In situ aberration-corrected environmental TEM: Reduction of model Co3O4 in H2 at the atomic level

Article abstract of DOI:10.1002/cctc.201300047

Understanding the dynamic evolution of structural changes in catalysts at the atomic level under controlled reaction conditions is one of the most challenging areas in heterogeneous catalysis. Here we present aberration-corrected environmental TEM at the

Full text of DOI:10.1002/cctc.201300047

CHEMCATCHEM
FULL PAPERS
FULL PAPERS
M. R. Ward, E. D. Boyes, P. L. Gai*
& – &&
&
In Situ Aberration-Corrected
Environmental TEM: Reduction of
Model Co O in H at the Atomic Level
3
4
2
Co O reduction is CoOl: By using ad-
vanced aberration-corrected environ-
mental TEM, the intricacies of the dy-
Tropsch process, have been investigat-
ed. Notably, the transformation process
at low pressure proceeds via an advanc-
ing interface between Co O and CoO
3
4
namic atomic scale reduction of Co O4
3
3
4
in H that are of interest in hydrogena-
regions. Co metal is observed at higher
pressure.
2
tion reactions, such as the Fischer–
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 8
&
1
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
DOI: 10.1002/cctc.201300047
In Situ Aberration-Corrected Environmental TEM:
Reduction of Model Co O in H at the Atomic Level
3
4
2
[
a]
Michael R. Ward, Edward D. Boyes, and Pratibha L. Gai*
Understanding the dynamic evolution of structural changes in
catalysts at the atomic level under controlled reaction condi-
tions is one of the most challenging areas in heterogeneous
catalysis. Here we present aberration-corrected environmental
TEM at the atomic level and electron diffraction of the reduc-
tion of model Co O catalysts in H to directly observe the dy-
Co O and CoO regions. The interface penetrates further into
3 4
the Co O crystal, with the CoO regions replacing the previous
3
4
Co O structure, with increasing reduction. The reduction to
3
4
CoO proceeds at approximately 2008C at rounded edges con-
taining atomic steps on the surfaces. The interfaces are ob-
served in crystals larger than approximately 15 nm in size but
not in smaller crystals, which indicates the rapid reduction of
smaller nanoparticles. The most dramatic changes are ob-
served at 3508C in larger crystals of size 50 nm.
3
4
2
namic phase evolution in the reduction process. New insights
into the reduction of Co O to the intermediate CoO include
3
4
the formation of an advancing atomic scale interface between
Introduction
Supported Co-based nanoparticles are of interest in a number
of hydrogenation reactions, such as the Fischer–Tropsch (F–T)
process for synthetic fuel technologies, and in environmental
control. The F–T process contains a series of chemical reactions
involving the hydrogenation of CO over catalysts that produce
various hydrocarbons according to Equation (1):
Promoter effects have been reported in Co-based catalysts,
[5–7]
which lead to an increase in the active site concentrations.
The model Co on carbon nanofibres suggests that the specific
activity of C5+ molecules decreases as the nanoparticle size de-
creases below 7 nm with an increase in the C molecule se-
1–3
[8,9]
lectivity.
The same group have also observed that the sup-
port is inert whereas more common supports such as Al O
2
3
can form mixed oxides with the supported nanoparticle
ð2n þ 1Þ H þ n CO ! C Hð þ Þ þ n H O
ð1Þ
2
n
2n
2
2
[8,9]
species.
The uncertainties surrounding the activity of Co-based cata-
lysts may originate from their activation processes. Co is rarely
in which n is an integer. The above reaction is a surface-cata-
[
1]
lysed polymerisation reaction. The Fe-based catalysts are re-
ported to be selective towards alkanes in the C1–4 range at
a high temperature of 3508C and a pressure of 25 mbar
available as a pure metal for the development of catalysts. In-
[10,11]
stead, an oxide, usually Co O ,
is reduced through pre-
3
4
treatment in H and N . The nature of the reduction of the
2
2
[
2]
(
2.5 kPa). The Co-based catalysts have been used because
oxide ultimately determines the range of reduced/oxidised Co
species in the catalyst. These species play a key role in the
structure, morphology and activity of the catalyst.
they offer a compromise between cost, efficiency and selectivi-
[3]
ty for linear hydrocarbons at a low temperature of 2258C.
The Co-based catalysts have been found to be highly selective
Both ex situ and in situ studies have revealed a two-stage re-
towards C5 hydrocarbons, which is beneficial for the synthesis
[12–16]
+
duction process of Co O :
3
4
[
3]
of diesel. On the basis of chemical methods and activity
measurements, active sites in the F–T process have been re-
Co O ! CoO ! Co
3
4
0
[3]
ported to be Co (Co metal). Typical F–T catalysts use Co
nanoparticles on supports such as Al O ; however, the metal
[17]
2
3
in which Co has a structure that is size dependent, with
smaller particles of size <30 nm favouring the face centred
cubic (fcc) phase. CoO is always observed in its rock salt phase
loading can often be above 33% to encourage a high level of
[
4]
activity and selectivity.
[18,19]
unless special sample preparations are used.
The unit cells
+
+
+
[
a] M. R. Ward, Prof. E. D. Boyes, Prof. P. L. Gai
The York-JEOL Nanocentre
for the three common fcc forms of Co O , CoO and Co are
3
4
University of York
shown in Figure 1.
Heslington, York YO10 5DD (UK)
E-mail: pratibha.gai@york.ac.uk
Exceptions to the above reduction sequence have been ob-
served by different groups. By using in situ XRD, Bulavchenko
+
[
] All authors belong to the Department of Physics, Prof. E. D. Boyes to the
Department of Electronics and Prof. P. L. Gai to the Department of
Chemistry.
[13]
et al. showed that at high H pressures [1 atm (101.3 kPa)],
2
the intermediate CoO phase is not observed on unsupported
Co O and Co (hexagonal close packed) forms directly from
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201300047.
3
4
Co O . In gas mixtures in which H partial pressure is low
3
4
2
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 8
&
2
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Co O , which leads to unusual intensities in {111} and {220}
3
4
reflections.
Here we have investigated the intricacies of the reduction of
Co O to CoO at the atomic level under controlled low H pres-
3
4
2
sure and at elevated temperatures by using aberration-correct-
ed environmental TEM (AC-ETEM) developed in-house by Gai
[24–26]
and Boyes.
Our aim is to directly observe and identify the
process by which Co O transforms to reduced phases in crys-
3
4
tals of different sizes. Our observations have revealed that the
evolution of CoO starts at approximately 2008C primarily at
Co O4 crystal edges composed of low-symmetry facets. The
3
transformation process is gradual, which always contains an in-
terface between Co O and CoO regions. We observed defects
3
4
only in crystals larger than 30 nm in size, with Co O present
3
4
up to approximately 4508C, whereas particles less than ap-
proximately 20 nm in size transformed completely to CoO by
4
008C. The atomic level insights into the dynamic reduction of
3 4
Figure 1. Unit cells of cubic Co, rock salt CoO and spinel Co O as labelled.
the model catalysts presented here are important to under-
stand the dynamic behaviour and activity of real supported
Co-based catalysts in hydrogenation reactions and will be
useful in the future catalyst design.
The models are approximately to scale. The larger blue balls are Co, and the
smaller red ones are O. Models were generated by using the VESTA
program.
[20]
[
1 Torr (133.3 Pa) of 10% H /N ] and at 400 8C, Li et al. re-
2 2
ported that Co is not observed in the in situ nanoanalysis with Results and Discussion
high-angle annular dark field scanning transmission electron
In situ reduction HRTEM and diffraction sequences
microscopy studies, and similar conclusions were reported in
[
13]
XRD studies by Bulavchenko et al. Conventional HRTEM has
shown that the reduction process is epitaxial between Co
Dynamic AC-ETEM images and the corresponding fast-Fourier-
transform (FFT) patterns of reaction sequences of a Co O cata-
3
4
[
12,14]
(
cubic), CoO and Co O ,
and the support increases the re-
lyst crystal at different temperatures during exposure to H2
in situ are shown in Figure 2. Co O was readily identified from
3
4
[13,15]
duction temperature owing to metal–support interactions.
3
4
[
12]
Dehghan et al. investigated the change in the structure of
a promoted 20 wt.% Co/0.5 wt.% Re/Al O catalyst at 3.4 mbar
other phases because of the presence of a large 0.47 nm {111}
atomic plane spacing. In the FFT, the {002} forbidden reflec-
tions are present, which indicates double diffraction. The crys-
tal shown in Figure 2a at room temperature is rounded on the
right-hand side, which indicates the presence of low-symmetry
atomic planes that contain kinks and steps. Kinks and steps act
as nucleation sites and encourage the adsorption and breaking
2
3
(0.34 kPa) up to 3608C by using in situ environmental TEM.
They found that Co metal forms at 3608C and CoO at tempera-
tures as low as 1508C. The promoter Re is believed to encour-
[
21]
age the reduction of CoO to Co.
To better understand the complex phase evolution at the
atomic level in the dynamic reduction of cobalt oxides initially
in H , we used model Co O catalysts. And, understanding the
[27]
[14]
of HꢀH bonds on Pt. Potoczna-Petru and Kepinski
also
suggested that rounded Co O crystals are more likely to con-
2
3
4
3
4
reduction of model catalysts can lead to deeper insights into
the operation of real catalysts. There are relatively few dedicat-
ed dynamic atomic scale electron microscopy studies on un-
supported model Co O catalysts. By using conventional TEM
tain defects on the surface. They referred to the work of Paryjc-
[15]
zak et al., in which the synthesis temperature of the Co O
3
4
crystals affected the onset of the reduction to CoO and the
synthesis temperature correlated with the nature of crystal
facets.
3
4
[
(
ex situ post-reaction studies on samples treated at 1 atm
[
14]
101.3 kPa)], Potoczna-Petru and Kepinski
suggested that
Increasing the temperature to 1508C resulted in no structur-
al changes in our observations. At approximately 2008C, the
initial signs of structural changes were observed. The image in
Figure 2b indicates that the crystal region previously with
a rounded edge as in Figure 2a has become more faceted. The
new surface is terminated by CoO {002}. It is not a perfect sur-
face and has some residual steps present (arrowed). The FFT
shows distorted Co O {002} and {004} reflections. The two re-
CoO and Co form on the crystals larger than 40 nm in size,
which possibly causes stacking faults and voids to appear in
the remaining Co O structure. They reported that high-tem-
3
4
perature reduction (>4008C) can result in the formation of Co
at the surface, which causes any CoO formed to reduce in an
autocatalytic fashion. Because this experiment was performed
ex situ, there appears to be no consensus on the possible for-
mation of stacking faults and their relationship to the reduc-
tion mechanism. Such defects were not observed by Dehghan
3
4
flections are in the vicinity of each other, which indicates some
“broadening” of the reflections. The two closely spaced reflec-
tions indicate the presence of CoO that has {002} spacings
within 1% of Co O {004}. The CoO region formed on the sur-
[
12]
[22]
et al.
Dieckmann
reported that there were no reliable
data on the defect structure of Co O ; however, Casas-Cabanas
3
4
3
4
[
23]
et al. showed that Co atoms can occupy interstitial sites in
face penetrates approximately 3.4 nm into the Co O crystal.
3 4
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 8
&
3
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
of the Co O structure. By allowing for the 6% difference be-
3
4
tween Co O and a doubled CoO unit cell, the mismatch could
3
4
lead to strain relaxation in this region.
Increasing the temperature up to 3008C caused the CoO
region to grow and penetrate deeper into the Co O crystal
3
4
whilst maintaining an essentially epitaxial grain boundary,
which is similar to the observation at 2008C. The distance it
penetrated into the Co O crystal was 6.1 nm at 3008C, which
3
4
is double that of at 2008C. Although the FFT in Figure 2c was
obtained from the region around the boundary, separate FFTs
(
not shown) confirmed that the left-hand side of the crystal is
CoO. At this temperature, we also observed similar events oc-
curring primarily at the edges of other rounded, differently
sized crystals. In contrast, flat facets (e.g. {111} and {200}) re-
mained unchanged during the reduction process. There seems
to be a correlation between low-symmetry faces and the re-
duction-onset temperature, which is consistent with previous
[14]
reports.
At 3508C, there is a dramatic change in the contrast of crys-
tals. The CoO region has moved further into the Co O crystal,
3
4
and the penetration is 15 nm deep. On the left-hand side of
the crystal, dramatic contrast can be observed. The previous
images suggest that there are multiple crystals here with slight
orientation differences between them. The unusual contrast
seen on the left-hand side of the image (and elsewhere) could
be due to double diffraction. The FFT demonstrates no addi-
tional spacings, which suggests that the effects observed are
the Moirꢁ fringes. The Moirꢁ fringe width D can be calculated
by using Equation (2).
2
2
D ¼ d d =½ðd ꢀd Þ þ d d q ꢁ
ð2Þ
1
2
1
2
1 2
in which d and d are the atomic plane distances for each
1
2
species responsible for the Moirꢁ fringes and q is the angle be-
tween them. The Moirꢁ fringes in Figure 2d generate effects in
[
1¯ 11] and [1 3¯ 1] directions. The Moirꢁ fringes are caused by
the overlap of the strong CoO {002} planes in the smaller crys-
tal and the Co O {113} planes in the large parent crystal.
3
4
At 4008C, the crystal that was initially Co O (as shown in
3
4
Figure 2a) is now completely CoO (as shown in Figure 2e). The
surrounding region is also CoO. The Moirꢁ fringe contrast seen
at 3508C was also observed at 4008C but only in regions
larger than 30 nm in size. The larger crystals had structures re-
sembling a core–shell structure, and CoO was the shell and
Co O the core. The resultant overlap produced Moirꢁ fringes
3
4
in the central regions of crystals larger than 20 nm in size. This
was not observed on smaller crystals, because the CoO trans-
formation was much more rapid.
Figure 2. Dynamic aberration-corrected environmental TEM image sequence
with the corresponding FFTs in H
showing the reduction of Co to CoO at the temperatures indicated. The
FFT in (a) shows Co ; in (b)–(d) the Co {004} reflections have become
broad because of the similar CoO {002} spacing. In (e), the FFT shows just
CoO with the Co originating from another crystal. The blue arrow shows
2
gas as a function of reaction temperature,
3
O
4
3
O
4
3 4
O
Increasing the temperature up to 5008C resulted in no dis-
cernible change in the structure, which suggested that the low
3
O
4
H pressure is insufficient to facilitate the reduction of CoO.
2
that the rounded edge of the crystal has become flat after heating and the
red arrow shows dimensions of the CoO region. Scale bar in image e is
Our results are consistent with the observations of Bulavchen-
[13]
10 nm.
ko et al. at low H partial pressure, who observed that Co
2
was not formed.
[14]
This is approximately six unit cells of CoO. The FFT indicates
that the lattice parameter varies across this region and that
there is an interface between the modified edge and the rest
The observations of Potoczna-Petru and Kepinski suggest-
ed that Co metal is formed in the middle of the modified struc-
[12]
tures; however, by using in situ microscopy, Dehghan et al.
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 8
&
4
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
showed that richer oxides were prevalent at the edges be-
{444} and CoO {222}). As in the cases in Figures 2 and 3, the
Co O and CoO crystallographic directions are essentially epi-
cause of residual O atoms re-oxidising CoO and Co to Co O4
3
3
4
and CoO, respectively, after H was removed from their micro-
taxial, with no clear rotation between the two crystals; that is,
{002} and {111} faces between the two materials are parallel
to each other.
2
scope and cooled overnight.
Nanoparticles less than 15 nm in size were observed to com-
pletely reduce to CoO at a lower temperature of 3508C. An ex-
ample is shown in the AC-ETEM images and the corresponding
FFTs (Figure 3), which illustrates a spherical nanoparticle ap-
proximately 12 nm in size. The crystal in Figure 3a has stepped
surfaces; however, the observations indicate that even after re-
duction, the stepped surfaces remain. In small particles
Notably, although there are no major structural changes up
to 3008C, the Co O {111} planes gradually lose their intensity
3
4
at the expense of other Co O {004} and CoO {002} reflections.
3
4
Electron diffraction intensities can vary as a function of crystal
thickness, but also because of the tilt of the crystal with re-
spect to the beam. The diffraction patterns shown in Figure 3
are approximately between 3
and 58 away from the exact
[
110] zone axis. The structure
factor for spinel materials pre-
dicts that {440} reflections
should be the most intense, fol-
lowed by {113}, {400} and {115}
reflections. At 3008C, the most
intense reflections are the {400}
reflections followed by {113} re-
flections. By using JEMs, which is
an HRTEM image simulation
package, and our aberration-cor-
rected electron microscope pa-
rameters (given in Table S1), we
simulated the electron diffrac-
tion patterns shown in Figur-
es S1 and S2. They indicate that
any major difference between
predicted and observed intensi-
ties can be due to the crystal tilt
rather than the presence of
2
Figure 3. Dynamic aberration-corrected environmental TEM image sequence in H gas as a function of the reac-
tion temperature with corresponding FFTs of a Co
forms completely to CoO at 3508C, but the larger nanoparticle it is connected to is still primarily Co
rounding amorphous layer is carbonaceous and is probably due to the contamination on the nearby Cu grid bar.
The scale bar in image a is 10 nm.
3
O
4
nanocatalyst approximately 12 nm in size. The particle trans- point defects as reported by
. The sur-
3
O
4
[23]
Casas-Cabanas et al.
The Co O4 {004} reflections
3
become significantly brighter
and broader at 4008C, which in-
(
ꢂ2 nm), the removal of low coordinated surface atoms to
dicates the occurrence of bulk transformations to CoO. By
4508C, the transformation is complete and the crystal has ori-
entated itself closer to the CoO [110] zone axis, which leads to
more typical intensities in different reflections.
minimise the surface energy and transport of the atoms be-
[
28]
tween particles leads to sintering and faceting. It is possible
that in the larger particle shown in Figure 3, primary faceting
has already occurred. In general, Co O particles larger than
3
4
2
0 nm in size, irrespective of their initial shape, became more
Co O to CoO transformation mechanism
3
4
faceted as they were reduced to CoO.
In Figure 4a, the ETEM image shows the central region of
In dynamic in situ reduction process at the atomic level, an in-
terface boundary is formed in most crystals. A close-up of such
an interface is shown in Figure 5. The crystal shown in Figure 5
is away from the [110] zone axis by approximately 48. The tilt
is such that one set of the {111} reflections is the strongest in
the FFT and diffraction patterns and thus {111} planes domi-
nate in the image.
a Co O4 crystal with a maximum width of approximately
3
5
0 nm. A sequence of the corresponding selected area electron
diffraction patterns at the temperatures indicated and the evo-
lution of reflections due to the reduction are shown in Fig-
ure 4b–e. There are no major changes at temperatures below
3
008C except for a slight change in the orientation of the crys-
tal, which results in small differences in the diffracted
intensities.
To understand the origin of the dark contrast observed be-
tween the Co O {111} planes at the top of Figure 5 and to ex-
3
4
At 3008C, weak CoO reflections can be identified in the dif-
fraction pattern. The differences in reciprocal lattice vectors are
plore the presence of Co O and CoO including the effect of
3 4
tilt on the contrast, JEMS image simulation package was used.
By using the Laue circle tool, it is possible to add tilt before
more apparent for the higher-order reflections (e.g. Co O4
3
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 8
&
5
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[29]
image calculations by using Co O (ICSD 28158 ) and CoO
3
4
(
ICSD 624575) [110] HRTEM images using settings correspond-
ing to our AC-ETEM images. The width of the crystal varies be-
tween 25 and 30 nm; therefore, we used this as the thickness
of the crystal for simulations. Our simulations have shown that
the dark contrast is due to the effect of thickness and defocus
(
Figure S3).
In smaller Co O crystals, the rapid transformation to CoO is
3
4
accompanied by the loss of surface O atoms and the formation
of anion vacancy sites at the surface that can provide highly
active sites for the dissociation of H . The lattice parameter of
2
Co O4 is 0.8084 nm, and the lattice parameter of CoO is
3
0.426 nm, as shown in Figure 1. By allowing for the doubling
of CoO unit cell, there is still approximately 6% mismatch be-
tween the two structures. Our preliminary analysis of Figure 5b
and its filtered image shown in Figure 5c reveals the presence
of dislocations around the interface boundary that can accom-
modate the strain at the interface due to the mismatch in the
1
transformation and indicates = [110] Burger’s (displacement)
2
vector for the defects. The existence of interfaces between
rock salt and spinel materials and in materials with small lattice
mismatch has been reported, for example between MgO and
[30]
[31]
Fe O4 and between NiO and Fe O , in which the lattice
3
3
4
mismatch is less than 1%. At these interface boundaries, anti-
phase boundaries (and not dislocations) have been reported.
In bimetallic metal nanoparticles, such as Pt–Pd nanoparticles
in which Pd and Pt have lattice parameters within 1%, stacking
[28]
faults and dislocations have been observed.
CoO to Co transformation
Figure 4. Dynamic selected area electron diffraction sequence in which
frame a is an aberration-corrected environmental TEM image of the area.
With the increase in temperature, the CoO reflections become more domi-
nant and the Co O reflections disappear at 4508C. The 5008C pattern is not
3 4
shown because it is virtually identical to that of 4508C, which implies that
During a scouting experiment using a higher pressure [3 mbar
0.3 kPa) inlet pressure], Co metal (fcc) formed at 4508C. The
AC-ETEM images illustrating Moirꢁ fringes caused by the over-
lap of the parallel CoO and Co {111} planes are shown in Fig-
ure 6a. In Figure 6a, Co is limited to the Moirꢁ fringe regions
as revealed by the filtration of the image by masking the Co
(
no Co is formed.
{
111} reflections. Co metal forms
small clusters on the crystals and
eventually spreads forming a Co
shell. The Co shell was also ob-
served in another crystal shown
in Figure 6b at 4508C, in which
the Moirꢁ fringes have become
more extensive.
Conclusions
We have presented direct atomic
level insights into the dynamic
reduction of model Co O cata-
3
4
lysts in controlled H2 environ-
ments at elevated temperatures.
The observations have revealed
that the transformation in crys-
tals larger than approximately
20 nm in size proceeds gradually
3 4
Figure 5. AC-ETEM images of the interface between Co O and CoO regions: a) at low magnification and b) at
high magnification. TEM simulations show that the contrast seen at the top of image b can be due to the effect
of thickness and defocus. The contrast in the central region in image b (which is the area around the interface) in-
3 4
dicates the presence of defects. The defects (dislocations) are revealed through the filtration of the Co O {222}
reflections as shown in image c in the FFT (not shown). The green and blue dots in image c signify the start and
end of the Burger’s circuit, respectively. Preliminary analysis suggests that the dislocations have Burger’s vector of
1
the type =
2
[110].
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 8
&
6
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[24–26]
pumping system.
This in-house AC-ETEM development fol-
lowed the pioneering development of atomic resolution ETEM by
[32,33]
Gai and Boyes.
This AC-ETEM enabled gas to be injected into the microscope, and
in the current stages of testing the pressure at the sample was ap-
proximately 0.1 mbar (0.01 kPa). The removal of aberrations from
the objective lens means that it is possible to operate the micro-
scope with the highest resolution without the need for through-
focal series, which is essential for real-time atomic resolution dy-
namic in situ studies. The AC-ETEM has a large objective lens pole-
piece, which enables the following holders to be used: A Gatan
6
28 single-tilt heating holder was used for in situ experiments. A
Figure 6. Aberration-corrected environmental TEM images obtained at
standard JEOL double-tilt holder was also used, but because of the
large-gap pole piece, tilts up to 258 in the X and Y axes were
possible.
4508C and during a separate experiment with a pressure 3 times greater
than that in the previous experiments. In this case, we observed the forma-
tion of Co metal on the surfaces in clusters at this high temperature. The
blue arrow in image b indicates a defect. Filtered FFT images show that
there are several dislocations near the yellow arrow, and the strained region
could be responsible for the dark contrast.
[34]
Careful in situ studies were performed with low electron beam to
minimise any beam effects, and the data were verified with calibra-
tion blank studies without the beam by using methods reported
[24–26]
previously.
Several dynamic experiments were performed with
sample areas away from the electron beam to minimise exposure,
and the areas were periodically checked under the beam at differ-
ent temperatures. Several minutes were allowed for temperatures
to stabilise. The temperature was increased in 508C steps up to
with an atomic interface between Co O and CoO regions,
3
4
which penetrates deeper into the crystal with the increase in
temperature. The formation of CoO is observed to proceed at
5
008C. We believe that the injection of gas into the specimen
2008C but only at the edges of rounded crystals, which sug-
chamber can cause a cooling effect on the thermocouple of the
heating holder. We repeated this experiment several times and
found that for a given current the temperature varied by a few
degrees.
gests that the weakly bound O should be removed from the
surface so to create surface sites for the dissociation of H . The
2
observation of CoO at operating temperatures of approximate-
ly 200–3508C has important implications in understanding the
catalytic activity of Co-based oxides used in hydrogenation re-
actions.
A JEOL 2011 TEM was also used for specimen screening. Dark-field
TEM was used to image defects in Co O . This was performed by
3
4
placing the in situ reduced sample into a double-tilt holder for
post-reduction analysis.
The observations of Co metal at higher temperatures of
4
508C and at higher pressures (ꢂ3 times used at which only
CoO is observed) suggest that higher temperature and higher
pressure are required for the transformation.
Acknowledgements
P.L.G. thanks the journal and the guest editors Prof. Dangsheng
Su and Prof. R. Schlogl for the invitation to contribute to this
Special Issue on Electron Microscopy for Catalysis. We thank the
Engineering and Physical Sciences Research Council, UK, for the
research funding and for a PhD studentship for M.R.W.
Experimental Section
The Co O powder was obtained from Aldrich. TEM and selected
3
4
area electron diffraction analysis of the powder showed primarily
Co O .
3
4
Specimens for conventional electron microscopy analysis were pre-
pared by dispersing a small amount of Co O powder in ethanol,
Keywords: aberration-corrected environmental TEM · cobalt ·
electron diffraction · Fischer–Tropsch · heterogeneous catalysis
3
4
which was then treated in an ultrasonic bath for 2 min. The solu-
tion was then deposited on to holey C grids. For in situ ETEM stud-
ies, unfilmed (empty) Cu grids were used because C films buckle at
high temperatures and react with samples under H2 at elevated
temperatures. The empty Cu grids were plasma-cleaned at 75%
power for 5 min with a Diener/Femto plasma cleaner to remove
any contamination. After depositing the Co O powder, the speci-
[1] a) C. K. Rofer-DePoorter, Chem. Rev. 1981, 81, 447–474; b) A. Datye
et al., Ind. Chem. Res. 2003, 42, 4001.
[2] A. P. Steynberg, R. L. Espinoza, B. Jager, A. C. Vosloo, Appl. Catal. A 1999,
1
86, 41–54.
[
[
3] E. Iglesia, Appl. Catal. A 1997, 161, 59–78.
4] R. Oukaci, A. H. Singleton, J. G. Goodwin, Appl. Catal. 1999, 186, 129–
3
4
mens were further plasma cleaned at 10% power for 30 s. Crystals
protruding over the edge of the grids were used in the in situ
studies.
144.
[
[
5] G. J. Haddad, B. Chen, J. J. G. Goodwin, J. Catal. 1996, 161, 274–281.
6] M. Rothaemel, K. F. Hanssen, E. A. Blekkan, D. Schanke, A. Holmen,
Catal. Today 1997, 38, 79–84.
For dynamic real-time in situ studies at the atomic level under con-
trolled gas and temperature conditions, we used a double AC-
[
[
[
7] F. Rohr, O. A. Lindvꢂg, A Holmen, E. A. Blekkan, Catal. Today 2000, 58,
2
47–254.
8] G. L. Bezemer, U. Falke, A. J. van Dillen, K. P. de Jong, Chem. Commun.
005, 731–733.
[24–26]
ETEM developed in-house by Boyes and Gai
at the University of York. The AC-ETEM with sub-angstrom (<
00 pm) resolution was developed by modifying a double aberra-
in the Nanocentre
2
1
9] G. L. Bezemer, J. H. Bitter, H. P. C. E. Kuipers, H. Oosterbeek, J. E. Hole-
wijn, X. Xu, F. Kapteijn, A. J. van Dillen, K. P. de Jong, J. Am. Chem. Soc.
2006, 128, 3956–3964.
tion-corrected JEOL 2200FS (S)TEM operating at 200 kV, which was
fitted with a large-gap objective lens pole piece and gas-tolerant
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 8
&
7
&
ÞÞ
These are not the final page numbers!

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[10] S. E. Colley, R. G. Copperthwaite, G. J. Hutchings, S. P. Terblanche, M. M.
[24] P. L. Gai, E. D. Boyes, Microsc. Res. Tech. 2009, 72, 153–64.
Thackeray, Nature 1989, 339, 129–130.
11] H. Schulz, Appl. Catal. A 1999, 186, 3–12.
[25] P. L. Gai, E. D. Boyes, J. Phys. 2010, 241, 012055–012060.
[26] P. L. Gai, E. D. Boyes, Handbook on Nanoscopy, Vol. 1 and 2 (Eds.: G. Van
Tendeloo, D. Van Dyke, S. J. Pennycook), Wiley, 2012.
[27] G. A. Somorjai, D. W. Blakely, Nature 1975, 258, 580–583.
[28] M. R. Ward, T. Hyde, E. D. Boyes, P. L. Gai, ChemCatChem 2012, 4, 1622–
1631.
[29] W. L. Smith, A. D. Hobson, Acta Crystallogr. Sect. B 1973, 29, 362–363.
[30] V. K. Lazarov, M. Weinert, S. A. Chambers, M. Gajdardziska-Josifovska,
Phys. Rev. B 2005, 72, 195401.
[
[
12] R. Dehghan, T. W. Hansen, J. B. Wagner, A. Holmen, E. Rytter, Ø. Borg,
J. C. Walmsley, Catal. Lett. 2011, 141, 754–761.
13] O. A. Bulavchenko, S. V. Cherepanova, V. V. Malakhov, L. S. Dovlitova,
A. V. Ishchenko, S. V. Tsybulya, Kin. Catal. 2009, 50, 192–198.
14] D. Potoczna-Petru, L. Kepinski, Catal. Lett. 2001, 73, 41–46.
15] T. Paryjczak, J. Rynkowski, S. Karski, J. Chromatogr. A 1980, 188, 254–
[
[
[
256.
[
[
16] H.-Y. Lin, Y.-W. Chen, Mater. Chem. Phys. 2004, 85, 171–175.
17] O. Kitakami, H. Sato, Y. Shimada, F. Sato, M. Tanaka, Phys. Rev. B 1997,
[31] A. Re cˇ nik, D. L. Carroll, K. A. Shaw, D. M. Lind, M. Rꢃhle, J. Mater. Res.
1997, 12, 2143–2151.
56, 13849–13854.
[32] P. L. Gai, K. Kourtakis, Science 1995, 267, 661.
[
[
18] R. W. Grimes, K. P. D. Lagerlof, J. Am. Ceram. Soc. 1991, 74, 270–273.
19] K. M. Nam, J. H. Shim, D.-W. Han, H. S. Kwon, Y.-M. Kang, Y. Li, H. Song,
W. S. Seo, J. T. Park, Chem. Mater. 2010, 22, 4446–4454.
[33] E. D. Boyes, P. L. Gai, Ultramicroscopy 1997, 67, 219.
[34] N. Shiju, K. Yoshida, E. D. Boyes, D. R. Brown, P. L. Gai, Catal. Sci. Technol.
2011, 1, 413–425.
[
[
20] P. Li, J. Liu, N. Nag, P. A. Crozier, Appl. Catal. A 2006, 307, 212–221.
21] K. M. Cook, S. Poudyala, J. T. Millerb, C. H. Bartholomewa, W. C. Hecker,
Appl. Catal. A 2012, 449, 69–80.
[22] R. Dieckmann, Solid State Ionics 1984, 12, 1–22.
[23] M. Casas-Cabanas, G. Binotto, D. Larcher, A. Lecup, V. Giordani, J.-M. Tar-
ascon, Chem. Mater. 2009, 21, 1939–1947.
Received: January 18, 2013
Published online on && &&, 0000
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 0000, 00, 1 – 8
&
8
&
ÞÞ
These are not the final page numbers!