Tuning enantioselectivity in asymmetric hydrogenation of acetophenone and its derivatives via confinement effect over free-standing mesoporous palladium network catalysts

Article abstract of DOI:10.1016/j.jcat.2014.03.004

The confinement effect on enantioselective hydrogenation of acetophenone and its derivatives over free-standing mesoporous Pd network catalysts was systematically studied. It was found for the first time that the enantiomeric excess (ee) could be effectiv

Full text of DOI:10.1016/j.jcat.2014.03.004

Journal of Catalysis 313 (2014) 113–126
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Tuning enantioselectivity in asymmetric hydrogenation of acetophenone
and its derivatives via confinement effect over free-standing mesoporous
palladium network catalysts
c
a
a
Yu Wang a,b, Na Su a, Lin Ye a, Yuanhang Ren a, Xueying Chen a, , Yujue Du , Zhenhua Li , Bin Yue ,
⇑
Shik Chi Edman Tsang d, Qiao Chen e, Heyong He a,
⇑
a Department of Chemistry and Shanghai Key Laboratory of Molecular Catalysis and Innovative Materials, Collaborative Innovation Center of Chemistry for Energy Materials,
Fudan University, Shanghai 200433, PR China
b School of Chemical and Environmental Engineering, Shanghai Institute of Technology, Shanghai 201418, PR China
c Shanghai Research Institute of Petrochemical Technology, SINOPEC, Shanghai 201208, PR China
d Wolfson Catalysis Centre, Department of Chemistry, University of Oxford, Oxford OX1 3QR, UK
e Department of Chemistry, School of Life Sciences, University of Sussex, Brighton BN1 9QJ, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
The confinement effect on enantioselective hydrogenation of acetophenone and its derivatives over free-
standing mesoporous Pd network catalysts was systematically studied. It was found for the first time that
the enantiomeric excess (ee) could be effectively tuned by altering the confinement effect optimized by
precise control of the topology, pore size, and lattice structure of mesoporous Pd catalysts. The double
gyroid structure with proper pore size and desired lattice structure formed by KBH4 reduction provided
suitable microenvironment to generate optimized confinement effect. The optimized catalyst exhibited
ee of 40–73% at 273 K under atmospheric pressure of H2. DFT study revealed that the major enantiomeric
product could be predicted by comparing relative energies of prochiral-R and -S complexes formed by
acetophenone derivatives with S-proline. The energetically favored complex led to the formation of
the corresponding enantiomer in excess upon hydrogenation, and ee was found to be linearly correlated
with the energy difference between prochiral-R and -S complexes.
Received 10 January 2014
Revised 5 March 2014
Accepted 8 March 2014
Keywords:
Confinement effect
Enantioselective hydrogenation
Mesoporous palladium
Acetophenone
Aromatic ketone
Proline
Ó 2014 Elsevier Inc. All rights reserved.
Molecular modeling
1. Introduction
the homogeneous counterparts [4], which have obvious advanta-
ges in green chemistry. Thus, heterogeneous enantioselective
The steady growing of high demand for efficacious and safe chi-
ral drugs has provided great inspiration to develop effective asym-
metric synthesis methods for preparing single-enantiomer
chemicals [1]. Heterogeneous enantioselective hydrogenation is
one of the most ideal methods for the synthesis of enantio-pure
compounds. It provides a versatile access to diverse types of
important chiral drug intermediates in pharmaceutical industries
such as chiral amines and alcohols by simply hydrogenating the
corresponding unsaturated prochiral imines and ketones [2].
Moreover, heterogeneous enantioselective hydrogenation process
is intrinsically atom-economical as compared to other reduction
methods [3]. In addition, heterogeneous catalysts possess many
technical processing advantages such as ease of handling,
separation, and recycling of solid catalysts in comparison with
hydrogenation is in the forefront of both academic and industrial
researches over recent years [2,5].
The success of developing effective heterogeneous enantiose-
lective catalysts relies on the structural optimization for a fine bal-
ance of reactivity and chiral selectivity in terms of conversion and
enantiomeric excess (ee). Here, we choose the enantioselective
hydrogenation of aromatic ketones (such as acetophenone) as a
model system for quantitatively evaluating the catalytic properties
of our newly developed free-standing mesoporous Pd network cat-
alysts in contrast to the conventional supported metal catalysts [6–
18]. Tungler et al. studied the enantioselective hydrogenation of
acetophenone over S-proline modified Pd/C, and the ee of 22%
was achieved at 78% conversion [6]. Perosa et al. studied the
enantioselective hydrogenation of acetophenone over cinchona-
modified Pt/C which gave 20% ee [7]. Baiker et al. studied the enan-
tioselective hydrogenation of acetophenone or ring-substituted
acetophenones over cinchonidine-modified Pt/Al2O3 [8,9] and
Rh/Al2O3 catalysts [10]. They found that the introduction of
⇑
Corresponding authors. Fax: +86 21 5566 5572.
E-mail addresses: xueyingchen@fudan.edu.cn (X. Chen), heyonghe@fudan.edu.cn
(H. He).
http://dx.doi.org/10.1016/j.jcat.2014.03.004
0021-9517/Ó 2014 Elsevier Inc. All rights reserved.

114
Y. Wang et al. / Journal of Catalysis 313 (2014) 113–126
electron-withdrawing functional groups in the aromatic ring in-
creased the enantioselectivity with up to 69% ee obtained in the
hydrogenation of 3,5-di(trifluoromethyl)acetophenone. Vetere
et al. investigated the enantioselective hydrogenation of acetophe-
none over chiral organotin-modified Pt/SiO2 catalysts which pro-
duced ca. 20% ee [11]. Basu et al. studied the asymmetric
hydrogenation of acetophenone over MCM-41 supported [Pt12
Pd surfaces. For a better understanding of the chiral recognition
on the Pd surface, the bonding and interaction of the chiral modi-
fier with acetophenone and its derivatives on a Pd surface were
also investigated using density functional theory (DFT) calcula-
tions. The structures of the enantio-differentiating intermediate
complexes were identified and their adsorption energies were
obtained.
(CO)24]
2ꢀ-derived catalysts, and 49% ee was achieved with the ace-
tophenone conversion of 40% [12]. Marzialetti et al. studied the
enantioselective hydrogenation of acetophenone on cinchoni-
dine-modified Ir/SiO2 catalyst, and up to 62% ee was claimed de-
2. Experimental
2.1. Catalyst preparation
spite the fact that it required
a very high cinchonidine
concentration [13]. Recently, it was reported that immobilized Ru
and Ir metal complexes with chiral diamine ligands exhibited good
enantioselectivity (60–96% ee) in the enantioselective hydrogena-
tion of aromatic ketones, but the leaching of Ru and Ir to the organ-
ic solvent may be a problem [14–20].
A series of mesoporous Pd network catalysts with 3D gyroid
structure and different pore sizes were prepared as follows. KIT-6
mesoporous silica templates were synthesized at different hydro-
thermal treatment temperatures of 308, 353, 373, and 403 K
according to the literature [33]. 1.00 g of KIT-6 was added into
40 mL of hexane and stirred vigorously for 30 min. An aqueous
solution of H2PdCl4 (0.56 M) was then added dropwise under vig-
orous stirring and ultrasonication. The volume ratio of the H2PdCl4
solution to the pore of the silica template was ca. 1.5. The excessive
solution was decanted and the resulting solids were dried over-
night at 393 K. The impregnation was repeated several times in or-
der to achieve high Pd loadings of ca. 25–55 wt.% with respect to
the silica template. The resulting powders were reduced by aque-
ous solution of KBH4 (0.3 M) or N2H4ꢁH2O at ambient conditions
or by H2 at 573 K for 2 h. The black solids obtained were treated
by hydrofluoric acid to dissolve the silica template. The products
were washed with distilled water until neutrality and then with
methanol three times. The catalysts were kept in methanol for
characterization and activity test.
It is known that the weak interaction between reactant and cat-
alyst support (such as van der Waals forces, hydrogen bonding, and
physical adsorption) is in the same order of magnitude as the en-
ergy difference between the two transition states of R- and S-prod-
ucts in chiral reactions [21]. Thus, the additional interactions with
the support can significantly affect the enantioselectivity. Recently,
Baiker and coworkers reported that by tuning the acid-base prop-
erties of the Al2O3 support, the enantioselectivity of hydrogenation
reaction can be altered [22,23]. Therefore, the presence of support
adds the complexity in the study of heterogeneous enantioselec-
tive hydrogenation reactions in elucidating the origin of the
enantioselectivity and the nature of the chiral recognition on metal
catalysts.
Recently, free-standing mesoporous metals with highly ordered
and interconnected networks have shown great potentials in het-
erogeneous catalysis owing to their synergetic advantages of na-
tive catalytic activity of the metal along with the large surface
area, well-defined mesostructures and controllable pore size distri-
butions [24–31]. For asymmetric hydrogenation reactions, free-
standing mesoporous metals have the following advantages.
Firstly, the effect of support is completely eliminated, which signif-
icantly reduces the complexity and difficulty in the study of the
mechanism of heterogeneous enantioselective hydrogenation
reactions. Secondly, the confinement effect within the mesoporous
metals can be precisely managed by controlling the pore shape,
pore size, and pore wall thickness of the original silica templates.
This provides us good opportunities to study the details of the con-
finement effect, allowing fine tuning of the enantioselectivity [21].
Previously, we reported that an S-proline-modified Pd nanoar-
ray catalyst with hexagonal mesostructure exhibited superior ee
over Pd black in the enantioselective hydrogenation of acetophe-
none due to the confinement effect [32]. However, the structural
factors influencing the confinement effect and their correlations
with the enantioselectivity remain to be clarified. The precise con-
trol of the nanoporosity and the lattice properties of the mesopor-
ous Pd catalyst may allow us to optimize the confinement effect
and further improve the enantioselectivity in the hydrogenation
process. Thus, the aim of the present work is to systematically
investigate the confinement effect over the free-standing mesopor-
ous Pd catalysts and to gain insights into enantioselective hydroge-
nation over such metal surfaces. We synthesized a series of free-
standing mesoporous Pd network catalysts with different topolo-
gies (space group of I4132, Ia3d, and P6mm), pore sizes, and lattice
structures by the hard templating method. With the enantioselec-
tive hydrogenation of acetophenone as a probe, the relationship
between the enantioselectivity and structure can be obtained.
Hydrogenation of ring-substituted acetophenones was also per-
formed for evaluating how the enantioselectivities were affected
by the interactions of prochiral compounds with the mesoporous
The parent KIT-6 templates, the Pd-loaded KIT-6 samples, and
the silica-free mesoporous Pd network catalysts were designated
as KIT-x, Pd-y/KIT-x, and Pd-x–y, respectively, where x refers to
the hydrothermal treatment temperature of original KIT-6 tem-
plate and y refers to the reducing agent. For comparison, ultrafine
Pd black and hexagonal Pd nanoarray [32] were also prepared by
similar procedure using KBH4 as the reducing agent. It should be
noted that the as-prepared mesoporous Pd network catalysts are
pyrophoric; thus, care must be taken to preclude them from air
exposure during sample handling and disposal.
2.2. Characterization
The chemical compositions of the catalysts were analyzed by
inductively coupled plasma-atomic emission spectroscopy (ICP-
AES, Thermo Elemental IRIS Intrepid). The FTIR spectra were re-
corded on a NEXUS 470 FTIR spectrometer. The BET specific surface
area and pore volume were measured by N2 adsorption at 77 K on a
Quantachrome Quadrasorb SI apparatus. Prior to the measurement,
the catalyst was transferred to a glass tube and degassed at 383 K
under N2 flow for 2 h. The pore volume was calculated from the
amount of N2 adsorbed at a relative pressure of 0.995. The pore size
distribution was calculated from the desorption branch of the iso-
therms using the Barrett–Joyner–Halenda (BJH) algorithm.
The active surface area was measured by pulsed CO chemisorp-
tion on a Micromeritics ChemiSorb 2750, assuming CO/Pd(s) = 0.6
and a surface area of 7.874 ꢂ 10ꢀ20 m2 per Pd atom [34]. The turn-
over frequency (TOF) was expressed as the number of acetophe-
none and its derivatives consumed per active surface Pd atom
per second.
The low-angle and wide-angle X-ray diffraction (XRD) patterns
were recorded on a Bruker D4 Endeavor X-ray diffractometer and a
Bruker AXS D8 Advance X-ray diffractometer, respectively, using
Cu Ka radiation. The catalyst with methanol was loaded in an in

Y. Wang et al. / Journal of Catalysis 313 (2014) 113–126
115
situ cell purged with Ar (99.9995%) during the analysis to avoid air
oxidation. The mean crystallite size of Pd was obtained from the
peak width of the Pd(111) reflection using the Scherrer equation
[35].
of 60 mL minꢀ1. Methanol was used as the solvent. The reaction
conditions were as follows: 0.02 mL of acetophenone, 0.40 g of S-
proline, 17 mL of methanol, 0.050 g of Pd catalyst, stirring rate of
1000 rpm to exclude diffusion effects, and reaction temperature
of 273 K. The products were identified by GC/MS. Conversion and
enantioselectivity were determined by a GC equipped with an Agi-
lent HP-CHIRAL (19091G-B213) capillary column and a FID detec-
tor. The enantioselectivity is expressed as ee (%) = 100 ꢂ |(R ꢀ S)|/
(R + S). The assignment of the absolute configuration of the product
was based on the retention times in GC analysis. Reproducibility of
ee was within 0.5%.
The mesostructure, selected-area electron diffraction (SAED)
pattern, and elemental analysis of the catalysts were acquired on
a JEOL 2011 transmission electron microscope (TEM) operated at
200 kV and fitted with an energy dispersive X-ray emission ana-
lyzer (EDX). High-resolution scanning electron microscopy
(HRSEM) images were taken on a Hitachi S-4800 ultrahigh-resolu-
tion scanning electron microscope operated at 20 kV equipped
with a cold field emission gun (FEG) and an in-lens electron optics.
The structure modeling of the mesoporous Pd replicas with the sin-
gle or double gyroid structure was carried out using the density
distribution function developed for gyroid mesostructure accord-
ing to the literature [36].
3. Results and discussion
3.1. Catalyst texture
Density functional theory (DFT) calculation of the adsorption
geometries and energies was performed using the SIESTA package
with numerical atomic orbital basis sets and Troullier–Martins
norm-conserving pseudopotentials [37,38]. The exchange–correla-
tion functional utilized is the generalized gradient approximation
PBE method, known as GGA-PBE [39]. A double-polarization basis
set (DZP) was employed. The orbital-confining cutoff radii were
determined from an energy shift of 0.01 eV. The energy cutoff for
the real space grid used to represent the density was set as 150
Rydberg. To further speed up calculations, the Kohn–Sham equa-
tions were solved by an iterative parallel diagonalization method
that utilized the ScaLAPACK subroutine pdsygvx with two-dimen-
sional block cyclically distributed matrix [40]. The Broyden method
was employed for geometry relaxation until the Cartesian forces on
each relaxed atom were all less than 0.05 eV Åꢀ1. A four-atomic-
layer periodic slab with top half of atoms allowed to relax was used
to simulate the Pd(111) surface. A (5 ꢂ 5) surface unit cell was
used to study the adsorptive conformations of S-proline with ace-
tophenone. The slabs were separated from their periodic images in
the direction perpendicular to the surface by interposing an ade-
quate amount of empty space (more than 15 Å). The Monkhorst–
Pack k-point sampling of (3 ꢂ 3 ꢂ 1) was utilized. The optimized
lattice constant was 3.95 Å for bulk Pd, which was consistent with
the experimental result [41]. Spin polarization was considered dur-
ing all the calculations. Adsorption energies were calculated as
follows:
The BET specific surface area, pore volume, pore diameter and
wall thickness of the parent KIT-6 templates prepared at different
hydrothermal treatment temperatures, are presented in Table 1.
The BET specific surface areas of all the KIT-6 are higher than
400 m2 gꢀ1. The pore volumes of the KIT-6 samples strongly de-
pend on the hydrothermal treatment temperature. Higher hydro-
thermal treatment temperature results in larger pore volume. It
is worth noting that for successful preparation of the mesoporous
Pd network catalyst, the exact pore volume of the parent KIT-6
templates has to be known, so the minimum volume of H2PdCl4
solution added can be adjusted accordingly. The wall thickness
was calculated according to the literature [42] based on the BJH
pore size obtained from the desorption branch of the isotherm
and the unit cell parameter obtained from XRD data. The wall
thickness of KIT-6 samples is also influenced by the hydrothermal
treatment temperature, decreasing from 5.8 nm at 308 K to 2.6 nm
at 403 K. The KIT-6 prepared at 308 K features the smallest meso-
pore diameter with the thickest wall, which is in line with the re-
ported results [33].
The nitrogen adsorption–desorption isotherms and the corre-
sponding pore size distribution curves of the KIT-6 samples are
shown in Fig. 1a and b, respectively. The isotherms of all the KIT-
6 samples (Fig. 1a) exhibit a type IV shape with a hysteresis above
P/P0 of 0.4, revealing the presence of mesopores. The relative pres-
sure at the appearance of the hysteresis loop increases with
increasing the hydrothermal treatment temperature, indicating
the increase in the pore dimensions. The narrow pore size distribu-
tion curves (Fig. 1b) indicate the uniform mesopores of the KIT-6
samples. The pore diameter of the KIT-6 samples increased from
4.0 to 9.8 nm when the hydrothermal treatment temperature
was increased from 308 to 403 K.
Eads ¼ Eproline—acetophenone complexþPdð111Þ ꢀ EPdð111Þ ꢀ Eproline
ꢀ Eacetophenone
where Eproline–acetophenone
is the total energy of the
complex+Pd(111)
coadsorbed S-proline and acetophenone on the Pd(111) surface,
EPd(111) is the energy of the isolated Pd(111) surface, Eproline and
Eacetophenone are the energy of the S-proline and acetophenone mol-
ecules in vacuum, respectively.
Fig. 2 displays the N2 adsorption–desorption isotherms and
pore size distribution curves of the replicated silica-free Pd cata-
lysts. Their BET specific surface areas, pore volumes and pore diam-
eters are summarized in Table 2. Generally, the BET specific surface
areas ranging from 45 to 80 m2 gꢀ1 were observed over the silica-
free Pd samples, which are significantly larger than the ultrafine Pd
black sample (23 m2 gꢀ1). ICP-AES analysis revealed that no boron
or potassium was detectable in the silica-free KBH4-reduced Pd
catalysts, excluding the contamination of Pd with potassium or
2.3. Activity test and product analysis
The activity test was carried out in a 100 mL glass flask with a
magnetic stirrer in H2 at atmospheric pressure with a flow rate
Table 1
Physicochemical properties of parent KIT-6 samples.
SBET (m2 gꢀ1
)
Vpore (cm3 gꢀ1
)
dpore (nm)
Unit cell parameter (nm)
Wall thicknessb (nm)
a
Sample
KIT-308
KIT-353
KIT-373
KIT-403
523
507
707
435
0.49
0.68
1.13
1.60
4.0
4.9
6.8
9.8
19.6
20.2
23.3
24.7
5.8
5.2
4.9
2.6
a
Calculated from the desorption branch of the isotherm using the BJH model.
Calculated according to Ref. [42].
b

116
Y. Wang et al. / Journal of Catalysis 313 (2014) 113–126
boron. As shown in Fig. 2a, the silica-free Pd samples also exhibited
a type IV isotherm with H3 hysteresis loops, indicative of their
mesoporous structure. For the Pd-308-KBH4 and Pd-353-KBH4
samples, only single narrow pore size distribution was observed
(Fig. 2b). The pore sizes are ca. 11–13 nm, which are much larger
than the wall thickness of the silica template (5.8–2.6 nm). The
presence of these large mesopores may relate to the special struc-
ture of the KIT-6 template [42,43]. It is well known that the 3D
bicontinuous gyroidal KIT-6 with space group of Ia3d consists of
two sets of interpenetrating mesopores and the pore interconnec-
tivity is influenced by the hydrothermal treatment temperature.
The low hydrothermal treatment temperature usually results in a
decreased interconnectivity [33]. Thus, during Pd replication from
KIT-6 prepared at low temperatures (KIT-308 and KIT-353), the
reduction and accretion of Pd occur in the two sets of mesopores
separately in KIT-6, forming uncoupled gyroid Pd networks with
the pore diameters of 13.1 and 11.0 nm, which are close to the
dimensions of two silica walls (5.8 and 5.2 nm) [42]. For the
Pd-373-KBH4 and Pd-403-KBH4 samples, although the large
mesopore at ca. 11–13 nm could be still observed, a new peak at
4.1 nm (Pd-373-KBH4) or 2.1 nm (Pd-403-KBH4) arose. Such a
bimodal pore size distribution is similar to the case of mesoporous
NiO and MoO2 replicas [44,45]. The diameter of the mesopores at
ca. 4.1 and 2.1 nm is close to the wall thickness of the parent
KIT-373 and KIT-403 templates. This represents a good quality of
replication, which is related to the higher interconnectivity in
KIT-6 prepared at high temperatures, allowing better mass
transportation [42,44].
Fig. 1. N2 adsorption–desorption isotherms (a) and pore size distribution curves (b)
of parent KIT-6 samples synthesized at different hydrothermal treatment
temperatures.
3.2. Topology
Fig. 3 shows the low-angle XRD patterns of the parent KIT-6
templates. All the KIT-6 samples exhibit typical diffraction peaks
that are characteristic of 3D gyroidal mesostructures [45]. The unit
cell parameters of the KIT-6 samples calculated based on the d(211)
-
spacing value from the XRD patterns are listed in Table 1. The unit
cell parameters are in the range of 19.6–24.7 nm and increase with
increasing the hydrothermal treatment temperature.
The low-angle XRD patterns of the silica-free Pd catalysts
strongly depend on the hydrothermal treatment temperature of
the parent silica templates. For KIT-6 synthesized at high temper-
atures (P373 K), similar diffraction patterns of the resulted Pd net-
work catalysts (Pd-373-KBH4 and Pd-403-KBH4) were achieved.
Fig. 4a shows the low-angle XRD patterns of samples at different
stages of impregnation using KIT-373 as a template. After the
incorporation of Pd in the parent KIT-373 template, the intensity
of all the diffraction peaks in Pd-KBH4/KIT-373 decreased dramat-
ically and only the strongest (211) reflection for Ia3d symmetry
[42] was clearly observed at 2h ꢃ 1.00°. However, by removing
the Pd networks from Pd-KBH4/KIT-373 using aqua regia, the dif-
fraction peaks of the KIT-373 silica template were fully recovered
despite a slight shift of all the diffraction peaks to larger angles.
Thus, the diffraction intensity difference between KIT-373 and
Pd-KBH4/KIT-373 is not due to the lowering of structural regularity
of the parent silica template. On the other hand, no significant
change of the diffraction peak was observed before (Pd-KBH4/
KIT-373) and after removal of KIT-6 template (Pd-373-KBH4).
Therefore, we can conclude that once loaded with Pd, the overall
diffraction peaks were dominated by the reflection from the Pd
structures, overshadowing the silica. The reduction in the diffrac-
tion peak intensity in the Pd-loaded samples may be attributed
to the different X-ray absorption characteristics of Pd compared
to silica [46]. Compared to the parent KIT-6, the (211) reflection
of the Pd-373-KBH4 sample shifted slightly to larger angles, indi-
cating a small shrinkage of the Pd mesostructure after replication.
Fig. 2. N2 adsorption–desorption isotherms (a) and pore size distribution curves (b)
of silica-free mesoporous Pd catalysts prepared using parent KIT-6 synthesized at
different hydrothermal treatment temperatures.

Y. Wang et al. / Journal of Catalysis 313 (2014) 113–126
117
Table 2
Physicochemical properties of silica-free mesoporous Pd catalysts.
a
Catalyst
SBET (m2 gꢀ1
)
Vpore (cm3 gꢀ1
)
dpore (nm)
Wall thicknessb (nm)
Pd-308-KBH4
Pd-353-KBH4
Pd-373-KBH4
Pd-403-KBH4
Pd-373-N2H4ꢁH2O
Pd-373-H2
78
65
78
46
51
39
23
0.47
0.38
0.49
0.27
0.31
0.38
0.19
13.1
11.0
4.1, 11.0
2.1, 13.3
12.2
4.3
5.3
7.4
9.7
7.4
7.4
–
14.5
42.5
Ultrafine Pd black
a
Calculated from the desorption branch of the isotherm using the BJH model.
Determined from TEM analysis.
b
Based on these XRD results, the aforementioned N2 physisorp-
tion data and the following TEM and HRSEM observations (domi-
nance of double gyroid networks), it is therefore reasonable to
infer that due to the large pore size and higher interconnectivity
of the KIT-6 samples prepared at high temperatures (KIT-373 and
KIT-403), the resulted mesoporous Pd networks replicated the
structure of the parent silica very well and the Ia3d symmetry
was preserved [47].
When KIT-6 samples prepared at low temperatures (6353 K)
were used as the templates, the diffraction patterns of the resulted
Pd-308-KBH4 and Pd-353-KBH4 are similar, but quite different
from those for Pd-373-KBH4 and Pd-403-KBH4, as shown in
Fig. 4b. Besides the (211) reflection at 2h ꢃ 1.10°, a new peak ap-
peared at 2h ꢃ 0.64°, which was retained even after the complete
removal of the silica template. This new diffraction peak can be in-
dexed to the forbidden (110) reflection for Ia3d symmetry [48].
The appearance of the (110) peak suggested the transformation
of the symmetry from cubic Ia3d to I4132 [42,48], which was fur-
ther verified by the dominance of single uncoupled networks in
the following TEM and HRSEM results.
Fig. 3. Low-angle XRD patterns and structure model (inset) of the parent KIT-6
samples synthesized at different hydrothermal treatment temperatures.
3.3. Phase and lattice structure
The wide-angle XRD patterns of the silica-free Pd catalysts re-
duced by KBH4, N2H4ꢁH2O and H2 are shown in Fig. 5. There are four
Pd diffraction peaks at 2h of 39.3–40.1°, 45.5–46.6°, 67.0–68.1° and
79.6–82.1°, which can be indexed to (111), (200), (220) and (311)
reflections of cubic fcc Pd (JCPDS No. 89-4897), respectively. The
lattice constant a of the Pd-373-KBH4, Pd-373-N2H4ꢁH2O and Pd-
373-H2 catalysts is 0.397, 0.393 and 0.389 nm, respectively. The lat-
tice constant of the Pd network reduced by H2 at 573 K is close to
that of the bulk Pd (0.389 nm, JCPDS No. 89-4897), while reduction
by KBH4 solution gives the largest lattice distortion. The crystallite
sizes of Pd in Pd-373-KBH4, Pd-373-N2H4ꢁH2O and Pd-373-H2
Fig. 4. Low-angle XRD patterns of (a) KIT-373, Pd-KBH4/KIT-373, silica-free Pd-373-
KBH4 and recovered KIT-373 after the removal of the Pd networks; (b) KIT-308, Pd-
KBH4/KIT-308 and silica-free Pd-308-KBH4 samples.
Fig. 5. Wide-angle XRD patterns of silica-free mesoporous Pd catalysts prepared
using different reducing agents.

118
Y. Wang et al. / Journal of Catalysis 313 (2014) 113–126
catalysts, the Pd-373-KBH4 catalyst has the highest degree of the
preferred orientation of the (111) plane.
catalysts calculated using Scherrer equation are ca. 5.0, 5.5 and
8.1 nm, respectively. The relative degree of preferred orientation
among crystal planes was calculated by evaluating the texture coef-
ficient (TC) using the Harris method [49]:
3.4. Mesoporous structure
I
=I0ðhklÞ
ðhklÞ
I
TC
¼ n P
ðhklÞ
n
Fig. 6 exhibits the representative TEM images of the replicated
silica-free Pd catalysts. It is clear that the preparation temperature
of KIT-6 can significantly affect the mesoporous structures of the
resulted Pd networks. The Pd-308-KBH4 sample (Fig. 6a and b)
was composed of 3D regular networks of Pd nanowires with the
diameter of ca. 4.3 nm. Only single uncoupled gyroid networks
with I4132 symmetry (Fig. 6e) analogous to the case of CMK-1
[36,50] could be observed. This observation is supported by the
appearance of the forbidden gyroid (110) diffraction peak in the
low-angle XRD patterns. EDX reveals that the nanowires are com-
posed of Pd with no detectable Si, indicating the complete removal
=I0ðhklÞ
ðhklÞ
i¼1
where I(hkl), I0(hkl) and n are the measured intensities of the meso-
porous Pd catalysts studied, the standard intensities for a com-
pletely random sample obtained from JCPDS and the number of
reflections, respectively. Calculation results show that the texture
coefficient of the (111) plane in the Pd-373-H2, Pd-373-N2H4ꢁH2O
and Pd-373-KBH4 catalysts is 1.00, 1.23 and 1.33, respectively. This
means that there is no preferred orientation in the Pd-373-H2 cata-
lyst, while the Pd-373-N2H4ꢁH2O and Pd-373-KBH4 catalysts have
the preferred orientation of the (111) plane. Among the above three
Fig. 6. TEM images of silica-free mesoporous Pd network catalysts. (a, b) Pd-308-KBH4, (c, d) Pd-353-KBH4, (e) skeletal model of a fragment of single gyroid with I4132
symmetry, (f) Pd-373-KBH4, (g) TEM image simulation of Pd-373-KBH4, (h, i) Pd-403-KBH4, (j) Pd-373-H2, (k) Pd-373-N2H4ꢁH2O and (l) skeletal model of a fragment of double
gyroid with Ia3d symmetry. Insets are the corresponding SAED patterns and Fourier diffractogram.

Y. Wang et al. / Journal of Catalysis 313 (2014) 113–126
119
of the silica template. The SAED pattern (the inset in Fig. 6a) dis-
3.5. Morphology
plays typical polycrystalline diffraction rings assignable to fcc
metallic Pd. The Pd-353-KBH4 sample (Fig. 6c and d) exhibits a
remarkable resemblance to the Pd-308-KBH4 with slightly larger
Pd nanowire diameter of ca. 5.3 nm.
HRSEM was further used to explore the morphology of the pre-
pared mesoporous Pd catalysts. The HRSEM images in Fig. 7 show
the typical mesoporosity of Pd catalysts replicated from KIT-6 tem-
plates prepared at different temperatures and reduced by different
reducing agents. As shown in Fig. 7a and b, uncoupled gyroid
networks with open pore system (I4132 symmetry) were clearly
observed in Pd-308-KBH4 and Pd-353-KBH4 catalysts. In contrast,
the Pd-373-KBH4 and Pd-403-KBH4 catalysts exhibited well-cou-
pled and interpenetrated double gyroid networks with Ia3d sym-
metry (Fig. 7c and d). As for the Pd-373-N2H4ꢁH2O (Fig. 7e) and
Pd-373-H2 (Fig. 7f), some disordered networks or large Pd particles
may be observed in addition to the ordered double gyroid net-
works, which are consistent with the above TEM observations.
Based on the HRSEM findings, the aforementioned N2 physi-
sorption, low-angle XRD and TEM results, it is indicated that our
impregnation method is effective for creating free-standing Pd
mesoporous networks. KIT-6 synthesized at high temperatures
can be replicated with better quality. The topology of the repli-
cated mesoporous Pd network catalysts can be tuned by changing
the hydrothermal treatment temperature of the parent KIT-6 silica
template. Specifically, when the hydrothermal treatment tempera-
ture of the KIT-6 is P373 K, double gyroid Pd networks with Ia3d
symmetry can be created, while when the hydrothermal treatment
temperature of the KIT-6 is 6353 K, single gyroid mesoporous Pd
KIT-6 samples prepared at high temperatures (P373 K) form
well-coupled double gyroid networks with even larger Pd nano-
wire diameters (Fig. 6f and h). The Fourier diffractogram (the insets
in Fig. 6f), the TEM image simulations (Fig. 6g) and the structure
modeling (Fig. 6l) clearly verify the Ia3d symmetry of such double
gyroid structures [47]. The diameters of the Pd nanowires in Pd-
373-KBH4 and Pd-403-KBH4 are 7.4 and 9.7 nm, respectively.
HRTEM image (Fig. 6i) shows clear lattice fringes with a spacing
of ca. 0.22 nm, which agrees well with the d(111) spacing of Pd.
The TEM images of Pd-373-H2 and Pd-373-N2H4ꢁH2O are also
shown in Fig. 6. As for the Pd-373-H2 sample (Fig. 6j), besides
the ordered double gyroid Pd networks, relatively large bulk Pd
particles were also observed, which was probably due to the
overflow of migrating Pd species to the external surface of the mes-
oporous silica template during the reduction process at a relatively
high temperature (573 K). For Pd-373-N2H4ꢁH2O (Fig. 6k), some
disordered networks rather than Pd particles were observed in
addition to the ordered double gyroid Pd networks. Based on
the TEM results, the regularity of the double gyroid Pd
network catalysts decreases in the order of Pd-373-KBH4 >
Pd-373-N2H4ꢁH2O > Pd-373-H2.
Fig. 7. HRSEM images of (a) Pd-308-KBH4, (b) Pd-353-KBH4, (c) Pd-373-KBH4, (d) Pd-403-KBH4, (e) Pd-373-N2H4ꢁH2O and (f) Pd-373-H2 catalysts.

120
Y. Wang et al. / Journal of Catalysis 313 (2014) 113–126
networks with I4132 symmetry are formed. Among different
reducing agents, the KBH4-reduced mesoporous Pd network cata-
lyst exhibits more ordered gyroid structure than the N2H4ꢁH2O
and H2-reduced samples.
adduct could not be formed under our present reaction conditions.
To further ensure that the reaction is not driven by kinetic resolu-
tion through condensation between S-proline and acetophenone,
we substituted S-proline with R-proline in the asymmetric hydro-
genation of acetophenone and found that the (S)-1-phenylethanol
instead of (R)-1-phenylethanol was formed in excess, verifying
that the current reaction is indeed a Pd surface driven asymmetric
reaction. The enantioselectivity achieved from our free-standing
gyroid Pd catalysts in the present work is probably determined
by the co-adsorbed S-proline and acetophenone on the metal sur-
face. Due to the large surface area of the mesoporous Pd catalysts
and the weak adsorption of S-proline on the Pd surface, a large
amount of S-proline is needed in order to ensure enough adsorbed
S-proline on the Pd surface for achieving high enantioselectivity.
On the other hand, if the S-proline concentration is too high, the
excessive S-proline probably also blocks the Pd active sites adja-
cent to the adsorbed S-proline, resulting in the decrease in both
ee and yield.
3.6. Enantioselective hydrogenation of acetophenone
3.6.1. Influence of proline on C@O selectivity and enantioselectivity
The S-proline plays a specific role in the enantioselective hydro-
genation of acetophenone. Without adding S-proline, the reactant
was almost completely converted to ethylbenzene over the meso-
porous Pd catalysts within 15 min and no enantioselectivity was
observed. After adding S-proline in the reaction mixture, the mes-
oporous Pd catalyst exhibited a lower activity, but with much in-
creased selectivity to 1-phenylethanol (C@O hydrogenation
product). With increasing the molar ratio of S-proline/Pd, both
the maximum yield of 1-phenylethanol and the enantioselectivity
increased and reached a maximum at the molar ratio of 7.4:1. Then
the enantioselectivity decreased dramatically with excessive
amount of S-proline (Fig. S1). Thus, in our present study, the molar
ratio of proline/Pd was kept to be 7.4:1 over all the mesoporous Pd
catalysts.
Tungler et al. [6] proposed that the S-proline acted as a chiral
auxiliary which reacted with acetophenone in solution to form a
chiral condensation adduct with high diastereoselectivity. In
their experiments, the methanolic solution of acetophenone and
S-proline was boiled for 5 min to prepare the condensation prod-
uct, which was evidenced by the disappearing of the C@O vibration
band at 1685 cmꢀ1 in acetophenone [32]. The hydrogenolysis of
the resultant adduct on the Pd surface led to the formation of the
chiral 1-phenylethanol. In such case, the ee is purely driven by
the formation of the chiral adduct, and the higher S-proline con-
centration will improve ee without significant effects on the yield.
However, in our case, the hydrogenation reaction started imme-
diately after the addition of acetophenone into the methanolic
solution of S-proline with the gyroid Pd catalysts at 273 K. There-
fore, the formation of S-proline-acetophenone adduct in solution
is less likely, which is evidenced by the FTIR observation of C@O
stretching mode in the reaction mixture (Fig. 8). Furthermore,
our Gaussian calculations (B3LYP functional and 6-31G* basis set)
[51] reveal that the formation of the condensation adduct between
acetophenone and S-proline is an endothermic process and the
Gibbs free energy difference between the condensation adduct
and the acetophenone and S-proline at 273 K without considering
the solvent effects is +23.4 kcal molꢀ1. Thus, the condensation
3.6.2. Theoretical calculations
For better understanding the chiral recognition of S-proline on
the Pd catalyst and to gain an insight into the reaction mechanism,
the adsorptions of S-proline, acetophenone and their interaction
conformations were calculated using the DFT methods. The
adsorption energies for both the enantio-differentiating prochi-
ral-R and prochiral-S intermediate complexes were quantitatively
evaluated. Judging from the above HRTEM images (Fig. 6i) and
the XRD analysis in Section 3.3, the crystalline plane preferred ex-
posed in the mesoporous Pd network catalysts is the Pd(111)
plane. Since it is not yet possible to mimic the Pd surface in the
mesoporous gyroid structure in the theoretical calculations, we
tentatively used flat Pd(111) surface with a (5 ꢂ 5) surface unit
cell as a model plane for the calculation.
The adsorption geometry of the phenyl ring in acetophenone on
Pd(111) is expected to be similar to that of benzene, dominated by
the p–d interaction between the phenyl ring and the Pd(111). For
benzene adsorbed on Pd(111), the bridge site was found to be the
most stable adsorption site followed by the hollow site [52,53]. Our
calculation results using the periodic slab are in accordance with
the literature [52,53]. Thus, based on the above adsorption results
of benzene, three reasonable adsorption modes of acetophenone
were considered as the starting geometries for optimization in
our calculations. In the first geometry, the adsorption is via the
bonding of keto-carbonyl moiety to the Pd surface (denoted as
the carbonyl adsorption mode), while the phenyl ring is on a hy-
brid between bridge and hollow sites. In the second geometry,
the benzene moiety is adsorbed on a bridge site (denoted as bridge
adsorption mode), which hinders the direct interaction of the car-
bonyl with the Pd surface atoms. The third one is similar to the sec-
ond one except that the benzene moiety is adsorbed on a hollow
site. Geometry optimization obtains only two stable adsorption
configurations (Fig. 9). One is the carbonyl adsorption mode and
the other is the bridge adsorption mode with corresponding
adsorption energies of ꢀ22.8 and ꢀ21.2 kcal molꢀ1, respectively.
Thus, the carbonyl adsorption structure is slightly more stable than
the bridge adsorption mode. The C@O bond in the carbonyl adsorp-
tion mode is elongated marginally (1.30 Å) as compared to that in
the bridge adsorption mode (1.26 Å).
The geometry and adsorption energy of the isolated S-proline
molecule adsorbed on the Pd(111) surface were also calculated
with three most likely adsorption structures. As an amino acid
molecule, both the pyrrolidine ring and the carboxylic group are
possible to bond directly on Pd. So, the first starting geometry is
the adsorption through the N atom in the pyrrolidine ring (denoted
as N-adsorption mode). The second one is the adsorption
through the hydrogen atoms in the pyrrolidine ring (denoted as
Fig. 8. FTIR spectra of (a) the methanolic solution of acetophenone, (b) the
methanolic solution of S-proline, (c) the methanolic mixture of S-proline and
acetophenone boiled at 363 K for 5 min, and (d) the reaction mixture of S-proline
and acetophenone mixed at 273 K.

Y. Wang et al. / Journal of Catalysis 313 (2014) 113–126
121
Fig. 9. Adsorption of acetophenone on the Pd slab. (a) Adsorption through the carbonyl; (b) Adsorption through benzene moiety (bridge).
H-adsorption mode). The third one is the adsorption through the
oxygen atoms in the carboxylate group (denoted as O-adsorption
mode). Since S-proline was adsorbed on Pd(111) as a zwitterion
based on XPS study [54], the N atom is saturated and the N-adsorp-
tion mode is not possible. We, therefore, only optimized the H- and
O-adsorption modes and the optimized geometries are shown in
Fig. 10. The adsorption energies for the H- and O-adsorption mode
are ꢀ10.4 and ꢀ14.1 kcal molꢀ1, respectively, indicating that the
adsorption through the carboxylic group in the form of a zwitter-
ion is the most favorable structure. Meanwhile, in the optimized
H-adsorption mode, S-proline is not present as a zwitterion, which
is inconsistent with the experimental results [54]. Therefore, we
only focused on the O-adsorption mode with the oxygen atoms
of the carboxylate group close to the Pd surface and the pyrrolidine
ring tilting above the carboxylate group in the following studies.
Taking the optimized adsorption geometries of acetophenone
and S-proline, the interaction conformations between the co-ad-
sorbed S-proline and acetophenone on Pd(111) surface were also
calculated in order to identify the influence of S-proline on the
adsorption of acetophenone. The optimized prochiral-R and pro-
chiral-S geometries are shown in Fig. 11. It should be noted that
by using acetophenone either in the carbonyl or bridge adsorption
mode as the starting geometry, the resultant optimized interaction
conformation is identical, which corresponds to the distorted
bridge adsorption of acetophenone associated with the O-adsorp-
tion of S-proline as shown in Fig. 11. More importantly, we also
found that the existence of a hydrogen bonding between the oxy-
gen atom of the carbonyl group in acetophenone and the hydrogen
atom of the amino group in S-proline, which could play a signifi-
cant role in the molecular chiral recognition. This hydrogen bond-
ing interaction forces the carbonyl group in acetophenone to be
away from the Pd surface. Meanwhile, this hydrogen bonding
interaction forces the S-proline molecule to be closer to the Pd sur-
face, resulting in the stronger interaction between S-proline and
the Pd surface. Such subtle adjusting of the adsorption geometry
of the acetophenone would be responsible for avoiding complete
hydrogenation to form ethylbenzene and more importantly, result-
ing in high enantioselectivity.
The adsorption energies of the prochiral-R and prochiral-S
interaction complex are ꢀ52.8 and ꢀ48.7 kcal molꢀ1, respectively,
which are more negative than the sum of the adsorption energies
of the separately adsorbed acetophenone and S-proline. The calcu-
lated distances between the carbonyl oxygen and the hydrogen of
the amino group in prochiral-R and prochiral-S complexes are 1.52
and 1.82 Å, respectively. These data suggest that there is an inter-
action between the co-adsorbed acetophenone and S-proline on
the Pd surface. The energy difference between the prochiral-R
and prochiral-S complexes is ꢀ4.1 kcal molꢀ1, indicating that the
prochiral-R complex is more stable and favors a higher fraction
coverage on Pd surface. Thus, after subsequent hydrogen additions,
(R)-1-phenylethanol is formed preferentially, which is in good
accordance with the experimental results.
Fig. 10. Adsorption of S-proline on the Pd slab. (a) Adsorption through the hydrogen atoms in the pyrrolidine ring; (b) Adsorption through the oxygen atoms in the
carboxylate group.

122
Y. Wang et al. / Journal of Catalysis 313 (2014) 113–126
Fig. 11. The interaction conformation of S-proline with acetophenone. (a) Prochiral-R interaction conformation, leading to (R)-1-phenylethanol upon hydrogenation;
(b) Prochiral-S interaction conformation, leading to (S)-1-phenylethanol upon hydrogenation.
3.6.3. Effect of Pd lattice structure on enantioselectivity
the ordered double gyroid networks, there are some disordered
networks or bulk Pd nanoparticles existed in the Pd-373-N2H4ꢁH2O
and Pd-373-H2 catalysts. The high quality gyroid structure in
Pd-373-KBH4 catalyst may also contribute to its superior enanti-
oselectivity as compared to the Pd-373-N2H4ꢁH2O and Pd-373-H2
catalysts.
The effect of lattice structures on the catalytic performance was
investigated by using the mesoporous Pd catalysts replicated from
the same KIT-373 template but prepared by different reduction
methods. The catalytic results for liquid phase enantioselective
hydrogenation of acetophenone with a fixed reaction time of
15 min are compiled in Table 3. When H2 was used as the reducing
agents for creating Pd networks, the fastest hydrogenation rate was
achieved with the 1-phenylethanol yield of 48.9% and the ee of ca.
22%. When N2H4ꢁH2O was used as the reducing agent, the yield of
1-phenylethanol is reduced to 24.2% while the ee is increased to
30%. When KBH4 was used as the reducing agent, the ee is further
increased to 34% with the yield of 1-phenylethanol reduced to
12.8%. Therefore, the enantioselectivities are in the sequence of
Pd-373-KBH4 > Pd-373-N2H4ꢁH2O > Pd-373-H2 catalyst, while the
1-phenylethanol yields are in the reverse order. As discussed above
in Section 3.3, the Pd-373-KBH4 catalyst has the highest preferred
orientation of the (111) plane among the three catalysts. Thus, the
Pd-373-KBH4 catalyst has higher density of flat surfaces with less
surface kinks and steps, which is possibly more appropriate to
accommodate the bulky prochiral complex that occupies minimum
13 surface atoms, resulting in the highest enantioselectivity. Also,
the above TEM and HRSEM results revealed that in addition to
3.6.4. Effect of template topology on enantioselectivity
The effect of the topology on the catalytic performance was
investigated by using the KBH4-reduced mesoporous Pd catalysts
replicated from the silica template synthesized at different temper-
atures. The catalytic results of the mesoporous Pd catalysts with
different topologies (space group I4132 and Ia3d) are also compiled
in Table 3. For comparison, the catalytic results of ultrafine Pd black
and hexagonal Pd nanoarray catalyst with the space group of P6mm
are also shown. The ee over ultrafine Pd black is ca. 20% with the
yield of 1-phenylethanol of ca. 82.0% after a reaction of 15 min. In
comparison, the ee over hexagonal Pd nanoarray catalyst is 28%,
which is much higher than that over ultrafine Pd black. The ee is
further increased to 31–42% over the Pd-x-KBH4 catalysts with sin-
gle or double gyroid structure. Generally, the enantioselectivity
over the mesoporous Pd catalysts is in the sequence of double gyro-
id structure (Ia3d) > single gyroid structure (I4132) > hexagonal
structure (P6mm) > ultrafine Pd black. Since the gyroid Pd-x-KBH4
catalysts and hexagonal Pd nanoarray catalyst were prepared under
the same reduction condition but using silica templates with
different topology, it is plausible to attribute the variation of
enantioselectivity to the different topology of the original silica
template. The unique spatial arrangement of the Pd crystallites in
the gyroid structure enhances the interaction conformation of the
co-adsorbed acetophenone and the chiral S-proline modifier to
form the R-isomer of 1-phenylethanol preferentially. In contrast,
the ultrafine Pd black catalyst without such steric constraint
exhibited the lowest enantioselectivity.
Table 3
Catalytic behaviors of mesoporous Pd catalysts in the enantioselective hydrogenation
of acetophenone.a
Catalyst
Topologyb
1-Phenylethanol yield (%)
ee (%)
Pd-308-KBH4
Pd-353-KBH4
Pd-373-KBH4
Pd-403-KBH4
I4132
I4132
Ia3d
Ia3d
Ia3d
Ia3d
P6mm
–
12.6
19.1
12.8
13.3
24.2
48.9
48.7
82.0
31 (R)
31 (R)
34 (R)
42 (R)
30 (R)
22 (R)
28 (R)
20 (R)
Pd-373-N2H4ꢁH2O
Pd-373-H2
Hexagonal Pd nanoarray
Ultrafine Pd black
3.6.5. Effect of pore size on enantioselectivity
a
Reaction conditions: 0.02 mL acetophenone, 0.40 g S-proline, 17 mL of metha-
The effect of pore size on the catalytic performance was further
investigated by using KBH4-reduced mesoporous gyroid Pd cata-
lysts with the same topology but different pore diameter (Table 3).
As for the single gyroid structures (Pd-308-KBH4 and Pd-353-
nol, T = 273 K, H2 flow rate of 60 mL minꢀ1
,
Wcat = 0.050 g, reaction
time = 15 min.
b
Topology derived based on the low-angle XRD, N2-physisorption, TEM and
HRSEM results.

Y. Wang et al. / Journal of Catalysis 313 (2014) 113–126
123
recorded (Fig. 12). The yield of 1-phenylethanol over the Pd-403-
KBH4 catalyst increased gradually to 100% at a reaction time of
150 min and remained constant at prolonged reaction time, clearly
demonstrating its excellent selectivity to the C@O group. The
enantioselectivity of the system is very stable with a value of
42%, which is much higher than that over the ultrafine Pd black
catalyst (ca. 20%), signifying the superior enantioselective property
of the Pd-403-KBH4. The stability of the enantioselectivity over the
Pd-403-KBH4 catalyst was further verified in three successive runs.
No reduction in yield and ee were observed and the mesostructure
of the used catalyst is maintained, which can be attributed to the
benefit of low reaction temperature and the superior structural
strength of the highly interconnected double gyroid Pd networks.
3.7. Enantioselective hydrogenation of ring-substituted acetophenone
derivatives
Fig. 12. The yield of 1-phenylethanol and the evolution of the enantioselectivity
(inset) as a function of the reaction time over the Pd-403-KBH4 catalyst. Reaction
conditions: 0.02 mL of acetophenone, 0.40 g S-proline, 17 mL of methanol,
T = 273 K, H2 flow rate of 60 mL minꢀ1, Wcat. = 0.050 g.
The catalytic performances of the Pd-403-KBH4 catalyst in the
enantioselective hydrogenation of ring-substituted acetophenone
derivatives were also investigated. As shown in Table 4, the ee of
the hydrogenation reaction decreases to below 16% when a H atom
in the aromatic ring is replaced by an electron-donating AOCH3
group. However, the ee increases to ꢄ73% when AH is replaced
by an electron-withdrawing ACF3 group. This suggests that the
mesoporous Pd catalyst may be promising in the enantioselective
hydrogenation of the acetophenone derivatives with electron-
withdrawing groups. For ACF3 substituted acetophenones, both
the ee value and the yield were affected by the substitution posi-
tions. From para, meta to ortho position, the ee increases from
42% to 73%, while the yield decreases dramatically. The TOF values
calculated on the basis of the active surface area of the Pd-4ꢀ031-
KBH4), they have only large pores and the change of the pore size
seems to have little effect on the enantioselectivity. Nearly the
same ee values were obtained over the Pd-308-KBH4 and Pd-
353-KBH4 catalysts with different pore size of ca. 13.1 and
11.0 nm, respectively. As for the double gyroid structure, the
enantioselectivity over Pd-403-KBH4 is 42%, which is higher than
that over Pd-373-KBH4 (34%). As discussed in Section 3.1, Pd-
373-KBH4 and Pd-403-KBH4 exhibit bimodal pore size distribution
with the small pore at ca. 4.1 or 2.1 nm, respectively, along with
the large pore at ca. 11–13 nm. The large pores in all the gyroid
structures (single and double) are expected to contribute to a sim-
ilar level of ee. Therefore, the additional enhancement of ee from
the double gyroid Pd catalysts may result from the presence of
small pores. Such enhancement can be attributed to the confine-
ment effects related to the small pores which generate stronger
steric constraints of the adsorbates and improve the intermolecu-
lar interaction between the co-adsorbed reactant and the chiral
modifier.
KBH4 catalyst as determined by CO chemisorption (13.3 m2 gPd
)
are also summarized in Table 4. In most cases, the R-enantiomer
is dominant except that in the case of OCH3(ortho)-substituted
acetophenone.
For a better understanding of how the substituent groups affect
the enantioselectivity and more importantly why the inversed chi-
ral recognition is occurred in the case of OCH3(ortho)-substituted
acetophenone, the interaction conformations of the co-adsorbed
substituted acetophenone and S-proline over the Pd(111) surface
were calculated using the DFT methods. The adsorption energies
3.6.6. Catalytic behavior of the optimized Pd-403-KBH4 catalyst
In order to have a better insight into the catalytic performance
of the optimized Pd-403-KBH4 catalyst, the reaction course was
Table 4
Catalytic behaviors of Pd-403-KBH4 catalyst in the enantioselective hydrogenation of acetophenone and ring-substituted acetophenones.a
TOF (Sꢀ1
)
Maximum ee (%)
Eads. (kcal molꢀ1
Prochiral-R
ꢀ52.8
)
D )
Ec (kcal molꢀ1
b
Reactant
Reaction time (min)
Yield (%)
Prochiral-S
15
13.3
2.2
1.8 ꢂ 10ꢀ3
4.5 ꢂ 10ꢀ6
42 (R)
73 (R)
ꢀ48.7
ꢀ4.1
ꢀ7.9
780
ꢀ64.6
ꢀ68.0
ꢀ56.7
ꢀ62.7
15
6.5
6.9 ꢂ 10ꢀ4
50 (R)
ꢀ5.3
60
19.6
3.7
5.2 ꢂ 10ꢀ4
1.3 ꢂ 10ꢀ5
42 (R)
8 (S)
ꢀ68.0
ꢀ69.9
ꢀ65.0
ꢀ70.3
ꢀ3.0
480
0.4
8.8 ꢂ 10ꢀ5
16 (R)
ꢀ73.3
ꢀ71.9
ꢀ1.4
o
240
12.2
OMe
a
b
c
Reaction conditions: 0.02 mL substrate, 0.40 g S-proline, 17 mL of methanol, T = 273 K, H2 flow rate of 60 mL minꢀ1, Wcat = 0.050 g.
Adsorption energies calculated using the SIESTA package with a double-polarization (DZP) basis set.
Energy difference between prochiral-R and prochiral-S intermediate complexes.

124
Y. Wang et al. / Journal of Catalysis 313 (2014) 113–126
Fig. 13. The prochiral-R (left) and prochiral-S (right) interaction conformation of S-proline with substituted acetophenone. (a) CF3(ortho)-substituted acetophenone; (b)
CF3(meta)-substituted acetophenone; (c) CF3(para)-substituted acetophenone; (d) OCH3(ortho)-substituted acetophenone; (e) OCH3(meta)-substituted acetophenone.
of the prochiral-R and prochiral-S interaction complexes and their
energy differences are shown in Table 4 and the optimized interac-
tion conformations are shown in Fig. 13. It was found that after
ring-substitution by either ACF3 or AOCH3, the interaction be-
tween the substituted acetophenones, S-proline and the Pd surface
was improved. In most cases, the adsorption energies for the pro-
chiral-R complexes are more negative than those for the prochiral-
S complexes, resulting in the dominant R-enantiomer. In contrast,
for OCH3(ortho)-substituted acetophenone, the adsorption energy
of the prochiral-S complex is slightly more negative than that of
the prochiral-R complex. Thus, after subsequent hydrogen
additions, the S-enantiomer is formed in excess, which is in good
accordance with the experimental results.
Further inspection of our experimental results and our DFT cal-
culations leads us to conclude that the enantio-differentiation in
the enantioselective hydrogenation of acetophenone derivatives

Y. Wang et al. / Journal of Catalysis 313 (2014) 113–126
125
ing the topology, pore size and lattice structure of the mesoporous
Pd network catalysts, the enantioselectivity of the reaction could
be tuned. The enantioselectivity over the mesoporous Pd catalysts
increases in the sequence of double gyroid structure (Ia3d) > single
gyroid structure (I4132) > 2D hexagonal structure (P6mm) > ultra-
fine Pd black. The relatively smaller pore diameter within the dou-
ble gyroid structures is favorable to achieve higher ee. Among H2,
N2H4ꢁH2O and KBH4, KBH4 was found to be the best reducing
agents to obtain high quality replicated gyroid Pd networks with
proper lattice structure, which is important to achieve high enanti-
oselectivity. The unique spatial arrangement of Pd crystallites in
the double gyroid structure with a proper pore size and the desired
lattice structure formed by KBH4 reduction provided suitable
microenvironment to generate optimized confinement effect,
which improved the enantio-differentiation. The optimized Pd-
403-KBH4 catalyst exhibits relatively high ee of ꢄ42% in the enan-
tioselective hydrogenation of acetophenone and ee up to 73% in the
ACF3 substituted acetophenone derivatives. DFT studies reveal that
the major enantiomeric product could be predicted by the relative
energy difference between the optimized prochiral-R and prochi-
ral-S intermediate complexes formed by the acetophenone and
its derivatives with S-proline. The energetically favored intermedi-
ate complex leads to the formation of the corresponding enantio-
mer in excess upon hydrogenation. The ee was linearly
correlated with the adsorption energy difference between prochi-
ral-R and prochiral-S complexes. Our work presents a new possibil-
ity for tuning the enantioselectivity by manipulating the
confinement effect in the free-standing mesoporous Pd catalysts
and provides useful information for understanding the chiral rec-
ognition mechanism, which is also helpful to provide guidance
for the design of highly efficient enantioselective metal catalysts
in heterogeneous asymmetric hydrogenation.
Fig. 14. The enantioselectivity in the hydrogenation of ring-substituted acetoph-
enones over the Pd-403-KBH4 catalyst as a function of the absolute value of the
calculated adsorption energy difference between the prochiral-R and prochiral-S
intermediate complexes.
over S-proline modified Pd catalysts could be correlated with the
relative adsorption energies of the optimized prochiral-R and pro-
chiral-S intermediate complexes. To demonstrate such a correla-
tion, a plot of ee versus the absolute value of the adsorption
energy difference between the prochiral-R and prochiral-S inter-
mediate complexes is shown in Fig. 14, which gives a good linear
relationship. The larger the absolute value of the adsorption energy
difference (DE) between the prochiral-R and prochiral-S interme-
diate complexes, the higher the ee obtained. Such correlation con-
firms that the achieved ee is determined by the chiral specific
interactions. It is worth noting that the
DE between the prochi-
ral-R and prochiral-S intermediate complexes we obtained in our
present theoretical calculations is not exactly equal to the differ-
Acknowledgments
ence in the Gibbs free energy (
to the size of the theoretical model used and the limits of the cur-
rent computing power, it is still infeasible to calculate the G in
DG) in the catalytic system. Due
This work was supported by the National Natural Science Foun-
dation of China (21343003, 20803009), the Shanghai Science and
Technology Committee (12ZR1401400) and the Ministry of Educa-
tion (the Scientific Research Foundation for the Returned Overseas
Chinese Scholars).
D
our present catalytic system. One would expect that, at 273 K, a
D
G of ca. 3 kcal molꢀ1 results in a predicted ee of ꢄ100%. However,
as shown in Fig. 14, much lower ee (ꢄ42%) was obtained in our
experiment compared to the predicted values (ꢄ100% ee) with
the
D DE
E of ca. 3 kcal molꢀ1. The observed difference between
Appendix A. Supplementary material
and
DG is probably related to the limitation of the present theoret-
ical calculation model, which is impossible to completely mimic
the real catalytic surface. The energetically favored complex leads
to the formation of the corresponding enantiomer in excess upon
hydrogenation. Similar theoretical calculations have earlier been
made for the Pt-cinchona system in the hydrogenations of keto-
pantolactone [55] and methyl pyruvate [56,57] by Baiker and
coworkers. In those studies a good agreement between the stabil-
ity of the prochiral-R and prochiral-S complexes and the stereo-
chemical outcome of the reactions were also found, where purely
thermodynamic considerations were used for predicting the major
enantiomer. It is interesting that this approach apparently also
works for our Pd-proline system to predict the major enantiomeric
product.
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcat.2014.03.004.
References
[1] E.N. Jacobsen, A. Pfaltz, H. Yamamoto, Comprehensive Asymmetric Catalysis,
Springer, Berlin, 1999.
[2] E. Klabunovskii, G.V. Smith, Á. Zsigmond, Heterogeneous Enantioselective
Hydrogenations – Theory and Practice, Springer, Dordrecht, 2006.
[3] R.A. Sheldon, Pure Appl. Chem. 72 (2000) 1233.
[4] G.V. Smith, F. Notheisz, Heterogeneous Catalysis in Organic Chemistry,
Academic Press, San Diego, 1999.
[5] H.U. Blaser, B. Pugin, Chiral Reactions in Heterogeneous Catalysis, Plenum
Press, New York, 1995.
[6] A. Tungler, T. Tarnai, T. Mathe, J. Petro, J. Mol. Catal. 67 (1991) 277.
[7] A. Perosa, P. Tundo, M. Selva, J. Mol. Catal. A – Chem. 180 (2002) 169.
[8] R. Hess, T. Mallat, A. Baiker, J. Catal. 218 (2003) 453.
[9] R. Hess, A. Vargas, T. Mallat, T. Burgi, A. Baiker, J. Catal. 222 (2004) 117.
[10] R. Hess, F. Krumeich, T. Mallat, A. Baiker, J. Mol. Catal. A – Chem. 212 (2004)
205.
[11] V. Vetere, M.B. Faraoni, G.F. Santori, J.C. Podesta, M.L. Casella, O.A. Ferretti, J.
Catal. 226 (2004) 457.
[12] S. Basu, M. Mapa, C.S. Gopinath, M. Doble, S. Bhaduri, G.K. Lahiri, J. Catal. 239
(2006) 154.
4. Conclusions
The factors influencing the confinement effect originated from
the nanopores such as the topology, pore size, and lattice structure
of the free-standing mesoporous Pd network catalysts and their
correlation with the catalytic performances in the enantioselective
hydrogenation of acetophenone and its derivatives were studied. It
was found that the confinement effect was very important in
determining the enantioselectivity of the reaction. By manipulat-
[13] T. Marzialetti, M. Oportus, D. Ruiz, J.L.G. Fierro, P. Reyes, Catal. Today 133
(2008) 711.
[14] H. Cheng, J. Hao, H. Wang, C. Xi, X. Meng, S. Cai, F. Zhao, J. Mol. Catal. A – Chem.
278 (2007) 6.

126
Y. Wang et al. / Journal of Catalysis 313 (2014) 113–126
[15] B. Tang, W. Xiong, E.R. Liu, Y. Jia, J.B. Wang, H. Chen, X.J. Li, Tetrahedron-
Asymmetry 19 (2008) 1397.
[16] H.Y. Jiang, H. Chen, R.X. Li, Catal. Commun. 11 (2010) 584.
[17] H.Y. Jiang, C.F. Yang, C. Li, H.Y. Fu, H. Chen, R.X. Li, X.J. Li, Angew. Chem. Int. Ed.
47 (2008) 9240.
[40] http://www.netlib.org/scalapack/scalapack_home.html.
[41] A. Eichler, F. Mittendorfer, J. Hafner, Phys. Rev. B 62 (2000) 4744.
[42] A. Rumplecker, F. Kleitz, E.L. Salabas, F. Schuth, Chem. Mater. 19 (2007) 485.
[43] J.K. Shon, S.S. Kong, J.M. Kim, C.H. Ko, M. Jin, Y.Y. Lee, S.H. Hwang, J.A. Yoon, J.N.
Kim, Chem. Commun. (2009) 650.
[18] C.F. Yang, H.Y. Jiang, J. Feng, H.Y. Fu, R.X. Li, H. Chen, X.J. Li, J. Mol. Catal. A –
Chem. 300 (2009) 98.
[44] F. Jiao, A.H. Hill, A. Harrison, A. Berko, A.V. Chadwick, P.G. Bruce, J. Am. Chem.
Soc. 130 (2008) 5262.
[19] C. Li, L. Zhang, Y. Du, X.L. Zheng, H.Y. Fu, H. Chen, R.X. Li, Catal. Commun. 28
(2012) 5.
[45] Y.F. Shi, B.K. Guo, S.A. Corr, Q.H. Shi, Y.S. Hu, K.R. Heier, L.Q. Chen, R. Seshadri,
G.D. Stucky, Nano Lett. 9 (2009) 4215.
[20] M.M. Liu, B.B. Fan, X.F. Shi, R.F. Li, Catal. Commun. 42 (2013) 20.
[21] C. Li, Catal. Rev. – Sci. Eng. 46 (2004) 419.
[22] F. Hoxha, B. Schimmoeller, Z. Cakl, A. Urakawa, T. Mallat, S.E. Pratsinis, A.
Baiker, J. Catal. 271 (2010) 115.
[23] F. Hoxha, E. Schmidt, T. Mallat, B. Schimmoeller, S.E. Pratsinis, A. Baiker, J.
Catal. 278 (2011) 94.
[24] S.C. Warren, L.C. Messina, L.S. Slaughter, M. Kamperman, Q. Zhou, S.M. Gruner,
F.J. DiSalvo, U. Wiesner, Science 320 (2008) 1748.
[25] S.C. Warren, U. Wiesner, Pure Appl. Chem. 81 (2009) 73.
[26] Y. Kuroda, Y. Yamauchi, K. Kuroda, Chem. Commun. 46 (2010) 1827.
[27] Y. Yamauchi, K. Kuroda, Chem. – Asian J. 3 (2008) 664.
[28] Y. Yamauchi, A. Takai, M. Komatsu, M. Sawada, T. Ohsuna, K. Kuroda, Chem.
Mater. 20 (2008) 1004.
[29] A. Takai, Y. Yamauchi, K. Kuroda, J. Am. Chem. Soc. 132 (2010) 208.
[30] Y. Doi, A. Takai, Y. Sakamoto, O. Terasaki, Y. Yamauchi, K. Kuroda, Chem.
Commun. 46 (2010) 6365.
[31] A. Takai, Y. Doi, Y. Yamauchi, K. Kuroda, J. Phys. Chem. C 114 (2010) 7586.
[32] X.Y. Chen, Z.Y. Lou, M.H. Qiao, K.N. Fan, S.C. Tsang, H.Y. He, J. Phys. Chem. C 112
(2008) 1316.
[33] T.W. Kim, F. Kleitz, B. Paul, R. Ryoo, J. Am. Chem. Soc. 127 (2005) 7601.
[34] V. Holler, K. Radevik, I. Yuranov, L. Kiwi-Minsker, A. Renken, Appl. Catal. B –
Environ. 32 (2001) 143.
[46] O. Bisi, O. Jepsen, O.K. Andersen, Phys. Rev. B 36 (1987) 9439.
[47] B.Z. Tian, X.Y. Liu, L.A. Solovyov, Z. Liu, H.F. Yang, Z.D. Zhang, S.H. Xie, F.Q.
Zhang, B. Tu, C.Z. Yu, O. Terasaki, D.Y. Zhao, J. Am. Chem. Soc. 126 (2004) 865.
[48] H.J. Shin, R. Ryoo, Z. Liu, O. Terasaki, J. Am. Chem. Soc. 123 (2001) 1246.
[49] C.S. Barrett, T.B. Massalski, Structure of Metals, Pergamon Press, Oxford, 1980.
[50] V. Saranathan, C.O. Osuji, S.G.J. Mochrie, H. Noh, S. Narayanan, A. Sandy, E.R.
Dufresne, R.O. Prum, Proc. Natl. Acad. Sci. 107 (2010) 11676.
[51] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman,
J.A. Montgomery, Jr., T. Vreven, K.N. Kudin, J.C. Burant, J.M. Millam, S.S. Iyengar,
J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G.A.
Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa,
M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J.E. Knox,
H.P. Hratchian, J.B. Cross, C. Adamo, J. Jaramillo, R. Gomperts, R.E. Stratmann,
O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, P.Y. Ayala, K.
Morokuma, G.A. Voth, P. Salvador, J.J. Dannenberg, V.G. Zakrzewski, S.
Dapprich, A.D. Daniels, M.C. Strain, O. Farkas, D.K. Malick, A.D. Rabuck, K.
Raghavachari, J.B. Foresman, J.V. Ortiz, Q. Cui, A.G. Baboul, S. Clifford, J.
Cioslowski, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.L.
Martin, D.J. Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M.
Challacombe, P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong, C. Gonzalez, J.A.
Pople, Gaussian 03, Revision B.04, Gaussian Inc., Pittsburgh, PA, 2003.
[52] F. Mittendorfer, C. Thomazeau, P. Raybaud, H. Toulhoat, J. Phys. Chem. B 107
(2003) 12287.
[35] B.E. Warren, J. Appl. Phys. 12 (1941) 375.
[36] L.A. Solovyov, V.I. Zaikovskii, A.N. Shmakov, O.V. Belousov, R. Ryoo, J. Phys.
Chem. B 106 (2002) 12198.
[37] J. Junquera, O. Paz, D. Sanchez-Portal, E. Artacho, Phys. Rev. B 64 (2001).
[38] N. Troullier, J.L. Martins, Phys. Rev. B 43 (1991) 1993.
[39] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865.
[53] C. Morin, D. Simon, P. Sautet, J. Phys. Chem. B 108 (2004) 5653.
[54] F. Gao, Y.L. Wang, L. Burkholder, W.T. Tysoe, Surf. Sci. 601 (2007) 3579.
[55] M. Schurch, O. Schwalm, T. Mallat, J. Weber, A. Baiker, J. Catal. 169 (1997) 275.
[56] T. Burgi, A. Baiker, J. Catal. 194 (2000) 445.
[57] A. Baiker, J. Mol. Catal. A – Chem. 163 (2000) 205.