Selective Stereochemical Catalysis Controlled by Specific Atomic Arrangement of Ordered Alloys

Article abstract of DOI:10.1002/cctc.201500808

Unprecedented surface stereochemistry governed by specific atomic arrangement of intermetallic compounds (ordered alloys) is shown. Hydrogen-mediated cis-trans alkene isomerization is selectively catalyzed without overhydrogenation to alkanes by Rh-based intermetallic compounds displaying an orthorhombic Pnma structure such as RhSb. Multiple characterization techniques combined with DFT calculations revealed that (2 1 1), (0 2 0), and (0 1 3) planes comprising one-dimensionally aligned Rh rows separated by Sb atoms (1D-planes) dominated RhSb surfaces. Systematic DFT calculations indicated that geometric constraints of these 1D-planes and steric hindrance from one alkyl group of cis-alkene limited hydrogen access to the alkenyl carbon to one direction, enabling one-atom hydrogenation for isomerization, but inhibiting two-atom hydrogenation for overhydrogenation.

Full text of DOI:10.1002/cctc.201500808

DOI: 10.1002/cctc.201500808
Full Papers
Selective Stereochemical Catalysis Controlled by Specific
Atomic Arrangement of Ordered Alloys
Shinya Furukawa,* Kazuyoshi Ochi, Hui Luo, Masayoshi Miyazaki, and Takayuki Komatsu*[a]
Unprecedented surface stereochemistry governed by specific
atomic arrangement of intermetallic compounds (ordered
alloys) is shown. Hydrogen-mediated cis–trans alkene isomeri-
zation is selectively catalyzed without overhydrogenation to al-
kanes by Rh-based intermetallic compounds displaying an or-
thorhombic Pnma structure such as RhSb. Multiple characteri-
zation techniques combined with DFT calculations revealed
that (211), (020), and (013) planes comprising one-dimension-
ally aligned Rh rows separated by Sb atoms (1D-planes) domi-
nated RhSb surfaces. Systematic DFT calculations indicated that
geometric constraints of these 1D-planes and steric hindrance
from one alkyl group of cis-alkene limited hydrogen access to
the alkenyl carbon to one direction, enabling one-atom hydro-
genation for isomerization, but inhibiting two-atom hydroge-
nation for overhydrogenation.
Introduction
Stereochemical transformations such as enantio, diastereo, and
regioselective conversions play key roles in numerous fields in-
cluding organic chemistry, pharmacology, enzymology, and in-
dustrial chemistry. Achieving these sophisticated transforma-
tions by simple and catalytic artificial methods continues to be
an intriguing area of research.[1]
Intermetallic compounds (ordered alloys) present specific
crystal structures and, thus, well-defined atomic arrangements
of their metal components[7] making them attractive candi-
dates for simple and ordered inorganic materials. Their highly
ordered bimetallic surface may provide an appropriate reaction
environment for controlling stereochemical transformations. A
simple geometric effect, dilution or isolation of active metals
by their inert metal counterpart (so-called ensemble effect),
has been known to inhibit undesired side reactions; for exam-
ple, PtÀSn for alkane dehydrogenation[8] and PdÀGa for alkyne
semihydrogenation.[9] This effect, however, does not contribute
to surface stereochemistry. To this day, no one has reported
stereochemical conversion governed by surface order. There-
fore, developing and understanding such a catalytic system is
highly innovative and will open new horizons for stereochemi-
cal catalysis, surface science, and materials science.
The development of an efficient catalyst for stereoselective
conversions generally requires a well-designed reaction envi-
ronment such as metal complexes bearing tailor-made organic
ligands.[2] In contrast, inorganic heterogeneous catalysts typi-
cally do not provide well-structured reaction sites because of
their heterogeneity, but offer potential advantages for practical
use such as durability, separability, and reusability. To introduce
such a well-structured reaction site to heterogeneous catalysts,
appropriate functionalizing organic molecules or metal com-
plexes are immobilized[3] or co-adsorbed[4] on the catalyst sur-
face. In this context, it is highly challenging to design effective
inorganic materials catalyzing stereochemical transformations
without the aid of organic components. Although micro or
mesoporous inorganic materials such as zeolites have been uti-
lized for stereo[5] and regioselective reactions,[6] their properties
relies on physical restriction by the pore wall typically in nano-
or mesoscale. Ideally, a reaction site displaying regularity at the
Angstrom-scale and achieving accurate molecular recognition
through chemical interactions is more appealing than these
materials.
We recently showed that SiO2-supported intermetallic RhSb
exhibited the highest selectivity in the hydrogen-mediated cat-
alytic cis–trans isomerization of stilbene (ST) compared to vari-
ous Rh-based intermetallic compounds and monometallic
Rh.[10] Hydrogen-mediated alkene isomerization generally re-
quires a reversible one-atom hydrogenation enabling CÀC ro-
tation, but must avoid the irreversible two-atom hydrogena-
tion producing alkane. Therefore, the ordered atomic arrange-
ment of the intermetallic compound likely exerts a specific ste-
reochemical effect on the hydrogenation. However, the nature
of the catalysis remains entirely unknown.
In this study, the origin of the specific catalytic properties of
RhSb and other intermetallic compounds was investigated in
detail in terms of space group and surface atomic arrangement
using multiple characterization techniques and density func-
tional theory (DFT) calculations. Herein, we introduce an inno-
vative concept in stereochemical conversion, in which specific
atomic arrangement of an ordered alloy governs surface ste-
reochemistry and selective catalysis.
[a] Dr. S. Furukawa, K. Ochi, H. Luo, M. Miyazaki, Prof. T. Komatsu
Department of Chemistry
Graduate School of Science and Engineering
Tokyo Institute of Technology
2-12-1-E1-10, Ookayama, Meguro-ku, Tokyo, 152-8550 (Japan)
Fax: (+81)3-3753-2602
E-mail: komatsu.t.ad@m.titech.ac.jp
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201500808.
ChemCatChem 2015, 7, 3472 – 3479
3472
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Results and discussion
Monometallic and intermetallic Rh-based catalysts displaying
¯
¯
different space groups (Fm3m (Rh), Pm3m (RhGa, RhIn, RhZn),
Pnma (RhGe, RhSb), P6/mmm (RhPb)) were synthesized by suc-
cessive impregnation using SiO2 as a support. Intermetallic
compound RuSb presenting the same crystal structure as RhSb
was also prepared. X-ray diffraction (XRD) analysis showed that
the desired intermetallic phase was obtained in single phase
for each compound (Figure S1). The H2-mediated isomerization
of cis-ST into trans-ST was performed in the presence of these
catalysts. When monometallic Rh/SiO2 was used, not only isom-
erization to trans-ST, but also overhydrogenation to diphenyl-
ethane (DPE) proceeded, lowering trans selectivity down to
ꢀ40% (Figure 1).
Figure 2. Selectivities to trans-alkenes during cis-ST and cis-MS isomeriza-
tions catalyzed by various SiO2-supported Rh- and Ru-based intermetallic
compounds and monometallic Rh under atmospheric pressure of H2. Selec-
tivities at 50% conversion are shown. The space group of the metallic phase
appears in parentheses.
to ST and MS, RhSb exhibited the highest selectivity (83%),
whereas the cubic compounds (RhZn, RhGa, and Rh) gave low
selectivities (50, 49, and 43%, respectively).
Next, the effect of Rh electronic state on selectivity was in-
vestigated by evaluating a) the binding energy of the Rh3d5/2
emission, b) d band center, and c) vibrational frequency of lin-
early adsorbed CO (Figure S3a–c). No correlation was observed
between the selectivities and these parameters (Figure S3d–f),
indicating that the selectivity does not depend on the elec-
tronic state of Rh.
Figure 1. Time courses of trans-ST selectivity, trans-ST yield, and DPE yield in
the liquid-phase hydrogen-mediated isomerization of cis-ST over Rh/SiO2
and RhSb/SiO2 catalysts.
Intermetallic compound RhSb was characterized in detail to
identify the geometric effect. Figure 3a shows a transmission
electron microscopy (TEM) image of a SiO2-supported interme-
tallic RhSb nanoparticle. The nanobeam diffraction (NBD) pat-
tern of this nanoparticle showed a set of diffraction spots cor-
responding to an intermetallic RhSb single crystal oriented
In contrast, the RhSb/SiO2 catalyst effectively inhibited the
overhydrogenation, affording 95% trans selectivity at full con-
version. The trans selectivities of the prepared catalysts at 50%
conversion were compared (Figure 2, red bars).
Intermetallic compounds RhIn, RhGa, and RhZn showed sim-
ilar selectivities to monometallic Rh but lower selectivities than
RhSb, RuSb, RhGe, and RhPb. Intermetallic RhSb exhibited the
highest selectivity (97%). Interestingly, the selectivity order ap-
peared to depend on the space group of the intermetallic
phases: Pnma (orthorhombic, RhSb, RuSb, RhGe)>P6/mmm
¯ ¯
along the [131] direction (Figure 3b). The equilibrium crystal
shape of RhSb (Figure 3c) was determined by Wulff construc-
tion[11] using surface energies (Table S1) of various RhSb low-
index planes (Figure S4) estimated by DFT calculation. The
RhSb crystal surface mainly comprised {211}, {020}, {013}, and
{102} planes and the crystal shape agreed with the TEM
image. Understanding of the geometric effect relies on an eval-
uation of exposed surface atomic arrangements. Figure 3d
shows atomic arrangements at equilibrium crystal surfaces.
Most planes displayed unique arrangements, in which Rh
atoms were one-dimensionally aligned (1D-plane; e.g., {211},
{020}, and {013}). The dominance of these 1D-planes may
result from the wafer-like structure of Pnma crystals. Although
bulk-terminated structures are shown here, note that almost
no surface reconstruction occurred during geometry optimiza-
tions (Figure S5). Outermost Sb atoms (typically 0.2 Š) were
only slightly lifted up. RhSb surfaces presented similar stability
¯
¯
(hexagonal, RhPb)>Pm3m, Fm3m (cubic, RhIn, RhGa, RhZn,
Rh), suggesting that geometry affects selectivity. To further ex-
amine this geometric effect, cis-b-methylstyrene (cis-MS), which
bears a less sterically hindered alkenyl moiety than cis-ST, was
isomerized. The resulting order of selectivity was similar to that
of cis-ST (Figure 2, green bars). Furthermore, the obtained se-
lectivities were lower than those of cis-ST for all catalysts, indi-
cating that the steric hindrance around the alkenyl moiety con-
tributed to the isomerization and/or hydrogenation. These re-
sults strongly suggest that geometric factors govern selectivity.
The isomerization of cis-2-butene, which is the least sterically
hindered inner alkene, was also assessed (Figure S2). Similarly
ChemCatChem 2015, 7, 3472 – 3479
www.chemcatchem.org
3473
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
presenting low CO coverage (1/8
full coverage for Rh) were adopt-
ed to exclude lateral interactions
between
CO
molecules
(Figure 4). Two geometrically dif-
ferent Rh sites (a and b,
Figure 4) were investigated for
the (211) plane.
Calculated frequencies for
these surfaces (2009, 2010, 2023,
and 2039 cmÀ1 for (020), (013),
(211)a, and (211)b, respectively)
agreed with the observed peak
(or shoulder) positions at low
coverage (ꢁ12 Pa), strongly con-
firming the dominance of these
surfaces. On the other hand, cal-
culations involving CO chemi-
sorption on bridge sites gave
considerably lower values (1908–
1922 cmÀ1, Table S2), which is
consistent with the absence of
bridged CO on RhSb. Similar
analyses were performed for
monometallic Rh/SiO2 and RhGa/
Figure 3. a) HR-TEM image of RhSb nanoparticles. b) NBD pattern focused on a single RhSb nanoparticle shown in
the center of (a). c) Equilibrium crystal shape of RhSb determined by Wulff construction. The crystal orientation
was equalized to that of the diffraction pattern (b). d) Crystal structure of RhSb showing the single unit cell (left)
and surface atomic arrangements displayed in (c) (right). e) HR-TEM image of another RhSb nanoparticle. f) Close-
up of the region delimited by the yellow square in (e). The crystal structure of RhSb viewed along the [001] direc-
tion is overlapped with the TEM image. Yellow and light blue lines represent experimental and reference d-spac-
ings, respectively.
to PdGa-like intermetallic compounds, which may originate
from covalent interactions between their metal components.[12]
Figure 3e and f show a high-resolution TEM image of anoth-
er RhSb nanoparticle and the magnification of the region des-
ignated by a yellow square, respectively. These images clearly
displayed lattice fringes corresponding to (020) planes. More-
over, the observed atomic arrays closely matched the corre-
sponding crystal structure viewed along the [001] direction.
Figure 3 f shows that RhSb crystal structure continues to the
outermost surface without any amorphous or deformed over-
layers. This indicates that surface degradation of RhSb nano-
crystals hardly occurs through TEM sample preparation using
organic solvent in the air. Crystal terminations by (020) and
(200) planes were also observed (3E), suggesting that these
planes were exposed. These results were consistent with the
exposure and dominance of 1D-planes in RhSb. In a wider-field
TEM image, similar polyhedral nanocrystals with 5~10 nm
sizes were mainly observed (Figure S6). However, a non-local
Figure 4. FT-IR spectra of CO chemisorbed on RhSb/SiO2 at various CO pres-
sures. Model structures of chemisorbed CO at low coverage (1/8 full cover-
age for Rh sites) and their calculated frequencies are also shown.
characterization is necessary to obtain information about the
entire RhSb catalyst. Therefore, a Fourier-transform infrared
(FT-IR) analysis involving CO chemisorption was subsequently
performed. Figure 4 shows FT-IR spectra of chemisorbed CO
on RhSb/SiO2 at various pressures. Some peaks assigned to lin-
early adsorbed CO[13] on Rh atoms appeared in the 2000–
2030 cmÀ1 region. No peak was observed below 1950 cmÀ1, in-
dicating the absence of two-fold species and, thus, large Rh
ensembles.[13] The peak positions remained almost unchanged
when the intensity was lower than half of the saturation value
(ꢁ12 Pa). Vibrational frequencies of CO chemisorbed on Rh
atop sites of major 1D-planes observed in RhSb ((211), (020),
and (013) planes) were estimated by DFT calculation. Models
SiO2. For Rh/SiO2, calculated frequencies for CO chemisorbed
on atop (2046 cmÀ1) and bridge sites (1930 cmÀ1) of Rh(111)
surface correctly reproduced reported experimental values
(2045 and 1925 cmÀ1, respectively,[14] Table S2). These calcula-
tions also agreed with experimental frequencies obtained in
this study at low coverage (Figure S7a), validating the compu-
tational approach. For RhGa/SiO2, the calculated frequencies of
CO chemisorbed on (100), (110), and (111) planes were con-
sistent with experimental results (Figure S7b).
ChemCatChem 2015, 7, 3472 – 3479
www.chemcatchem.org
3474
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
We now discuss the geometric role of 1D-plane on the selec-
tive isomerization. Hydrogen atom diffusion on the 1D-plane
surface may be restricted to the direction of the Rh row.
Indeed, on the RhSb(020) surface, DFT calculated activation
barriers for hydrogen diffusion pathways amounted to 138–
209 kJmolÀ1 across the Sb row, but 32–52 kJmolÀ1 along the
Rh row (bridge-atop-bridge) (Figure 5).
because Rh and Sb rows were not straight but slightly
zigzagged on the RhSb(020) surface, the adsorbate orientation
was also geometrically distinguishable: methyl moieties faced
left (cisL) or right (cisR) in cis-2-butene, whereas the trans-2-
butene molecule lay parallel (trans//) or across the Rh row
(transX). In the present study, possible adsorption geometries
of 2-butene (cis/trans, p/di-s, and orientations) and half-hydro-
genated isobutyl intermediates were systematically considered
(optimized structures are shown in Figure S8). The presence
and absence of co-adsorbed hydrogen atoms was also taken
into account. Figure 6 shows the adsorption energies of 2-bu-
tenes in various modes and representative optimized struc-
tures.
Figure 5. a) Hydrogen atom diffusion pathways over RhSb(020) and (211)
surfaces and b) their corresponding energy diagrams. Geometry optimized
stationary adsorption sites are designated as a–e and 1–8 (no stationary ad-
sorption site was found between 2 and 3 or 6 and 7). Total energies of
RhSb(020)-H(a) and RhSb(211)-H(5) slabs were set to zero. Energy barriers
of representative steps (kJmolÀ1): a!b (209), b!a (138), c!d (32), e!d
(52), 2!3 (63), 3!2 (76),4!5 (152), 6!7 (124), 7!6 (127).
Figure 6. Adsorption energies of cis and trans-2-butenes on bare and hydro-
gen-adsorbed (designated by H0) RhSb(020) surfaces for various adsorption
modes. Representative structures (side and top views), in which Rh, Sb, C,
and H atoms appear in black, blue, gray, and white, respectively, are also
shown.
The activation barriers for the Rh row exceeded the reported
values for a monometallic Rh(111) surface (ꢀ10 kJmolÀ1),[15]
which is probably due to the absence of stable hollow sites
and, consequently, the necessity to pass through relatively un-
stable atop sites.[15b] Similar calculations performed for the
RhSb(211) surface provided significantly higher energy barri-
ers, even for the pathway along Rh rows (63–127 kJmolÀ1,
Figure 5). This may be attributed to the longer Rh-Rh distances
obtained on RhSb(211) than on (020) in addition to the pres-
ence of intercepting Sb atoms in the Rh rows. On the basis of
these results, further analysis focused on RhSb(020) as an
active surface.
Surprisingly, di-s adsorptions were energetically unfavorable
compared to their p counterparts by ꢀ40 kJmolÀ1 regardless
of the presence of hydrogen, in contrast with alkene adsorp-
tion on Pt(111) and Pd(111).[18] Because the RhÀC bond length
should be shorter for the di-s adsorption than for the p ad-
sorption, repulsion between methyl moieties and the slightly
uplifted Sb atoms may have caused molecular deformation
and/or destabilization. cis Conformations were more favorable
than trans isomers regardless of orientation and bonding style.
This may stem from the ability of cis isomers to tilt their molec-
ular plane toward the surface to reduce steric repulsion, which
is unavoidable in trans conformations.[19] The presence of ad-
sorbed hydrogen gave more positive adsorption energies, as
reported for the Pt(111) surface,[19b] without changing their
order for various modes. This increase in adsorption energy
may result from the hydrogen coverage-induced decrease in
Finally, the stereochemistry on 1D-planes was assessed by
investigating the adsorption geometries of cis and trans-2-bu-
tenes as a model alkenes on the RhSb(020) surface and the
energy profile of the isomerization. Generally, alkene exhibit p
and di-s adsorptions on metallic surfaces. The conventional
Horiuti–Polanyi mechanism[16] suggests that the conversion of
di-s-bonded alkenes into secondary alkyl intermediates is
a key step in alkene isomerization and hydrogenation. Howev-
er, p-bonded alkenes may also participate in the hydrogena-
tion as has been proposed for Pt(111) surfaces.[17] Moreover,
ChemCatChem 2015, 7, 3472 – 3479
www.chemcatchem.org
3475
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
surface energy. We calculated adsorption energies of 2-butenes
also for a bare (211) surface. Much more positive adsorption
energies than those for (020) surface were obtained (14–20
and 37–58 kJmolÀ1 for p and di-s modes, respectively, Fig-
ure S9). This unfavored adsorption is probably due to its small
surface energy (Table S1), i.e., being stable surface. Our at-
tempt for temperature-programed desorption of 2-butenes on
RhSb/SiO2 resulted in rapid desorption upon evacuation even
at room temperature (data not shown) and failed to estimate
an experimental adsorption energy. However, this rapid de-
sorption is consistent with the positive or slightly negative ad-
sorption energies in Figure 6 and S9. Figure 7 shows the
energy diagram for 2-butene isomerization over RhSb(020)
with several adsorbate structures. This diagram addresses cisL–
trans// (solid lines) and cisR–transX isomerizations (dotted lines)
involving p adsorption via isobutyl intermediates and CÀC ro-
tation. Typically, hydrogen atoms are assumed to access the al-
kenyl carbons from the bottom (H0-TS1-H, see yellow arrow in
2). In cis conformations, one of the two alkenyl carbon atoms
and the neighboring Rh-Rh bridge site are sterically blocked
by one methyl group. Therefore, the hydrogen atom preferen-
tially accesses the “open” side of the neighboring bridge site
and/or the alkenyl carbon atoms rather than their blocked
side. In the cisL conformer, hydrogen access to the blocked
side of the neighboring RhÀRh bridge site (yellow arrow in 2’)
requires skeletal deformation (methyl rotation) and tilting of 2-
butene at transition state TS1, resulting in a higher energy bar-
rier (66 kJmolÀ1, green solid line through 2’) than for the open
side (49 kJmolÀ1, blue solid line through 2). A similar situation
is also observed for the cisR conformation (61 kJmolÀ1, blue
dotted line through 7). Hydrogen attack to cisL-2-butene gen-
erates an isobutyl intermediate (4) via transition state TS2 with
an activation energy of 44 kJmolÀ1. Isomerization to trans//-2-
butene proceeds via CÀC rotation (TS3, Ea =10 kJmolÀ1) and
subsequent b-H (orange) elimination (TS2’, Ea =18 kJmolÀ1).
The blocking methyl group may also hinder additional hydro-
gen attack to the isobutyl intermediate from the opposite side,
minimizing overhydrogenation to alkane. The formed trans//-2-
butene can desorb directly into the gas phase because of the
absence of activation barrier. Hydrogen atoms do not access
trans//-2-butene (6) because both alkenyl carbon atoms are
blocked by the methyl groups. Both alkenyl carbon atoms are
open in transX-2-butene, making the two-atom hydrogenation
possible. However, this process is limited because transX-2-
butene adsorption is less energetically favorable than other
conformations (Figure 6 and 7). Applying Boltzmann distribu-
tion to adsorption energy difference between cisL and transX
(described below; Ecis ÀEtrans =DE=12 kJmolÀ1) yields 45-fold
L
X
difference in adsorbate ratio even at 373 K, which is sufficient
to ignore the reverse reaction [Eq. (1)]:
expðE =RTÞ
½cisLÀpŠ=½cisLðgފ
cisL
¼
¼ expðDE=RTÞ
ð1Þ
½transXÀpŠ=½transXðgފ expðEtrans =RTÞ
X
The isomerization process via diÀs bonded 2-butenes was
also considered but their hydrogenations exhibited dramatical-
ly higher activation energies (75–132 kJmolÀ1, Figure S10) than
those of p-bonded alkenes. Di-s bonded species thus dis-
played significantly higher energy barriers for adsorption and
hydrogenation, ruling out their contribution to this catalysis.
Thus, several combined effects play a critical role on the se-
lective cis–trans isomerization of 2-butene on the RhSb(020)
surface. a) Hydrogen atom diffusion on the surface is restricted
to the direction of the one-dimensionally aligned Rh row.
b) cis-2-Butene adsorption is considerably more favorable than
that of trans-2-butene. c) In cis conformations, steric hindrance
from the methyl group limits hydrogen atom access to one
side of the alkene (Scheme 1a).
The second effect promotes the desorption of the resulting
trans isomer in addition to enhancing the third effect, which
enables the one-atom hydrogenation and isomerization, while
inhibiting the two-atom hydrogenation (Scheme 1c). This
highly selective isomerization
catalysis relies on the specific
and regular surface atomic ar-
rangement of the orthorhombic
Pnma intermetallic compound.
The characteristic zigzag align-
ment of the Rh atoms also con-
tributes to the surface stereo-
chemistry. A straight alignment
may result in equal steric hin-
drance of the alkenyl carbon
atoms from two alkyl groups,
preventing geometric differen-
tiation between open or
blocked. Note that the overall
selectivity should be a weighted
superposition of the different
Figure 7. Energy diagram for the cis–trans isomerization of 2-butene on the RhSb(020) plane and adsorbate struc-
tures (side and top views, 1–8). Total energy of gas-phase trans-2-butene and hydrogen-adsorbed RhSb(020) slab
was set to zero. Hydrogen atoms involved in the isomerization appear in yellow and orange. The green line corre-
sponds to hydrogen access to the alkenyl carbon of cisL-2-butene from the top, which is sterically hindered by the
methyl group.
contribution of each plane; for
RhSb, (211), (020), and (013)
planes. As discussed above,
since (211) plane can be regard-
ChemCatChem 2015, 7, 3472 – 3479
www.chemcatchem.org
3476
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
because contribution of corners and edges cannot be ignored
for nanoparticles smaller than 10 nm.[20] However, the wafer
like crystal structure of Pnma RhSb should exposes 1D-like sur-
face for most Miller indices excepting {200} planes. The con-
cept of selective isomerization on 1D-plane is simple and
would essentially be applicable to the actual surface.
Conclusions
The origin of the stereochemical cis–trans alkene conversion
on Rh-based intermetallic compounds was investigated using
multiple techniques combined with DFT calculations. The
trans-alkene selectivity strongly depends on geometric factors,
such as space group of the intermetallic phases and alkene
size, instead of their electronic states. Highly selective interme-
tallic compounds, such as RhSb, exhibit an orthorhombic
Pnma structure, of which surfaces are dominated by planes
consisting of one-dimensionally aligned Rh rows separated by
Sb (1D-planes) with minimum surface relaxation. Geometric
constraints on 1D-planes and steric hindrance from one alkyl
group of cis-alkene effectively limit hydrogen access to the al-
kenyl carbon to one direction, which enables the one-atom hy-
drogenation hence isomerization, while inhibiting the two-
atom hydrogenation to alkane. This unprecedented surface
stereochemistry largely relies on the unique and highly or-
dered surface atomic arrangement of the intermetallic com-
pounds, which is rare in other inorganic materials. This study
provides the first example of selective stereochemical conver-
sion controlled by surface order. The gained insight is expected
to open new horizons for several fields, such as catalytic
chemistry, surface science, stereochemistry, and materials sci-
ence.
Scheme 1. Geometric restriction of hydrogen access to cis-2-butene on a) 1D
and b) Rh-terminated planes such as RhSb (020) and RhGa(100), respective-
ly. c) Selective cis–trans alkene isomerization without overhydrogenation to
alkane.
ed as almost inactive due to the significantly high hydrogen
diffusion barriers and positive 2-butene adsorption energies,
its contribution to the catalysis may be negligible. Therefore,
overall selectivity depends strongly on the nature of (020) and
(013) planes. For (013) surface, the nature and surface stereo-
chemistry are likely to be similar to those for (020) surface be-
cause its surface energy (Table S1) and atomic arrangement
(Figure 3 and S4) are almost identical to those of (020) surface.
This means that the overall selectivity is governed by a portion
of Pnma crystal facets.
Experimental Section
For monometallic Rh, because hydrogen atoms can access
the alkenyl moiety from several orientations, the two-atom hy-
drogenation easily occurs, explaining the low selectivity. Simi-
Catalyst preparation
Silica-supported Rh or Ru catalysts (Rh/SiO2, Ru/SiO2) were pre-
pared by a pore-filling impregnation. An aqueous Rh(NO3)3 (N.E.
ChemCat Corp., 99%) or RuCl3 solution (Furuya Metal Co., Ltd.,
99%) was added to dried silica gel (Cariact G-6, Fuji Silysia Ltd.)
The metal solution volume was calculated to fill silica gel pores
and achieve 3 wt% metal loading. The mixture was sealed over-
night at room temperature using a piece of plastic film, dried over
a hot plate at 373 K, and calcined under dry air at 723 K for 4 h.
The catalyst was subsequently reduced under H2 flow
(60 mLminÀ1, 99.9995%, Taiyo Nippon Sanso Corp.) for 2 h at 823
(Rh) or 723 K (Ru).
¯
larly for other cubic Pm3m intermetallic compounds such as
RhGa, RhIn, and RhZn, the two-atom hydrogenation may pro-
ceed on Rh-terminated (100) surfaces (Scheme 1b and Fig-
ure S7b). In this catalytic system, selectivity also depends on
the degree of steric hindrance from the alkyl moiety of ad-
sorbed alkene, which is consistent with the deviations in selec-
tivity between cis-ST and cis-MS isomerizations (Figure 2). The
phenyl group of ST displays a significantly higher steric effect
than the methyl group of MS, giving rise to higher selectivity.
We reported in the previous work that RhSb was almost inac-
tive for semihydrogenation of alkyne to alkene. This supports
that RhSb surface catalyzes one-atom hydrogenation but not
two-atom hydrogenation. More likely, for alkynes on 1D-plane,
hydrogen access to the alkynyl carbons may be blocked com-
pletely by two alkyl moieties on an extension of the CÀC triple
bond.
Silica-supported intermetallic catalysts (RhM/SiO2 with M=Ga, Ge,
In, Pb, Sb, Zn; RuSb/SiO2) were prepared by successive impregna-
tions of Rh/SiO2 or Ru/SiO2. The RhSb catalyst was obtained by
adding an aqueous SbCl3 solution (99%, Kanto) to Rh/SiO2 to
adjust the Rh/Sb atomic ratio to 1. The mixture was sealed by
a piece of plastic film overnight at room temperature, dried over
a hot plate, and reduced under H2 flow at 1073 K for 1 h. Other in-
termetallic catalysts (RhGa, RhGe, RhIn, RhPb, and RhZn) were pre-
pared by a similar procedure at various reduction temperatures
using Ga(NO3)3·8H2O (Wako, 99%, 1073 K), (NH4)2GeF6 (Aldrich,
99%, 1073 K), InCl3·4H2O (Wako, 723 K), Pb(NO3)2 (Kanto, 99%,
Although we here demonstrated the case of an ideal sur-
face, slight surface modification might be caused in the actual
reaction condition by other factors such as solvent. Moreover,
some nanoparticles may not show an ideal Wulff polyhedron
ChemCatChem 2015, 7, 3472 – 3479
www.chemcatchem.org
3477
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
1023 K), and Zn(NO3)2·2H2O (Wako, 99%, 1073 K), respectively. Bulk
intermetallic catalysts (RhM; M=Ga, Ge, In, Sb) were prepared by
arc melting of pure metal beads at a 1:1 atomic ratio under Ar at-
mosphere. The resulting ingots were crushed into fine powder in
the air.
nected to a glass circulation system. The catalyst was reduced
under H2 flow at 673 K for 0.5 h, evacuated at the same tempera-
ture for 0.5 h, and cooled to room temperature. Next, a spectrum
was recorded as a baseline for subsequent measurements. A 101–
103 Pa CO was introduced stepwise at room temperature. All spec-
tra were acquired at a 4 cmÀ1 resolution.
Reaction conditions
Computational Details
Isomerizations of cis-stilbene (cis-ST) and cis-b-methylstyrene (cis-
MS) were carried out in a 100 mL three-neck round bottom flask
equipped with a reflux condenser and a gas balloon. Prior to the
reaction, the catalyst (100 mg) was reduced under flowing H2
(60 mLminÀ1) for 1 h in the reactor heated to 673 K using a mantle
heater. The reactor was subsequently cooled to room temperature
and the atmosphere was completely replaced by dry Ar. A THF so-
lution (5.0 mL) of cis-ST or cis-MS (5 mmol, TCI, 99%) was added to
the reactor through a silicone septum. The reaction was initiated
at 298 K by flowing H2 at 100 mLminÀ1 for 1 min and 5 mLminÀ1
afterward. Products present in the liquid phase were quantified
using a flame ionization detection gas chromatograph (Shimadzu,
GC-14B) equipped with a capillary column (GL Sciences, TC-70,
0.25 mm”60 m). The trans-alkene selectivity at 50% conversion
was determined using 50 mg of catalyst and 1 mmol of cis-alkenes.
Periodic density functional theory (DFT) calculations, except fre-
quency calculations, were performed using the CASTEP code[21]
with Vanderbilt-type ultrasoft pseudopotentials[22] and a revised
Perdew–Burke–Ernzerhof exchange-correlation functional (RPBE)[23]
based on the generalized gradient approximation (GGA). The
plane-wave basis set was truncated at a kinetic energy of 370 eV. A
Fermi smearing of 0.1 eV was utilized. The reciprocal space was
À1
sampled using a k-point mesh with a typical spacing of 0.04 Š
generated by the Monkhorst–Pack scheme.[24]
Geometry optimizations were performed in supercell structures
using periodic boundary conditions. Surfaces were modeled using
6 atomic layer-thick metallic slabs. Surface energies were calculated
using (1”1) unit cells with a 10 Š vacuum region separating the
two surfaces under symmetry restrictions. When any adsorbate
(CO, H, and/or cis- or trans-2-butenes) was included, (2”1)
(RhSb(211) and RhSb(013)), (2”2) (RhSb(020)), or (3”3) (Rh(111),
RhGa(100), RhGa(110), and (RhGa(111)) unit cells with a 13 Š
vacuum region were used without symmetry restrictions. Atomic
coordinates were fully relaxed, while the lattice constants were
fixed. Convergence criteria comprised a) a self-consistent field
cis-2-Butene was isomerized in a continuous fixed-bed flow reactor
(quartz, I.D.=6 mm). Prior to the reaction, the catalyst (10 mg) was
reduced under H2 flow (60 mLminÀ1) in the reactor at 623 K for 1 h
and allowed to cool to 373 K under He flow (60 mLminÀ1). The re-
action was initiated by flowing the 2:1:15 cis-2-butene/H2/He reac-
tion mixture (total flow rate: 90 mLminÀ1) at 373 K. Products pres-
ent in the gas phase were analyzed using a thermal conductivity
detection gas chromatograph (Shimadzu, GC-8A) equipped with
a packed column (VZ-10, 3 mm”6 m).
(SCF) tolerance of 2.0”10À6 eV/atom, b) an energy tolerance of
1.0”10À5 eV/atom, c) a maximum force tolerance of 0.05 eVŠ
,
À1
and d) a maximum displacement tolerance of 1.0”10À3 Š for struc-
ture optimization and energy calculation.
The adsorption energy was defined as Equation (2):
Characterization
Ead ¼ EAÀSÀðES þ EAÞ
ð2Þ
Crystal structures were examined by powder X-ray diffraction
(XRD) under CuKa radiation using a Rigaku RINT2400 apparatus. Dif-
ference XRD patterns were obtained by subtracting the silica sup-
port pattern from those of the supported catalysts to give a flat
baseline. Transmission electron microscopy (TEM) measurements
were conducted using a JEOL JEM-2010F instrument at an acceler-
ating voltage of 200 kV. All TEM specimens were prepared by soni-
cating the samples in tetrachloromethane and dispersing them on
a copper grid by an ultrathin carbon film supported. Nanobeam
diffraction patterns were acquired at a camera distance of 30 cm.
The camera constant was calibrated using a gold standard for re-
flection index determination. X-ray photoelectron spectra (XPS)
were recorded using an ULVAC PHI 5000 VersaProbe spectrometer.
Catalysts were pressed into a pellet and reduced under H2 flow
(60 mLminÀ1) at 873 K in a quartz reactor for 1 h. Spectra were ob-
tained under AlKa X-ray radiation using the C 1s peak (248.8 eV) to
calibrate the binding energy. Valence band XPS measurements
were performed using unsupported bulk catalysts because their
supported counterparts produced extremely low valence band in-
tensities. Bulk catalysts exhibited similar selectivities to their corre-
sponding supported catalysts. For example, bulk RhSb showed
96% trans selectivity at 50% conversion of cis-ST. The d band
center (Cd) was defined as the position of the vertical line bisecting
the d band area. Fourier-transform infrared (FT-IR) spectra of ad-
sorbed CO were obtained using a JASCO FT/IR-430 spectrometer in
For which EAÀS is the energy of the slab with the adsorbate, EA is
the total energy of the free adsorbate, and ES is the total energy of
the bare slab. The adsorption energy with a hydrogen-adsorbed
slab was calculated using the total energy of hydrogen-adsorbed
slab (ESH) instead of ES.
Surface energies of various planes, except RhSb(200), were esti-
mated using Equation (3):
1
ð3Þ
g ¼ lim
½EsÀNEBŠ
n!1 2 A
For which EB is the energy of bulk unit cell, A is the surface area, n
is the number of layers, and N is the number of unit cells in the
slab. Surface energy calculations were conducted for densely
packed low-index RhSb(111), (102), (112), (211), (200), (202),
(020), and (013) planes. These surface energies converged within
6 or 7 atomic layers for several surfaces (Table S1). Unlike other
planes, RhSb(200) displays one surface consisting of Rh atoms
only (RhSb(200)Rh) and another one composed of Sb atoms only
(RhSb(200)Sb). This procedure gave averaged surface energies for
RhSb(200)Rh (gRh) and RhSb(200)Sb (gSb) instead of individual contri-
butions. Individual surface energies can be estimated using a non-
stoichiometric slab model in which both surfaces are terminated
by an identical plane (6). According to literature[11b,c] for example,
transmission mode. A self-supporting catalyst wafer (50 mgcmÀ2
was placed in a quartz cell equipped with CaF2 windows and con-
)
ChemCatChem 2015, 7, 3472 – 3479
www.chemcatchem.org
3478
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[2] a) J. Mlynarski, S. Bas, Chem. Soc. Rev. 2014, 43, 577–587; b) T. Ohkuma,
J. Synth. Org. Chem. Jpn. 2007, 65, 1070–1080.
gRh for the Rh-terminated slab is described as Equation (4):
[3] a) Z. An, J. He, Y. Dai, C. G. Yu, B. Li, J. He, J. Catal. 2014, 317, 105–113;
b) J. M. Thomas, R. Raja, Acc. Chem. Res. 2008, 41, 708–720; c) M. Nasr-
Esfahani, I. Mohammadpoor-Baltork, A. R. Khosropour, M. Moghadam, V.
Mirkhani, S. Tangestaninejad, H. A. Rudbari, J. Org. Chem. 2014, 79,
1437–1443.
[4] a) M. Studer, H. U. Blaser, C. Exner, Adv. Synth. Catal. 2003, 345, 45–65;
b) C. J. Baddeley, Top. Catal. 2003, 25, 17–28.
[5] a) E. J. Creyghton, S. D. Ganeshie, R. S. Downing, H. Vanbekkum, J.
Chem. Soc. Chem. Commun. 1995, 1859–1860; b) E. J. Creyghton, S. D.
Ganeshie, R. S. Downing, H. vanBekkum, J. Mol. Catal. A 1997, 115, 457–
472.
[6] a) S. S. Kim, T. J. Pinnavaia, R. Damavarapu, J. Catal. 2008, 253, 289–294;
b) K. Smith, S. D. Roberts, G. A. Ei-Hiti, Org. Biomol. Chem. 2003, 1,
1552–1559.
[7] J. H. Westbrook, R. L. Fleischer, Magnetic, Electrical and Optical Properties
and Applications of Intermetallic Compounds, Wiley, 2000.
[8] a) Z. Nawaz, F. Baksh, J. Zhu, F. Wei, J. Ind. Eng. Chem. 2013, 19, 540–
546; b) S. Scir›, G. Burgio, C. Crisafulli, S. Minico, Appl. Catal. A 2004,
274, 151–157.
[9] a) M. Armbrüster, M. Behrens, F. Cinquini, K. Fçttinger, Y. Grin, A.
Haghofer, B. Klçtzer, A. Knop-Gericke, H. Lorenz, A. Ota, S. Penner, J.
Prinz, C. Rameshan, Z. Revay, D. Rosenthal, N. Rupprechter, P. Sautet, R.
Schlçgl, L. D. Shao, L. Szentmiklosi, D. Teschner, D. Torres, R. Wagner, R.
Widmer, G. Wowsnick, ChemCatChem 2012, 4, 1048–1063; b) M.
Armbrüster, K. Kovnir, M. Behrens, D. Teschner, Y. Grin, R. Schlçgl, J. Am.
Chem. Soc. 2010, 132, 14745–14747; c) K. Kovnir, M. Armbrüster, D.
Teschner, T. V. Venkov, L. Szentmiklosi, F. C. Jentoft, A. Knop-Gericke, Y.
Grin, R. Schlogl, Surf. Sci. 2009, 603, 1784–1792.
[10] S. Furukawa, A. Yokoyama, T. Komatsu, ACS Catal. 2014, 4, 3581–3585.
[11] a) G. Wulff, Z. Kristallogr. Mineral. 1901, 34, 449–530; b) N. Moll, A. Kley,
E. Pehlke, M. Scheffler, Phys. Rev. B 1996, 54, 8844–8855; c) A. Dannen-
berg, M. E. Gruner, A. Hucht, P. Entel, Phys. Rev. B 2009, 80, 245438.
[12] D. Rosenthal, R. Widmer, R. Wagner, P. Gille, M. Armbruster, Y. Grin, R.
Schlogl, O. Groning, Langmuir 2012, 28, 6848–6856.
[13] K. Tomishige, K. Asakura, Y. Iwasawa, J. Catal. 1994, 149, 70–80.
[14] A. C. Yang, C. W. Garland, J. Phys. Chem. 1957, 61, 1504–1512.
[15] a) M. Mavrikakis, J. Rempel, J. Greeley, L. B. Hansen, J. K. Norskov, J.
Chem. Phys. 2002, 117, 6737–6744; b) B. S. Bunnik, G. J. Kramer, J. Catal.
2006, 242, 309–318.
[16] J. Horiuti, M. Polanyi, Trans. Faraday Soc. 1934, 30, 1164–1172.
[17] P. S. Cremer, X. C. Su, Y. R. Shen, G. A. Somorjai, J. Am. Chem. Soc. 1996,
118, 2942–2949.
[18] A. Valcµrcel, A. Clotet, J. M. Ricart, F. Delbecq, P. Sautet, Surf. Sci. 2004,
549, 121–133.
[19] a) I. Lee, F. Delbecq, R. Morales, M. A. Albiter, F. Zaera, Nat. Mater. 2009,
8, 132–138; b) F. Delbecq, F. Zaera, J. Am. Chem. Soc. 2008, 130, 14924–
14925.
[20] a) H. Graoui, S. Giorgio, C. R. Henry, Philos. Mag. B 2001, 81, 1649;
b) S. W. Wang, L. M. Falicov, A. W. Searcy, Surf. Sci. 1984, 143, 609.
[21] M. D. Segall, P. J. D. Lindan, M. J. Probert, C. J. Pickard, P. J. Hasnip, S. J.
Clark, M. C. Payne, J. Phys. Condens. Matter 2002, 14, 2717–2744.
[22] D. Vanderbilt, Phys. Rev. B 1990, 41, 7892–7895.
[23] a) B. Hammer, L. B. Hansen, J. K. Norskov, Phys. Rev. B 1999, 59, 7413–
7421; b) Y. K. Zhang, W. T. Yang, Phys. Rev. Lett. 1998, 80, 890–890.
[24] H. J. Monkhorst, J. D. Pack, Phys. Rev. B 1976, 13, 5188–5192.
[25] a) N. Govind, M. Petersen, G. Fitzgerald, D. King-Smith, J. Andzelm,
Comput. Mater. Sci. 2003, 28, 250–258; b) T. A. Halgren, W. N. Lipscomb,
Chem. Phys. Lett. 1977, 49, 225–232.
1
2 A
ð4Þ
gRh
¼
½EsÀNSb mRhSbÀDN mRhŠ
For which NSb is the number of Sb atoms in the slab, DN is the
excess of Rh atoms relative to Sb atoms, mRhSb is the chemical po-
tential of bulk RhSb, and mRh is the chemical potential of Rh, re-
spectively.
Because of the bulk intermetallic compound stability against de-
composition, the chemical potential mRh should lie within this
range:
m
RhðbulkÞÀjDHRhSbj ꢁ mRh ꢁ mRhðbulkÞ
ð5Þ
Therefore, gRh is expected to range within:
1
2 A
½EsÀNSb mRhSbÀDNðmRhðbulkÞÀjDHRhSbjފ ꢁ gRh
ð6Þ
1
ꢁ
½EsÀNSb mRhSbÀDNmRhðbulkފ
2 A
For which DHRhSb is the formation enthalpy of RhSb.
The transition state (TS) search was performed using the complete
LST/QST method.[25] A linear synchronous transit (LST) maximiza-
tion was performed, followed by an energy minimization in the di-
rections conjugating to the reaction pathway. The approximated
TS structure was subjected to a quadratic synchronous transit
(QST) maximization with conjugate gradient minimization refine-
ments. This cycle was repeated until a stationary point was found.
The convergence criterion for TS calculÀa1tions was set to a root-
mean-square force tolerance of 0.10 eVŠ
.
Frequency calculations of adsorbed CO molecules were conducted
from geometry optimized structures using the DMol3 code.[26]
These calculations involved the RPBE functional, a double-numeric
quality basis set with polarization functions (DNP, comparable to
Gaussian 6-311G**)[27] with a real-space cutoff of 4.2 Š, DFT semi-
core pseudopotential core treatment,[28] and a Fermi smearing of
0.1 eV. The SCF convergence was accelerated using the iterative
scheme proposed by Kresse and Furthmüller.[29] The partial Hessian
matrix including C and O atoms was computed to evaluate the
harmonic frequencies for adsorbed CO. All computed harmonic fre-
quencies were scaled by an empirical factor of 1.0624, which corre-
sponds to the ratio of experimental[30] and computed values for
gas-phase CO (2143 cmÀ1/2017 cmÀ1).
Acknowledgements
This work was supported by a Japan Society for the Promotion
of Science KAKENHI (Grant Number 23360353). We are grateful to
the Center for Advanced Materials Analysis of Tokyo Institute of
Technology for the aid of TEM observations.
[26] B. Delley, J. Chem. Phys. 1990, 92, 508–517.
[27] Y. Inada, H. Orita, J. Comput. Chem. 2008, 29, 225–232.
[28] B. Delley, Phys. Rev. B 2002, 66, 155125.
[29] G. Kresse, J. Furthmuller, Phys. Rev. B 1996, 54, 11169–11186.
[30] D. Scarano, S. Bordiga, C. Lamberti, G. Spoto, G. Ricchiardi, A. Zecchina,
C. O. Arean, Surf. Sci. 1998, 411, 272–285.
Keywords: atomic arrangement · intermetallic compound ·
isomerization · stereochemistry · surface chemistry
[1] a) S. Oudeyer, J. F. Briere, V. Levacher, Eur. J. Org. Chem. 2014, 6103–
6119; b) Y. L. Liu, J. S. Yu, J. Zhou, Asian J. Org. Chem. 2013, 2, 194–206;
c) X. Wang, K. Sun, Y. H. Lv, F. J. Ma, G. Li, D. H. Li, Z. H. Zhu, Y. Q. Jiang,
F. Zhao, Chem. Asian J. 2014, 9, 3413–3416.
Received: July 22, 2015
Published online on October 26, 2015
ChemCatChem 2015, 7, 3472 – 3479
www.chemcatchem.org
3479
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim