Combinatorial Strategies to find New Catalysts for Asymmetric Hydrogenation Based on the Versatile Coordination Chemistry of METAMORPhos Ligands

Article abstract of DOI:10.1002/cctc.201500621

To extend the toolbox and find improved catalysts, anionic METAMORPhos ligands and neutral amino-acid-based ligands were used separately and in mixtures to form Rh complexes used in the asymmetric hydrogenation of eight industrially relevant substrates. S

Full text of DOI:10.1002/cctc.201500621

DOI: 10.1002/cctc.201500621
Full Papers
Combinatorial Strategies to find New Catalysts for
Asymmetric Hydrogenation Based on the Versatile
Coordination Chemistry of METAMORPhos Ligands
FrØdØric G. Terrade,[a] Alexander M. Kluwer,*[b] Remko J. Detz,[b] Zohar Abiri,[b]
Alida M. van der Burg,[a] and Joost N. H. Reek*[a]
To extend the toolbox and find improved catalysts, anionic
METAMORPhos ligands and neutral amino-acid-based ligands
were used separately and in mixtures to form Rh complexes
used in the asymmetric hydrogenation of eight industrially rel-
evant substrates. Spectroscopic studies showed that under the
catalytic conditions, the mononuclear complex with two differ-
ent ligands (the heterocombination) is the main complex in so-
lution if both the anionic and neutral ligands have the same
chirality. If the neutral ligand and the anionic ligand have the
opposite chirality at the P atom, monometallic and bimetallic
heterocomplexes were detected by NMR spectroscopy and MS.
For the majority of substrates evaluated in this study, higher
enantioselectivities were obtained if the complexes used were
based on the heterocombination of an anionic and a neutral
ligand compared to respective homocombinations. After we
found the initial leads, higher turnover numbers and enantio-
selectivities could be obtained easily by further exploring fo-
cused ligand libraries. The superior activity of the complexes
based on the different ligands is highlighted by their robust-
ness: significant divergence from a 1:1 ratio between the li-
gands does not lower the selectivity of the catalyst, although
more of the competing homocomplexes are formed under
these conditions.
Introduction
Biologically active compounds often contain one or more ste-
reocenters, and the fragrance, pharmaceutical, and crop pro-
tection industries rely heavily on synthetic strategies that lead
to single-enantiomer compounds.[1,2] Following the pioneering
work of Horner, Kagan, Knowles, and Noyori on asymmetric hy-
drogenation, the first industrial-scale synthesis of an optically
enriched drug (l-dopa) was implemented in the early 1970s.[3]
Many other commercial successes were achieved subsequently,
and today homogeneous asymmetric hydrogenation is one of
most proven methodologies to obtain enantiomerically pure
molecules on a multiton scale as exemplified by the recent
dinuclear,[5c] ligand assisted,[5d] etc.) cannot be predicted in ad-
vance, which introduces yet another uncertainty in the predic-
tion of catalytic outcomes. For these reasons, high-throughput
screening is often the method of choice to find a catalyst
system that will yield the product in a sufficiently high enantio-
meric excess (ee) with a sufficient turnover number (TON) and
rate for industrial application.[6] Cationic Rh complexes that
bear one bidentate or two monodentate neutral phosphorus li-
gands are often the most efficient for the asymmetric hydroge-
nation of C=C bonds. Bidentate ligands generally form more
rigid, well-defined complexes, and it was, therefore, long
thought that these complexes were superior to their mono-
dentate counterparts. However, monodentate ligands are often
easier and more cost-effective to synthesize.[7] In addition, they
are particularly suitable for high-throughput combinatorial cat-
alysis: n monodentate ligands can give rise to n(n+1)/2 unique
combinations that consist of n “homocomplexes” (complexes
with two of the same ligands) and n(nÀ1)/2 “heterocomplexes”
(complexes with two different ligands). If two different ligands
(L1 and L2) are combined with a metal precursor, a mixture of
three complexes can be formed: two homocomplexes [(L1)2Rh]
and [(L2)2Rh] and one heterocomplex [(L1)(L2)Rh], which are
often obtained in a 1:1:2 statistical ratio.[7b] As these systems
are (usually) in a thermodynamic equilibrium, the product is
only produced in higher selectivity if the heterocomplex is
more active and selective than the corresponding homocom-
plexes.[7c]
launch of
a new l-menthol production unit of BASF
(20000 tons/year).[1b,4] Despite the multitude of publications on
asymmetric hydrogenation,[3f] we are far from able to design
the best catalyst for a given substrate rationally. Besides the
general issue of small energy differences in the enantiodiscri-
minating reaction step, the particular reaction mechanism that
a hydrogenation catalyst follows (Halpern,[5a] anti-Halpern,[5b]
[a] Dr. F. G. Terrade, A. M. van der Burg, Prof. Dr. J. N. H. Reek
Van’t Hoff Institute for Molecular Sciences
University of Amsterdam
Science Park 904, 1098 XH Amsterdam (The Netherlands)
E-mail: j.n.h.reek@uva.nl
[b] Dr. A. M. Kluwer, Dr. R. J. Detz, Z. Abiri
InCatT B.V.
Science Park 904, 1098 XH Amsterdam (The Netherlands)
E-mail: sander.kluwer@incatt.nl
The concept of supramolecular ligands is based on the self-
assembly of ligand building blocks to form supramolecular bi-
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201500621.
ChemCatChem 2015, 7, 3368 – 3375
3368
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
dentate ligands. Such systems combine the advantage of an
easy synthesis of the building blocks and the rigidity of biden-
tate ligands. If sufficiently strong interactions between the
ligand building blocks are present, selective heterobidentate li-
gands are formed if the two building blocks and the metal pre-
cursor are mixed in solution. To form such supramolecular bi-
dentate ligands, different interactions[8] have been employed,
such as hydrogen bonding,[9] metal–ligand interaction,[10] and
ionic interactions.[11] Strategies based on hydrogen bonding
have shown that a single hydrogen bond can be sufficient to
achieve efficient self-assemblies,[12b] and hydrogen-bonded sys-
tems have even been used in solvents that compete for hydro-
gen bonding.[9f] Metal–ligand interactions have also been ap-
plied successfully as a tool to form rigid well-defined self-as-
semblies that result in selective catalysts. Ionic interactions
have been used frequently to construct supramolecular host–
guest structures,[13] but there are only a few examples in which
such interactions are used for the assembly of supramolecular
bidentate ligands.[11] Gennari and co-workers combined anionic
(which have a carboxylate moiety) and cationic ligands (which
have an ammonium moiety) with a cationic Rh center for the
asymmetric hydrogenation of methyl 2-acetamidoacrylate. The
formation of heterocomplexes was favored moderately as indi-
cated by 31P NMR spectra, and a slightly higher level of enan-
tioselectivity was obtained if a mixture of ligands was applied
compared to the corresponding homocombinations.[11a] More
recently, Pfaltz et al. made a small library of phosphites and
phosphoramidites equipped covalently with noninteracting
anion moieties that were used in combination with neutral li-
gands and a cationic Rh precursor for asymmetric hydrogena-
tion.[11e] They showed that the formation of the neutral hetero-
complexes was favored over that of the anionic or the cationic
homocomplexes and that in some cases, the neutral com-
plexes were more stereoselective. This approach is particularly
interesting as only one ligand needs to be functionalized with
an anionic moiety; the second ligand can be virtually any neu-
tral monodentate ligand.
Scheme 1. Anionic METAMORPhos ligands LanÀ and neutral ligands Lneu
used in this study. The BINOL moiety of Lan1À, Lan4À, Lneu1, Lneu2, and
Lneu3 is in the R form. The other ligands were used in both enantiomeric
forms.
used, various mono- and dinuclear complexes can form in so-
lution, which represents an additional diversity factor for com-
binatorial catalysis. In the current study, the catalytic properties
of the new complexes are compared with the pure mono- and
dinuclear analogues for the hydrogenation of industrially rele-
vant substrates. We demonstrate that for most substrates, the
heterocombinations display improved catalytic performance
compared to the homoligated counterparts. This work pro-
vides a new system to access diverse sets of catalysts easily
and broadens the toolbox for asymmetric hydrogenation.
In 2008, our group introduced METAMORPhos ligands based
on 1,1’-bi-2-naphthol (BINOL) as efficient anionic ligands for
asymmetric hydrogenation.[5c,14a–b] METAMORPhos ligands are
effectively phosphorus ligands equipped covalently with
weakly interacting sulfonamide anions (Scheme 1). Recently, it
was shown that dinuclear Rh complexes based on anionic li-
gands Lan2À and Lan1À are dianionic in the resting state, that
is, four negative charges are located on ligands, two positive
charges are located on the metal centers, and the overall
charge is balanced by two positive charges of the triethylam- Results and Discussion
monium counterions (Scheme 2a).[14c] These complexes display
Choice of ligands
a high efficiency for a number a substrates such as methyl-2-
acetamidoacrylate, but also for the challenging cyclic enam-
ides. They show unrivalled selectivity for the hydrogenation of
difficult tetrasubstituted cyclic enamides.[5c] Herein, we expand
the scope of the application of METAMORPhos ligands (LanÀ)
by using them in combination with neutral amino-acid-based
ligands[12] (Lneu; Scheme 1), which are modular derivatives of
MONOPhos-type ligands.[15] Depending on the characteristics
of the METAMORPhos ligands and phosphoramidite ligands
Triethylammonium salts of METAMORPhos ligands Lan1À–
Lan5À were synthesized in both enantiomeric forms by the
simple condensation of 1,1’-binaphthyl-2,2’-diyl phosphoro-
chloridate (BINOL P-Cl) with sulfonamide in the presence of
triethylamine according to reported procedures.[5c,16a] The elec-
tronic properties of the ligand were varied by changing the
substituent on the S atom (from weakly electron-withdrawing
4-butylphenyl to strongly electron-withdrawing trifluorometh-
ChemCatChem 2015, 7, 3368 – 3375
www.chemcatchem.org
3369
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
van der Waals radii (3.22 Š) and
suggests the presence of lone
pair–p interactions.[17] This inter-
action is difficult to observe in
solution, but it is expected to
also give additional stabilization
to the bimetallic structure if the
complex is dissolved. As a result
of the anticipated stabilization
effect of the close proximity of
these ligands in the dinuclear
structure, we expect that the
functionalization of the BINOL
moieties should disturb the for-
mation of such complexes.
Indeed, steric bulk on the 3,3’-
positions of the BINOL unit
(Lan3À) leads to the formation
of
a
mononuclear complex
(Scheme 2b) as indicated by the
doublet as the only signal that
can be observed in the 31P NMR
spectrum (Figure S5). The use of
octahydro-BINOL (Lan4À), which
has a similar bulk to the parent
BINOL, again results in a dinu-
clear complex (Scheme 2a) as
evidenced by the pattern in the
31P NMR spectrum (Figure 1a).
The coordination of Lneu li-
gands to [Rh(nbd)2]BF4 yields
Scheme 2. a) Negatively charged dinuclear complexes that arise from the coordination of METAMORPhos ligands
LanÀ to [Rh(nbd)2]BF4; b) negatively charged mononuclear complexes that arise from the coordination of bulky
METAMORPhos ligands LanÀ to [Rh(nbd)2]BF4; c) cationic complexes that arise from the coordination of amino-
acid-based phosphoramidite ligands Lneu to [Rh(nbd)2]BF4; d) complexes that can be formed if [Rh(nbd)2]BF4 is
exposed to a mixture of LanÀ and Lneu.
mononuclear
cationic
com-
yl) and the steric properties were tuned by introducing differ-
ent groups on the 3,3’-position of the BINOL moiety.
The neutral ligands Lneu1–Lneu5 were synthesized from
BINOL P-Cl and commercially available amino acids [(R)- and
(S)-leucine, -valine, -glycine, and -phenylalanine] according to
reported procedures.[12]
plexes, as expected for phosphoramidite ligands (Scheme 2c)
and as evidenced by the doublets observed in the 31P NMR
spectra (see Figure S4 for the spectrum of [Rh(Lneu3)2(nbd)]+
or Figure 1b for the spectrum of [Rh(Lneu3)2]+).
Spectroscopic studies
Before we applied mixtures of these ligands in catalysis, we
first explored their coordination chemistry under various condi-
tions. The coordination behavior of unfunctionalized BINOL-
METAMORPhos ligands (Lan1À and Lan2À) has been well es-
tablished (Scheme 2a).[5c,14a,c] If Lan1À (or Lan2À) is mixed with
[Rh(nbd)2]BF4 (nbd=norbornadiene) under a hydrogen atmos-
phere, a dianionic dinuclear complex is formed selectively. The
31P NMR spectrum of such a complex displays characteristic
signals that consist of a doublet-of-doublets at dꢀ135 ppm
(chelating ligand) and an AA’BB’XX’ pattern at dꢀ115 ppm
(bridging ligand). The crystal structures of these dinuclear com-
plexes exhibit very short distances between the BINOL moiet-
ies connected to geminal P atoms:[14c,17a] the distance between
the O atom from the BINOL moiety of the bridging ligand and
the C atom from the BINOL moiety of the chelating ligand is
less than 3.1 Š, which is significantly lower than the sum of the
Figure 1. 31P{1H} NMR spectra (recorded under 10 bar of hydrogen) of a) the
dinuclear homocomplex (HNEt3)2[Rh2((R)-Lan4)4], b) the homocomplex
[(Rh(R)-Lneu3)2]BF4, and c) the mononuclear heterocomplex [Rh((R)-
Lneu3)((R)-Lan4)].
ChemCatChem 2015, 7, 3368 – 3375
www.chemcatchem.org
3370
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Next, we studied the complexes formed if a 1:1 mixture of
various Lneu and LanÀ ligands are mixed with the Rh(nbd)
precursor under a nitrogen atmosphere. The 31P NMR spectrum
of the precatalyst made from a mixture of (R)-Lan4À and (R)-
Lneu3 shows two doublet-of-doublets, which indicates the for-
mation of the neutral heterocomplex [Rh(Lan4)(Lneu3)(nbd)].
These signals integrate to approximately 75%, and the signals
of [Rh(Lan4)2(nbd)]À and [Rh(Lneu3)2(nbd)]+ integrate to only
1.5 and 3%, respectively. Three unassigned doublets, which do
not appear in the spectra of the solutions that only contain ho-
mocomplexes, are also present (Figure S4). Next, we investigat-
ed the complexes that form if the nbd complexes are submit-
ted to 10 bar of H2. Under these conditions, a mononuclear
heterocomplex is the major species, as indicated by the two
doublet-of-doublets at d=149.8 and 134.4 ppm (Figure 1). The
spectrum displays a typical JPÀP coupling of 45.5 Hz, in line
with a complex in which the P atoms are cis to each other. MS
confirms the formation of a mononuclear heterocomplex, and
the triethylammonium adduct HNEt3[Rh(Lan4)(Lneu3)] is clear-
ly visible in the spectrum (m/z=1120.2). In addition, a dinuclear
[Rh2(Lan)2(Lneu)2] heterocomplexes complexes was confirmed
by MS (see Supporting Information). These coordination ex-
periments show that the use of mixture of ligands results in
the formation of different complexes, and the type of complex
formed (mono-/dinuclear, neutral/cationic) depends on the
ligand structures. This suggests that the use of ligand mixtures
gives rise to diverse sets of complexes that expand the library
of complexes for combinatorial catalysis.
Choice of substrates
To explore the potential in asymmetric hydrogenation of the
ligand mixtures based on METAMORPhos ligands, we chose
a diverse library of industrially relevant substrates with various
functional groups and coordination abilities (Scheme 3). A is
species can be identified by MS, that is, (H)2[Rh2(Lan4)3-
+
(Lneu3)]
and its acetonitrile adduct (eluent used for MS
measurements). This species may be present in a small amount
in solution as barely noticeable signals can be seen in the
31P NMR spectrum (dꢀ140 and 155 ppm; Figure 1c). In the
1H NMR spectrum of the heterocombination, hydride signals
that do not appear in the spectra of the homocombinations
are observed. A well-defined hydride signal at d=À21.8 ppm
(ddd,
JRhÀH =26.1 Hz) can be assigned tentatively to
HNEt3+[HRh(Lan4)(Lneu3)]À or [HRh(Lan4)(Lneu3)] (Figure S7).
These hydrides are present in a very small amount (~1%) ac-
cording to the relative integration.
Scheme 3. Industrially relevant substrates subjected to asymmetric hydroge-
nation.
We performed the analogous NMR spectroscopy of com-
plexes made from a 1:1 mixture of LanÀ and Lneu ligands that
bear the opposite chirality at their BINOL moieties ((S)-Lan3À
and (R)-Lneu3). Under a nitrogen atmosphere, the homocom-
plex [Rh(Lan)2(nbd)]À is present in a small amount (4%) and
[Rh(Lneu)2(nbd)]+ cannot be detected. Six other signals that
correspond to three different heterocomplexes are present in
the 31P NMR spectrum: three doublet-of-doublets, a doublet,
and a doublet-of-triplets. This doublet (assigned to JRhÀP) of
a triplet (assigned to JPÀP) indicates a complex with three (or
four) P ligands on one Rh atom: the nbd ligand is displaced
even in the absence of hydrogen; this has not been observed
for the homocombinations and reflects the difference in the
steric properties of the heterocomplexes compared to the ho-
mocomplexes. If 10 bar of hydrogen is applied to this mixture,
both homocomplexes [Rh(Lan)2]À and [Rh(Lneu)2]+ are formed
together with the heterocomplexes. The broadness of the sig-
nals and the fact that they are partially superimposed prevents
a proper interpretation of the spectrum. According to the rela-
tive integrations, the homocomplexes/heterocomplexes ratio is
roughly 1:1. Notably, two broad doublets are present at d=
118.7 and 113.3 ppm (JPÀRh =231 and 201 Hz, respectively),
which are typical shifts for bridging METAMORPhos ligands of
dinuclear complexes (e.g., Figure 1a).[5c,14b–f] The existence of
mononuclear [Rh(Lneu)(Lan)] complexes and dinuclear
a trisubstituted anionic substrate that is a precursor for chiral
carboxylic acid derivatives reported as building blocks for the
synthesis of an antitumoral prodrug[18a] and several adenosine
antagonists.[18b] The hydrogenation product of B (known as the
Roche ester) is a broadly applicable synthon used, for example,
to make the antitumoral drugs tedanolide[18c] and discodermo-
lide.[18d] Cyclic enamides C and D belong to the important class
of aminotetralin precursors. They can be used for the synthesis
of many drugs such as Sertraline (antidepressant) and Rotigo-
tine (used to treat Parkinson’s disease). Enamide E is a precursor
for the synthesis of CGP-55845, a GABA-B antagonist.[18e] F, G,
and H are precursors of b-amino acids. H is also precursor in
the synthesis of the VLA-4 antagonist S9059 (used in the treat-
ment of relapsing multiple sclerosis).[3f]
Initial ligand screening
For the initial evaluation of the combinatorial approach that in-
volves the METAMORphos ligands, we performed experiments
in which six ligand combinations based on two different anion-
ic ligands and one neutral ligand were applied. The anionic
ligand Lan2À was used in its two enantiomeric forms. A rela-
tively high catalyst concentration (2 mm) and low substrate-to-
catalyst ratio were used as initial screening was focused on the
determination of the enantioselectivity induced by the various
ChemCatChem 2015, 7, 3368 – 3375
www.chemcatchem.org
3371
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
catalyst mixtures. The preparation of the reaction mixtures, the
hydrogenation reaction (20 bar), and sampling for analysis
were performed by using an Accelerator SLT workstation from
Chemspeed Technologies (48 experiments in total). As expect-
ed, the substrates are produced in variable enantioselectivies,
which depends strongly on the type of ligand employed. The
most selective ligand mixture changes with the substrate
(Table 1).
neutral ligand. The use of pure anionic ligands leads to much
lower ee values and conversions.
For the hydrogenation of substrate H, the highest ee (50%)
is obtained with a mixture of (R)-Lneu1 and (R)-Lan2À, and
the highest conversion is achieved with a mixture of (R)-Lneu1
and (S)-Lan2À. Homocombinations of the anionic ligands lead
to racemic products, and homocombinations of the neutral
ligand lead to the product with a modest ee and low conver-
sion.
Substrate A is converted with the highest ee if a mixture of
(S)-Lan2À and (R)-Lneu1 is applied. The enantiomeric form of
the product is determined by the chirality of the BINOL unit of
Lan2À as the mixture (S)-Lan2À/(R)-Lneu1 leads to 21% ee and
the mixture (R)-Lan2À/(R)-Lneu1 leads to 43% ee of the oppo-
site enantiomer. The use of the neutral ligand leads to the
lowest conversion in which the racemic product is formed. The
homocomplex of (R)-Lan1À is as selective as the homocomplex
of (R)-Lan2À, but it is much less active: the conversion is 20%
lower with (R)-Lan1À.
For substrate B, all combinations give full conversion. The
homocomplexes give poor ee values (<10%). Interestingly, the
heterocombination of (S)-Lan2À and (R)-Lneu1 leads to the
highest ee (52%), whereas the combination of (R)-Lan2À and
(R)-Lneu1 leads to a racemic product.
Focused ligand optimization
Iterative procedures in combinatorial catalysis are typically
used to converge rapidly to an active and selective cat-
alyst.[6a,9e,16] We were curious to know if our system could be
employed in such an evolutionary approach. We extended the
ligand library by structural changes on the BINOL moiety for
the anionic METAMORPhos ligand. The neutral ligands were
varied systematically at their amino acid moiety. The robotic
workstation was used to perform 94 hydrogenations in parallel,
but the preparation of the catalyst solutions was performed in
a glovebox to minimize catalyst decomposition by air. For this
second run, the catalyst concentration was decreased to 1 mm,
the substrate concentration was increased to 100 mm, and the
other parameters were kept constant. Neutral and anionic li-
gands of opposite chirality at the BINOL unit were used for the
hydrogenation of substrates A and B, and ligands with the
same chirality at the BINOL were used for the reaction of C, F,
and H. As substrates D, E, and G were hydrogenated more effi-
ciently with homocombinations of ligands for the first run,
they were not considered for this optimization run.
For the hydrogenation of C, the combination of (R)-Lan2À
and (R)-Lneu1 leads to the highest ee (64%) with full conver-
sion. The other combinations show good conversions and low
ee values (25% or less).
For the hydrogenation of substrate D, the homocombination
of the neutral ligand leads to the highest ee by far (À88%).
The anionic complexes give very low ee values, and the hetero-
combinations lead to intermediate results.
In the case of substrate E, the three homocomplexes give
similar ee values (43%), and all the heterocombinations lead to
poor ee values, which indicates more active but less enantiose-
lective heterocomplexes.
The results presented in Figure 2 establish that varying the
amino acid moiety allows the fine-tuning of the catalytic
system (minor contribution), whereas changing the BINOL unit
of the METAMORPhos ligand has the biggest influence (major
contribution) on the catalytic outcome. The effect of these
contributions on the asymmetric hydrogenation of substrates
A–H will be discussed below.
For substrate A, if the unfunctionalized BINOL unit of META-
MORPhos is replaced with the methylated version Lan3À, the
ee is improved (from <50 to 73%), however, the conversion
decreases (from >89 to <62%). If the bigger trimethylsilyl
functionalized ligand Lan5À is used, the conversion decreases
For substrate F, the highest selectivity is obtained with het-
erocombination of anionic and neutral ligands in which the
BINOL moieties have the same enantiomeric form ((R)-Lan2À/
(R)-Lneu1; 76%). The application of the homocomplex of
Lneu1 also leads to a good ee, whereas the complex based on
the anionic ligands gave a low ee.
Substrate G is hydrogenated with the highest conversion
(89%) and the highest ee (76%) by the homocomplexes of the
Table 1. Initial hydrogenation results.
A
B
C
D
E
F
G
H
ee[a] Conv.[a] ee[b] Conv.[b] ee[b] Conv.[b] ee[b] Conv.[b] ee[b] Conv.[b] ee[b] Conv.[b] ee[c] Conv.[b] ee[c] Conv.[b]
[%]
[%]
[%]
[%]
[%] [%]
[%]
[%]
[%]
[%]
[%]
[%]
[%]
[%]
[%] [%]
(R)-Lan1À
18
20
80
98
À7 >99
20 >99
3
13
18
85
91
90
2
>99
À43 >99
À4 >99
À43 >99
À8
97
1
29
À2 73
(R)-Lan1À (R)-Lneu1
(R)-Lan2À
À26 >99
À43 >99
À5 51
20
7
17 >99
21 >99
À10 >99
7
>99
À4
À76
À57
À72
98
96
96
93
À5 12
<1 49
50 49
29 78
(R)-Lan2À (R)-Lneu1
2
>99
64 >99
<1
25
À33 >99
À33 >99
À88 >99
6
>99
À58 56
À37 43
À76 89
(S)-Lan2À (R)-Lneu1 À43 >99
52 >99
À6 >99
94
99
24 >99
À43 >99
(R)-Lneu1
1
50
10
4
Conditions: [Rh]=2 mm, [L]total =4.4 mm, [S]=50 mm, reaction time=18 h, pressure=20 bar, solvent=CH2Cl2. [a] Determined by chiral GC after methyla-
tion (see Supporting Information). [b] Determined by chiral GC. [c] Determined by chiral HPLC.
ChemCatChem 2015, 7, 3368 – 3375
www.chemcatchem.org
3372
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
lyst: in combination with Lan2À
and Lan4À, the use of Lneu2
and Lneu3 lead to the highest
conversion (>53%), the use of
Lneu1 and Lneu5 lead to inter-
mediate results (ꢀ40% conver-
sion), and the use of Lneu4
leads to the lowest conversion
(30–32%). Except for Lneu4,
which leads to a lower ee (by
~20%), the choice of the neutral
ligand does not have a major
impact on the ee in combination
with Lan2À and Lan4À: 87 or
88% ee are obtained with Lan2À
and 87–90% ee are obtained
with Lan4À.
Figure 2. Results of focused ligand optimization. For each ligand combination, the ee [%] is given on the left and
the conversion [%] is given on the right in brackets; [Rh]=1 mm, [L]=2.2 mm, [S]=100 mm, reaction time=18 h,
pressure=20 bar, solvent=CH2Cl2.
Ligand ratio study
to less than 12%. For the neutral ligands, Lneu4 gives the
lowest ee, Lneu5 gives the lowest conversion, Lneu2 leads to
the highest ee, and Lneu1 shows the highest conversion (com-
bined with Lan2À or Lan3À). In the case of substrate B, Lan3À
also gives better results in terms of enantioselectivity, and for
this substrate the conversion remains high. Bulky Lan5À also
gives a reasonable ee and full conversion, which is because B
is a geminally disubstituted alkene with low hindrance. In com-
bination with Lan2À, Lneu5 gives good results. In combination
with Lan3À, Lneu5 and Lneu3 lead to the highest ee (94–
95%). Lneu4 leads to the lowest ee if combined with Lan2À.
For the hydrogenation of C, the use of METAMORPhos with an
octahydro-BINOL backbone Lan4À leads to the highest enan-
tioselectivity and conversion (90% ee and 97% conversion if
combined with Lneu1); the enantioselectivity with this ligand
is on average 10% higher than that of the parent BINOL ligand
(Lan2À) at a similar conversion. Surprisingly, Lan5À gives the
product of opposite chirality (À13 to À31% ee). This result
cannot be because of the homocomplex of the amino-acid-
based ligand, as the homocomplex of Lneu1 leads to the
product with +25% ee (Table 1). For the hydrogenation of C,
the best neutral ligand is Lneu1, both in combination with
Lan2À and Lan4À. Both Lan2À and Lan4À in combination with
any neutral ligand, give full conversion for the hydrogenation
of substrate F. The use of Lan2À leads to the highest ee (À80
to À85%), followed by Lan4À (À75 to À83%). Lan3À is slightly
less selective (72–79% ee) and significantly less active. The
strong influence of the bulkiness of the BINOL moiety on the
conversion of this substrate is further illustrated by the poor
conversion obtained with the trimethylsilyl-BINOL-based
ligand. Again, changing the neutral ligand only has a minor in-
fluence on the ee if Lan2À is used as the anionic ligand. If it is
used in combination with Lan4À, the choice of the neutral
ligand shows a more profound effect: Lneu1 gives À75% ee
and Lneu2 gives À83% ee.
If mixtures of monodentate phosphorus ligands are used,
a mixture of complexes is expected to be formed. As such, the
optimal ratio to favor the presence of complex with two differ-
ent ligands (heterocombination) is 1:1. For this reason, in most
combinatorial studies, a ligand ratio of 1:1 is used. However, if
the mixture of complexes is not statistical or if one homocom-
plex is significantly more active than the heterocomplex, the
optimal ratio to obtain the highest selectivity can be far from
this 1:1 ratio.[6a,7] To gain an insight into these aspects, we per-
formed catalytic experiments in which the ligand ratios were
varied. For each substrate, we used the ligand combination
that gave the optimal ee and conversion (Figure 2): (R)-Lneu1/
(S)-Lan3À for A, (R)-Lneu3/(S)-Lan3À for B, (R)-Lneu1/(R)-
Lan4À for C, (R)-Lneu3À/(R)-Lan2À for F, and (R)-Lneu3/(R)-
Lan4À for H.
For each substrate, the ee and the conversion as a function
of the mole fraction of ligand LanÀ (the total concentration of
ligand [LanÀ]+[Lneu] was kept constant) is shown in Figure 3.
The plot obtained for substrate A shows a maximum ee if the
fraction of LanÀ was between 20 and 70%. The conversion
shows a broad maximum around the 1:1 ratio of LanÀ and
Lneu.
These results indicate clearly that for this catalytic system,
the heterocombination is significantly more active and selec-
tive as the results are relatively insensitive to the ligand ratio
used. The plot obtained for substrate B is substantially differ-
ent: although the ee is high between 40 and 70% of LanÀ,
below 40% the ee decreases rapidly. This suggests that the ho-
mocomplex of Lneu, which forms the product of opposite
enantioselectivity, shows significant activity compared to the
heterocomplex. In the case of substrate C, the plot shows that
the homocomplex of Lneu has a poor selectivity and a negligi-
ble activity compared to the heterocomplex.[19] Already with
10% of LanÀ, the ee reaches a plateau of ~90% ee. The
heterocombination is more active than the homocomplex of
LanÀ as the maximum conversion was obtained between 50
and 70% LanÀ.
The use of Lan2À and Lan4À gives the highest ee values and
conversions for the hydrogenation of substrate H. The amino
acid moiety has a strong influence on the activity of the cata-
ChemCatChem 2015, 7, 3368 – 3375
www.chemcatchem.org
3373
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Conclusions
The potential of METAMORPhos ligands in combinatorial catal-
ysis has been demonstrated. These ligands can form both
mononuclear [ML2] and dinuclear [M2L4] complexes that are
active in asymmetric hydrogenation. Here we expanded the
application of METAMORPhos ligands by using mixtures of li-
gands in combination with neutral phosphoramidite ligands as
a new combinatorial strategy to arrive at a diverse set of cata-
lysts. In situ spectroscopic studies show the formation of vari-
ous mononuclear and dinuclear heterocomplexes. If the anion-
ic ligand (Lan) and neutral ligand (Lneu) have the same chirali-
ty, mononuclear neutral heterocomplexes are by far the major
species, although monoanionic dinuclear heterocomplexes
have also been detected. If Lan and Lneu have the opposite
chirality, mixtures of complexes are obtained. As a result of
their modularity, METAMORPhos ligands are highly suitable for
iterative procedures, which are commonly used in high-
throughput screening. For most substrates hydrogenated in
this study, the application of the mixture of ligands resulted in
a more enantioselective formation of the product compared to
the use of the pure homocomplexes.
Acknowledgements
We acknowledge the Dutch National Research School Combina-
tion Catalysis Controlled by Chemical Design (NRSC-Catalysis) for
financial support and Dr. ir. J. I. v.d. Vlugt, Prof. Dr. B. de Bruin,
Dr. W. I. Dzik, and Dr. P. Dydio for fruitful discussions.
Keywords: asymmetric catalysis · high-throughput screening ·
hydrogenation · rhodium · supramolecular chemistry
[1] a) Asymmetric Catalysis on Industrial Scale: Challenges, Approaches and
Solutions (Eds.: H.-U. Blaser, E. Schmidt), Wiley-VCH, Weinheim, 2004,
pp. 1–15; b) D. J. Ager, A. H. M. de Vries, J. G. de Vries, Chem. Soc. Rev.
2012, 41, 3340–3380.
[2] For crop protection and fragrance industries enantioenriched (not
enantiopure) compounds are often sufficient.
[3] a) L. Horner, H. Siegel, H. Büthe, Angew. Chem. Int. Ed. Engl. 1968, 7,
942; Angew. Chem. 1968, 80, 1034; b) T. P. Dang, H. B. Kagan, J. Chem
Soc. Chem. Commun. 1971, 481; c) W. S. Knowles, M. J. Sabacky, B. D.
Vineyard, J. Chem. Soc. Chem. Commun. 1972, 10–11; d) R. Noyori,
Angew. Chem. Int. Ed. 2002, 41, 2008–2022; Angew. Chem. 2002, 114,
2108–2123; e) W. S. Knowles, Acc. Chem. Res. 1983, 16, 106–112; f) for
some recent reviews, see: P. Etayoa, A. Vidal-Ferran, Chem. Soc. Rev.
2013, 42, 728–754; J. J. Verendel, O. Pàmies, M. DiØguez, P. G. Ander-
sson, Chem. Rev. 2014, 114, 2130–2169.
Figure 3. Results of the ligand ratio optimization. [Rh]=0.5 mm,
[L]total =2.2[Rh], [S]=200*[Rh], reaction time=18 h, pressure=20 bar,
solvent=CH2Cl2.
The plot obtained for substrate F shows full conversion for
all experiments, except if the homocomplex of LanÀ is em-
ployed. The maximum ee is obtained as a plateau at a propor-
tion of LanÀ between 40 and 90%, which indicates that the
heterocomplex is more active and selective. For the hydroge-
nation of H, the homocomplex of Lneu is significantly more
active than the heterocombination, and the homocomplex of
LanÀ is almost inactive. As the homocomplex of Lneu gives
poor enantioselectivity, the highest ee is obtained if
[Rh(Lneu)2]+ is virtually absent in the mixture (this already
occurs if the LanÀ/Lneu ratio is 1).
[4] B. Schäfer, Chem. Unserer Zeit 2013, 47, 174–182.
[5] a) J. Halpern, Science 1982, 217, 401–407; b) H.-J. Drexler, W. Baumann,
T. Schmidt, S. Zhang, A. Sun, A. Spannenberg, C. Fischer, H. Buschmann,
D. Heller, Angew. Chem. Int. Ed. 2005, 44, 1184–1188; Angew. Chem.
2005, 117, 1208–1212; c) F. W. Patureau, S. de Boer, M. Kuil, J. Meeuwis-
sen, P.-A. R. Breuil, M. Siegler, A. L. Spek, A. J. Sandee, B. de Bruin, J. N. H.
Reek, J. Am. Chem. Soc. 2009, 131, 6683–6685; d) S. E. Clapham, A. Had-
zovic, R. H. Morris, Coord. Chem. Rev. 2004, 248, 2201–2237.
[6] a) J. A. F. Boogers, U. Felfer, M. Kotthaus, L. Lefort, G. Steinbauer, A. H. M.
de Vries, J. G. de Vries, Org. Process Res. Dev. 2007, 11, 585–591; b) M. J.
Johansson, S. Berglund, Y. Hu, K. H. O. Andersson, N. Kann, ACS Comb.
Sci. 2012, 14, 304–308; c) F. Ausfelder, L. A. Baumes, D. Farrusseng,
Catal. Today 2011, 159, 1; d) J. Meeuwissen, M. Kuil, A. M. van der Burg,
ChemCatChem 2015, 7, 3368 – 3375
www.chemcatchem.org
3374
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
A. J. Sandee, J. N. H. Reek, Chem. Eur. J. 2009, 15, 10272–10279; e) C.
Jäkel, R. Paciello, Chem. Rev. 2006, 106, 2912–2942.
Dekker, C. G. De Koster, P. W. N. M. Van Leeuwen, J. N. H. Reek, Chem.
Commun. 2006, 1700–1702.
[7] For key examples of successful monodentate phosphorus ligands, see:
a) M. van den Berg, R. M. Haak, A. J. Minnaard, A. H. M. de Vries, J. G.
de Vries, B. L. Feringa, Adv. Synth. Catal. 2002, 344, 1003–1007; b) M. T.
Reetz, Angew. Chem. Int. Ed. 2008, 47, 2556–2588; M. T. Reetz, Angew.
Chem. 2008, 120, 2592–2626; c) C. Monti, C. Gennari, U. Piarulli, J. G.
de Vries, A. H. M. de Vries, L. Lefort, Chem. Eur. J. 2005, 11, 6701.
[8] For reviews on supramolecular bidentate ligands, see: a) J. Meeuwissen,
J. N. H. Reek, Nat. Chem. 2010, 2, 615–621; b) R. Bellini, J. I. van der V-
lugt, J. N. H. Reek, Isr. J. Chem. 2012, 52, 613–629; c) B. Breit, Angew.
Chem. Int. Ed. 2005, 44, 6816–6825; Angew. Chem. 2005, 117, 6976–
6986; d) S. Carboni, C. Gennari, L. Pignataro, U. Piarulli, Dalton Trans.
2011, 40, 4355–4373.
[9] a) M. Durini, E. Russotto, L. Pignataro, O. Reiser, U. Piarulli, Eur. J. Org.
Chem. 2012, 5451–5461; b) L. Pignataro, C. Bovio, M. Civera, U. Piarulli,
C. Gennari, Chem. Eur. J. 2012, 18, 10368–10381; c) V. Agabekov, W.
Seiche, B. Breit, Chem. Sci. 2013, 4, 2418–2422; d) U. Gellrich, J. Huang,
W. Seiche, M. Keller, M. Meuwly, B. Breit, J. Am. Chem. Soc. 2011, 133,
964–975; e) J. Wieland, B. Breit, Nat. Chem. 2010, 2, 832–837; f) A. T.
Straub, M. Otto, I. Usui, B. Breit, Adv. Synth. Catal. 2013, 355, 2071–
2075.
[10] a) N. C. Thacker, S. A. Moteki, J. M. Takacs, ACS Catal. 2012, 2, 2743–
2752; b) T. Besset, D. W. Norman, J. N. H. Reek, Adv. Synth. Catal. 2013,
355, 348–352; c) M. Kuil, P. E. Goudriaan, P. W. N. M. van Leeuwen,
J. N. H. Reek, Chem. Commun. 2006, 4679–4681; d) M. Kuil, P. E. Gou-
driaan, A. W. Kleij, D. M. Tooke, A. L. Spek, P. W. N. M. van Leeuwen,
J. N. H. Reek, Dalton Trans. 2007, 2311–2320; e) V. F. Slagt, M. RÅder,
P. C. J. Kamer, P. W. N. M. van Leeuwen, J. N. H. Reek, J. Am. Chem. Soc.
2004, 126, 4056–4057; f) P. E. Goudriaan, M. Kuil, X.-B. Jiang, P. W. N. M.
van Leeuwen, J. N. H. Reek, Dalton Trans. 2009, 1801–1805; g) S. A.
Moteki, K. Toyama, Z. Liu, J. Ma, A. E. Holmes, J. M. Takacs, Chem.
Commun. 2012, 48, 263–265.
[14] a) F. W. Patureau, M. Kuil, A. J. Sandee, J. N. H. Reek, Angew. Chem. Int.
Ed. 2008, 47, 3180–3183; Angew. Chem. 2008, 120, 3224–3227; b) F. W.
Patureau, M. A. Siegler, A. L. Spek, A. J. Sandee, S. JugØ, S. Aziz, A. Ber-
kessel, J. N. H. Reek, Eur. J. Inorg. Chem. 2012, 496–503; c) F. G. Terrade,
M. Lutz, J. N. H. Reek, Chem. Eur. J. 2013, 19, 10458–10462; d) F. G. Ter-
rade, M. Lutz, J. I. van der Vlugt, J. N. H. Reek, Eur. J. Inorg. Chem. 2014,
1826–1835; e) S. Oldenhof, B. de Bruin, M. Lutz, M. A. Siegler, F. W. Pa-
tureau, J. I. van der Vlugt, J. N. H. Reek, Chem. Eur. J. 2013, 19, 11507–
11511.
[15] a) M. van den Berg, B. Feringa, A. J. Minnaard, J. G. de Vries (DSM IP
Assets B. V.), US 6989461, 2006; b) E. Alberico, W. Baumann, J. G. de V-
ries, H.-J. Drexler, S. Gladiali, D. Heller, H. J. W. Henderickx, L. Lefort,
Chem. Eur. J. 2011, 17, 12683–12695.
[16] a) A. M. Kluwer, R. J. Detz, Z. Abiri, A. M. van der Burg, J. N. H. Reek, Adv.
Synth. Catal. 2012, 354, 89–95.
[17] a) CCDC 931395 and 931396 contain the supplementary crystallograph-
ic data for this paper. These data can be obtained free of charge from
The Cambridge Crystallographic Data Centre; b) T. J. Mooibroek, P.
Gamez, J. Reedijk, CrystEngComm 2008, 10, 1501–1515.
[18] a) T. Suzuki, S. Hisakawa, Y. Itoh, S. Maruyama, M. Kurotaki, H. Nakagawa,
N. Miyata, Bioorg. Med. Chem. Lett. 2007, 17, 1558–1561; b) J. M. Hitch-
cock, S. M. Sorensen, M. W. Dudley, N. P. Peet (Merrell Pharmaceuticals
Inc.), US 5840729, 1998; N. P. Peet, N. L. Lentz (Merrell Pharmaceuticals
Inc.), US 5047534, 1991; N. P. Peet, N. L. Lentz, M. W. Dudley, A. M. L.
Ogden, D. R. McCarty, M. M. Racke, J. Med. Chem. 1993, 36, 4015–4020;
N. P. Peet, N. L. Lentz, E. C. Meng, M. W. Dudley, A. M. L. Ogden, D. A. De-
meter, H. J. R. Weintraub, P. Bey, J. Med. Chem. 1990, 33, 3127–3130;
c) A. B. Smith III, C. M. Adams, S. A. Barbosa Lodise, A. P. Degnan, Proc.
Natl. Acad. Sci. USA 2004, 101, 12042–12047; d) M. Freemantle, Chem.
Eng. News 2004, 82, 33–35; e) S. J. Mickel (Ciba-Geigy AG), European
Patent EP 0543780, 1993.
[11] a) L. Pignataro, B. Lynikaite, J. Cvengros, M. Marchini, U. Piarulli, C. Gen-
nari, Eur. J. Org. Chem. 2009, 2539–2547; b) K. Ohmatsu, M. Ito, T. Kunie-
da, T. Ooi, J. Am. Chem. Soc. 2013, 135, 590–593; c) K. Ohmatsu, M. Ito,
T. Kunieda, T. Ooi, Nat. Chem. 2012, 4, 473–477; d) H. Gulyµs, J. Benet-
Buchholz, E. C. Escudero-Adan, Z. Freixa, P. W. N. M. van Leeuwen, Chem.
Eur. J. 2007, 13, 3424–3430; e) D. J. Frank, A. Franzke, A. Pfaltz, Chem.
Eur. J. 2013, 19, 2405–2415.
[19] Notably, in some cases (especially for substrate C), the METAMORPHos
homocombination gives a slightly higher selectivity than that shown in
Table 1. This is explained by the sensitivity of the reaction towards air
and moisture, especially at the small scale we worked on. For the ex-
periments summarized in Table 1, the catalysts were not prepared in
a glovebox. These differences do not affect the conclusions drawn from
Table 1 nor the conclusions of this study.
[12] a) P.-A. R. Breuil, J. N. H. Reek, Eur. J. Org. Chem. 2009, 6225–6230; b) P.-
A. R. Breuil, F. W. Patureau, J. N. H. Reek, Angew. Chem. Int. Ed. 2009, 48,
2162–2165; Angew. Chem. 2009, 121, 2196–2199.
Received: June 2, 2015
[13] a) T. S. Koblenz, H. L. Dekker, C. G. de Koster, P. W. N. M. van Leeuwen,
J. N. H. Reek, Chem. Asian J. 2011, 6, 2431–2443; b) T. S. Koblenz, H. L.
Published online on August 20, 2015
ChemCatChem 2015, 7, 3368 – 3375
www.chemcatchem.org
3375
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim