High-resolution in situ and ex situ TEM studies on graphene formation and growth on Pt nanoparticles

Article abstract of DOI:10.1016/j.jcat.2011.10.008

The formation of graphene layers on MgO-supported Pt nanoparticles was studied by both in situ and ex situ high-resolution transmission electron microscopy (HRTEM). The HRTEM images indicate that graphene sheets grow from steps in the surface of Pt nanoparticles. The subsequent morphology of the graphene sheets is a strong function of Pt particle size. For particles less than ～6 nm in diameter, the graphene sheets form nanotubes or move from the surface of Pt particles and accumulate on the MgO support. Complete particle envelopment by multiple graphene layers was only observed for particle greater than ～6 nm in diameter. The observed dependence of graphene morphology on Pt nanoparticle size and shape is associated with the strain energy generated between graphene layers during their growth and the overall free energy of the graphene-Pt system.

Full text of DOI:10.1016/j.jcat.2011.10.008

Journal of Catalysis 286 (2012) 22–29
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
High-resolution in situ and ex situ TEM studies on graphene formation
and growth on Pt nanoparticles
Zhenmeng Peng a, Ferenc Somodi a, Stig Helveg d, Christian Kisielowski b, Petra Specht c, Alexis T. Bell a,
⇑
a Department of Chemical and Biomolecular Engineering, University of California, Berkeley, CA 94720, United States
b Joint Center for Artificial Photosynthesis, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720, United States
c Department of Materials Science and Engineering, University of California, Berkeley, CA 94720, United States
d Haldor Topsøe A/S, Nymøllevej 55, DK-2800 Kgs. Lyngby, Denmark
a r t i c l e i n f o
a b s t r a c t
Article history:
The formation of graphene layers on MgO-supported Pt nanoparticles was studied by both in situ and ex
situ high-resolution transmission electron microscopy (HRTEM). The HRTEM images indicate that graph-
ene sheets grow from steps in the surface of Pt nanoparticles. The subsequent morphology of the graph-
ene sheets is a strong function of Pt particle size. For particles less than ꢀ6 nm in diameter, the graphene
sheets form nanotubes or move from the surface of Pt particles and accumulate on the MgO support.
Complete particle envelopment by multiple graphene layers was only observed for particle greater than
ꢀ6 nm in diameter. The observed dependence of graphene morphology on Pt nanoparticle size and shape
is associated with the strain energy generated between graphene layers during their growth and the
overall free energy of the graphene-Pt system.
Received 6 August 2011
Revised 11 October 2011
Accepted 13 October 2011
Available online 21 November 2011
Keywords:
Platinum
Nanoparticles
Graphene
In situ experiment
High-resolution electron microscopy
Ó 2011 Elsevier Inc. All rights reserved.
1. Introduction
dehydrogenation and C–C bond cleavage of hydrocarbons on the
metal surfaces. The nascent carbon atoms may then diffuse from
Supported platinum nanoparticles are widely used to catalyze
the transformation of alkanes via reactions such as dehydrogena-
tion [1–4], dehydroaromatization [5,6], isomerization [7,8], and
hydrogenolysis [9,10]. A by-product of these reactions is the depo-
sition of coke, which can lead to changes in product distribution
and, more importantly, to catalyst deactivation [11,12]. For these
reasons, there has been an ongoing interest in finding ways to sta-
bilize the activity of supported Pt nanoparticles and to understand
the processes by which coke forms and accumulates on the surface
of such particles. Good progress toward the first of these objectives
has been achieved by alloying Pt with elements such as Sn, Ga, and
In [3,4,13]. However, relatively little is known about the mecha-
nism of carbon formation on Pt.
Transmission electron microscopy (TEM) has proven to be an
invaluable tool in understanding the deposition of carbon on 3d
transition metals (Ni, Fe, and Co) catalysts [14–22] and has been
used to make deductions about the elementary processes involved.
Notwithstanding the valuable information obtained from these ef-
forts, the mechanism of carbon formation and growth continues to
be the subject of intense discussion. In general, carbon atoms can
be produced by the adsorption and subsequent stepwise
the adsorption sites to the centers for graphene growth among
one of two routes. The first route involves dissolution of carbon
into the bulk of the metal nanoparticles. Once the metal is satu-
rated, carbon segregates to the metal surface where the formation
of graphene initiates. The carbon formation on these metals is thus
controlled by bulk solubility and diffusion of carbon atoms, as evi-
denced by change in lattice parameter, the formation of carbides at
elevated temperatures, and kinetic studies [19–23]. The alternative
perspective is that the transport of carbon is mediated by surface
or sub-surface diffusion on the metal nanoparticles so the surface
properties alone are sufficient to describe the graphene growth
[15,24–26]. As concerns the growth centers, recent high-resolution
transmission electron microscopy (HRTEM) studies indicate that
these are associated with step sites in the surface of Ni and Fe/Co
nanoparticles [15,17–19]. The preference for the step-mediated
growth by Ni has been explained by the site-dependent adsorption
energies of carbon obtained by density functional theory (DFT)
[15,27].
In contrast to Ni, Fe, and Co, only a small number of TEM studies
on carbon formation on Pt have been reported [28–31]. Previous
work suggests that the dissolution and bulk diffusion mechanism
of carbon deposition does not apply to Pt because of the very low
solubility of carbon in Pt [32], and it has been proposed, therefore,
that carbon deposition on Pt proceeds via a surface diffusion
⇑
Corresponding author. Fax: +1 510 642 4778.
E-mail address: bell@cchem.berkeley.edu (A.T. Bell).
0021-9517/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2011.10.008

Z. Peng et al. / Journal of Catalysis 286 (2012) 22–29
23
mechanism [26,33]. However, to the best of our knowledge, exper-
imental evidence to support this transport mechanism as well as
the surface processes involved in the assembly of graphene has
not been presented.
differentially pumped environmental cell. The microscope was
operated in TEM mode at an acceleration voltage of 300 kV, with
an activated spherical aberration corrector and with an electron
beam dose rate in the range of 104–106 eꢁ/nm2 s incident on the
CCD camera. Specimens were prepared by crushing and dispersing
dry catalyst powders onto a plasma-cleaned stainless steel grid.
The grid was mounted in a Gatan 628 single-tilt heading holder
with an isotropic information transfer out to at least 0.14 nm.
The specimens were first reduced by heating the sample in 1 mbar
H2 to 500 °C for 30–60 min, after which the sample was cooled to
475 °C. The deposition of graphene was initiated by the removal of
hydrogen and introduction of 1.3 mbar isobutene. TEM images and
time-lapsed TEM image series were acquired in situ of samples
during the exposure to isobutene once the sample drift had leveled
The aim of the present study was to investigate the formation of
graphene layers on Pt nanoparticles supported on cubic magne-
sium oxide (MgO). Both in situ and ex situ high-resolution TEM
(HRTEM) were used for this purpose. In situ characterization of
the working catalyst was undertaken in order to obtain direct
observation of the formation and growth of graphene sheets under
the reaction condition. These efforts were complemented by ex
situ characterization of carbon deposits using HRTEM to obtain
additional information about the structure of the deposited carbon.
Using this approach, a systematic investigation was undertaken of
the effects of Pt particle size and shape on the deposition of
graphene.
out, usually in less than 10 min.
A low electron dose of
1.5 ꢂ 105 eꢁ/nm2 s was used for in situ observations of carbon
deposition in order to minimize the effects of the electron beam
on the process of carbon generation and the structure of the carbon
formed. For the single TEM images, the exposure time was 0.5 s.
For the time series, each image was taken over a period of 0.1 s.
The image series are represented with an improved signal-to-noise
ratio by adding five consecutive images after making accurate
alignments and appropriate indexing of the images.
2. Experimental
2.1. Synthesis of Pt nanoparticles
Platinum nanoparticles with average diameters of 3.5 and
6.1 nm were prepared by reduction of platinum acetylacetonate
(Pt(acac)2) dissolved in octyl ether (OE) by 1,2-hexadecanediol
(HDD) in the presence of oleylamine (OAm) and oleic acid (OA).
All experiments were conducted under Ar using a standard Schlenk
line. To produce 3.5 nm Pt particles, platinum acetylacetonate
(Pt(acac)2, 97%, Aldrich, 0.05 g or 0.125 mmol) was dissolved in a
mixture of OE (99%, Aldrich, 2 mL), HDD (90%, Aldrich, 0.2 g or
0.77 mmol), OAm (70%, Aldrich, 0.2 mL), and OA (99%, Aldrich,
0.2 mL) in a 25-mL three-neck flask. The resulting solution was
heated to 290 °C at a rate of 5 °C/min and maintained at this tem-
perature for 10 min before cooling down to ambient room temper-
atures. To make 6.1-nm Pt nanoparticles, Pt(acac)2 (0.05 g or
0.125 mmol) dissolved in OE (1 mL) was injected into the organic
mixture after it was heated to 290 °C [34]. A large amount of anhy-
drous ethanol was added to the reaction mixture to aid in the sep-
aration of Pt nanoparticles by centrifugation at 3000 rpm for 5 min.
The solid material produced in this manner was dispersed in 3 mL
of anhydrous toluene, forming a stable colloidal solution.
2.4. Deposition of carbon for ex situ characterization
In a typical procedure, 25 mg of catalyst was loaded into a
quartz tube reactor and heated to 600 °C at 15 °C/min in 20% H2
in He using a three-zone furnace at total pressure of an atmo-
sphere. The catalyst was maintained at this temperature for 1 h be-
fore being exposed to a mixture of ethane (C2H6), H2, and He. The
ratio of the volumetric flow rates of C2H6 and H2 was fixed at
1:1.25, and total flow rate was maintained at 60 cm3/min. The
reaction was terminated after a fixed period of time by switching
the reacting gases to pure He and cooling the sample down to room
temperature. The samples were then removed from the reactor and
stored in a vacuum desiccator.
2.5. Ex situ characterization by HRTEM, EELS, and Raman spectroscopy
Transmission electron microscopy (TEM) images were taken
using a FEI Tecnai 12 microscope with an accelerating voltage of
120 kV. High-resolution TEM (HRTEM) characterizations were con-
ducted on the TEAM 0.5 high-resolution microscope operated at
80 kV at the National Center for Electron Microscopy (NCEM)
[36]. This instrument is a modified FEI Titan 80-300 microscope
equipped with a special high-brightness Schottky field emission
electron source, a gun monochromator, a high-resolution GIF Tri-
diem energy filter, and two CEOS (correlated electron optical sys-
tems) hexapole-type spherical aberration correctors. Scanning
transmission electron microscopy (STEM) and electron energy loss
spectroscopy (EELS) data were collected on a FEI Titan S80-300
microscope operated at 300 kV and under the high-angle annular
dark field (HAADF) mode. Raman spectra were recorded using a
Kaiser Optical HoloLab series 5000 Raman spectrometer equipped
with a Nd:YAG laser source.
2.2. Preparation of MgO-supported Pt catalysts
MgO (<50 nm, Aldrich) was calcined in air at 1060 °C for 30 min
in order to form MgO nanocubes [35]. A suspension of 0.2 g of MgO
in 10 mL of anhydrous toluene was stirred under an inert atmo-
sphere, after which an appropriate volume of the Pt suspension
(containing 0.01 g Pt) was added. The mixture was stirred over-
night, and the resulting product was precipitated out by adding
three times the volume anhydrous ethanol. MgO-supported Pt
was then separated by centrifugation. The solid material was
heated at 300 °C for 1 h in air to remove the surface capping agents
and then reduced in an atmosphere of H2 in Ar (10 vol.%) at 600 °C
for 1 h.
A sample of 1.4-nm-diameter Pt nanoparticles supported on
MgO was prepared by incipient wetness impregnation. Pt(acac)2
(0.02 g or 0.05 mmol) was dissolved in anhydrous toluene (1 mL)
and added drop wisely onto the calcined MgO support (1 g) under
continuous stirring. The solid was dried under vacuum and then
reduced in H2/Ar (10 vol.%) by raising the temperature to 600 °C
at 5 °C/min and maintaining at 600 °C for 1 h.
3. Results and discussion
Fig. 1 shows TEM images of the calcined MgO powders used to
support Pt. Calcination at 1060 °C produced almost perfect cubes
about 30 nm on edge. The lattice fringes are parallel to the surface
of each cube and have a spacing of 2.11 Å, indicating that the MgO
cubes are bounded by (200) planes (PDF 87-0652). Fig. 2a shows a
TEM image of the as-synthesized Pt nanoparticles, and Fig. 2b
2.3. In situ HRTEM
In situ TEM experiments were carried out using an image aber-
ration-corrected Titan 80-300 ETEM microscope equipped with a

24
Z. Peng et al. / Journal of Catalysis 286 (2012) 22–29
Fig. 1. (a) TEM and (b) HRTEM images of the cubic MgO support.
Fig. 3. In situ TEM images of supported Pt nanoparticles at (a) low and (b) high
magnifications reduced in 1 mbar H2 at 500 °C, taken with an electron dose of: (a)
5.0 ꢂ 104 and (b) 2.5 ꢂ 106 eꢁ/nm2 s.
shape of the Pt nanoparticles. As in Fig. 2d, Fig. 3b shows clear evi-
dence for (111) lattice fringes.
Carbon formation on Pt under the microscope conditions was
studied by switching the composition of the gas fed to the sample
holder from H2 to a hydrocarbon. Although most light hydrocarbons
are thought to form carbon on metal surfaces via a similar mecha-
nism [14], the growth conditions can be quite different reflecting
differences in the carbon chemical potential,
lC, of the gaseous
hydrocarbon. Attempts to generate a carbonaceous deposit in the
microscope using 1.1 mbar ethane were unsuccessful at tempera-
tures up to 575 °C, in contrast to ex situ experiments in which eth-
ane was used to generate carbon deposits [38]. Equally
unsuccessful attempts were made using 1.3 mbar ethene, a princi-
pal source of carbonaceous deposits used for the growth of carbon
nanostructures on 3d transition metals [39]. The findings indicate
that the chemical potential is too low for graphene formation at
the pressure in the mbar range in the microscope. However, previ-
ous work shows that isobutene readily produced carbonaceous
deposits on Pt under mild reaction conditions [40]. Accordingly, car-
bon deposition was indeed observed when 1.3 mbar isobutene was
introduced in the microscope with the sample heated to 475 °C.
Fig. 4 presents a series of co-added images taken during the
deposition of carbon on Pt/MgO. The first image (Fig. 4a) was re-
corded after about 10 min of exposure of the Pt particle to isobu-
tene, whereas the next three images (Fig. 4b–d) were taken 1 s,
1.2 min, and 3.2 min after the first image. Details on the initial car-
bonaceous layers may be seen in Fig. S1. While evidence for the lat-
tice fringes of Pt can be observed in Fig. 4, they are not as clearly
evident as those seen in Fig. 3b. The reason for this is the greater
motion of the sample occurring after the change over from H2 to
isobutene, coupled with lower signal-to-noise ratio of the heavier
molecules and the slight blurring of the fringes upon the addition
of sequential images taken with low electron beam exposure.
The deposition of carbonaceous layers is clearly evidenced by the
appearance of multiple layers growing outward from the surface
of Pt particle. Fig. 4a suggests that growth of the carbonaceous lay-
ers initiates from the particle surface, most likely at the Pt steps, as
indicated by the red arrows, and terminates either on the Pt sur-
face or on the support, as indicated by the blue arrow. No changes
in particle size or morphology were observed during carbon depo-
sition, indicating that the multiple carbon layers formed without
strongly affecting the compact and facetted shape of the Pt nano-
particles. The measured spacing of the graphene layers was around
3.5 Å and indexed to the (002) plane of graphite. This interlayer
spacing is larger than that for well-crystallized graphitic carbon,
3.35 Å, (PDF 13-0148), suggesting that the carbonaceous deposit
is turbostratic [14]. The somewhat higher interlayer spacing for
the carbonaceous deposit formed in situ may also be due to graph-
ene layers not being fully devoid of hydrogen.
Fig. 2. (a) TEM and (b) size distribution of as-synthesized Pt nanoparticles and (c
and d) TEM and HRTEM images of the supported Pt nanoparticles.
shows the distribution of Pt particle sizes. Most of the particles are
uniform in size and faceted in shape. Statistical analysis of particle
size distribution gives an average diameter of 6.1 0.8 nm. An im-
age of Pt nanoparticles supported on MgO is presented in Fig. 2c. It
is evident that sintering of the Pt particles did not occur during cal-
cination to remove the capping agents. The absence of sintering is
attributed to the interactions of the Pt particles with the support
[37]. The cubic shape of the MgO support and the low Z numbers
of the Mg and O enable clear images of the Pt nanoparticles to be
acquired, many of which can be observed in profile at the edges
of the MgO cubes. A high-resolution image of a representative Pt
particle is shown in Fig. 2d. Lattice fringes with a periodic spacing
of 2.27 Å corresponding to the Pt(111) plane are clearly visible
(PDF 70-2057). The observation of a straight surface with atomic
resolution suggests that the organic molecules present on the par-
ticle surface after synthesis have been removed without altering
the surface morphology of the particle.
In situ HRTEM observations of the Pt/MgO (dPt = 6.1 nm) cata-
lyst were initiated by reducing the catalyst at 500 °C for 30 min
in 1 mbar H2. Fig. 3 shows representative TEM images of the sup-
ported Pt particles under the H2 atmosphere. The image of the
sample taken in situ (Fig. 3a) is very similar in appearance to that
taken ex situ (Fig. 2a). Faceted particles with distinct edges were
observed, suggesting a minimal effect of in situ reduction in the

Z. Peng et al. / Journal of Catalysis 286 (2012) 22–29
25
Fig. 6 shows representative ex situ HRTEM images of Pt particles
observed after Pt/MgO (dPt = 6.1 nm) had been used for the dehy-
drogenation of ethane for 1, 2, 10, and 60 min. All Pt nanoparticles
about 6 nm in diameter and larger were completely covered by
layers of graphene, the number of which increased with increasing
time of exposure to ethane dehydrogenation. The spacing between
these layers is around 3.4 Å, consistent with that of graphite. After
1 min of exposure to the reaction mixture (Fig. 6a), two to three
graphene layers were formed. The number of graphene layers grew
with further time under dehydrogenation conditions, but the rate
of growth between 10 and 60 min was clearly slower than upon
initial exposure of the catalyst to reaction conditions. The decrease
in the rate of carbon deposition is very likely due to the inhibition
of atomic carbon formation on the surface of Pt nanoparticles by
the accumulating layers of graphene.
Raman and EELS spectroscopy were used to confirm the attribu-
tion of the deposits seen in Fig. 6 to graphene. Raman spectra of Pt/
MgO (dPt = 6.1 nm) taken after 1 min and 60 min of use for ethane
dehydrogenation are shown in Fig. 7. Two broad peaks at around
1600 and 1350 cmꢁ1 were observed after 1 min of reaction. The
first of these features corresponds to the G band associated with
the longitudinal optical (LO) phonon mode of graphite, and the sec-
ond feature corresponds to the D band due to the structural disor-
der and the presence of carbon edges [42–44]. Both peaks
increased in intensity and became better defined after 60 min of
reaction, indicating a greater amount of deposited carbon. The
broader shape of the G band compared to that observed for crystal-
line graphite suggests that defects are present among the graphene
layers deposited during ethane dehydrogenation [43]. The pres-
ence of both D and G bands with comparable intensities is indica-
tive of structural disorder in the graphene layers [44]. The slight
downscale shift of the D and G bands to 1593 and 1341 cmꢁ1 with
increasing exposure to reaction conditions can be attributed to an
increase in the number of graphene layers [43]. EELS spectra col-
lected on individual Pt particles provide further evidence for the
deposition of carbon. Fig. 8 shows the EELS K-edge region of
Fig. 4. Time-lapsed HRTEM images of the growing graphene layers on Pt recorded
in situ under exposure of 1.3 mbar iso-butene at 475 °C (a: 0 s; b: 1 s; c: 1.2 min;
and d: 3.2 min), taken with an electron dose of 1.5 ꢂ 105 eꢁ/nm2 s.
The sample observed in situ under exposure to isobutene was
characterized ex situ using the TEAM 0.5 microscope. A represen-
tative image is shown in Fig. 5. It should be noted that particle
shown is one that had not been exposed to the effects of the elec-
tron beam during in situ studies. The carbonaceous layers seen in
Fig. 5 are somewhat better ordered than those seen in Fig. 4, and
the number of layers is smaller than the number seen in Fig. 4d.
These differences suggest that carbon growth kinetics observed
in situ was influenced by the presence of the electron beam and
that further optimization of the beam dosage is needed to extract
quantitative information from such time series [41].
Fig. 6. HRTEM images of the graphene layers grown on Pt nanoparticles after (a) 1,
(b) 2, (c) 10, and (d) 60 min of reaction at 600 °C in the forming gas,
VC2H6:VH2:VHe = 12:15:33 mL/min, taken with an electron dose of around
3 ꢂ 106 eꢁ/nm2 s.
Fig. 5. Ex situ HRTEM image of the grown graphene layers on Pt after being exposed
under 1.3 mbar iso-butene for 60 min at 475 °C in the in situ experiment, taken
with an electron dose of 7.8 ꢂ 104 eꢁ/nm2 s.

26
Z. Peng et al. / Journal of Catalysis 286 (2012) 22–29
Fig. 9. HRTEM images of the grown carbon on the shaped Pt particles at the (a)
early and (b) late stages, VC2H6:VH2:VHe = 12:15:33 mL/min, T = 600 °C, taken with
an electron dose of around 3 ꢂ 106 eꢁ/nm2 s.
environment. On such particles, carbon nanotubes (CNTs), rather
than graphene shells, were observed. Fig. 9a shows a HRTEM image
of one CNT in the early stage, which appears to grow from the Pt
steps via a base-growth mechanism [18]. Notably, though, the tip
of this Pt particle remained free of carbon throughout the growth
process (see Fig. 9b). We believe that detachment of the graphene
layers from the tip of the Pt particle is due to the high strain in the
layers imposed by the increasing surface curvature as the tip of the
particle is approached.
Fig. 7. Raman spectra of the worked catalyst after (a) 1 and (b) 60 min of reaction at
600 °C in the forming gas, VC2H6:VH2:VHe = 12:15:33 mL/min.
To further address the role of strain on the growth of graphene
layers, the effect of the Pt particle size on the growth of graphene
layers was examined. Fig. 10 shows images illustrating the pat-
terns of graphene layer growth on Pt nanoparticle with diameters
of around 4 and 2 nm taken after 1, 2, and 60 min of time on
stream. Carbon deposition on the 4-nm particles resulted in a rel-
ative small number of graphene layers, which with time formed a
capped nanotube and then grew away from the Pt particle (see
Fig. 10a–c). In contrast, on the 2-nm particles, the newly formed
graphene layers appear to immediately slough off onto the MgO
support. Thus, the pattern of graphene layer growth on small Pt
particles is distinctly different from that seen on larger (>6 nm)
Pt particles, which build up an increasing number of graphene lay-
ers that envelope the particle with increasing time under reaction
conditions. The different patterns of the graphene growth on Pt
may help explain the observed effects of particle size on the activ-
ity, selectivity, and degree of coking of supported Pt for a number
of reactions [12,48,49].
Fig. 8. EELS spectrum of one single Pt particle after 60 min of reaction at 600 °C in
the forming gas, VC2H6:VH2:VHe = 12:15:33 mL/min, STEM image of the Pt particle
shown in the inset.
carbon. The peak centered at around 297 eV corresponds to the
transition from the 1s core level to the r⁄ band and the small
shoulder peak at around 288 eV can be assigned to the 1s to the
p⁄ transition [45]. Both of these features are characteristic of
graphene layers.
The results of this investigation clearly demonstrate that the
formation of graphene on the surface of Pt nanoparticles initiates
at steps in the particle surface and that the extent of graphene for-
mation under identical reaction conditions is particle size depen-
dent. For particles larger than ꢀ6 nm, particles become
completely enveloped in multiple layers of graphene; for particles
between 6 and 2 nm, graphene nanotubules are formed, and for
particles ꢃ2 nm, graphene layers formed on the particle surface
are transported to the support. It is useful to examine these obser-
vations in light of a simple model for graphene growth at step sites
on transition metal surfaces [33]. The authors of this study show
based upon DFT calculations of the adsorption energies for carbon
atoms that carbon atoms adsorbed on terrace surfaces are unstable
to those adsorbed at step edges. For metals such as Ni and Co, the
carbon adsorption energy is further decreased upon incorporation
of the atom into a finite graphene layer adsorbed at the step edge.
But for metals such as Ru and Rh, the transfer of a carbon atom to
such a finite graphene layer is unfavorable. The origin of this differ-
ence in behavior is the strain energy, Estrain, resulting from epitaxial
lattice mismatch between the metal step edge and graphene. This
mismatch is minimal for Ni and Co but becomes substantial for Ru
and Rh and is even higher for Pt. A thermodynamic model for
As noted in the discussion of Fig. 4, the formation of carbona-
ceous sheets appears to initiate at steps in the surface of Pt nano-
particles. Further evidence for a step-mediated growth of graphene
on Pt is provided by Fig. S2 where it is seen that the first several
layers of graphene bind preferentially to Pt steps. The finding that
graphene grows out from Pt steps is similar to the observation of
graphene growth from methane or acethylene decomposition on
Ni nanoparticles [14,15,17,18]. In contrast to what was observed
in those studies, the Pt particles retain their compact and facetted
shape as the layers of carbon accumulate in the present experi-
ments. However, surface restructuring was observed on some of
the Pt nanoparticles as evidenced by the disappearance of clear fac-
ets and edges. These changes are probably associated with a de-
crease in the melting point of surface Pt and its interaction with
the carbonaceous species [46,47]. Moreover, a minor fraction of
the Pt particles were irregularly shaped either as a result of the
synthesis or the interaction with the alkene dehydrogenation

Z. Peng et al. / Journal of Catalysis 286 (2012) 22–29
27
Fig. 10. HRTEM images of the grown carbon on (a–c) 3.5 and (e and f) 1.4 nm Pt nanoparticles after (a and d) 1, (b and e) 2, and (c and f) 60 min of reaction at 600 °C in the
forming gas, VC2H6:VH2:VHe = 12:15:33 mL/min, taken with an electron dose of around 3 ꢂ 106 eꢁ/nm2 s.
graphene growth is then developed for the purpose of identifying
the critical number of carbon atoms that must be accumulated into
a nascent graphene island in order to nucleate the continued
growth of the sheet. The resulting expression is
and wrinkles. Consistent with these findings, the graphene-
Pt(111) separation and the separation of two graphene sheets
are found to be ꢄ3.30 Å, nearly identical to the interlayer spacing
of graphite, 3.36 Å [51]. On the basis of this evidence, it is reason-
able to conclude that Estretch would be considerably smaller than
0.55 eV. If it assumed that Estretch = 0.30 eV, the value of Ir, a metal
that also exhibits a weak graphene–metal interaction [52], then NC
for Pt is estimated to be 21 and the corresponding value of Nedge is
5. This means that in order to nucleate a graphene layer the step
edges should contain more than five Pt atoms or be about
1.4 nm (5 ꢂ 0.278 nm) in length. Consistent with these calcula-
tions, we observe that particles with average diameters smaller
than about 2 nm do not form much carbon, since only a fraction
of such particles will be sufficiently large to nucleate graphene
growth at 600 °C. A similar effect of metal particle size on the rate
of carbon deposition has been reported previously for Ni particles
[27]. We note, however, that with increasing temperature the va-
rffiffiffiffiffiffiffiffi
Ntot
3
D
G ¼ ꢁNtotDlC þ 2
Eedge þ EstretchNtot
;
ð1Þ
where
D
G is the total free energy change for a semi-hexagonal
graphene island; Ntot is the total number of atoms in the graphene
island; lC is the carbon chemical potential, i.e., the difference in
D
the chemical potentials of carbon in the precursor state and in
graphene; Eedge is the energy cost per carbon atom in the perimeter
of the island relative to that in an infinite graphene sheet; and
Estretch is the cost associated with stretching the finite-sized graph-
ene layer along the {101} zigzag direction to an epitaxial match
with the step edges as a function of the total number of carbon
atoms. The authors of Ref. [33] propose that Estretch can be estimated
from the value of Estrain, the strain energy of an infinite graphene
layer stretched (or compressed) at its edge to match the M–M
distance along the zigzag {101} step of the metal. The critical value
of Ntot above which graphene growth on a metal will be stable, NC, is
lue of
DlC increases and hence the value of NC decreases, from
which it is concluded that the average size of Pt nanoparticles on
which graphene formation can occur declines with increasing tem-
perature. Moreover, as the value of
DlC becomes larger when iso-
butene instead of ethane is used for carbon deposition,
D
G
determined by letting dDG/dNtot = 0.
decreased correspondingly, thereby facilitating carbon growth, in
agreement with our in situ TEM observations. Similarly, if ele-
ments are alloyed with Pt, e.g., Sn, Ga, In, which cause an increase
in Estrain or that just block the steps for carbon, then the value of NC
will increase and the temperature required to nucleate graphene
growth on a Pt bimetallic particles of a given size will increase
[27].
The picture of graphene formation described earlier is also con-
sistent with the TEM images presented in Figs. 6, 9 and 10, which
show the distance between the Pt particle and the first graphene
and between adjacent graphene–graphene layers is ꢀ3.5 Å. These
large distances are similar to those reported in surface science
studies on graphene formation on Pt(111) [51]. Of particular note
is the large spacing between graphene layers seen in Fig. 9. Further
evidence of weak interaction between Pt and the graphene basal
plane is the presence of numerous folds and wrinkles in the
Application of Eq. 1 to determine NC for Pt, assuming that the
precursor for carbon deposition is C2H6, reveals an interesting sit-
uation. The value of
DlC at 600 °C in the present case is 0.40 eV
[50], whereas the values of Eedge and Estrain are estimated to be
0.80 eV and 0.55 eV, respectively [33]. Because of the value of
Estrain for Pt exceeds the DlC, DG is always positive, suggesting that
the nucleation of graphene could not occur. The authors of Ref.
[33] point out, however, that the assumption that Estretch ꢄ Estrain
may overestimate the energy associated with larger, finite-sized
graphene islands, because the perimeter atoms away from the step
edge may have more degrees of freedom to match the metal lat-
tice. By contrast, evidence from a number of sources indicates that
graphene has only a small interaction with a Pt(111) surface and
that the growth of graphene on this surface is not epitaxial [51–
57]. Because of this weak interaction, stress relaxation of graphene
sheets formed on Pt(111) can be relieved by the formation of folds

28
Z. Peng et al. / Journal of Catalysis 286 (2012) 22–29
of any change in the lattice parameters of Pt during the course of
carbon deposition supports the proposal that carbon deposition oc-
curs via a surface diffusion mechanism. The progress of graphene
layer growth depends on both the size and shape of the Pt parti-
cles. Particles of greater than about 6 nm in diameter become
enveloped in layer of turbostratic graphene, the number of which
grows with the time of exposure of the sample to hydrocarbon.
For smaller Pt particles, the graphene layers can form carbon nano-
tubes or sheets that slough off onto the support. The size depen-
dency of carbon growth is attributed to the accommodation of
strain energy generated in the graphene layers and the minimiza-
tion of overall free energy in the growth process. Pt particles of
similar size but varied shape catalyzed the formation of different
carbon structures under the same condition, revealing other con-
tributing factors besides the traditional understanding on their
growth. The systematic study on size and shape effects on the car-
bon deposition may be instructive in developing new catalyst in-
tended for both carbon growth and suppression. In particular, we
suggest that the observed suppression of carbon deposition on
bimetallic Pt nanoparticles containing Sn, Ga, or In as the second
element may be due to the presence of these elements at the edges
of step in the metal particles, where they may inhibit the nucle-
ation of graphene layer growth.
Acknowledgments
The authors express their appreciation for the opportunity to
obtain in situ HRTEM images at Haldor Topsøe A/S and to Sven Ull-
mann for his contribution to this effort. This work was supported
by a grant from Chevron Energy Technology Company. The authors
acknowledge support of the National Center for Electron Micros-
copy, Lawrence Berkeley Lab, which is supported by the US Depart-
ment of Energy under Contract DE-AC02-05CH11231.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2011.10.008.
Fig. 11. Schematic illustration of graphene layer growth on Pt particles of
increasing size: (a and b) graphene growth from onto the (111) surface of Pt
initiated at a [101] step; (c and d) envelopment of Pt particles by graphene for
particles greater than ꢀ6 nm in diameter; (e and f) formation of graphene
nanotubes on Pt particles of 2–6 nm; (g and h) formation of graphene sheets and
their migration to the support for Pt particles less than 2 nm in diameter.
References
[1] D.J. Trevor, R.L. Whetten, D.M. Cox, A. Kaldor, J. Am. Chem. Soc. 107 (1985)
518–519.
[2] S. Vajda, M.J. Pellin, J.P. Greeley, C.L. Marshall, L.A. Curtiss, G.A. Ballentine, J.W.
Elam, S. Catillon-Mucherie, P.C. Redfern, F. Mehmood, P. Zapol, Nat. Mater. 8
(2009) 213–216.
[3] V. Galvita, G. Siddiqi, P.P. Sun, A.T. Bell, J. Catal. 271 (2010) 209–219.
[4] G. Siddiqi, P.P. Sun, V. Galvita, A.T. Bell, J. Catal. 274 (2010) 200–206.
[5] R.J. Davis, E.G. Derouane, Nature 349 (1991) 313–315.
[6] Z. Paal, P. Tetenyi, J. Catal. 30 (1973) 350–361.
[7] Y. Barron, D. Cornet, G. Maire, F.G. Gault, J. Catal. 2 (1963) 152–155.
[8] I. Lee, F. Zaera, J. Am. Chem. Soc. 127 (2005) 12174–12175.
[9] Y. Barron, G. Maire, J.M. Muller, F.G. Gault, J. Catal. 5 (1966) 428–445.
[10] R.D. Cortright, R.M. Watwe, J.A. Dumesic, J. Mol. Catal. A – Chem. 163 (2000)
91–103.
[11] G.F. Froment, Catal. Rev.-Sci. Eng. 50 (2008) 1–18.
[12] Z. Paal, A. Wootsch, R. Schlogl, U. Wild, Appl. Catal. A – Gen. 282 (2005) 135–
145.
graphene layers after they are formed, features that can be clearly
seen in Figs. 9 and 10.
The changes in the form of graphene growth with Pt particle
size can be explained in terms of the balance between the free en-
ergy gained upon the formation of graphene and the interactions
between graphene layers and the strain in the graphene layers in-
duced by bending the layers so that they can contour the surface of
particles on which they are formed. While interlayer interactions
will work to stabilize the growth of graphene, bending of the layers
will offset this stabilization. It is for this reason that the form of
graphene growth is dependent on the size of the Pt nanoparticles.
This point is illustrated by the schematics of graphene layer growth
shown in Fig. 11.
[13] F. Humblot, J.P. Candy, F. Le Peltier, B. Didillon, J.M. Basset, J. Catal. 179 (1998)
459–468.
[14] K.P. De Jong, J.W. Geus, Catal. Rev.-Sci. Eng. 42 (2000) 481–510.
[15] S. Helveg, C. Lopez-Cartes, J. Sehested, P.L. Hansen, B.S. Clausen, J.R. Rostrup-
Nielsen, F. Abild-Pedersen, J.K. Nørskov, Nature 427 (2004) 426–429.
[16] A.T. Bell, Nature 456 (2008) 185–186.
[17] S. Hofmann, R. Sharma, C. Ducati, G. Du, C. Mattevi, C. Cepek, M. Cantoro, S.
Pisana, A. Parvez, F. Cervantes-Sodi, A.C. Ferrari, R. Dunin-Borkowski, S. Lizzit,
L. Petaccia, A. Goldoni, J. Robertson, Nano Lett. 7 (2007) 602–608.
[18] M. Lin, J.P.Y. Tan, C. Boothroyd, K.P. Loh, E.S. Tok, Y.L. Foo, Nano Lett. 6 (2006)
449–452.
[19] J.A. Rodriguez-Manzo, M. Terrones, H. Terrones, H.W. Kroto, L.T. Sun, F.
Banhart, Nat. Nanotechnol. 2 (2007) 307–311.
[20] R.T.K. Baker, P.S. Harris, R.B. Thomas, R.J. Waite, J. Catal. 30 (1973) 86–95.
[21] R. Sharma, E. Moore, P. Rez, M.M.J. Treacy, Nano Lett. 9 (2009) 689–694.
4. Conclusions
Both in situ and ex situ HRTEM experiments were used to inves-
tigate carbon formation upon exposure of MgO-supported Pt nano-
particles to hydrocarbons. TEM observations suggest that the
formation of graphene layers initiates at low-coordination number
sites located at step on the surface of Pt nanoparticles. The absence

Z. Peng et al. / Journal of Catalysis 286 (2012) 22–29
29
[22] H. Yoshida, S. Takeda, T. Uchiyama, H. Kohno, Y. Homma, Nano Lett. 8 (2008)
2082–2086.
[40] P.S. Cremer, X.C. Su, Y.R. Shen, G. Somorjai, J. Chem. Soc.-Faraday Trans. 92
(1996) 4717–4722.
[23] R.T.K. Baker, M.A. Barber, P.S. Harris, F.S. Feates, R.J. Waite, J. Catal. 26 (1972)
51–62.
[41] S.B. Simonsen, I. Chorkendorff, S. Dahl, M. Skoglundh, J. Sehested, S. Helveg, J.
Am. Chem. Soc. 132 (2010) 7968–7975.
[24] F. Abild-Pedersen, J.K. Nørskov, J.R. Rostrup-Nielsen, J. Sehested, S. Helveg,
Phys. Rev. B 73 (2006) 115419.
[42] M.S. Dresselhaus, A. Jorio, R. Saito, Annu. Rev. Condens. Matter Phys. 1 (2010)
89–108.
[25] O.V. Yazyev, A. Pasquarello, Phys. Rev. Lett. 100 (2008) 156102.
[26] S. Hofmann, G. Csanyi, A.C. Ferrari, M.C. Payne, J. Robertson, Phys. Rev. Lett. 95
(2005) 036101.
[43] A.C. Ferrari, Solid State Commun. 143 (2007) 47–57.
[44] L.M. Malard, M.A. Pimenta, G. Dresselhaus, M.S. Dresselhaus, Phys. Rep.-Rev.
Sec. Phys. Lett. 473 (2009) 51–87.
[27] H.S. Bengaard, J.K. Nørskov, J. Sehested, B.S. Clausen, L.P. Nielsen, A.M.
Molenbroek, J.R. Rostrup-Nielsen, J. Catal. 209 (2002) 365–384.
[28] W.T. Owens, N.M. Rodriguez, R.T.K. Baker, J. Phys. Chem. 96 (1992) 5048–5053.
[29] N.M. Rodriguez, P.E. Anderson, A. Wootsch, U. Wild, R. Schlogl, Z. Paal, J. Catal.
197 (2001) 365–377.
[30] R.A. Cabrol, A. Oberlin, J. Catal. 89 (1984) 256–266.
[31] T.S. Chang, N.M. Rodriguez, R.T.K. Baker, J. Catal. 123 (1990) 486–495.
[32] B.J. Kang, J.H. Mun, C.Y. Hwang, B.J. Cho, J. Appl. Phys. 106 (2009) 104309.
[33] S. Saadi, F. Abild-Pedersen, S. Helveg, J. Sehested, B. Hinnemann, C.C. Appel, J.K.
Nørskov, J. Phys. Chem. C 114 (2010) 11221–11227.
[34] X.W. Teng, D. Black, N.J. Watkins, Y.L. Gao, H. Yang, Nano Lett. 3 (2003) 261–
264.
[35] K. Itatani, K. Koizumi, F.S. Howell, A. Kishioka, M. Kinoshita, J. Mater. Sci. 23
(1988) 3405–3412.
[36] C. Kisielowski, B. Freitag, M. Bischoff, H. van Lin, S. Lazar, G. Knippels, P.
Tiemeijer, M. van der Stam, S. von Harrach, M. Stekelenburg, M. Haider, S.
Uhlemann, H. Muller, P. Hartel, B. Kabius, D. Miller, I. Petrov, E.A. Olson, T.
Donchev, E.A. Kenik, A.R. Lupini, J. Bentley, S.J. Pennycook, I.M. Anderson, A.M.
Minor, A.K. Schmid, T. Duden, V. Radmilovic, Q.M. Ramasse, M. Watanabe, R.
Erni, E.A. Stach, P. Denes, U. Dahmen, Microsc. Microanal. 14 (2008) 469–477.
[37] J.K.A. Clarke, M.J. Bradley, L.A.J. Garvie, A.J. Craven, T. Baird, J. Catal. 143 (1993)
122–137.
[45] S. Tomita, M. Fujii, S. Hayashi, K. Yamamoto, Chem. Phys. Lett. 305 (1999) 225–
229.
[46] S. Horch, H.T. Lorensen, S. Helveg, E. Laegsgaard, I. Stensgaard, K.W. Jacobsen,
J.K. Nørskov, F. Besenbacher, Nature 398 (1999) 134–136.
[47] E.C. Garnett, W.J. Liang, P.D. Yang, Adv. Mater. 19 (2007) 2946–2950.
[48] M.S. Kumar, D. Chen, J.C. Walmsley, A. Holmen, Catal. Commun. 9 (2008) 747–
750.
[49] R.M. Rioux, H. Song, J.D. Hoefelmeyer, P. Yang, G.A. Somorjai, J. Phys. Chem. B
109 (2005) 2192–2202.
[50] A.B. Preobrajenski, M.L. Ng, A.S. Vinogradov, N. Martensson, Phys. Rev. B 78
(2008) 073401.
[51] P. Sutter, J.T. Sadowski, E. Sutter, Phys. Rev. B 80 (2009) 245411.
[52] J. Coraux, A.T. N’Diaye, M. Engler, C. Busse, D. Wall, N. Buckanie, F. Heringdorf,
R. van Gastel, B. Poelsema, T. Michely, New J. Phys. 11 (2009) 023006.
[53] D.E. Starr, E.M. Pazhetnov, A.I. Stadnichenko, A.I. Boronin, S.K. Shaikhutdinov,
Surf. Sci. 600 (2006) 2688–2695.
[54] M. Sasaki, Y. Yamada, Y. Ogiwara, S. Yagyu, S. Yamamoto, Phys. Rev. B 61
(2000) 15653–15656.
[55] M. Enachescu, D. Schleef, D.F. Ogletree, M. Salmeron, Phys. Rev. B 60 (1999)
16913–16919.
[56] T.A. Land, T. Michely, R.J. Behm, J.C. Hemminger, G. Comsa, Surf. Sci. 264 (1992)
261–270.
[38] R.S. Vincent, R.P. Lindstedt, N.A. Malik, I.A.B. Reid, B.E. Messenger, J. Catal. 260
(2008) 37–64.
[57] Z.P. Hu, D.F. Ogletree, M.A. Vanhove, G.A. Somorjai, Surf. Sci. 180 (1987) 433–
459.
[39] K. Takehira, T. Ohi, T. Shishido, T. Kawabata, K. Takaki, Appl. Catal. A – Gen. 283
(2005) 137–145.