Letter
pubs.acs.org/NanoLett
Enhancing the Hydrogen Activation Reactivity of Nonprecious Metal
Substrates via Confined Catalysis Underneath Graphene
Yinong Zhou,† Wei Chen,†,‡ Ping Cui,*,† Jiang Zeng,†,§ Zhuonan Lin,∥,† Efthimios Kaxiras,‡
and Zhenyu Zhang*,†
†
International Center for Quantum Design of Functional Materials (ICQD), Hefei National Laboratory for Physical Sciences at the
Microscale, and Synergetic Innovation Center of Quantum Information and Quantum Physics, University of Science and Technology
of China, Hefei, Anhui 230026, China
‡
Department of Physics and School of Engineering and Applied Sciences, Harvard University, Cambridge, Massachusetts 02138,
United States
§
Beijing Computational Science Research Center, Beijing 100094, China
∥
School of Physics and Technology, Wuhan University, Wuhan 430072, China
ABSTRACT: In the hydrogen evolution reaction (HER), the reactivity as a
function of the hydrogen adsorption energy on different metal substrates
follows a well-known volcano curve, peaked at the precious metal Pt. The
goal of turning nonprecious metals into efficient catalysts for HER and other
important chemical reactions is a fundamental challenge; it is also of
technological significance. Here, we present results toward achieving this goal
by exploiting the synergistic power of marginal catalysis and confined
catalysis. Using density functional theory calculations, we first show that the
volcano curve stays qualitatively intact when van der Waals attractions
between a hydrogen adatom and different metal (111) surfaces are included.
We further show that the hydrogen adsorption energy on the metal surfaces
is weakened by 0.12−0.23 eV when hydrogen is confined between graphene
and the metal surfaces, with Ni exhibiting the largest change. In particular, we
find that the graphene-modified volcano curve peaks around Ni, whose bare
surface already possesses moderate (or marginal) reactivity, and the corresponding HER rate of graphene-covered Ni is
comparable to that of bare Pt. A hydrogen adatom has high mobility within the confined geometry. These findings demonstrate
that graphene-covered Ni is an appealing effective, stable, and economical catalytic platform for HER.
KEYWORDS: density functional calculations, heterogeneous catalysis, hydrogen evolution reaction, confined catalysis, graphene
H
catalytic surface against sustainable reactivity; if the binding is
too weak, hydrogen will not be readily captured from its
dissolved phase by the catalyst.3
Among the various conceptual approaches for catalyst design,
confined catalysis is attracting increasing attention.4−13 In the
present work, confined catalysis refers to a situation where
confinement of the whole or parts of the catalyst, reactants, and
products in a restricted space, enhances the reactivity due to
modification of the relevant electronic states by the boundary
conditions. In particular, the discoveries of an ever-increasing
number of different classes of two-dimensional (2D) materials
offer new opportunities for optimizing confined catalysis, as
demonstrated experimentally for CO oxidation in the confined
geometry formed by a layer of graphene or hexagonal boron
nitride on Pt(111).7,8 Confined catalysis using 2D materials has
the added advantage that typically the 2D overlayer is highly
ydrogen is an ideal fuel with zero emission of pollutants
or greenhouse gases. The hydrogen evolution reaction
(HER) using metal catalysts is an efficient way to produce
hydrogen; in this reaction, protons from solution combine with
electrons at metal electrodes to form atomic H and then
combine with each other to become H2 via associative
desorption. The ability of a given metal to catalyze HER is
usually measured experimentally by the exchange current
density, i0. In the Tafel equation,1 i0 is defined as the current
at zero overpotential, which is the rate of hydrogen evolution
per surface area of the electrode when the electrode potential is
at equilibrium of the reaction. In a seminal study, it was pointed
out that the measured exchange current as a function of the
calculated hydrogen adsorption energy (ΔEH) follows a
volcano-shaped curve, peaked at the precious metal Pt.2
Furthermore, the adsorption free energy of hydrogen (ΔGH*)
was identified as a good descriptor for HER, based on the fact
that the binding between a H adatom and the metal surface
cannot be too strong or too weak for the reaction to proceed. If
the binding is too strong, the chemisorbed H will poison the
© 2016 American Chemical Society
Received: May 20, 2016
Revised: August 15, 2016
Published: September 2, 2016
6058
DOI: 10.1021/acs.nanolett.6b02052
Nano Lett. 2016, 16, 6058−6063
Letter
Nano Letters
calculations were spin polarized. We used the climbing image
nudged elastic band (CI-NEB) method24 to determine the
potential energy barriers of the hydrogen diffusion processes,
with five intermediate images constructed along each pathway,
and performed proper convergence tests to ensure that the
central findings are robust.
In order to minimize the lattice mismatch between graphene
and metals, we adjusted the lattice constant of a (2 × 2)
supercell of graphene to match that of a (2 × 2) supercell for
Ni and Cu, or a (√3 × √3) supercell for Pd, Rh, Pt, Au, and
Ag,25,26 as illustrated in Figure 1. This treatment ensures that
selective in its permeability, allowing small atoms and molecules
to permeate while blocking other undesirable species that are
likely to poison the catalyst.
In these areas and more broadly in surface chemistry, a
fundamental challenge is how to convert nonprecious metals or
materials into efficient catalysts for reactions of industrial
significance. Given the recent advances in confined catalysis, it
is particularly intriguing to explore whether the proper choice
of a 2D overlayer would be able to convert a nonprecious metal
into an effective catalyst for important chemical reactions such
as HER or CO oxidation. Toward this goal, we exploit the
concept of marginal catalysis, that is, a catalyst with moderate
reactivity becoming more reactive through controlled variations
of the reaction conditions or environments. In what concerns
HER, nonprecious metal candidates for marginal catalysts are
those located close to the peak of the volcano curve. As a
compelling example, a recent study has shown that Ti-doping
turns Al into a noble-metal-like catalyst for low-temperature
activation of molecular hydrogen.14
In this Letter, we investigate the catalytic performance of
graphene-covered metals for HER, using density functional
theory (DFT) calculations. We explore several representative
transition and noble metals, including Ni, Pd, Rh, Pt, Cu, Au,
and Ag. We first revisit the previously established volcano curve
of HER on metal (111) surfaces by including van der Waals
(vdW) corrections, which are now readily available in
computational modules. We confirm that the volcano curve is
qualitatively unchanged when we include vdW interactions, a
natural contribution to the energy, between the H adatom and
the metal substrates. Next, we calculate the adsorption free
energies of hydrogen at the interfaces between a graphene layer
and the metal substrates. We find that the graphene layer
generally weakens the H binding energy by different amounts
for different metals, thereby introducing significant variations in
the expected volcano curve. In particular, the Ni surface, which
exhibits moderate HER rate in bare form, shifts to be close to
the peak of the graphene-modified volcano curve. We also
confirm that the high mobility of H on bare Ni(111) is largely
preserved in the confined geometry, a necessary requirement in
the proposed reaction scheme. These findings show that
graphene-covered Ni is an effective, stable, and economical
catalytic platform for HER.
For the DFT calculations, we used the Vienna ab initio
simulation package (VASP)15 with the projector-augmented
wave (PAW) pseudopotentials16,17 and the generalized gradient
approximation (GGA) of Perdew−Burke−Ernzerhof (PBE)18
to treat the exchange-correlation potential of electrons. For the
vdW corrections, we used the DFT-D2 method,19,20 which has
been shown to yield a reasonable description of the graphene/
metal (Gr/M) interactions.21,22 We chose a cutoff energy of
400 eV to expand the electron wave functions in a plane-wave
basis set and a 7 × 7 × 1 Monkhorst−Pack k-point grid,23 to
achieve sufficient accuracy in the total energy. We obtained the
lattice constants of the metals and monolayer graphene via
structural optimization. We modeled the metal substrates by
slabs of five atomic layers, with the bottom two layers fixed in
their respective bulk positions and all the other atoms fully
relaxed until the forces on each atom were smaller in magnitude
than 0.01 eV/Å. We modeled the Gr/M system by placing a
monolayer of graphene on top of the metal surface, with the
lattice of graphene adjusted to match that of the metal. We used
a vacuum region of more than 16 Å to eliminate the couplings
between neighboring slabs. In the present study, all the
Figure 1. Top and side views of the H adsorption configurations on
graphene-covered metal surfaces: (a) and (c) for Ni and Cu; (b) and
(d) for Pd, Rh, Pt, Au, and Ag substrates. The (2 × 2) and (√3 ×
√3) supercell of the metal substrate is indicated by the black lines in
(a) and (b), respectively.
the hydrogen coverages per surface area are essentially
equivalent for each investigated system. The range of the
lattice mismatch between graphene and metal substrates is 0.0−
5.4%. The most stable contact geometries were obtained by
structural optimization of many different initial configurations.
For Cu(111) and Ni(111) surfaces, the carbon atoms in
graphene are located at the atop and hollow sites of the metal
substrates, whereas for the rest of the metals, the atop and
bridge sites are preferred. To find the most stable adsorption
site of H on each metal substrate in the absence or in the
presence of a graphene layer, we consider different highsymmetry positions, including the fcc hollow, hcp hollow, atop,
and bridge sites of the metal substrate. The hydrogen
adsorption energy on a bare metal is defined by2
1
ΔE H = E(metal + H) − E(metal) − E(H 2)
(1)
2
where E(metal+H) and E(metal) are the total energies of metal
plus a H adatom and bare metal, respectively, and E(H2) is the
total energy of a hydrogen molecule in the gas phase. Here, we
have taken the limit of low hydrogen coverage, with only one H
adatom in the supercell system. However, the volcano curve is
not expected to change when higher coverages are considered.2
The corresponding adsorption free energy is calculated as2
ΔG H * = ΔE H + ΔEZPE − T ΔSH = ΔE H + 0.24 eV
(2)
where T is the surface temperature, ΔEZPE is the difference in
zero point energy of hydrogen vibration between the adsorbed
state and gas phase, and ΔSH is a measure of the corresponding
6059
DOI: 10.1021/acs.nanolett.6b02052
Nano Lett. 2016, 16, 6058−6063
Letter
Nano Letters
entropy change, with ΔEZPE − TΔSH = 0.24 eV at T = 300 K, a
reasonable, well established approximation.2 When the metal
surface is covered by a graphene layer, eq 1 is modified to
1
ΔE HGr = E(Gr/metal + H) − E(Gr/metal) − E(H 2)
2
generally enhance the binding between H and metal, and the
overall volcano curve is shifted to the left in the ΔGH* scale,
with the new peak still located at the most reactive metal Pt
(ΔGH* = −0.21 eV) as established experimentally. The values
of the exchange current density i0 are from the experimental
data,2 and the fitted relationship between log i0 and ΔGH* can
now be described as
(3)
and the corresponding adsorption free energy in the presence
of the graphene overlayer is calculated as
ΔG HGr*
=
ΔE HGr
+ ΔEZPE − T ΔSH =
ΔE HGr
i0 ∝ exp(−|ΔG H * − (− 0.21 eV)| /kT )
+ 0.24 eV
(5)
where k is the Boltzmann constant and the peak is positioned at
Pt. Clearly, the classic volcano curve stays qualitatively intact
when the vdW interactions are included. In particular, the
relative locations of the different metals are essentially
unaffected.
Next, we explore the HER reactivity of each metal covered by
a layer of graphene, with the H adatom confined at the interface
of the Gr/M system. The adsorption free energies ΔGGr
H* in the
Gr/M structures are shown in Figure 2c. Generally, the
graphene overlayer weakens the adsorption free energy of
hydrogen on the metal surfaces by about 0.12−0.23 eV, with Ni
showing the largest shift to the right in the ΔGH* scale,
changing its relative location with respect to its neighbors. The
hydrogen adsorption free energies in the Gr/M systems show
pronounced differences from those in Figure 2b, strongly
indicating that the Gr/M systems will possess distinctly
different HER rates from the bare metal systems.
Experimental values of HER rates for the Gr/M systems are
not available, and we expect that the HER rates are
predominantly determined by the ΔGGr
H* values. We use the
established volcano curve shown in Figure 2b for the bare metal
systems to estimate the exchange current densities in the Gr/M
systems, with the optimal adsorption free energy of H and the
prefactor approximately unchanged. Such approximations are
reasonable, because (a) the graphene cover is unlikely to
change the fundamental qualitative nature of the HER
mechanism on the metal surface, leading to the optimal
adsorption free energy of H essentially unchanged; (b) the
HER rate depends less sensitively on the prefactor, whose
influence by graphene is even further limited due to the much
stronger interaction of H with metal than with graphene. The
results are shown in Figure 2c. The Gr/M systems with M = Ni,
Pd, and Rh are located near the peak of the volcano curve, that
is, highly reactive for HER, with comparable exchange current
densities. We expect the absolute values of the peaked HER
rates for Gr/Ni, Gr/Pd, and Gr/Rh to be comparable to that of
bare Pt, which is no longer the most reactive substrate for HER
underneath a graphene overlayer. Among these three
candidates, Ni is the only nonprecious metal, thereby fulfilling
our goal of finding an efficient and nonprecious-metal catalyst
for HER by exploiting the synergistic power of marginal
(4)
In the present study, we first calculate the hydrogen
adsorption energy on different bare metal (111) surfaces with
and without vdW interactions, and the corresponding ΔGH*,
shown in Figure 2a and b, respectively. The vdW corrections
Figure 2. Fitted volcano-shaped dependence of log(i0) in HER on the
calculated adsorption free energy of hydrogen on different catalytic
metal (111) surfaces (a) without and (b) with vdW corrections,
respectively. The volcano curve in (c) indicates the anticipated HER
rates when the H adatom is confined underneath a graphene layer. A
schematic view of the system considered is shown on the right of each
curve. It is notable that the log(i0) data in the top two subfigures are
from experimental measurement, while that in the bottom are from
theoretical prediction.
Table 1. Structural and Energetic Parameters for the Bare Metals and Gr/M Systemsa
dGr−metal w/o H (Å)
dGr−metal w/ H (Å)
ΔGGr
H* − ΔGH* (eV)
ΔEGr/M (eV)
ΔEH/M (eV)
Ni
Pd
Rh
Pt
Cu
Au
Ag
2.14
3.14
0.23
0.16
0.07
2.96
3.13
0.13
0.04
0.09
3.05
3.23
0.13
0.04
0.09
3.07
3.20
0.14
0.02
0.12
2.98
3.21
0.13
0.03
0.10
3.14
3.22
0.19
0.02
0.17
3.07
3.25
0.12
0.03
0.09
a
dGr−metal w/o H is the distance between graphene and metals without H at the interfaces; dGr−metal w/ H is the distance between graphene and
metals with H at the interfaces; ΔGGr
H* − ΔGH* is the difference between the hydrogen adsorption free energies without and with a graphene
overlayer, which can be decomposed into the changes in Gr/M interactions (ΔEGr/M) and those in H/M interactions (ΔEH/M).
6060
DOI: 10.1021/acs.nanolett.6b02052
Nano Lett. 2016, 16, 6058−6063
Letter
Nano Letters
observe that the presence of graphene changes H from the fcc
site (favored in H/Ni) to the originally unfavorable hcp site
(favored in Gr/H/Ni). Furthermore, we notice that graphene
pushes H to be closer to Ni by about 0.07 Å, a structural change
also reflected in the slightly more significant charge depletion
around H in Figure 3b. The closer contact between H and Ni
increases their mutual overlap of electron clouds, resulting in
the enhanced repulsion part of the interaction. Additionally, we
find that the nature of the weakened H−metal bond by
graphene is generic for all the metals investigated here.
The volcano curve in HER suggests that the interaction
between the H adatom and a metal catalyst has to be optimal
for achieving high reactivity, that is, this interaction cannot be
too strong or too weak. Hydrogen is quite strongly adsorbed on
the bare Ni surface with adsorption energy −0.41 eV (relative
to a H molecule in the gas phase), a value reduced to −0.18 eV
in graphene-covered Ni. As a consequence, the interplay
between graphene, hydrogen, and Ni interactions leads to the
optimization of hydrogen adsorption, rendering the composite
Gr/Ni system an effective nonprecious-metal catalyst for HER.
Here we note that within the framework of the volcano curve,
we have implicitly assumed that the Tafel mechanism is the
dominant mechanism for H2 production under the graphene
cover. This assumption is made more valid because larger
reaction intermediates favoring other production mechanisms
are more difficult to penetrate into the confined space between
graphene and metal.
In addition to the beneficial modifications that an overlayer
graphene produces on the relevant HER processes, another
significant advantage is the protection of the highly reactive
catalytic metal surfaces from being poisoned by other atomic or
molecular species present in the environment. As a related
matter, likely poisoning effects of graphene on metal catalysts
have also been considered in previous studies, due to its
chemical inertness and physical blockage.28−30 However,
structural imperfections in graphene, including island
edges,31−35 grain boundaries,36−39 and wrinkles,40 are inevitable
during growth or transfer processes. These imperfect structures
provide fast channels for the diffusion of molecules into the Gr/
M interfaces, which is a prerequisite for confined catalysis to
occur. Furthermore, a recent experimental study reported that
small atoms and molecules such as protons can easily penetrate
through the one-atom-thick graphene,41 making the Gr/M
systems permeable for potential HER processes. Our detailed
calculations also indicate that hydrogen adsorption at the
subsurface sites of Gr/Ni is energetically less stable than that at
the interface by 0.62 eV per H. Therefore, H strongly prefers to
stay at the interface rather than to diffuse into the subsurface or
bulk sites. Given these considerations, an important issue in
considering the Gr/Ni catalyst for HER is whether hydrogen
atoms can diffuse rapidly at the Gr/Ni interface. To address this
question, we calculate the diffusion barriers of a H adatom on
the bare Ni surface and at the Gr/Ni interface. The results are
shown in Figure 4, where the H adatom moves between an hcp
hollow and an fcc hollow site. The diffusion barrier is 0.15 eV
on the bare Ni surface and 0.20 eV at the Gr/Ni interface,
respectively, slightly increased by the graphene overlayer. By
closely examining the detailed structures, we observe that the
graphene overlayer pushes the H atom toward Ni by about 0.07
Å at the hcp and fcc sites while pulling it away from the surface
by about 0.04 Å at the saddle point of the diffusion. Such
variations of the H vertical positions are caused by the
confinement effect and the potential energy surface overlap
catalysis and confined catalysis. We also note that Ni will stay as
the only most reactive nonprecious metal for HER even if the
ΔGGr
H* value for the maximal rate deviates by ∼ ±0.1 eV from
the experimentally determined value of −0.21 eV for bare
metals (Figure 2b). These analyses collectively characterize the
Gr/Ni system as an appealing effective and economical catalytic
platform for HER.
In the following, we examine the underlying mechanisms for
the weakened adsorption of hydrogen in the Gr/M systems.
First, the weakened adsorption of H at the Gr/M interface is
partly due to the energy penalty caused by lifting the graphene
overlayer to establish enough space for the adsorbed H. Table 1
shows that the vertical distance between a carbon atom in the
graphene overlayer and the top layer of a given metal substrate,
dGr−metal, is increased when a H adatom is trapped at the
interface. The vertical distances are in the range of 3.13−3.25 Å
when H is present at the Gr/M interface, and the changes in
this distance from the structures without H range from 0.08 to
1.00 Å. In particular, because of the stronger attraction between
graphene and Ni,27 the distance between graphene and Ni
without H adsorption is much shorter than the other systems,
and the distance increase between graphene and Ni due to H
adsorption is much larger (1.00 Å). Accordingly, the change in
the Gr/M interaction of the Gr/Ni system (0.16 eV) is also the
largest among all the Gr/M systems considered (Table 1).
Second, the chemical binding between hydrogen and the
metal surface is also weakened by the graphene overlayer, which
can be illustrated through detailed structure and charge density
analysis. The essence of a catalytic process is that a catalyst
assists the bond breaking or formation involved in a chemical
reaction during which the changes in the electrostatic force and
sharing or transfer of electrons associated with the presence of
the catalyst play an important role. Although we find that the
interaction between graphene and H is quite weak (H is
separated from graphene by 2.29 Å) in the Gr/H/Ni system,
the graphene overlayer alters the interaction between H and the
underlying metal surface. Figure 3a shows the adsorption
Figure 3. Charge density differences of: (a) H/Ni and (b) Gr/H/Ni.
Blue and yellow isosurfaces indicate electron depletion and electron
accumulation, respectively, with the same isosurface values of 0.0025
e/bohr3 in (a) and (b). Cut-off planes across the H adatoms are added
in both figures. The numbers in the colorbar are in e/bohr3.
configuration and charge density difference of H on Ni(111),
calculated by the total charge density of H on Ni(111)
subtracting the charge density of the isolated hydrogen atom
and that of the bare Ni(111) substrate. Figure 3b shows the
corresponding structure and charge density difference of H in
Gr/Ni. In both cases, electrons are found to transfer from the
H adatom to the first-layer Ni. By comparing the most stable
adsorption structures of H with and without graphene, we
6061
DOI: 10.1021/acs.nanolett.6b02052
Nano Lett. 2016, 16, 6058−6063
Letter
Nano Letters
graphene could also be applicable by using transition metal
dichalcogenides (TMDs)42 or other 2D materials. In particular,
it has been shown that enhanced H binding can be achieved on
top of a MoS2 overlayer adsorbed on different metal substrates,
favoring enhanced HER processes.42 It would be interesting to
examine how the corresponding HER rates will change
underneath such a TMD overlayer and to investigate whether
a different nonprecious metal could be tuned to be most
reactive for HER in that case.
■
Figure 4. Minimum energy paths of H adatom diffusion on the bare
Ni surface (left) and at the Gr/Ni interface (right).
AUTHOR INFORMATION
Corresponding Authors
*E-mail: [email protected].
*E-mail: [email protected].
between graphene and Ni, and are very likely to account for the
increased H diffusion barrier in the presence of graphene. In a
recent study,35 we have shown fast diffusion of carbon atoms at
the interfaces of Gr/Cu(111) and of Gr/Ni(111), thereby
facilitating catalytic growth of the second-layer graphene
underneath the first layer. Here, we find that H diffuses even
faster than C at the Gr/Ni interface, which is reasonable given
the smaller atomic radius of H atom. The state with two
individually adsorbed H atoms is energetically more stable than
H2 underneath the graphene overlayer. The fast diffusion of H
thus enables a continuous H supply around the structurally
imperfect sites of graphene, where the H−H coupling and
subsequent H2 desorption occurs more efficiently. Collectively,
H ions are able to penetrate into the confined region, and the
activated H atoms underneath graphene are also able to diffuse
fast to the active sites for rapid H2 formation and desorption.
Because the graphene overlayer is impermeable to H 2
molecules, the structural imperfections that exist inevitably
and usually abundantly in graphene are the active sites for H2
desorption.
In the field of surface catalysis, novel catalysts are structurally
designed and modified from their elemental materials in many
different ways, such as alloying and coating, to enhance their
catalytic performance for desirable reactions. In most situations,
the reaction mechanisms need to be unaffected upon the
structural change of the elemental catalyst, which means that
the tuning capacity of catalysis is generally quite limited.
Therefore, in the design of new catalysts, the elemental
materials should be selected among those showing at least some
noticeable (or marginal) catalysis. As demonstrated in our
present case study, because the reaction rate depends
exponentially on the adsorption energy, a relatively moderate
tuning of the adsorption strength by the graphene overlayer is
able to turn Ni (with marginal catalysis) into a highly reactive
catalyst for HER.
To conclude, our systematic study has identified the
graphene-covered Ni surface as an effective nonpreciousmetal catalyst for HER. The graphene overlayer weakens the
adsorption free energy of a H adatom on the metal surface,
shifting Ni to be close to the peak of the volcano curve for
HER. We demonstrated that a H adatom has a very small
diffusion barrier at the interface of the Gr/Ni system. We
therefore predict that the confined space between the Ni(111)
surface and graphene overlayer will have high catalytic
performance for HER. The present study introduces an
appealing new catalytic system to produce hydrogen, which
we expect to stimulate experimental validations and promote
related technological developments in green energy. The
approach to tune the reactivity of metal catalysts using
Author Contributions
Y.Z. and W.C. contributed equally to this work.
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS
This work was supported in part by the NSFC (Grant Nos.
61434002, 11634011, 11374273, and 11504357), and the
NKBRPC (Grant No. 2014CB921103). W.C. and E.K. are also
partially supported by the Integrated Mesoscale Architectures
for Sustainable Catalysis, an Energy Frontier Research Center
funded by the U.S. Department of Energy, Office of Science,
Basic Energy Sciences under Award # DE-SC0012573.
■
REFERENCES
(1) Petrii, O. A.; Nazmutdinov, R. R.; Bronshtein, M. D.; Tsirlina, G.
A. Electrochim. Acta 2007, 52, 3493−3504.
(2) Nørskov, J. K.; Bligaard, T.; Logadottir, A.; Kitchin, J. R.; Chen, J.
G.; Pandelov, S.; Stimming, U. J. Electrochem. Soc. 2005, 152, J23−J26.
(3) Nørskov, J. K.; Bligaard, T.; Rossmeisl, J.; Christensen, C. H. Nat.
Chem. 2009, 1, 37−46.
(4) Pan, X.; Fan, Z.; Chen, W.; Ding, Y.; Luo, H.; Bao, X. Nat. Mater.
2007, 6, 507−511.
(5) Pan, X.; Bao, X. Acc. Chem. Res. 2011, 44, 553−562.
(6) Sanlés-Sobrido, M.; Pérez-Lorenzo, M.; Rodríguez-González, B.;
Salgueiriño, V.; Correa-Duarte, M. A. Angew. Chem., Int. Ed. 2012, 51,
3877−3882.
(7) Yao, Y.; Fu, Q.; Zhang, Y. Y.; Weng, X.; Li, H.; Chen, M.; Jin, L.;
Dong, A.; Mu, R.; Jiang, P.; Liu, L.; Bluhm, H.; Liu, Z.; Zhang, S. B.;
Bao, X. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 17023−17028.
(8) Zhang, Y.; Weng, X.; Li, H.; Li, H.; Wei, M.; Xiao, J.; Liu, Z.;
Chen, M.; Fu, Q.; Bao, X. Nano Lett. 2015, 15, 3616−3623.
(9) Jiao, F.; Li, J.; Pan, X.; Xiao, J.; Li, H.; Ma, H.; Wei, M.; Pan, Y.;
Zhou, Z.; Li, M.; Miao, S.; Li, J.; Zhu, Y.; Xiao, D.; He, T.; Yang, J.; Qi,
F.; Fu, Q.; Bao, X. Science 2016, 351, 1065−1068.
(10) Andersen, M.; Hornekær, L.; Hammer, B. Phys. Rev. B: Condens.
Matter Mater. Phys. 2014, 90, 155428.
(11) Ferrighi, L.; Di Valentin, C. Surf. Sci. 2015, 634, 68−75.
(12) Doyle, A. D.; Montoya, J. H.; Vojvodic, A. ChemCatChem 2015,
7, 738−742.
(13) Ferrighi, L.; Datteo, M.; Fazio, G.; Di Valentin, C. J. Am. Chem.
Soc. 2016, 138, 7365−7376.
(14) Chopra, I. S.; Chaudhuri, S.; Veyan, J. F.; Chabal, Y. J. Nat.
Mater. 2011, 10, 884−889.
(15) Kresse, G.; Furthmüller, J. Phys. Rev. B: Condens. Matter Mater.
Phys. 1996, 54, 11169−11186.
(16) Blöchl, P. E. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50,
17953−17979.
(17) Kresse, G.; Joubert, D. Phys. Rev. B: Condens. Matter Mater. Phys.
1999, 59, 1758−1775.
6062
DOI: 10.1021/acs.nanolett.6b02052
Nano Lett. 2016, 16, 6058−6063
Letter
Nano Letters
(18) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77,
3865−3868.
(19) Grimme, S. J. Comput. Chem. 2006, 27, 1787−1799.
(20) Bučko, T.; Hafner, J.; Lebègue, S.; Á ngyán, J. G. J. Phys. Chem. A
2010, 114, 11814−11824.
(21) Sławińska, J.; Dabrowski, P.; Zasada, I. Phys. Rev. B: Condens.
Matter Mater. Phys. 2011, 83, 245429.
(22) Voloshina, E.; Dedkov, Y. Phys. Chem. Chem. Phys. 2012, 14,
13502−13514.
(23) Methfessel, M.; Paxton, A. T. Phys. Rev. B: Condens. Matter
Mater. Phys. 1989, 40, 3616−3621.
(24) Henkelman, G.; Uberuaga, B. P.; Jónsson, H. J. Chem. Phys.
2000, 113, 9901−9904.
(25) Hamada, I.; Otani, M. Phys. Rev. B: Condens. Matter Mater. Phys.
2010, 82, 153412.
(26) Vanin, M.; Mortensen, J. J.; Kelkkanen, A. K.; Garcia-Lastra, J.
M.; Thygesen, K. S.; Jacobsen, K. W. Phys. Rev. B: Condens. Matter
Mater. Phys. 2010, 81, 081408.
(27) Chen, H.; Zhu, W. G.; Zhang, Z. Y. Phys. Rev. Lett. 2010, 104,
186101.
(28) Wolf, E. E.; Alfani, F. Catal. Rev.: Sci. Eng. 1982, 24, 329−371.
(29) Bartholomew, C. H. Appl. Catal. A: Gen. 2001, 212, 17−60.
(30) Peng, Z.; Somodi, F.; Helveg, S.; Kisielowski, C.; Specht, P.;
Bell, A. T. J. Catal. 2012, 286, 22−29.
(31) Sutter, P.; Sadowski, J. T.; Sutter, E. A. J. Am. Chem. Soc. 2010,
132, 8175−8179.
(32) Mu, R. T.; Fu, Q.; Jin, L.; Yu, L.; Fang, G. Z.; Tan, D. L.; Bao, X.
H. Angew. Chem., Int. Ed. 2012, 51, 4856−4859.
(33) Jin, L.; Fu, Q.; Yang, Y.; Bao, X. Surf. Sci. 2013, 617, 81−86.
(34) Zhang, X.; Wang, L.; Xin, J.; Yakobson, B. I.; Ding, F. J. Am.
Chem. Soc. 2014, 136, 3040−3047.
(35) Chen, W.; Cui, P.; Zhu, W. G.; Kaxiras, E.; Gao, Y. F.; Zhang, Z.
Y. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 91, 045408.
(36) Duong, D. L.; Han, G. H.; Lee, S. M.; Gunes, F.; Kim, E. S.;
Kim, S. T.; Kim, H.; Ta, Q. H.; So, K. P.; Yoon, S. J.; Chae, S. J.; Jo, Y.
W.; Park, M. H.; Chae, S. H.; Lim, S. C.; Choi, J. Y.; Lee, Y. H. Nature
2012, 490, 235−239.
(37) Hsieh, Y.-P.; Hofmann, M.; Chang, K.-W.; Jhu, J. G.; Li, Y.-Y.;
Chen, K. Y.; Yang, C. C.; Chang, W.-S.; Chen, L.-C. ACS Nano 2014,
8, 443−448.
(38) Liu, L.; Siegel, D. A.; Chen, W.; Liu, P.; Guo, J.; Duscher, G.;
Zhao, C.; Wang, H.; Wang, W.; Bai, X.; McCarty, K. F.; Zhang, Z. Y.;
Gu, G. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 16670−16675.
(39) Chen, W.; Chen, H.; Lan, H. P.; Cui, P.; Schulze, T. P.; Zhu, W.
G.; Zhang, Z. Y. Phys. Rev. Lett. 2012, 109, 265507.
(40) Zhang, Y.; Fu, Q.; Cui, Y.; Mu, R.; Jin, L.; Bao, X. Phys. Chem.
Chem. Phys. 2013, 15, 19042−19048.
(41) Hu, S.; Lozada-Hidalgo, M.; Wang, F. C.; Mishchenko, A.;
Schedin, F.; Nair, R. R.; Hill, E. W.; Boukhvalov, D. W.; Katsnelson,
M. I.; Dryfe, R. A. W.; Grigorieva, I. V.; Wu, H. A.; Geim, A. K. Nature
2014, 516, 227−230.
(42) Chen, W.; Santos, E. J. G.; Zhu, W. G.; Kaxiras, E.; Zhang, Z. Y.
Nano Lett. 2013, 13, 509−514.
6063
DOI: 10.1021/acs.nanolett.6b02052
Nano Lett. 2016, 16, 6058−6063