J. Quesada et al. / Journal of Catalysis 377 (2019) 133–144
141
manufacturer), the analysis was focused on the platinum 4d
region, instead of the common utilized 4f one, to avoid the interfer-
ence with the aluminum 2p region characterized by similar bind-
ing energies of approximately 74 eV [39,40]. The binding energy
signal consists of two differentiated areas corresponding to the
deactivation rate constant, and
a
is the exponent that specifies
the dependence of the deactivation rate on the acetone concentra-
tion (C ) affecting deactivation. Considering: (i) first-order deactiva-
tion kinetics (d = 1); (ii) the independence of the deactivation rate
on the C = 0) because of the differential conditions of operation
(C almost constant with the reaction time); and (iii) the relation
A
A
(a
4
d doublet pairs (4d5/2 and 4d3/2). Both peaks observed at 313.9
A
0
and 330.9 eV are related to the 4d5/2 and 4d3/2 doublets of Pt
between the acetone reaction rate (r) and the catalyst activity (Eq.
(2)), the acetone reaction rate can be described as follows (Eq. (3)):
4
+
[
41]. In addition, an almost negligible peak associated to Pt arises
at 317.8 eV [42]. As a consequence, it is confirmed that the plat-
inum is present as Pt oxidation state as nanoparticles (Pt /Pt
r ¼ r
0
ꢁ a
ð2Þ
0
0
4+
XPS 4d5/2 area ratio of 16.4), in agreement with its easy reducibility
after thermal treatment even at low temperature [43]. Further-
more, no mixed-oxides crystallites of titanium and platinum are
formed comparing the nickel material. This fact could be explained
by two reasons: (i) as a noble metal, platinum is hardly involved in
mixed structures with other metals; and (ii) platinum atoms are
much larger than the titanium ones, hence hindering their inser-
r ¼ r
0
ꢁ expðꢂk
d
A^ ꢁ tÞ
ð3Þ
being r0 the acetone reaction rate at zero time. Furthermore, the
evolution of the acetone conversion can be depicted in terms of
the deactivation rate constant (Eq. (4)), taking into account the rela-
tion between the reaction rate and conversion through the space
velocity (SV) (Eq. (5)):
tion into the TiO
Regarding the titanium 2p region of the Pt/TiO
Fig. 9c), in contrast to the Ni/TiO , the same peaks as for the parent
TiO are noticed, otherwise they arise at lower binding energies
457.7 and 463.4 eV). This last fact is proposed to be directly
related to the presence of the well-distributed platinum nanopar-
ticles on the TiO . However, it can be assumed that the peak at
lower binding energy (i.e., 2p3/2 doublet) is formed by the contribu-
tion of two overlapping peaks (456.9 and 458.2 eV), suggesting
that unlike titanium species in terms of coordination exist on the
surface. This would be caused by the dissimilar degree of interac-
tion of the titanium atoms with the platinum nanoparticles (inter-
face and non-interface titanium atoms).
2
to generate a new hybrid phase.
x ¼ x
r ¼ SV ꢁ x
where x
0
ꢁ expðꢂk
d
ꢁ tÞ
ð4Þ
ð5Þ
2
XPS analysis
(
2
2
(
0
is the acetone conversion at zero time.
The calculated k values are included in Fig. 3. Regarding the
d
2
catalytic stability of the parent TiO
(0.058 ks ) decreases in 32% comparing to the value obtained in
absence of H
to the worse molecule-stabilization capacity of this material
because of the H presence, as explained above, which hinders
the strong adsorption of oligomers and coke formation on the cat-
alytic surface resulting in softer deactivation. Although differential
reaction conditions cannot be considered in this case when using
2
in presence of H
2
, the k
d
ꢂ1
ꢂ1
2
(0.086 ks ). This improvement might be related
2
Hydrogen chemisorption analyses of both Pt/TiO
are in line with the results previously observed by other techniques.
The H uptake per metal atom is two orders of magnitude lower in
the case of the Ni/TiO
for Ni/TiO and Pt/TiO
lower active metallic surface (1.33 and 51.96 m ꢁgmetal, for Ni/TiO
and Pt/TiO , respectively). Accordingly, the calculations obtained
2 2
and Ni/TiO
Pt/TiO
2
(high acetone conversion), the k
d
are determined in inert
ꢂ1
ꢂ4
ꢂ1
2
conditions (0.096 ks ) and in the reducing ones (2.5 ꢁ 10 ks ).
ꢂ29
ꢂ27
3
ꢂ1
2
(7.41 ꢁ 10
and 7.83 ꢁ 10
m H
2
ꢁatom
,
In the last case, value obtained is lower than the threshold deemed
ꢂ1
2
2
, respectively), therefore confirming its
by Herrmann and Iglesia (no deactivation when k
[11].
d
< 0.05 ks
)
2
ꢂ1
2
2
The analysis in the case of Ni/TiO
the co-presence of more than one deactivation causes. Concerning
the faster deactivation observed with the Ni/TiO comparing with
parent TiO in inert conditions, it would be due to the Ni sites
(resulting from Ni insertion on the TiO surface) that act as Lewis
2
is more complex, because of
in terms of mean particle size and metal dispersion cannot be con-
sidered since, as it was observed, not all the nickel atoms are taking
part in metallic nanoparticles. However, the mean particle size and
2
2+
2
metal dispersion determined for the Pt/TiO
respectively) are very similar to those ascertained by HRTEM
1.6 nm and 70%, respectively). Furthermore, another interesting
2
(1.5 nm and 75%,
2
acid sites but with lower strength that the corresponding Ti sites
that they are replacing. Following this line, the acid-basic pairs in
(
2
+
difference between both bifunctional catalysts is noticed by doing
a second hydrogen chemisorption cycle. Whereas the capacity of
which Ni takes part show lower ability to favor the dehydration
of the aldol by the E mechanism (Fig. S8; step (iii) and (iv)), pro-
2
Pt/TiO
2
to chemisorb hydrogen seems to be kept in the second cycle
moting the stable adsorption of aldol intermediates. This phe-
nomenon must be added to the common deactivation process of
ꢂ27
3
ꢂ1
with only 3% of loss (7.58 ꢁ 10
Nm H
2
ꢁatom ), the chemisorbed
ꢂ29
hydrogen is reduced in 20% in the case of the Ni/TiO
2
(5.91 ꢁ 10
2
the acid-basic surfaces (TiO , in this case) in this type of reaction
3
ꢂ1
Nm H
2
ꢁatom ). This drop of the capacity to chemisorb hydrogen
because of subsequent aldol condensation and oligomerization.
2
+
at room temperature (308 K in this case) in subsequent analysis
cycles is closely associated with a phenomenon known as Strong
Metal-Support Interaction (SMSI) [44]. As it was reported [44],
the SMSI is quite common for metal-supported titania catalysts
and, mainly, considered as simple site blocking when nickel as
metal, but being able to contribute to lower catalytic activity.
Due to the low proportion of Ni sites, the former contribution
only affects deactivation at low TOS. In fact, the assumption of
these two deactivation mechanisms in the case of the Ni/TiO
2
in
absence of H is confirmed by the lines of fit for the acetone reac-
2
tion rate with the time-on-stream (Fig. 3). It can be noticed that
below 20 ks the overall deactivation would be due to the combina-
tion of the two mechanisms proposed. However, over 20 ks, it
2
+
seems that all the Ni sites are deactivated, and therefore only
the stable adsorption on TiO support is the responsible of the
deactivation. In fact, the deactivation rate constant (k ) above 20
3.3. Analysis of the catalytic stability
2
d
In order to model and evaluate the catalyst deactivation operat-
ing in both presence and absence of H
kinetics was considered (Eq. (1)) [45]:
ks can be considered the same than that corresponding to parent
2
, a power-law deactivation
ꢂ1
TiO
2
in inert conditions (0.085 and 0.086 ks , respectively). The
experimental confirmation of these hypotheses requires a deep
analysis of the evolution of solid surface at reaction conditions.
TPO and DRIFTS experiments were realized with the aim of
completing the gas-phase assessment by analyzing the solid-
phase (i.e., the chemical species adsorbed on the catalyst surface
a
d
r
d
¼ ꢂda=dt ¼ k
d
ꢁ C ꢁ a
ð1Þ
A
where r is the deactivation rate, a is the catalyst activity, t is the
time-on-stream (TOS), d is the deactivation rate order, k is the
d
d