52
J. Sun et al. / Journal of Catalysis 306 (2013) 47–57
EXAFS Pd K-edge spectra for a Pd foil, 2Pd10Fe/C catalyst under
100% H2 at room temperature, and the fit for 2Pd/10Fe/C. It can
be seen from Fig. 5 that the spectrum for 2Pd10Fe/C differs signif-
icantly from that of a monometallic Pd with the appearance of a
peak at 2.15 Å (not phase corrected), suggesting the formation of
PdFe alloy. The data and fit using Pd–Pd and Pd–Fe scattering paths
are also shown in r-space in Fig. 5 and the fitting results are listed
in Table 2. The Pd–Pd bond distance (2.726 Å) was significantly
shorter than bulk Pd (2.751 Å), especially considering that the
spectrum was collected under 100% H2 which would promote the
formation of Pd-hydride and increase the Pd–Pd bond distance.
In addition, the fitted Pd–Fe coordination number was 2.7 and
the Pd–Fe bond distance (2.608 Å) was significantly shorter than
that of Pd–Pd (2.726 Å). These results clearly show that Pd formed
an alloy with Fe which confirms the STEM/EDS and TPR results.
Further information on the structure of the nanoparticles from
the EXAFS results cannot be extracted due to the broad particle size
and composition distributions (from STEM/EDS).
DFT calculations were used to better understand the interaction
between Pd and Fe. It has been reported that Pd presents a strong
surface segregation preference when acting as an atom impurity
within a Fe host while a Fe impurity atom within a Pd host shows
a preference for strong anti-segregation [63,64]. From this, it is ex-
pected that when the host material in a PdFe alloy is either Fe or
Pd, the surface composition will tend to favor an enrichment of
Pd. However, no theoretical work has been undertaken to deter-
mine if the surface segregation behavior is enhanced or inverted
when examining a layer of one type of metal as an impurity in
the other as a host. To this end, two configurations for the
Fe(110) surface were constructed. One scenario had a pure Pd
top layer and the other had a perfectly mixed PdFe alloy within
the top two layers, as shown in Fig. 1a. The surface segregation en-
ergy for the Fe(110) host system, as defined in Section 3, was
found to be ꢂ0.275 eV/Pd, which implies that Pd atoms prefer to
form a pure layer on the surface of the Fe(110) host rather than
mixing itself with the subsurface Fe atoms. This observation is in
line with our STEM/EDS observations on large-sized Fe particles
(20–30 nm). Similar calculations were performed with the
Pd(111) surface acting as the host material. Because the Fe impu-
rity within a Pd host shows a strong tendency toward anti-segrega-
tion, the Fe impurity layer will prefer to be located somewhere
within the host material, away from the surface. However, nothing
is known about the preferred Fe layer location within the subsur-
face Pd layers. Therefore, the most favorable subsurface layer in
which to place a pure Fe layer had to be determined. Calculations
were performed with the pure Fe layer in different layers within
the Pd(111) host, and these results showed that the most stable
configuration for the pure Fe layer was when it was placed in the
second layer of the Pd(111) host (see supporting materials). This
preferential configuration for the pure Fe layer within the Pd host
was then compared with a perfectly mixed system as examined for
the Fe(110) host case above. These configurations are shown in
Fig. 1b. For the Pd host situation, the surface segregation energy
was found to be ꢂ0.286 eV/Fe, showing that Fe atoms prefer to
form a pure layer in the second layer of Pd(111) surface. The sur-
face segregation results presented above confirm that the surface
of the PdFe alloy will be enriched in Pd with Fe beneath. These
observations differ somewhat to the STEM/EDS results for the very
small particles (2–3 nm) with Pd as host, in which Pd core and
Fe2O3 shell structure were observed. This could be due to the oxi-
dation of Fe which migrates to the surface of Pd under air during
STEM sample preparation.
To summarize, STEM characterizations showed that the
2Pd10Fe/C catalyst has both larger Fe particles modified by Pd
patches and smaller Pd–Fe nanoparticles without monometallic
Pd or isolated Fe being observed. The results from TPR, STEM/
EDS, and EXAFS characterization reveal the formation of Pd–Fe al-
loy for the 2Pd10Fe/C catalyst. Finally, DFT calculations suggest
that the surface of Pd–Fe alloy is enriched in Pd and the subsurface
is enriched in Fe.
4.2. Hydrodeoxygenation of guaiacol
The supported metal catalysts were evaluated at temperatures
from 250 °C to 450 °C under atmospheric pressure. Blank-reactor
showed a negligible activity for guaiacol at temperatures below
350 °C. When the reaction temperature is increased from 350 to
450 °C, the conversion of guaiacol to catechol is observed. How-
ever, this conversion was still very low, increasing from 0.2% to
6.3% with negligible deoxygenated products being formed
(Table S1). The pure carbon support also showed negligible activity
at 250 °C, but became significantly more active at higher tempera-
tures, e.g., reaching 29.1% and 65.1% guaiacol conversions at 350 °C
and 450 °C, respectively. The major products with this catalyst
were found to be catechol and phenol (Fig. S4). Separate experi-
ments at low conversions on the carbon support (data not shown)
showed that catechol is the primary product formed via demethyl-
ation of guaiacol, while phenol was a secondary product formed by
HDO of catechol. This observation suggests the facile demethyla-
tion of guaiacol in the absence of metal catalysts, consistent with
the weaker Me–OAr bond strength (247 kJ/mol, demethylation)
compared with the C(Ar)–OMe bond (356 kJ/mol, demethoxylation)
in the guaiacol molecule [12]. Therefore, although the carbon
support is active in converting guaiacol to catechol and phenol at
350–450 °C, it does not further catalyze the deoxygenation of
phenol as evidenced by the lack of benzene and toluene/trimethyl
benzene (TMB) produced at these temperatures.
The acidity of the support has been found to have a key effect on
the reaction of methyl substitution in HDO of guaiacol [17,19,42].
In this study, no detectable methylated guaiacol and only very
small amounts of cresol were observed. This is most likely due to
the significantly lower acidity of the carbon support when com-
pared with the acidity of metal oxides supports like Al2O3 and H-
beta [17–19]. These acidic, metal oxide supports can cause both
unimolecular and biomolecular transalkylation as well as HDO by
the metal catalysts, resulting in various methylated products such
as cresol, methylated guaiacol, toluene, and trimethylbenzene.
4.2.1. Carbon-supported precious metal catalysts
The carbon-supported metal catalysts exhibited significant
activity and yields to deoxygenated products (i.e., phenol, oxy-
gen-free aromatic compounds, cyclohexanone, and cyclohexanol)
as shown in Fig. 6 and Table S2. Among the monometallic pre-
cious metal catalysts at 250 °C (Fig. 6A), catalyst activity (as mea-
sured by guaiacol conversion) increased in the order of: 5Pd/C
(50.4%) < 5Pt/C (87.7%) < 5Ru/C (95.5%). Phenol was the major
Table 2
EXAFS fitting results for the 2% Pd–10% Fe/C catalyst at the PdK edge. EXAFS spectra were collected at 25 °C in 100% H2 flow. The numbers in parentheses indicate the statistical
error in the most significant digit obtained from the fit in Artemis (e.g., 5.2(7) ꢄ 5.2 0.7).
r2PdꢂPd (Å2)
r2PdꢂFe (Å2)
NPd–Pd
RPd–Pd (Å)
NPd–Fe
RPd–Fe (Å)
D
E0 (eV)
Reduced v2
2Pd10Fe/C
5.2(7)
2.726(5)
0.009(2)
2.7(6)
2.608(7)
0.005(2)
1.7(5)
9.7