Outstanding Methane Oxidation Performance of Pd-Embedded Ceria
Catalysts Prepared by a One-step Dry Ball-Milling Method
Maila Danielis,[a] Sara Colussi,[a] Carla de Leitenburg,[a] Lluís Soler,[b] Jordi Llorca,[b] and
Alessandro Trovarelli*[a]
[a] M. Danielis, dr. S. Colussi, prof. C. de Leitenburg, prof. A. Trovarelli
Polytechnic Department. University of Udine. Via del Cotonificio 108, 33100 Udine, Italy
[b] dr. L. Soler, prof. J. Llorca
Institute of Energy Technologies, Department of Chemical Engineering and Barcelona Research
Centre in Multiscale Science and Engineering. Universitat Politècnica de Catalunya, EEBE.
Eduard Maristany 10-14, 08019 Barcelona, Spain
Abstract: By carefully mixing Pd metal nanoparticles with CeO
2
polycrystalline powder under dry conditions a new
unpredicted arrangement of the Pd-O-Ce interface is obtained, where an amorphous shell containing Palladium species
dissolved in Ceria is covering a core of CeO
2
particles. The robust contact that is generated at nanoscale, along with
mechanical forces generated during mixing, promotes the redox exchange between Pd and CeO
2
and creates highly
reactive and stable sites constituted by PdO
x
embedded into CeO
2
surface layers. This specific arrangement
outperforms conventional Pd/CeO
2
reference catalysts in methane oxidation, by lowering light-off temperature by more
than 50 degrees and boosting reaction rate. The origin of the outstanding activity is traced back to the structural
properties of the interface, modified at nanoscale by mechano-chemical interaction, and it is unraveled by a combined
set of experimental data including high resolution transmission electron microscopy and supported by recent
computational studies.
The increasing concern over the abundant emissions of greenhouse gases from motor vehicles
is pushing towards the development of more efficient catalysts for their abatement. This is
particularly true for natural gas fueled vehicles, for which the exponential growth of the
market and the concern for methane global warming potential urge for the design of catalytic
systems with improved activity at low temperature and higher resistance to deactivation
under operating conditions.[1] An efficient low temperature activation of the CH4 molecule
would also be a significant advancement in the field of methane utilization, an issue that is
now attracting several efforts due to the increased supplies of shale gas and the consequent
availability of natural gas as a feedstock.[2] Pd-based formulations are the most effective for
methane oxidation, and the use of ceria as support confers additional benefits to the catalysts
due to its unique redox features and to the level of Pd-Ce interfacial interactions.[3] It is
reported from experimental[3b, 4] and computational[5] studies that an enhanced Pd-ceria
interaction can improve significantly the catalytic activity of these materials. In particular, the
presence of Pd into ceria lattice can lead to the formation of highly reactive Pd2+/4+ sites
which show lower methane activation barriers compared to isolated PdOx units. Interestingly,
differently from what happens on PdOx clusters, on these ionic Pd species methane activation
proceeds via hydrogen abstraction, a route that is potentially very important for methane
utilization.[5a]
We have already reported the increased activity of a Pd/CeO2 catalyst prepared by solution
combustion synthesis in which the substitution of Pd2+ ions into ceria lattice caused the
formation of ordered arrays of oxygen vacancies and highly reactive undercoordinated oxygen
atoms.[6] More recently we also investigated the milling of CeO2-based materials with carbon,
originating a 2D carbon layer, covering the ceria particles, and improving the interfacial redox
exchange between the two materials.[7] Here, by combining the above mentioned
approaches, we use a controlled one-step dry milling procedure where Pd metal nanoparticles
are put in contact with ceria particles to prepare a methane oxidation catalyst that
outperforms traditional Pd/CeO2 due to the unique structural arrangements that characterize
metal/support interface at nanoscale. In addition, avoiding the use of Pd nitrate or chloride
solution significantly reduces waste generation ensuring lower environmental impact. The
characteristics of the preparation method, the properties and performances of the catalyst
have been investigated in detail; the data suggest a correlation between the unusual
morphology developed at nanoscale and the high catalytic activity observed, and this
correlation is supported by most recent theoretical simulations.[5]
The mechanically mixed samples (denoted with M) were prepared by mixing together metallic
Pd nanoparticles with CeO2 powder aged at 1173 K in a mini ball mill, to obtain a nominal Pd
loading of 1% wt. (PdCeM). Reference catalysts with the same nominal Pd loading were also
synthesized by incipient wetness impregnation on the same CeO2 support (PdCeIW) and by
solution combustion synthesis (PdCeSCS). An additional catalyst, where PdO nanoparticles
were used in substitution of Pd metal in the milling procedure, was also prepared for
comparison (PdOCeM). To check the effect of the support, comparison of PdCeM with Pd
supported on ZrO2 (PdZrM) prepared by the same procedure was also carried out. The details
on samples preparation and the milling parameters are reported in the Supporting
Information, along with the detailed description of testing conditions and characterization
methods. Stability under reaction conditions and durability were tested in comparison to
PdCeIW following six reaction cycles up to 1173 K and monitoring time on stream behavior
also under hydrothermal conditions. Comparison of X-ray diffraction and temperature
programmed reduction profiles of PdCeM, PdCeIW and PdOCeM are reported in Figure S1 and
S2. The light-off curves of methane combustion under lean atmosphere for the different Pd-
ceria formulations are shown in Figure 1, along with the corresponding cooling part of the
cycle.
In these experiments the catalysts were cycled under reaction conditions between room
temperature and 1173 K for 2 heating/cooling cycles and the second cycle was selected as
representative of the catalytic behavior, unless otherwise stated. Interestingly, the overall
performance of the M samples strongly depends on the nature of the palladium precursor.
PdCeM has a much higher activity than PdOCeM in the whole temperature range, and its
behavior outperforms significantly that of the impregnated sample and of the catalyst made by
solution combustion synthesis, which was reported as one of the best literature examples.[3b,
6] This can be seen either by the lower light-off temperature or by the higher reaction rate
measured for the PdCeM sample in a recycle reactor, as reported in Table 1.
Figure 1. Light-off curves for Pd-CeO2 catalysts (2nd cycle). Solid line, closed symbols: heating
part of the cycle; dashed line, open symbols: cooling part of the cycle. Conditions: GHSV ca.
200000 h-1, 0.5% CH4, 2% O2, He to balance.
Table 1. Physico-chemical properties and activity parameters for methane combustion.
Table 1. Physico-chemical properties and activity parameters for methane
combustion.
Sample
Pd loading
(wt%)
[a]
Surface area
(m
2
/g)
T10
[b]
(K)
Reaction rate
[c]
(mol/g
Pd
*s)
PdCeM
0.81
3.2
564
208
PdOCeM
0.80
4.1
700
16
PdCeIW
0.97
2.3
619
32
PdCeSCS
0.93
5.9
581
112
[a] measured by ICP elemental analysis; [b] temperature for 10% conversion;
[c] measured at 623 K in a recycle reactor.
The better performance of PdCeM is maintained also in the cooling portion of the cycle, where
the drop of activity due to the dynamics of Pd-PdO transformation follows the order
PdCeM<PdCeSCS<PdOCeM<PdCeIW. The catalyst stability has been successfully checked over
six light-off cycles (Figure S3) where it can be seen that PdCeM shows a stable CH4 conversion
behavior from the third cycle onward. Time on stream behavior has also been investigated
both under reaction conditions and in the presence of large excess of water (Figure S4), which
is known to accelerate deactivation in Pd-based catalysts.[8] Catalysts prepared by milling are
more stable with an overall activity loss of ca. 25% after 24 h on stream, compared with a loss
of 70% observed in reference PdCeIW.
Light-off activity behavior of PdCeM is also affected by the modification of milling parameters.
It is observed that, increasing milling intensity, the overall light-off profiles of the catalyst shift
to higher temperatures, indicating a drop of reaction rates compared to our standard PdCeM
sample (see Figure S5).
Also, milling under loose conditions without using balls does provide catalysts active in a first
cycle at low temperature but suffering of dramatic deactivation, due to the lack of nanoscale
interactions between Pd and ceria and extensive sintering at higher temperatures (Figure S6).
This shows that the choice of our milling parameters for PdCeM sample (10 minutes milling at
an oscillation frequency of 15 Hz with a ball to powder ratio of 10) optimizes activity
performances.
Methane oxidation light-off profiles were also collected on the corresponding ZrO2 supported
catalysts, and they did not evidence any significant difference among samples prepared by
impregnation and by mechanical milling (Figure S7). This strongly supports the fact that the
origin of the unique activity of PdCeM must be found in a specific characteristic of the Pd-CeO2
interface that is promoted during milling, and not in a more general behavior of samples
prepared by a mechanochemical procedure.
In parallel to light-off cycles, temperature programmed oxidation experiments (TPO, Figure 2)
showed for fresh PdCeM the oxidation of Pd at low temperature (cycle 1), and the presence of
at least three PdO decomposition peaks (cycles 2 and 3), indicating the coexistence of different
palladium species. Quantitative re-oxidation of Pd during cooling occurs in one single peak at
ca. 910 K. A comparison with PdCeIW, which presents only two oxygen release peaks, is shown
in Figure S8.
Figure 2. TPO profiles for PdCeM over three consecutive heating/cooling cycles. Solid line:
heating ramp; dotted line: cooling ramp.
Details on the Pd-CeO2 morphology were obtained by HRTEM analysis on the fresh samples.
Figure 3 (A,B) shows the surface of PdCeIW where, as expected, small Pd nanoparticles (ca. 2
nm) are well dispersed over a clean ceria surface. In contrast, PdCeM is characterized by ceria
crystallites that are covered by an amorphous layer measuring between 2 and 5 nm in
thickness(Figure 3 C-F). This amorphous shell is compact and perfectly defined, following the
perimeter of the ceria particles (Figure 3C-D). Interestingly, the ceria crystallites covered by
this layer present a more rounded morphology, suggesting that the spreading of Pd by
mechanical mixing affects their surface. Ceria nanoparticles show well-defined lattice fringes at
3.12 and 2.71 Å in both samples, which correspond to the (111) and (200) crystallographic
planes of CeO2, respectively. In Figure 3D the EDX analysis of the shell is reported, which
contains both ceria and Pd (the Cu signal originates from the TEM grid), indicating that the
shell is comprised of a mixed Pd-Ce phase.
In addition to this amorphous layer, some smaller particles measuring less than 5 nm are also
detected, mostly decorating the ceria crystallites (Figure 3E,F). Lattice fringe analysis of these
nanoparticles (see insets “c” and “d” corresponding to FT images of “a” and “b”, respectively,
in Figure 3F) shows fringes at 2.25 and 1.95 Å. They correspond to the (111) and (200)
crystallographic planes of Pd metal. Noticeably, these Pd nanoparticles are embedded in the
amorphous shell, which suggests that the shell is produced by the distribution of Pd metal
nanoparticles over the ceria support upon mechanical mixing. Figure 3E represents nicely this
situation: a very small nanoparticle (less than 2 nm) diffuses into the shell at the edges, still
preserving some crystallographic order (see the FT image in inset “a” showing fringes of Pd
metal).
Figure 3. HRTEM images of PdCeIW (A,B) and PdCeM (C-F).
Remarkably, this morphology is not observed on PdOCeM and PdZrM where well defined PdO
and Pd nanoparticles are respectively found over the support (see Figure S9 and S10 for
representative samples). Moreover, the mechanical milling of pure CeO2 under the same
conditions does not induce any change of the surface (Figure S11) indicating that surface
amorphisation takes place only upon mixing of Pd and ceria.
