5
6
J. Xu et al. / Applied Catalysis A: General 514 (2016) 51–59
than that of Pd foil, indicating that Pd was highly dispersed on the
support [58], in good agreement with the TEM result. When the
reduction temperature was increased to 500 C, the Pd–Pd coordi-
conversion or the CH OH selectivity (see Fig. S2), indicating that
the catalyst possessed high stability.
3
◦
To further study the effect of Pd loading on the catalytic perfor-
mance, the detailed structures of these catalysts were examined.
As shown in Fig. 2, the PdZn phase could only be observed at
nation number decreased to 3.6, while that of Pd–Zn increased to
7
.3. The RPd–Pd elongated to about 2.87 Å, while the RPd–Zn remained
◦
at around 2.60 Å. Both the two bond distances were very close to
those in the intermetallic PdZn alloy. The NPd–Zn/NPd–Pd ratio of 2
suggested that the PdZn alloy was formed with a face-centered
tetragonal structure. This result provides a direct evidence of the
complete transformation of Pd into PdZn alloy when the catalyst
2ꢂ = 41.23 (indicating the (1 1 1) lattice plane) over the samples
with Pd loadings above 2 wt%. Reducing the Pd loading led to a
decrease in peak intensity. For the catalysts with Pd loadings lower
than 2 wt%, the PdZn phase was not detectable, likely due to the fact
that the crystal size of PdZn was less than 2 nm, which was beyond
the detection limit of XRD technique.
◦
reduced at 500 C.
To summarize this section briefly, the XRD (Fig. 2) and HRTEM
H -TPR profiles of the Pd/ZnO/Al O catalysts with different Pd
2 2 3
(
Fig. 3) results provided direct evidences of the formation of PdZn
loadings were depicted in Fig. 6. In most cases, the peaks below
80 C for all catalysts were ascribed to the reduction of PdO to
metallic Pd [62,63]. The amount of H2 consumption was calcu-
lated. The results showed that PdO could be completely reduced
at such temperatures. Meanwhile, a small amount of ZnO species
◦
alloy upon the reduction of Pd/ZnO/Al O catalyst. In situ DRIFT
2
3
◦
spectra revealed that the catalyst reduced at 300 C contained both
monometallic Pd and PdZn alloy, and the content of PdZn alloy was
estimated to be 34% based on the XAS analysis. The metallic Pd
was completely transformed into PdZn alloy when the reduction
temperature was increased to 500 C, which was accompanied by
an increase in methanol selectivity (see Fig. 1B and Table 3). For
example, the CH OH selectivity at 180 C was 79.4% over the cata-
lyst reduced at 300 C, while this value increased to 85.5% when the
reduction temperature was raised to 500 C. The change in selectiv-
ity could be explained by the successive transformation of metallic
Pd into PdZn alloy. It was proposed that during the Pd/ZnO reduc-
tion, PdO is first reduced, and then the adjacent ZnO is reduced
to form PdZn alloy [59–61]. The higher the reduction temperature
was, the greater extent of alloy would be formed, and hence the
enhanced methanol selectivity.
was also reduced. The ZnO/Al O3 support was also detected as ref-
2
◦
erence. It was found that the ZnO species started to be reduced at
◦ ◦
602 C on bare ZnO/Al O support, but this peak shifted to 562 C
2 3
◦
and even low temperatures when Pd was supported on the surface
of ZnO/Al O , which can be ascribed to the spillover of hydrogen
3
◦
2
3
◦
species from Pd to the support [64]. To be noted, the hydrogen
◦
consumption peaks appeared at high temperatures (>763 C) over
the Pd supported catalysts were resulted from the surface reaction
(methanation or rWGS) of H2 with CO , which derived from the
2
decomposition of the interlayer carbonates [36,65].
The curve-fitting results of EXAFS spectra were summarized in
Table 2, which provided the evidence of coexistence of Pd0 and
PdZn. Moreover, we found that a small amount of Pd–O species co-
existed in the 0.5 wt% and 2 wt% Pd catalysts. The H -TPR results
2
3.3. Effect of the Pd loading
indicated that PdO species could be completely reduced to metallic
Pd below 80 C. Therefore, the Pd–O species existed in the 0.5 wt%
◦
The influence of the Pd loading on the catalytic performance
and 2 wt% Pd catalysts were possibly resulted from the migration
of partially reduced ZnOx onto the surface of Pd particles at the
interface [32]. A further reduction of the catalyst at higher temper-
atures led to the formation of PdZn alloy, possibly in a pathway like
Pd/ZnO → Pd(Hy)/ZnO → Pd(Hy)/ZnOx/ZnO → PdZn/ZnO. Based on
this assumption, three kinds of Pd-containing species were identi-
fied in these catalysts: Pd in PdZn alloy, Pd modified by ZnOx and
was investigated. Prior to the experiment, the content of Pd was
determined by ICP spectrophotometry. The results showed that
the measured Pd loadings were slightly smaller than that of the
nominal composition (Table 4). It can be noted from Fig. 1 that
the decrease in Pd loading led to an increase in CH OH selectivity,
3
albeit at the expense of CO2 conversion. We further compared the
0
CH OH selectivity over the 2 wt%, 5 wt% and 7.5 wt% Pd/ZnAl-HT
unaffected Pd . However, the ratio of these Pd-containing species
3
catalysts with reactions conducted under similar CO2 conversion
levels by verifying the space velocity. As shown in Table S1, the
couldn’t be directly determined with XAFS characterization, due to
the existence of Pd–O band.
CH OH selectivities over all catalysts were within a similar range.
In summary of this section, the XRD results indicated that the
particle size of Pd species decreased with decreasing the Pd loading.
In addition to PdZn alloy, the EXAFS results further revealed that
ZnOx islands modified Pd species formed when the Pd loading was
3
The TOF values (calculated on the basis of the number of surface
Pd atoms on the ZnO/Al O3 support) of the catalysts with varied
2
Pd loadings were obtained by determining the Pd dispersion using
hydrogen–oxygen titration method. The data in Tables 3 and 4 show
that the Pd dispersion decreased with increasing the Pd loading. At
less than 2 wt%. However, the CH OH selectivity remained nearly
3
the same when the reaction was conducted under similar CO con-
2
◦
the same reaction temperature, i.e., at 220 C, the calculated TOF
versions. Therefore, it arrived at a conclusion that Pd modified by
ZnOx islands was also the active site for methanol synthesis.
Various hypotheses for the reaction mechanism have been pre-
sented during the past decades; however, there is still controversy
over some important issues. One of the major dispute topics is
related to whether methanol synthesis and RWGS reactions are
evolving in parallel pathways, sharing a common intermediate, or
whether the methanol formation proceeds by sequential RWGS and
CO hydrogenation reactions. By conducting H/D exchange experi-
ments, Kunkes et al. [66] proved that the methanol synthesis and
RWGS reactions proceeded on different surface sites in a parallel
manner. It was reported that, for the steam reforming of methanol
(MSR, the reverse reaction of methanol synthesis from CO2 and
hydrogen), the selectivity of the Pd catalysts was significantly
enhanced upon the formation of Pd alloys such as, Pd–Zn, Pd–Ga
and Pd–In by the reduction of Pd/ZnO, Pd/Ga O3 and Pd/In O3
−
2
−2 −1
s when
values increased slightly from 0.75 × 10 to 1.07 × 10
the Pd loading decreased from 5 wt% to 0.5 wt%. The apparent acti-
vation energies (Ea) of CO hydrogenation over these catalysts were
2
also estimated, as shown in Fig. S1. The estimated Ea values over the
5
wt% and 2 wt% Pd/ZnO/Al O catalysts were 59.9 and 64.8 kJ/mol,
2 3
respectively, very close to those reported previously [31,36]. How-
ever, the estimated Ea value over the 0.5 wt% Pd/ZnO/Al O catalyst
2
3
was 71.5 kJ/mol, slightly higher than that over other catalysts. Since
Ea was calculated based on CO2 conversion, which contained both
the methanol synthesis and the rWGS reaction. The higher Ea value
estimated over the 0.5 wt% catalyst than that estimated over the
5
wt% and 2 wt% catalysts might be attributed to the decreased CO
formation rate through rWGS reaction.
The stability experiment was tested over the 5 wt%
◦
Pd/ZnO/Al O catalyst reduced at 300 C. After 36 h time on
2
3
2
2
◦
stream at 200 C, no deactivation was observed in either the CO2
[67,68]. Lin et al. [69] reported several possible pathways of MSR