Angewandte
Research Articles
Chemie
Further HAADF-STEM characterization of the Ru/
Fe O /Al O heterodimer sample after catalysis demonstrated
Supporting Information, Table S1. The best fit model to the
EXAFS data confirms that the fresh catalyst comprises both
metallic Ru and Ru oxide, and the coordination numbers are
consistent with the Ru being present as nanoparticles.
However, after reduction, the RuꢀO scattering path at
2
3
2
3
an unexpected structural transformation of the catalyst. Here,
we observed the presence of a core–shell structure, with
a high-contrast core surrounded by a lower-contrast shell
material of the nanoparticles supported on alumina (Fig-
ure 4). EDS-mapping suggested that the core material is
mostly composed of Ru, encapsulated by Fe and C, whereas
the shell mostly of Fe and O. These maps suggest that
ruthenium was encapsulated by the iron phase during the
reduction and/or catalytic reaction. EDS line scans (Support-
ing Information, Figure S10) and EDS tomography (Support-
ing Information, Videos S1, S2) of the fresh and the post-
catalysis heterodimer sample further demonstrate the core–
shell nature of the nanoparticles in the spent catalyst.
2.01 ꢀ disappeared, indicating that the sample completely
reduced to metallic Ru. At the same time, while the reduced
and the post-catalysis samples mostly consist of one dominant
contribution owing to RuꢀRu at 2.66 ꢀ, the two samples have
non-negligible contributions at 2.58 ꢀ (Supporting Informa-
tion, Figures S12–S14, Table S1). Previous studies identified
[
18]
that this distance corresponds to the RuꢀFe scattering path.
Thus, the fact that we see appearance of RuꢀFe contribution
in the reduced and the post-catalysis samples is consistent
with Ru encapsulation by Fe, in agreement with the EDS
characterization. The EDS maps in Figure 4 suggest that
while the core mostly comprised of metallic Ru, the presence
of the RuꢀFe contribution in the EXAFS modelling suggests
formation of alloying of Ru surface layers with Fe. While
XANES at the Ru edge does not show significant differences
between the reduced catalyst and metallic Ru, the EXAFS
analysis clearly indicates the appearance of the RuꢀFe
contribution after reduction of the catalyst (Supporting
Information, Figures S12–S14, Table S1). We believe that
the interaction between Ru and Fe is only pronounced in
EXAFS because it is a fraction of surface Ru atoms that
would interact with Fe, and the XANES spectrum at the Ru
K-edge is relatively insensitive to charge transfer (s!p). In
a Ru nanoparticle of an average size of 4.8 nm, the fraction of
surface atoms is about 16%. This value is consistent with the
obtained RuꢀFe coordination numbers (Supporting Informa-
tion, Table S1). EXAFS fitting at the Fe edge, however, did
not show FeꢀRu contribution in the reduced sample probably
owing to the even smaller fraction of Fe interacting with Ru
(ca. 3%) in the core–shell structure.
Figure 4. a) Representative HAADF-STEM image and b)to f) energy
dispersive X-ray spectroscopy maps of b) C, c) O, d) superposition of
Al, Fe, and Ru, e) Fe, and f) Ru in the post-catalysis heterodimer
sample.
We also observed similar core–shell structures in the pure
iron oxide sample after catalysis (Supporting Information,
Figure S9). Lattice fringe analysis clearly demonstrates that
the shells are crystalline, with lattice constants corresponding
to either g-Fe O3 or Fe O4 (Supporting Information, Fig-
To understand whether the encapsulation occurs during
the reductive pretreatment or during reaction, we reduced the
heterodimers at 3008C and characterized the sample using
TEM. These micrographs indicate the presence of thin
overlayers of an iron phase, in this case iron oxide (Support-
ing Information, Figure S11). This observed behavior is
consistent with a strong metal–support interaction effects
between a reducible oxide and Group 8 noble metals, where
the reduced oxide phase covers the surface of a supported
2
3
ure S9). The presence of an iron oxide shell evidenced by
(S)TEM and EDS in the spent catalysts is likely the result of
oxidation following exposure to air rather than a realistic
representation of the catalyst structure under reaction con-
ditions. Indeed, in situ Fe K-edge XANES showed that the
heterodimer sample predominantly consisted of metallic iron
and iron carbide phases with no observable contribution from
g-Fe O3 (Supporting Information, Figure S15). However,
2
[
17]
metal as a result of more favorable surface energy. Upon
during the in situ experiments we observed that upon cool
down in the reaction mixture, the contribution owing to
reduction of surface g-Fe O to FeO , the latter migrates on
2
3
x
the ruthenium metal surface, leading to encapsulation of the
ruthenium. Such transformation of ruthenium–iron oxide
heterodimers into core–shell structures has not been reported
before.
metallic iron decreased, while that of g-Fe O3 increased
2
(Supporting Information, Figure S15). Thus, we surmise that
the active phase of the catalyst consists mainly of metallic iron
and iron carbide phases which, upon exposure to air, oxidize
and form an iron oxide shell. This result is easily understood
given the susceptibility of iron to oxidation. EDS maps
(Figure 4) are consistent with this claim since we found a non-
negligible carbon signal arising from the core of the nano-
structures (Figure 4c). The overall characterization by com-
We further studied the Ru/Fe O /Al O sample by mod-
2
3
2
3
eling in situ Ru K-edge EXAFS data of the fresh, reduced,
and post-catalysis samples. The EXAFS data are shown in the
Supporting Information, Figures S12, S13 and the numerical
results of the modeling for each sample are presented in the
&
&&&
ꢀ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2019, 58, 2 – 9
These are not the final page numbers!