Molecular Insights into the Ligand-Reactant Interactions of Pt Nanoparticles Functionalized with α-Amino Acids as Asymmetric Catalysts for β-Keto Esters

Article abstract of DOI:10.1002/cctc.201900238

The asymmetric hydrogenation of ?-keto esters over α-amino acid-functionalized Pt nanoparticles was explored in order to expand our understanding of ligand-reactant interactions underlying the chiral induction. A comprehensive investigation aimed at the quantification of the nonlinear effects demonstrated that for most of the ligands and reactants enantiodifferentiation is determined by 1 : 1 ligand-reactant interaction. However, attachment of phenyl substituents to the ligands or reactants likely involves the formation of more intricate intermediate complexes. We have shown that the asymmetric bias is sensitive to even small changes in the geometry of the ligand. Additionally, we have found that alkali metal cations, which balance the negative charge of the ligand's carboxyl group and originate from the metal hydroxide used for ligand functionalization, play a key role in the process of chiral induction. As the nature of the cation can be varied by simply changing the metal hydroxide used during functionalization, this finding opens an additional possibility to control the stereoselectivity by tuning the ligand-reactant interaction.

Full text of DOI:10.1002/cctc.201900238

Accepted Article
Title: Molecular Insights into the Ligand-Reactant Interactions of Pt
Nanoparticles Functionalized with α-Amino Acids as Asymmetric
Catalysts for β-Keto Esters
Authors: Anda Šulce, Nico Mitschke, Vladimir Azov, and Sebastian
Kunz
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemCatChem 10.1002/cctc.201900238
Link to VoR: http://dx.doi.org/10.1002/cctc.201900238
A Journal of
www.chemcatchem.org

ChemCatChem
10.1002/cctc.201900238
Molecular Insights into the Ligand-Reactant Interactions of Pt
Nanoparticles Functionalized with α-Amino Acids as Asymmetric
Catalysts for β-Keto Esters
[
a]
[b]
[c]
[a]
Anda Šulce, Nico Mitschke, Vladimir Azov, Sebastian Kunz*
[a]Institute of Applied and Physical Chemistry (IAPC), Center for Environmental Research and
Sustainable Technology, University of Bremen, Leobener Straße 6, 28359 Bremen, Germany
[b]Institute for Chemistry and Biology of the Marine Environment (ICBM), University of Oldenburg,
Carl-von-Ossietzky-Straße 9-11, 26111 Oldenburg, Germany
[c]Department of Chemistry, University of the Free State, PO Box 339, Bloemfontein 9300, South
Africa
*
Corresponding Author:
PD Dr. Sebastian Kunz
University of Bremen
Institute of Applied and Physical Chemistry (IAPC)
D-28359 Bremen
Tel: +49-421-218 63187
eMail: SebKunz@uni-bremen.de
1
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
Table of Contents
Understanding of ligand-reactant interactions: In the asymmetric hydrogenation of ß-keto esters
over α-amino acid-functionalized supported Pt nanoparticles the stereoselectivity is mainly
determined by a 1:1 ligand-reactant interaction and further influenced by small geometric changes in
the ligand structure as well as by the alkali metal cation coordinated to the ligand.
2
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
Abstract
The asymmetric hydrogenation of ß-keto esters over α-amino acid-functionalized Pt nanoparticles
was explored in order to expand our understanding of ligand-reactant interactions underlying the
chiral induction. A comprehensive investigation aimed at the quantification of the nonlinear effects
demonstrated that for most of the ligands and reactants enantiodifferentiation is determined by 1:1
ligand-reactant interaction. However, attachment of phenyl substituents to the ligands or reactants
likely involves the formation of more intricate intermediate complexes. We have shown that the
asymmetric bias is sensitive to even small changes in the geometry of the ligand. Additionally, we
have found that alkali metal cations, which balance the negative charge of the ligand’s carboxyl
group and originate from the metal hydroxide used for ligand functionalization, play a key role in the
process of chiral induction. As the nature of the cation can be varied by simply changing the metal
hydroxide used during functionalization, this finding opens an additional possibility to control the
stereoselectivity by tuning the ligand-reactant interaction.
3
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
Introduction
Due to the key role of supported precious metal catalysts in multiple industrial processes, the control
of selectivity of such catalysts is of fundamental importance for the improvement of resource
utilization.[1] Nanoparticles (NPs), which are used for the preparation of supported metal catalysts,
exhibit a variety of surface atoms (e.g. corner, edge, and terrace atoms) of different chemical
[
2]
character and activity. Furthermore, groups of adjacent surface atoms can form ensembles capable
[
3]
of distinctive reactivities. This leads to a broad variety of catalytically different sites on the surface
of a NP.[4] As all these different sites can transform reactants in different ways, the control of
selectivity is generally challenging for nanoparticle-based catalysts and requires strategies to modify
the catalytic surface.
The most successful approach to improve selectivities by modification of supported catalysts is the
use of bimetallic systems, which led e.g. to the development of lead free gasoline production
[
5]
[6]
process. Incorporation of a second metal into a NP causes changes of its electronic structure. This
in turn can alter binding energies of reactants as well as of intermediates and, as a result, can lead to
improved selectivities.[7] Furthermore, the number, size, and chemical identity of active atomic
ensembles can be directly manipulated, in particular when a catalytically active metal is diluted with
a catalytically less active or inactive metal.[8]
A more recent approach that is showing increasing impact in catalysis research constitutes the use of
organic molecules that are not soluble in the reaction medium and are strongly bound to the particle
[
9]
surface, also referred to as ligands, to control the catalyst selectivity. Similar as incorporation of a
second metal into a NP, the binding of ligands can electronically and geometrically alter the surface
and, as a result, tune the catalytic properties.[10] However, in contrast to bimetallic NPs, the use of
ligands is not restricted to modifications of the catalyst nature. In addition, interactions between the
ligand shell and the reactants are possible.[11,12] In this way, the reactivity of certain functionalities
can be inhibited parallel to the simultaneous increase of the reactivity of other functional
groups.[13,14] This feature opens the possibility for a completely unexplored potential to tune and
improve the chemoselectivity of supported catalysts in a way thus far only known for homogenous
catalysts.[15] Even chiral information can be transferred from the ligand shell to the activated
reactant,[16] which allows for stereoselective conversions. Supported ligand-functionalized catalysts
with more than 80% enantiomeric excess for specific reactants have been recently reported.[12]
Therefore, this relatively novel catalyst type is a significant contribution to the field of asymmetric
catalysis by metal nanoparticles, which is currently dominated by (i) dispersed colloidal non-
supported or supported NPs stabilized by phosphorous-based ligands known from homogeneous
catalysis[17] and (ii) supported NPs covered with chiral modifiers, i. e. chiral molecules soluble in the
reaction medium, especially chinchona alkaloids.[18]
4
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
Clearly, the key to utilize the enormous potential of ligands for supported catalysts is to understand
the synergy between ligands and reactants and the way the interactions between the two can be
utilized to tune catalyst selectivity. A fundamental question that needs to be answered in this context
is: does the reactant interact with a single ligand molecule forming a 1:1 complex, which governs the
stereochemistry of the product? Or does the ligand shell act as a collective entity, forming an
environment that determines the molecular interactions with the reactant like, e.g., the cavities in
the porous structure of a zeolite.[19] The later aspect would strongly hamper a rational catalyst design
because lateral interactions between ligands, which are challenging to rationalize and predict, would
play a dominant role in the control of selectivity. From a historical perspective, most groups have
primarily tried to understand and control selectivity by assuming 1:1 ligand-reactant interactions
rather than considering the ligand shell as an environment within which the reactant activation and
conversion take place. However, experimental evidence supporting that molecular reaction control
by ligands on NPs is determined by 1:1 ligand-reactant interactions is lacking.
Motivated by this problem and inspired by the success of groups from the field of homogeneous
catalysis, we here present a study on so-called nonlinear effects (NLE) that probes 1:1 ligand-reactant
interaction hypothesis as the selectivity controlling factor. Surprisingly, we have found that for most
of the investigated ligand-reactant combinations the control of stereoselectivity seems to be
primarily determined by a 1:1 ligand-reactant interaction. Based on this novel mechanistic insight, we
further investigated the influence of small structural changes of the ligand on the stereoselectivity of
the catalyst. Similar as in homogeneous catalysis, the stereoselectivity and, accordingly, the ligand-
reactant interactions were found to be very sensitive to even small geometric variations in the ligand
structure. Finally, we report new fundamental insights into the interaction of amino acid based
ligands with Pt NPs. Evidence is presented that in the active form of the catalyst a cation is
coordinated to the ligand’s carboxyl group. This cation, which can be easily varied during the process
of catalyst preparation, alters the ligand-reactant interaction, eventually allowing the tuning of the
catalyst’s stereoselectivity.
5
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
Results and Discussion
Validation of the 1:1 Stoichiometry for the Ligand-Reactant Interaction Model
We have previously introduced an interaction model for α-amino acid ligands and β-keto ester
reactants that predicts the origin of stereoselectivity to be determined by a 1:1 ligand-reactant
complex.[12,20] This model has been very effective for understanding the origin of stereoselectivity and
giving guidance to enhance the asymmetric bias. However, experimental evidence has been missing
to support the hypothesis that the selectivity of ligand-functionalized nanoparticle catalysts is mainly
determined by 1:1 ligand-reactant complexes. In order to address this fundamental question, the
presence of possible nonlinear effects (NLE) was explored. Testing for NLE is a sensitive tool in
homogeneous catalysis to probe if stereoselectivity is determined by a 1:1 ligand-reactant
complex.[21,22] For this purpose, NPs were functionalized with given mixtures of the R- and S-
enantiomers of proline (PRO), tert-leucine (tert-LEU), and phenylglycine (PHG) (see Fig. 1 for
structures) and then deposited onto γ-Al
2
O
3
. The supported ligand-functionalized Pt NPs, which are
1
.2 ± 0.3 nm in size (determined previously by transmission electron microscopy),[14,16] were tested as
catalysts for the hydrogenation of the simplest ß-keto ester methyl acetoacetate (MAA) and the
sterically more demanding methyl 4,4-dimethyl-3-oxovalerate (MDOV) and ethyl 3-oxo-3-
phenylpropanoate (EOPP) (see Fig. 1). The resulting enantiomeric excess (ee) of the product was
plotted vs. the ee of the ligand used for functionalization of the NPs. A NLE is considered to be
present if the resulting plot deviates significantly from a linear function.
Figure 1. Catalytic hydrogenation process and structures of reactants and ligands.
6
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
Taking into account the preparation procedure of the ligand-functionalized Pt NPs and our so far
achieved knowledge about the hydrogenation of ß-keto esters over these catalysts, there are likely
three possible reasons for the occurrence of NLEs in these systems. The first concerns the catalyst
structure only. The ratio of the two ligand enantiomers in solution, used to prepare the
functionalized NPs, does not necessarily reflect the actual enantiomer ratio on the particle surface.
Depending on the solvent and the ligand nature, aggregates may form due to ligand-ligand
interactions. For example, the formation of ligand H-bonded dimers, a known phenomenon for
carboxylic acids, can lead to three combinations RR, SS, and RS. While RR and SS are enantiomers
that are hence energetically identical, RS is diastereomeric to RR and SS. As a result, RS exhibits a
different energy than SS and RR. Although formation of H-bonded dimers of amino acids in our case
is rather unlikely due to strong basicity of the media, similar processes may lead to
enantiodiscriminative formation of other types of aggregates in the solution, e.g. micelles in the case
of amino acids with lipophilic residues. This means that the ratio of “single, free” ligands can deviate
from the total ratio in solution. Assuming that only “single, free” ligands can be bound to the NPs,
but not dimers or aggregates, the resulting ligand ee on the particle surface will vary from that in
solution. Moreover, on the NP surface adjacent ligands may interact with each other. Such lateral
interaction will be sensitive to the ligand configuration and influence the overall energetics of the
ligand shell formation. During functionalization the ligand shell with the lowest energy will be
formed. If ligand-ligand interactions are strong, the ligand ee on the surface may certainly not be the
same as that in solution.
The second reason for a NLE can be that the origin of enantiodifferentiation is not exclusively based
on a 1:1 interaction complex of the reactant with a ligand. A 1:1 complex leads to two diastereomeric
transition states. As these two interaction modes have different free energies, the overall reaction
becomes stereoselective with the resulting ee being determined by the difference in free energy of
the two transition states (ΔΔG‡).[23] If, for instance, the two 1:1 ligand-reactant complexes experience
a chiral environment such as an adjacent ligand or an adsorbed reactant (note that the adsorbed
prochiral reactant is chiral due to its interaction with a ligand) they further change in energy. The
resulting energy changes for the two states are however different as they become diastereomeric.
This results in a nonlinearity for the product ee when plotted vs. the ligand ee. The stronger this plot
deviates from linearity the more complex is the interaction of the reactant with the ligand shell.
Additional insights in the structure of the catalytic active species reveal the consideration of curve
shapes. Kagan et al. have developed mathematic models (ML
n
where M is the metal center and L
n
refer to n ligand molecules) for the calculation of curve shapes by taking into account the relative
reactivity and the relative concentrations of homochiral and heterochiral complexes.[24,25] These
models are helpful as a guide for the further interpretation of the observed NLEs.
7
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
The third possibility for a NLE is that the product ratio is not merely determined by two
diastereomeric reaction pathways proceeding via two ligand-reactant complexes, but is also
influenced by additional, completely different pathways. Since these other reaction pathways will
show different dependencies for the product ee as the ligand ee on the surface is being varied, a very
complex, unpredictable relation between product ee and ligand ee must be expected. Based on
recent kinetic investigations, we can however exclude this possibility.
In homogeneous catalysis, the magnitude of NLEs depends strongly on the structure of reactant and
ligand.[26] The possibility of deviations from a simple 1:1 ligand-reactant complex is expected to
increase with complexity of the ligand and reactant. The results obtained for MAA, which is the
smallest and least complex of the explored reactants, are hence discussed first (see Fig. 2).
a)
b) 50
c)50
4
3
2
1
0
4
3
2
1
0
0
0
0
0
40
30
20
10
0
0
0
0
0
0
20 40 60 80 100
ee(L-PRO) / %
0 20 40 60 80 100
0 20 40 60 80 100
ee(L-tert-LEU) / %
ee(L-PHG) / %
Figure 2. Relation between the product ee and the ligand ee for hydrogenation of MAA over Pt NPs
functionalized with PRO (a), tert-LEU (b), and PHG (c). The solid lines are fits of the experimental data that
serve merely as guides to the eye.
For PRO- and tert-LEU-functionalized Pt NP catalysts, the ee clearly shows a linear dependence and
no NLE. We thus conclude for these materials: i) the ratio of the two ligand configurations on the
surface does not vary from the ratio in the solution used to functionalize the NPs, ii) as expected, the
hydrogenation of MAA seems to proceed via only two suggested diastereomeric reaction pathways
that iii) are indeed determined by a 1:1 ligand-reactant interaction. For PHG-functionalized Pt NPs, a
minor NLE is found. Given the fact that no NLE was obtained for MAA hydrogenation over the other
two catalytic systems, we assume that in the case of PHG-functionalized Pt NPs the ratio of the two
ligand configurations on the catalyst surface deviates from that used for the catalyst preparation. As
above mentioned, such artifacts are expected to arise from ligand-ligand interactions in solution or
8
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
on the surface during preparation. In this context, the phenyl group may play a decisive role.
Functionalization is performed in water, a very polar solvent that poorly solvates nonpolar groups
(e.g. hydrocarbons, aromats). Likely, the hydrophilic phenyl substituents of PHG ligands may interact
preferentially with each other (the strength of π-π-interactions is up to 12 kJ/mol for a pair of
interacting benzene rings[27]), a phenomenon known from surfactants that drives the formation of
micelles. Positive NLE effect (asymmetric amplification, higher product ee in relation to ligand ee)
implies that the catalyst surfaces were modified by enantiomerically enriched ligand.
For hydrogenation of the very bulky reactant MDOV, no NLE is obtained over PRO-Pt NPs with
respect to the limit of experimental accuracy, while for tert-LEU-Pt NPs a modest NLE is found (see
Fig. 3).
a) 70
b) 8
c) 70
60
50
40
30
20
10
0
0
0
6
5
4
3
2
1
0
0
0
0
0
0
0
7
60
5
4
3
2
1
0
0
0
0
0
0
0
20 40 60 80 100
0 20 40 60 80 100
0 20 40 60 80 100
ee(L-PRO) / %
ee(L-tert-LEU) / %
ee(L-PHG) / %
Figure 3. Relation between the product ee and the ligand ee for hydrogenation of MDOV over Pt NPs
functionalized with PRO (a), tert-LEU (b), and PHG (c). The solid lines are fits of the experimental data that
serve merely as guides to the eye.
MDOV is much bulkier reactant than MAA that exhibits a significantly greater steric demand than the
latter due to tert-butyl group. This increases the probability of lateral interactions and in turn reduces
the likelihood of a 1:1 ligand-reactant complex as the sole stereoselectivity determining factor.
However, since the NLE for MDOV hydrogenation over tert-LEU-Pt NPs is only modest,
enantiodifferentation still seems to be determined mainly by a 1:1 ligand-reactant complex. The
simple symmetric curve shape suggests that additionally there are ligand-reactant complexes which
[
24]
interact with one adjacent ligand or reactant molecule (ML model ). In homogeneous catalysis,
2
London dispersion interactions are known to be an important stereoselectivity governing factor
when utilizing bulky reactants and ligands.[28] In fact, it has recently been shown that such
9
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
interactions also play a decisive role for our catalytic systems when ligands and reactants with bulky
alkyl substituents are applied.[12] We, therefore, assume that the slight deviation from a linear
behavior may result from attractive van der Waals interactions (London dispersion) between the tert-
butyl substituent of the reactant or ligand of the ligand-reactant interaction complex and second
adjacent ligand or reactant molecule. For PHG-functionalized Pt NPs, a stronger deviation from a
linear dependence is observed. This indicates that deviations from a 1:1 model may become
significant. We have recently shown experimental evidence for the presence of π-π-interactions
between aromatic ligands and reactants for the here discussed catalytic systems.[12] It is hence
assumed that the increasing deviation from a 1:1 ligand-reactant interaction is driven by the
presence of such π-π-interactions that may allow for lateral interactions with additional adjacent
ligand or reactant molecules. An interesting aspect is the unconventional curve shape. There is
clearly a positive NLE, but only until an ee value of 80% is reached for PHG ligand. Then a linear
relationship occurs until enantiomerically pure PHG is used. A similar behavior has been reported for
the asymmetric oxidation of sulfide in the presence of diethyl tartrate where a strong negative NLE
was prevailing up to a value of 70%, but the linearity was regained by using diethyl tartrate of optical
purity higher than 70%.[21] An explanation for this behavior may be interaction of some 1:1 ligand-
[
24]
reactant complexes with three adjacent ligand or reactant molecules (ML
4
model ), i.e. considering
(S) . However, as
the configuration there are five possible combinations (R) , (S) , (R) S, (S) R, and (R)
4
4
3
3
2
2
concluded above for the reactant MAA, the ligand ee of PHG on the surface is expected to deviate
from the ligand ee in solution. Hence, the stronger NLE for MDOV hydrogenation over PHG-Pt NPs
compared to that over tert-LEU-Pt NPs is likely to result from two effects occurring in parallel. As the
two effects cannot be distinguished from each other, we do not further stress the discussion of these
results but rather conclude that this is the first of the here reported ligand-reactant combinations
that shows a significant NLE.
The results obtained for the hydrogenation of EOPP over PRO-Pt NPs differ from those for the two
other reactants (compare Fig. 4a to 2a and 3a). While no NLE was concluded for MAA and MDOV, a
minor NLE is seen for EOPP. The effect is still small enough justifying the conclusion that the
stereoselectivity is predominantly determined by a 1:1 ligand-reactant complex. However, the result
reveals an interesting aspect considering the previously discussed results for PHG-functionalized NPs.
It seems that the presence of phenyl substituents within the ligand or the reactant increases the
probability of NLEs (see. Fig 2c and 3c), most likely due to π-π-interactions between the aromatic
moieties.
1
0
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
a)60
b) 70
c)7
0
0
6
5
4
3
2
1
0
0
0
0
0
0
0
5
0
0
0
0
0
0
6
4
3
2
1
50
4
3
2
1
0
0
0
0
0
0
20 40 60 80 100
0 20 40 60 80 100
0 20 40 60 80 100
ee(L-PHG) / %
ee(L-PRO) / %
ee(L-tert-LEU) / %
Figure 4. Relation between the product ee and the ligand ee for hydrogenation of EOPP over Pt NPs
functionalized with PRO (a), tert-LEU (b), and PHG (c). The solid lines are fits of the experimental data that serve
merely as guides to the eye.
The NLE of tert-LEU-functionalized Pt NPs is slightly more pronounced as that of PRO-Pt NPs, but still
moderate. This implies that a 1:1 interaction is still sufficient to explain the asymmetric control for
EOPP hydrogenation over these NPs. Interestingly, the NLE curves obtained for the hydrogenation of
EOPP over PRO- and tert-LEU-Pt NPs are not symmetrical (i. e. the maximum is not at 50% ee). Such
complicated curve shapes may originate from a lateral interaction of some of 1:1 ligand-reactant
[
24]
complexes with at least two further ligand or reactant molecules (ML
3
or ML model ).
4
The NLE of PHG-functionalized Pt NPs is clearly more pronounced as those for PRO- and tert-LEU-Pt
NPs, but as for MDOV we cannot deconvolute the contribution of the catalyst’s surface ligand ee. We
therefore do not further discuss the results obtained for PHG-Pt NPs in combination with EOPP.
To conclude, the NLE test experiments suggest that for ligands and reactants without bulky
substituents, the sole stereoselectivity determining factor is 1:1 ligand-reactant interaction. For
ligands and reactants possessing sterically demanding alkyl substituents, the stereoselectivity is still
predominantly determined by a 1:1 ligand-reactant complex but effects due to lateral interactions of
these 1:1 complexes with adjacent ligand or reactant molecules become evident. More pronounced
NLEs and thus stronger deviations from 1:1 complexes as the stereoselectivity controlling factor are
observed for ligands and reactants with phenyl substituents, likely due to the ability to build up π-π
interactions. Furthermore, evidence was found that the presence of a phenyl substituent within the
ligand influences even the catalyst preparation. As a result, the ligand ee on the particle surface
deviates from that used for NP functionalization, which leads to complex systems and diminishes our
predicting capacity for these systems.
1
1
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
Effect of Geometric Changes within the Ligand
As previously reported, ß-keto ester reactants and α-amino acid ligands are privileged combinations
to achieve high stereoselectivities with supported ligand-functionalized NP catalysts. We could
explain this selectivity by a 1:1 ligand-reactant interaction model, which has now been supported by
NLE tests (see above). As demonstrated experimentally and by means of computational methods,
amine bound hydrogen substituents become acidic as amines are bound to late d-metals (e.g. Pt).
Such protons coordinate and activate ketones leading to an enhanced reaction rate.[29] An ester can
coordinate in two ways to the ligand’s N-H proton: either with the ester group being oriented
towards the side of the ligand that binds the COOH group or vice versa, thus affording two
diastereomeric adsorption modes (compare Fig. 7). The pronounced stereoselectivity, which can be
achieved for β-keto ester reactants in combination with α-amino acid ligands, stems from the fact
that in one of the two diastereomeric adsorption modes the reactant’s ester group can interact with
the ligand’s carboxyl group. This leads to a so-called two-point binding and a stabilization of this
adsorption mode compared to its diastereomeric counterpart. The stereoselectivity is determined by
‡
the difference in free energy (ΔΔG ) of the two transition states formed from these diastereomeric
1
:1 adsorption complexes. An enhanced two-point binding mode will lead to a stronger
enantiodifferentation.[20]
As the stereoselectivity is determined by molecular interactions between the ligand and the reactant,
the ee should be sensitive to small changes in the geometry of the ligand-reactant interaction
complex. In order to verify this hypothesis, a four- and a six-membered ring analogues to PRO were
tested as ligands for Pt NPs (see Fig. 5 for structures). These three ligands exhibit quite similar steric
demand, but due to ring tensions the distance and angle between the amino and the acid group
differ, which may cause geometric variations within the ligand-reactant interaction mode.
1
2
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
O
*
N
OH
H
L-azetidine-2-carboxylic acid
n = 4
OH
*
N
H
O
L-proline (PRO)
n = 5
*
OH
N
H
O
L-pipecolic acid
n = 6
Figure 5. Effect of the ring size of the ligand on the stereoselectivity. Structures of the ligands are shown on the
right. The solid lines are fits of the experimental data that serve merely as guides to the eye.
The obtained stereoselectivities (see Fig. 5) clearly show that PRO is the best ligand irrespective of
the reactant structure. This seems reasonable considering the fact that PRO is also the best chiral
organocatalyst used in multiple asymmetric transformations.[30] For example, List et al. have
compared PRO with its four- and six-membered ring analogues in the direct asymmetric aldol
reaction between acetone and a variety of aldehydes with the result that the pyrrolidine ring is
essential for an efficient stereoselective conversion.[31] A further similarity to observations in
organocatalytic reactions[32] is the fact that the ligand with six-membered ring shows a low sensitivity
to steric bulkiness of the reactant unlike both other ligands, which produce much higher ee for the
bulky reactants MDOV and EOPP than for MAA.
Therefore, the high deviations of the ees obtained for the hydrogenation of ß-keto esters over Pt NPs
functionalized with PRO compared to its four- and six-membered ring analogues indicate that the
ligand-reactant interaction and the resulting stereoselectivity are very sensitive to the distance and
angle between the amino and the carboxyl groups of the ligand.
1
3
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
The Role of Alkali Metal Ions
-
2
The chemical form of amino acids can vary from neutral (NH RCOOH) to anionic (NH
2
RCOO ), to
cationic (NH +
RCOOH), or to zwitterionic (NH +
RCOO ), depending on the pH of the solution. In
-
[33]
3
3
aqueous solutions at pH = 7 amino acids appear as zwitterions with protonated amino group, which
is lacking the free lone pair at the nitrogen required for coordination to the late transition metal
surface.[34] Therefore, alkaline conditions are needed for the deprotonation of the NH +
case of PRO) functionality. For aqueous solution of PRO, the pK value for the deprotonation of
carboxyl group is 1.99, whereas the pK group is 10.60. This
value for the deprotonation of the NH +
(or NH +
in
3
2
a
a
2
means that the carboxyl group of the amino acid is already fully deprotonated, when the pH value
becomes high enough for the deprotonation of the amino group. This trend generally holds for all
amino acids.
We have recently showed that the pH of the functionalizing solution during the ligand
functionalization step influences the catalytic properties of the resulting catalysts[12,20] and, thus,
decided to further explore the state of the carboxylic group of the ligand. In our preparation
procedure, ligand functionalization is performed under basic conditions and the supported ligand-
functionalized Pt NPs are washed twice with ethanol to remove excess ligands. Successful removal of
non-binding ligands was demonstrated by NMR spectroscopy.[14] In early studies, when only a rather
low ee of 16% was obtained, we found this “preconditioning” of the as-prepared catalysts to be
essential. When catalyst rinsing was performed with acetone the reaction turned out to be much less
stereoselective. From this finding it was concluded that a protic solvent is of fundamental importance
to achieve meaningful stereoselectivities, which may be related not only to the removal of excess
ligands but also to removal of the excess NaOH and, possibly, partial re-protonation of the ligand’s
carboxyl group. For this reason, we thus far proposed in our ligand-reactant interaction model that
the carboxyl group is protonated, which allowed us reasonable mechanistic understanding.
Since these early studies we intensively worked on the functionalization of the NPs to improve
stereoselectivities and have achieved it with great success: ees of more than 80% are today feasible.
However, both ligand and NaOH concentrations used in our present preparation procedure vary
significantly from the initial recipe and are much higher (∆c(PRO) = 7.5 mM, ∆c(NaOH) = 27.5 mM).
This finding encouraged us to test the influence of catalyst rinsing prior to catalytic applications with
acidic protic (acetic acid), protic (ethanol), and non-protic (acetone) solvents on the stereoselectivity
(see Table 1). It is seen that the highest ee is reached by using catalysts rinsed with ethanol. The ees
obtained for the hydrogenation over Pt NPs rinsed with acetic acid or acetone are significantly lower.
In order to verify that the changes in stereoselectivity are not caused by a loss of surface-bound
ligands during samples rinsing, the chemoselectivity of the hydrogenation of EOPP was investigated.
This reactant possesses a phenyl group and a C=O group that may both undergo hydrogenation.
1
4
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
While PRO-Pt NPs are highly chemoselective toward the unsaturated alcohol (= exclusive
hydrogenation of the C=O group), over “ligand-free” Pt NPs undesired hydrogenation of the phenyl
substituent occurs resulting in a chemoselectivity of about 63% for the formation of the target
unsaturated alcohol.[20] If any of the solvents used for the samples rinsing causes a removal of
surface-bound ligands, a decrease in chemoselectivity was expected. For the hydrogenation of EOPP
over PRO-Pt NPs rinsed with different solvents, the stereoselectivity shows a similar behavior as in
the case of MAA (see Table 1) and a chemoselectivity of more than 99% is obtained without
production of any side products. This indicates that changes in stereoselectivity are not caused by
formation of ligand-free surface sites, but the surface remains densely covered with ligands.
Table 1. Sodium content of PRO-Pt NPs rinsed with different solvents and the corresponding enantioselectivity
for the hydrogenation of MAA and EOPP. An experimental error of ±2% has to be taken into account for the ee.
+
Additional, the Na /Pt molar ratio was calculated.
+
rinsing
Pt
Na+
[wt%] [mol]
Pt*
Na
Na /Pt
MAA
ee [%]
40
EOPP
ee [%]
72
[
wt%]
[mol]
3.95∙10-6
2.08∙10-4 52.55
5.43∙10-5 12.40
6.03∙10-5 14.36
2.76∙10-5 9.54
2
2
1
2
*
x ethanol
1.25
6.58
1.59
1.87
0.65
4.37∙10-6
4.19∙10-6
2.89∙10-6
x acetone
1.28
1.30
0.68
33
29
6
44
39
34
x acetic acid, 1x ethanol
x acetic acid, 1x ethanol
ligand-functionalized Pt atoms taking into account that ligand coverage of PRO-Pt NPs is 85%
Our results show that ethanol is the best solvent for catalyst rinsing to obtain high stereoselectivities.
Rinsing with acetic acid, which should be accompanied by at least partial reprotonation of the
ligand’s carboxyl group, leads to a decrease of the stereoselectivity. This implies that, in the case of
our standard catalyst preparation procedure, the negative charge of the carboxyl group of the ligand
+
is not balanced by a proton but presumably by Na that originates from the NaOH used for
functionalization. To test for this hypothesis, we determined the sodium concentrations of the PRO-
Pt NPs rinsed with different solvents (see Table 1). It can be seen that the sodium content of catalysts
rinsed with acetone or acetic acid is significantly lower than that of Pt NPs rinsed with ethanol. Thus,
+
we conclude that the high Na content of PRO-Pt NPs rinsed with ethanol is essential for reaching of
a high ee. If the particles are rinsed with acetic acid, the sodium ions present in the catalysts are
likely to be at least partially displaced with hydrogen, lowering its stereoselectivity. The second
rinsing with acetic acid lowers also the platinum content of the catalyst, most likely due to the partial
removal of Pt-NPs from the support, as the acetic acid turns dark after catalyst rinsing. On the other
hand, the high molar content of sodium ions in the case of ethanol-washed catalyst indicates that,
1
5
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
+
-
+
beside Na coordinated to COO group of the ligand, there should be excessive Na content in the
catalysts, which presumably resides on the support material.
To further test if the ligand actually exists in the form of a sodium salt on the metal surface, three
different alkali metal hydroxides (LiOH, NaOH, and KOH) were tested during particle functionalization
phase with PRO (see experimental section for details and see Fig. S1 for metal content in samples).
Alkali metal cations are all singly charged, but they are clearly different in size. If the cation takes part
in the ligand-reactant complex, change of the cation is expected to alter stereoselectivity. This
hypothesis was probed by application of catalysts prepared with LiOH, NaOH, and KOH for the
asymmetric hydrogenation of MAA.
Catalyst prepared from particles functionalized in the presence of NaOH led to a significantly higher
ee than LiOH and KOH (see Fig. 6). This result indicates that the counter ion of the metal hydroxide
used for the ligand functionalization process does not just remain a passive additive to the catalyst,
but plays an important role in the enantiodifferentiating process. Hence, the ion must contribute to
the ligand-reactant complex and, furthermore, its size plays a decisive role in the catalytic process.
This fact clearly supports the hypothesis that the charge of carboxyl group is not balanced by a
proton but by a cation originating from the metal hydroxide.
1
6
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
5
4
3
2
1
0
0
0
0
0
0
+
Na
+
K
+
Li
0
12 . 24680 76
95
114 133
ionic radius / pm
Figure 6. Effect of different alkali ions used for the functionalization of Pt NPs with PRO on the stereoselectivity.
The solid line is a fit of the experimental data that serve merely as a guide to the eye. The ionic radii were taken
from the literature.[35]
The functionalization of the Pt NPs is not the only preparation step where NaOH is present. Also the
synthesis of “ligand-free” Pt NPs that are subsequently functionalized with PRO is performed in the
presence of NaOH (see experimental section). To clarify that the cation being relevant for the
enantiodifferentiation originates from the ligand functionalization step and not from the synthesis of
the “ligand-free” NPs, we performed NP synthesis using LiOH followed by functionalization in the
presence of LiOH-, NaOH-, and KOH-PRO-solutions. No difference in the ees was found if LiOH or
NaOH were used for the synthesis (see Table S2). Accordingly, the cation used for the ligand
functionalization step is essential for the control of the stereoselectivity, but not the cation applied
for the synthesis of “ligand-free” NPs during the earlier stages of catalyst preparation.
Based on the above discussed experimental evidence that the carboxyl group is definitely not
protonated but coordinates a cation to maintain the charge balance, our previously introduced
model of reactant-ligand interactions should be modified accordingly. In previous studies, we
proposed a hydrogen bond between the C=O group of the ester and the OH group of the amino
acid.[12] However, as discussed above, it seems much more plausible that the deprotonated carboxyl
+
group is counter balanced by Na cation. As a result, the carboxylic group of amino acid and the C=O
1
7
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
+
group of the ester will attract the Na cation by electrostatic interactions forming two-point binding
complex (see Fig. 7a). This modification allows us to explain the strong variations of the
+
+
+
stereoselectivity when Na is replaced by either Li or K . As shown above for varying the ring size of
the ligands, the asymmetric bias is very sensitive to even small variations of bond lengths and angles.
+
+
+
The diameter of a Li cation is about 25% smaller than that of a Na cation, while a K cation exhibits a
6% greater diameter. Either changes, decrease or increase of cation size, may lead to destabilization
2
of the two-point binding complex (Fig. 7a) compared to the ligand-reactant complex formed with
+
Na . On the other hand, the diastereomeric binding mode (Fig. 7b) will not be strongly influenced by
the metal cation. As a consequence, the free energy of the transition state formed from the two-
‡
point binding mode will increase, which in turn causes a decrease of ΔΔG and the ee, as the size of
the cation is varied.
a)
b)
O
O
R
R
H
*
*
H O
H O Na
N
N
H
H
H
Na
R''
O
O

R''
O
O
O
R'
R'
O
Figure 7. Ligand-reactant interaction model featuring two diastereomeric H-bonding interaction modes
between α-amino acid (blue) and β-keto ester (black) on Pt NP surface (grey balls). The two-point binding mode
(
a, leads to the product with R configuration) is lower in free energy than the one-point binding mode (b, leads
to the product with S configuration). Please note that the molecule of β-keto ester does not lie flat, but tilted to
the surface and H is attached from the “bottom” side of the molecule affording R-isomer for the two-point
2
binding mode (a). Color codes for the attractive interactions: green – H-bonding interaction, red – London
dispersion interaction between R substituent of the ligand and R’ substituent of the reactant, brown –
Coulombic interaction of sodium cation.
+
For the case of ß-keto esters, Na seems to give the optimal fit between ligand and reactant.
However, for other reactants a different cation size may be more beneficial for the asymmetric bias.
If this holds, the choice of the cation may serve as an additional degree of freedom to fine-tune
+
ligand-reactant interactions and, in this way, enantiodiscrimination. The first example where Li was
+
indeed found to be better for the asymmetric bias than Na is the hydrogenation of the smallest α-
keto ester methyl pyruvate (see Fig. 8).
1
8
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
Figure 8. Hydrogenation of methyl pyruvate over PRO-functionalized Pt NPs coordinating different alkali metal
cations.
If NaOH or KOH are used for the functionalization of Pt NPs with PRO, (S)-enantiomer of the product
alcohol is primarily formed (ee = 5%). By using LiOH the configuration of the product is inverted from
(S)- to the (R)-enantiomer and an ee of 10% is obtained. Even though ees are very modest, the results
demonstrate the possibility to tune asymmetric biases by the variation of the cation. Further studies
+
+
are in progress to find examples where Li or K will allow for a stronger enantiodiscrimination
compared to Na+.
1
9
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
Conclusions
In this work, we presented new fundamental insights into the ligand-reactant interactions in the
asymmetric hydrogenation of ß-keto esters over α-amino acid-functionalized Pt nanoparticle
catalysts. Previously, we proposed a 1:1 ligand-reactant interaction model. Now experimental
evidence is provided from studies on nonlinear effects (NLE) that, at least in the absence of phenyl
substituents, the stereoselectivity is indeed mainly determined by 1:1 ligand-reactant complexes. In
contrast, introduction of a phenyl substituent within the ligand or the reactant leads to more
complex systems with rather unpredictable NLEs.
We have demonstrated that the ligand-reactant interaction and the resulting stereoselectivity are
very sensitive to small variations in distance and angle between the amino and the carboxyl groups of
the ligand. Furthermore, we have presented evidence that the charge of the ligand’s carboxyl group
is balanced by coordinating to a metal cation. The nature of this cation plays a decisive role in the
enantiodifferentiating process and we suppose that it may serve as an additional degree of freedom
to fine-tune ligand-reactant interaction and in this way stereoselectivity of the catalyst. More
detailed investigations to explore this hypothesis are currently in progress.
2
0
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
Experimental Section
Synthesis of “Ligand-Free” Pt Nanoparticles
Ligand-free” Pt NPs were prepared according to a slightly modified protocol of the ethylene glycol
EG) method, first described by Wang et al.[36] In a 250 mL glass flask, H
PtCl ∙H O (0.25 g, 40% metal,
“
(
2
6
2
ChemPur) was dissolved in ethylene glycol (25 mL, EG, ≥ 99.5%, Fluka) and a 0.5 M NaOH (99.6%,
VWR) solution in EG (25 mL) was added. For specific experiments, the same amount of LiOH (98%,
Alfa Aesar) was used instead of NaOH. The resulting mixture was vigorously stirred at 500 rpm and
heated to 150°C using a preheated oil bath. After about 5 min the color of the solution changed from
yellow to black indicating the formation of Pt NPs. To ensure quantitative reduction of the precursor,
the reaction mixture was kept at 150°C for 1.5 h and then cooled to ambient temperature. The
average particle size determined by transmission electron microscopy is 1.2 nm ± 0.3 and is
maintained in all following preparation steps.[14,16] 1 M aq. HCl (50 mL, VWR) was added to
precipitate the “ligand-free” Pt NPs. The precipitate was separated from the supernatant solvent by
centrifugation and washed once with 1 M aq. HCl (50 mL). Finally, the Pt NPs were re-dispersed in
cyclohexanone (100 mL, 99.9%, VWR).
Preparation of Ligand-Functionalized Pt NPs
For the functionalization of the “ligand-free” Pt NPs, a previously established phase transfer method
was used.[16] First, alkaline aqueous ligand solutions were prepared. The ligand and NaOH
concentrations needed for each ligand to obtain maximum stereoselectivity were determined
previously.[12,20] In order to optimize the concentration of LiOH and KOH (≥85%, Roth), Pt NPs
functionalized with L-proline at different hydroxide concentrations were tested for the
hydrogenation of methyl acetoacetate (MAA, see Fig. 1 for structure). The ligands and concentrations
used are: L-tert-leucine (tert-LEU, > 98%, TCI) with 20 mM NaOH and 16 mM ligand, L-phenylglycine
(PHG, 99%, Acros Organics) with 30 mM NaOH and 22 mM ligand, S-(-)-azetidine-2-carboxylic acid
with 40 mM NaOH and 16 mM ligand, L-Pipecolic acid with 50 mM NaOH and 16 mM ligand, and L-
proline (PRO, > 99%, TCI) with 40 mM NaOH, 40 mM LiOH or 30 mM KOH and 16 mM ligand (see Fig.
1
for ligand structures). The ligand solution (400 mL) was added to the previously prepared
dispersion of Pt NPs in cyclohexanone (100 mL). This gives a ligand-to-Pt ratio of 12.3 – 16.9,
depending on the particular ligand, to ensure saturation of the particle surface with ligands. The
resulting two phase mixture was vigorously stirred. The stirring time needed for a successful phase
transfer of the particles from the organic into the aqueous phase is 1 h for Pt NPs functionalized with
PRO and its analogs and 1.5 h for tert-LEU- and PHG-Pt NPs.[12] During this mixing, the initially black
organic phase turned clear while the aqueous phase became black. The black aqueous phase that
contains the ligand-functionalized NPs was separated from the organic phase in a separation funnel.
2
1
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
Deposition of Ligand-Functionalized Pt NPs
For the deposition of the as-prepared Pt NPs onto support material, Al
2
O
3
(Puralox SCCa 150/200;
Sasol, grain size = 200-500 µm) was added to the particle dispersion to give a nominal metal loading
of 2 wt%. The solvent was removed using a rotary evaporator (p = 20 mbar; T = 45°C). The supported
ligand-functionalized Pt NPs were rinsed with ethanol (99.9%, VWR) two times to remove free
ligands, dried under vacuum for about 2 min, and then applied for catalytic experiments. For specific
experiments, acetic acid (99%, VWR) or acetone (99.9%, VWR) were used for samples rising.
In order to determine the actual Pt loadings of the catalysts, the samples were digested with freshly
prepared aqua regia and analyzed by atomic absorption spectroscopy (AAS, Carl Zeiss Technology
AAS 5 FL). The Pt loading is approximately 0.6 wt% Pt for PHG-Pt NPs and 1.2 wt% for Pt NPs
functionalized with other ligands used in this work. For specified samples, also the sodium and
potassium content was determined. Experimental capabilities to determine the lithium were not
available.
Catalytic Hydrogenation
Hydrogenation experiments of methyl acetoacetate (MAA, 99%, Sigma Aldrich), methyl 4,4-dimethyl-
3
-oxovalerate (MDOV, >95.0%, TCI), ethyl 3-oxo-3-phenylpropanoate (EOPP, ≥97.0%, Sigma Aldrich),
and methyl pyruvate (MP, >97%, TCI) were carried out in five in-house designed stainless steel
autoclaves. The autoclaves were connected to the same H (Linde 5.0) gas line and the same
2
thermostat in order to perform five catalysis experiments with different reactants and/or catalysts in
parallel under identical experimental conditions. In a typical experiment each autoclave was loaded
with dioxane (9 mL, ≥99.9%, Carl Roth), reactant (1 mL) and catalyst (0.2 g). The autoclaves were
flushed three times with hydrogen and filled to a pressure of 20 bar. The experiments were
performed at a temperature of 293 K and at a stirring rate of 800 rpm. A stirring speed of 800 rpm
was determined to be required to eliminate artifacts that may arise from external diffusion
limitations. After the catalysis experiment (≥ 19 h for full conversion), the reaction mixtures were
centrifuged to separate the liquid-phase products from the catalyst. We have seen by carrying out
experiments for our previous publications[12,20] that for all ligands and reactants investigated in this
work the ee does not vary in a conversion range of 1 to 100%.
Product Analysis of Catalytic Experiments
The reaction mixtures were analyzed by a gas chromatograph (Perkin Elmer GC-Clarus 580) equipped
with a flame ionization detector (FID). For the analysis, the samples were diluted in 1:2 ratio with
acetone (99.9%, VWR). Injections were performed at 200°C with Helium (Linde, 5.0) as carrier gas.
2
2
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
In order to determine the enantiomeric excess (ee), a Lipodex E column (Macherey-Nagel, 27 m
length, 0.25 mm inner diameter, 0.25 µm film thickness) was used. Prior to analyzing the
hydrogenation product of MDOV the samples were derivatized with N-methylbis(trifluroacetamide)
(
25 µL, 97%, abcr) at 60°C for about 4 h. For the ee determination of MAA and MDOV, the GC oven
was held at 85°C for 20 min, then heated to 180°C at a rate of 30°C min-1 and kept at this
temperature for 10 min. For the ee determination of EOPP the oven temperature was held at 80°C
-
1
for 5 min and then increased to 139°C at a rate of 0.5°C min . After further heating to 180°C at a rate
-
1
of 30°C min the temperature was held for 11 min. For the ee determination of MP, the oven was
-
1
held at 70°C for 20 min, then heated to 180°C at a rate of 30°C min and kept at this temperature for
0 min.
1
Acknowledgements
S. K. acknowledges the “Fonds der Chemischen Industrie” (FCI) for financial support through a Liebig
Research Grant. The authors further thank the DFG (Deutsche Forschungsgemeinschaft) for financial
support (KU 3152/4-1).
Conflict of Interest
The authors declare no conflict of interest.
Keywords
Ligand-functionalized nanoparticles, ligand-reactant interactions, nonlinear effects, selectivity
control, stereoselective heterogeneous catalysis.
[
[
1] a) L. Wu, T. Moteki, A. A. Gokhale, D. W. Flaherty, F. D. Toste, Chem 2016, 1, 32–58; b) G. Centi,
S. Perathoner, Catal. Today 2003, 77, 287–297;
2] P. Serp, K. Philippot, Nanomaterials in Catalysis, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim,
Germany, 2013.
[
[
3] J. H. Sinfelt, Bimetallic catalysts. Discoveries, concepts and applications, WILEY, 1983.
4] a) S. Mostafa, F. Behafarid, J. R. Croy, L. K. Ono, L. Li, J. C. Yang, A. I. Frenkel, B. R. Cuenya, J. Am.
Chem. Soc. 2010, 132, 15714–15719; b) D. Gao, H. Zhou, J. Wang, S. Miao, F. Yang, G. Wang, J.
Wang, X. Bao, J. Am. Chem. Soc. 2015, 137, 4288–4291;
[
[
[
[
5] J. H. Sinfelt, Rev. Mod. Phys. 1979, 51, 569–589.
6] F. Gao, D. W. Goodman, Chem. Soc. Rev. 2012, 41, 8009–8020.
7] N. M. Wilson, P. Priyadarshini, S. Kunz, D. W. Flaherty, J. Catal. 2018, 357, 163–175.
8] a) J. H. Sinfelt, J. L. Carter, D. J. C. Yates, J. Catal. 1972, 24, 283–296; b) S. Siahrostami et al., Nat.
Mater. 2013, 12, 1137–1143;
2
3
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201900238
[
[
9] a) Y. Wang et al., J. Am. Chem. Soc. 2016, 138, 3278–3281; b) P. Liu, R. Qin, G. Fu, N. Zheng, J.
Am. Chem. Soc. 2017, 139, 2122–2131;
10] a) S. T. Marshall, M. O'Brien, B. Oetter, A. Corpuz, R. M. Richards, D. K. Schwartz, J. W. Medlin,
Nat. Mater. 2010, 9, 853–858; b) A. J. McCue, F.-M. McKenna, J. A. Anderson, Catal. Sci. Technol.
2
2
015, 5, 2449–2459; c) F.-M. McKenna, R. P. K. Wells, J. A. Anderson, Chem. Commun. 2011, 47,
351–2353;
[
[
11] K. R. Kahsar, D. K. Schwartz, J. W. Medlin, J. Am. Chem. Soc. 2014, 136, 520–526.
12] A. Šulce, J. Backenköhler, I. Schrader, M. D. Piane, C. Müller, A. Wark, L. C. Ciacchi, V. Azov, S.
Kunz, Catal. Sci. Technol. 2018, 8, 6062–6075.
[
13] M. Makosch, W.-I. Lin, V. Bumbálek, J. Sá, J. W. Medlin, K. Hungerbühler, J. A. van Bokhoven, ACS
Catal. 2012, 2, 2079–2081.
[
[
[
14] I. Schrader, J. Warneke, J. Backenköhler, S. Kunz, J. Am. Chem. Soc. 2015, 137, 905–912.
15] S. Kunz, Top. Catal. 2016, 59, 1671–1685.
16] S. Kunz, P. Schreiber, M. Ludwig, M. M. Maturi, O. Ackermann, M. Tschurl, U. Heiz, Phys. Chem.
Chem. Phys. 2013, 15, 19253–19261.
[
17] a) T. Yasukawa, H. Miyamura, S. Kobayashi, Chem. Soc. Rev. 2014, 43, 1450–1461; b) I. Cano, M.
J.-L. Tschan, L. M. Martínez-Prieto, K. Philippot, B. Chaudret, van Leeuwen, Piet W. N. M., Catal.
Sci. Technol. 2016, 6, 3758–3766; c) A. Gual, C. Godard, K. Philippot, B. Chaudret, A. Denicourt-
Nowicki, A. Roucoux, S. Castillón, C. Claver, ChemSusChem 2009, 2, 769–779;
18] a) F. Meemken, A. Baiker, Chem. Rev. 2017, 117, 11522–11569; b) X. Zuo, H. Liu, C. Yue, J. Mol.
Catal. A: Chem. 1999, 147, 63–72;
[
[
[
[
19] R. Gounder, E. Iglesia, Acc. Chem. Res. 2012, 45, 229–238.
20] I. Schrader, S. Neumann, A. Šulce, F. Schmidt, V. Azov, S. Kunz, ACS Catal. 2017, 7, 3979–3987.
21] C. Puchot, O. Samuel, E. Dunach, S. Zhao, C. Agami, H. B. Kagan, J. Am. Chem. Soc. 1986, 108,
2
353–2357.
[
[
22] T. Satyanarayana, S. Abraham, H. B. Kagan, Angew. Chem., Int. Ed 2009, 48, 456–494.
23] P. J. Walsh, M. C. Kozlowski, Fundamentals of asymmetric catalysis, Univ. Science Books,
Sausalito, Calif., 2009.
[
24] D. Guillaneux, S.-H. Zhao, O. Samuel, D. Rainford, H. B. Kagan, J. Am. Chem. Soc. 1994, 116,
9
430–9439.
[
[
[
[
25] C. Girard, H. B. Kagan, Angew. Chem. 1998, 110, 3088–3127.
26] Y. K. Chen, A. M. Costa, P. J. Walsh, J. Am. Chem. Soc. 2001, 123, 5378–5379.
27] M. O. Sinnokrot, E. F. Valeev, C. D. Sherill, J. Am. Chem. Soc. 2002, 124, 10887–10893.
28] a) A. Armstrong, R. A. Boto, P. Dingwall, J. Contreras-García, M. J. Harvey, N. J. Mason, H. S.
Rzepa, Chem. Sci. 2014, 5, 2057–2071; b) E. Lyngvi, I. A. Sanhueza, F. Schoenebeck,
Organometallics 2015, 34, 805–812;
[
29] K. Abdur-Rashid, S. E. Clapham, A. Hadzovic, J. N. Harvey, A. J. Lough, R. H. Morris, J. Am. Chem.
Soc. 2002, 124, 15104–15118.
[
[
[
30] B. List, Tetrahedron 2002, 58, 5573–5590.
31] B. List, R. A. Lerner, C. F. Barbas, J. Am. Chem. Soc. 2000, 122, 2395–2396.
32] P. H.-Y. Cheong, H. Zhang, R. Thayumanavan, F. Tanaka, K. N. Houk, C. F. Barbas, Org. Lett. 2006,
8
, 811–814.
[
[
[
[
33] S. M. Barlow, R. Raval, Surf. Sci. Rep. 2003, 50, 201–341.
34] M. D. Fryzuk, C. D. Montgomery, Coord. Chem. Rev. 1989, 95, 1–40.
35] E. Riedel, C. Janiak, Anorganische Chemie, 9. Auflage ed., de Gruyter, Berlin, 2015.
36] Y. Wang, J. Ren, K. Deng, L. Gui, Y. Tang, Chem. Mat. 2000, 12, 1622–1627.
2
4
This article is protected by copyright. All rights reserved.