beck, Z. Naturforsch. B 1998, 53, 101; c) E.
Farkas, H. Csꢄka, J. Inorg. Biochem. 2002, 89, 219; d) C. J. Milios, E.
Manessi-Zoupa, S. P. Perlepes, A. Terzis, C. P. Raptopoulou, Trans.
Metal Chem. 2002, 27, 864.
[16] C28H42ClN2O2RhS, Mr =609.06, crystal size 0.05ꢉ0.03ꢉ0.01 mm,
monoclinic, C2, a=18636(18), b=9.410(10), c=19.132(9) ꢊ, b=
112.69(3)8, volume=3095(5) ꢊ3, Z=4, 1calcd =1.307 gcmꢀ3
,
m=
synchrotron, lꢃ0.9083 ꢊ, T=100 K, 2qmax =60.588,
measured =9168, Nunique =2381, Rint =0.1982, R,=0.0696, wR=0.1573.
0.730 mmꢀ1
,
N
The diffraction data were collected on beam-line I911-5 at the
Swedish synchrotron facility MaxLab, Lund (Sweden). The data was
obtained with a MARCCD diffractometer by using f scans at a de-
tector distance of 35 mm. The structure was solved by using stan-
dard structure invariant direct methods using SHELXS97 (G. M.
Sheldrick, Acta Crystallogr. Sect. A 2008, 112–122) and refined with
full matrix least-squares calculations using SHELXL97 (G. M. Shel-
[8] a) C. P. Casey, H. Guan, J. Am. Chem. Soc. 2007, 129, 5816; b) S. En-
thaler, B. Hagemann, G. Erre, K. Junge, M. Beller, Chem. Asian J.
2006, 1, 598; c) S. Enthaler, G. Erre, M. Kin Tse, K. Junge, M.
Beller, Tetrahedron Lett. 2006, 47, 8095.
[9] a) S. Hashiguchi, A. Fujii, J. Takahara, T. Ikariya, R. Noyori, J. Am.
Chem. Soc. 1995, 117, 7562; b) A. Fujii, S. Hashiguchi, N. Uematsu,
T. Ikariya, R. Noyori, J. Am. Chem. Soc. 1996, 118, 2521.
11208
www.chemeurj.org
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 11197 – 11209

Asymmetric Transfer Hydrogenation of Ketones
FULL PAPER
drick, SHELXL97, Program for the Refinement of Crystal Struc-
tures, University of Gçttingen, Gçttingen (Germany), 1994). Anom-
alous dispersion correction terms for the used wavelength
(0.9083 ꢊ) were obtained from WCROMER (a) S. Brennan, http://
www.ccp14.ac.uk/ccp/ccp14/ftp-mirror/cross-section-sfac/pub/cross-
section_codes/; b) L. Kissel, R. H. Pratt, Acta Crystallogr. Sect. A
1990, 46, 170–175).
[22] a) M. Yamakawa, I. Yamada, R. Noyori, Angew. Chem. 2001, 113,
2900; Angew. Chem. Int. Ed. 2001, 40, 2818; b) R. Noyori, M. Yama-
kawa, S. Hashiguchi, J. Org. Chem. 2001, 66, 7931; c) M. Yamakawa,
H. Ito, R. Noyori, J. Am. Chem. Soc. 2000, 122, 1466; d) D. A.
Alonso, P. Brandt, S. J. M. Nordin, P. G. Andersson, J. Am. Chem.
Soc. 1999, 121, 9580; e) P. Brandt, P. Roth, P. G. Andersson, J. Org.
Chem. 2004, 69, 4885.
[17] CCDC 712178 contains the supplementary crystallographic data for
this paper. These data can be obtained free of charge from The
Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/
data_request/cif.
[18] R. L. Chowdhury, J.-E. Bꢀckvall, J. Chem. Soc. Chem. Commun.
1991, 1063.
[19] For a recent discussion on the influence of base in ruthenium-cata-
lyzed transfer hydrogenations, see: W. Baratta, K. Siega, P. Rigo,
Chem. Eur. J. 2007, 13, 7479.
[23] For an excellent review on hydrogen transfer mechanisms, see:
J. S. M. Samec, J.-E. Bꢀckvall, P. G. Andersson, P. Brandt, Chem.
Soc. Rev. 2006, 35, 237.
[24] J.-W. Handgraaf, E. J. Meijer, J. Am. Chem. Soc. 2007, 129, 3099.
[25] a) K. J. Haack, S. Hashiguchi, A. Fujii, T. Ikariya, R. Noyori,
Angew. Chem. 1997, 109, 297; Angew. Chem. Int. Ed. Engl. 1997, 36,
285; b) F. K. Cheung, C. Lin, F. Minissi, A. L. Crivillꢃ, M. A.
Graham, D. J. Fox, M. Wills, Org. Lett. 2007, 9, 4659.
[20] J. Wettergren, E. Buitrago, P. Ryberg, H. Adolfsson, Chem. Eur. J.
2009, 15, 5709.
Received: February 27, 2009
Revised: July 15, 2009
[21] P. Kuzmic, Anal. Biochem. 1996, 237, 260.
Published online: September 11, 2009
Chem. Eur. J. 2009, 15, 11197 – 11209
ꢁ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11209