Journal of the American Chemical Society
ARTICLE
Figure 5A. As expected, shape does not exert a considerable
effect on TOF for MBY transformation since both (111) and
(100) planes behave similarly. Cubo-octahedra, however, are
slightly disfavored due to the larger fraction of edge atoms (or
σ2 sites), which are considerably less active in the hydrogena-
tion of MBY relative to plane atoms (or σ1 sites). The most
significant size effect is observed in the 3ꢀ20 nm range, after
which the reaction becomes size and shape independent with
respect to TOF. Nanocrystals smaller than 3 nm in size start to
lose their bulk properties,2,25 and, consequently, the catalytic
behavior may not follow the prediction. This region is depicted
with dotted lines.
If, on the other hand, the MBY transformation rate is
referenced to the total amount of Pd, then both size and shape
effects change substantially (Figure 5B). While there would be no
difference between cubic or octahedral nanocrystals, cuboctahe-
dra will be significantly less active due to their low dispersions.
This difference, however, can be neglected for particles larger
than 15 nm in size, after which the reaction becomes again size
and shape independent. Therefore, if the activity of a catalyst is
used as the only optimization criterion, either cubes or octahedra
of roughly ∼5 nm in size should be the best choice for the
catalyst.
Selectivity is a very, if not the most, important catalytic
property for most applications. Therefore, a dual selectivityꢀ
activity criterion for optimization seems to be well-justified. To
achieve this, we had to combine eqs 8ꢀ16 with the set of
empirically obtained constants (Table 3) and the statistics of
surface atoms for the nanocrystals in study. This allowed us to
obtain simulation curves analogous to those shown in Figures S6
and 4 for cubic, octahedral, and cuboctahedral nanocrystals with
sizes in the range 3ꢀ50 nm (Figure S7, Supporting Information).
From such simulations, the selectivity toward MBE at 95%
conversion can be estimated for all different nanocrystals:
were prepared with cubic, octahedral, and cuboctahedral shapes
using a solution-phase method, with PVP serving as a stabilizing
agent. The observed activity and selectivity suggested that two
types of active sites were involved in the catalysis, which differ in
coordination numbers and are located on planes and edges,
respectively. A two-site LangmuirꢀHinshelwood kinetic model
allowed for an accurate description of the experimentally ob-
served activity and selectivity. Semihydrogenation to MBE was
found to occur on both types of sites, but the reactivity depended
on the coordination number of the atoms. Edge atoms were
4-fold less active in the semihydrogenation as compared to the
plane atoms. Overhydrogenation to MBA occurred solely on the
edge atoms presumably due to increased adsorption strength of
the alkene. Selectivity was then linked to the fraction of edge sites
on each type of nanocrystal. Kinetic simulations pointed toward
∼3ꢀ5 nm cubic nanocrystals as an optimal catalyst for the
highest productivity of MBE.
Metal nanocrystals with tuned sizes and shapes prepared via
colloidal techniques can be considered as a new generation of
model catalysts, which allow overcoming the material and pres-
sure gaps in catalysis and complementing single-crystal studies,
which inherently lack the complexity of industrial catalysis. The
approach shown in this work provides a powerful tool for rational
catalyst design for industrially relevant chemical reactions under
real conditions.
’ ASSOCIATED CONTENT
S
Supporting Information. Statistics of surface atoms for
b
common fcc crystal shapes, HRTEM imaging and particle size
distribution of the nanocrystals, XPS spectra of CUB6 sample,
TEM image of cubic nanoparticles after the reaction, complete
LangmuirꢀHinshelwood mechanism for the hydrogenation of
2-methyl-3-butyn-2-ol, kinetic modeling with one set of adsorp-
tion constants, and kinetic simulation results to estimate the
productivity of the target product MBE. This material is available
CE
C0Y ꢀ CY
SE ¼
ð17Þ
Figure 5C shows the results of the simulations. Selectivity
toward MBE increases monotonically with particle size in the
order cubes > octahedra > cuboctahedra. Figure 5B and C can be
combined (Figure 5D) to simultaneously optimize selectivity
and activity, showing that the reaction is indeed structure
sensitive in the size range studied. In this case, the productivity
of the target product is plotted against particle size for all three
shapes. As a result, two optimal nanocrystals can be proposed. If
the productivity of MBE is to be maximized, then cubic nano-
crystals of approximately 3ꢀ5 nm would be the best choice. If, on
the other hand, a pure selectivity criterion is judged as more
appropriate, then larger cubic nanocrystals should be chosen. In
this case, a compromise between productivity and selectivity
must be reached.
’ AUTHOR INFORMATION
Corresponding Author
lioubov.kiwi-minsker@epfl.ch
’ ACKNOWLEDGMENT
This work was supported by the Swiss National Science
Foundation (grant no. 200021-118067 to L.K.M.) and the U.S.
National Science Foundation (DMR-0804088 to Y.X.). The
authors thank Marco Cantoni (EPFL-SB-CIME) for the high-
resolution TEM images and Nicolas Xanthopoulos (EPFL-SB-
CIME) for the XPS measurements.
The structure sensitivity of a chemical reaction is specific for
each catalytic system.25 Therefore, such a modeling-simulation
approach combined with the experimental kinetic data obtained
from uniform, well-defined metal nanocrystals gives a powerful
tool for rational catalyst design for a given chemical reaction.
’ REFERENCES
(1) Boudart, M. Chem. Rev. 1995, 95, 661.
(2) Che, M.; Bennett, C. O. Adv. Catal. 1989, 36, 55.
(3) Molnar, A.; Sarkany, A.; Varga, M. J. Mol. Catal. A 2001,
173, 185.
(4) Van Hardeveld, R.; Hartog, F. Surf. Sci. 1969, 15, 189.
(5) Freund, H. J. Top. Catal. 2008, 48, 137.
(6) Silvestre-Albero, J.; Rupprechter, G.; Freund, H. J. J. Catal. 2006,
240, 58.
’ CONCLUSIONS
We have studied the structure sensitivity of the water-assisted
selective hydrogenation of MBY over uniform, unsupported Pd
nanocrystals with different sizes and shapes. The Pd nanocrystals
(7) Silvestre-Albero, J.; Rupprechter, G.; Freund, H. J. Chem. Commun.
2006, 1, 80.
12793
dx.doi.org/10.1021/ja204557m |J. Am. Chem. Soc. 2011, 133, 12787–12794