Reduction enhancement of Fe2O3 in physical mixtures with Pt/mordenite via Pt migration or 'hydrogen spillover'

Article abstract of DOI:10.1039/a800200b

Mixing Fe₂O₃ with Pt/NaMor of similar grain size, followed by grinding in a mortar and calcination in O₂, leads to a remarkable enhancement of the reducibility of the Fe₂O₃ with hydrogen. The TPR profile of such mixtures is virtually identical with that of Fe₂O₃ onto which Pt was deposited chemically. It is concluded that in the ground and calcined mixtures Pt migration from the zeolite to the iron oxide is crucial. Upon varying the amount of deposited Pt in Pt/Fe₂O₃ between 0.001% and 1%, TPR profiles are obtained showing two discrete peaks characterizing a Pt promoted and an unpromoted reduction of Fe₂O₃ respectively. No Pt migration occurs in mixtures of prereduced Pt/NaMor with Fe₂O₃; this shows that surface migration of Pt⁰ clusters is negligible, but transport of PtO₂ either through the gas phase or via the surface is likely. Pt migration is also detectable at room temperature in mixtures stored for weeks in a moist atmosphere; in this case the data suggest surface migration of hydrated Pt²⁺ ions; the TPR profiles are distinctly different from those of the mixtures calcined in O₂. TPR also permits discrimination between the promotion of oxide reduction by migrating Pt and 'true' hydrogen spillover. The latter phenomenon requires transport of H atoms via protons and electrons and is realized with powder mixtures containing a semiconducting oxide, such as TiO₂. Its TPR signature is a broad peak located between those for unpromoted and Pt promoted reduction. Physical mixtures of Fe₂O₃ and Pt/NaMor catalyze the reduction of acetic acid vapor to acetaldehyde via a Mars-van Krevelen mechanism. In this case Pt migration helps to regenerate oxygen vacancies in the Fe₃O₄ surface, whereas direct contact of CH₃CO₂H vapor with Pt results in the formation of methane and higher hydrocarbons. The promoting effect of Pt is not observed after prereduction of Pt/NaMor, because Pt⁰ does not migrate effectively under the conditions used.

Full text of DOI:10.1039/a800200b

View Article Online / Journal Homepage / Table of Contents for this issue
Reduction enhancement of Fe O in physical mixtures with
2 3
Pt/mordenite via Pt migration or “hydrogen spilloverÏ
Gerd Frohlich and Wolfgang M. H. Sachtler*
V . N. Ipatie† L aboratory, Center for Catalysis and Surface Science, Department of Chemistry,
Northwestern University, Evanston, IL 60208, USA
Mixing Fe O with Pt/NaMor of similar grain size, followed by grinding in a mortar and calcination in O , leads to a remark-
2
3
2
able enhancement of the reducibility of the Fe O with hydrogen. The TPR proÐle of such mixtures is virtually identical with that
2
3
of Fe O onto which Pt was deposited chemically. It is concluded that in the ground and calcined mixtures Pt migration from
the zeolite to the iron oxide is crucial. Upon varying the amount of deposited Pt in Pt/Fe O between 0.001% and 1%, TPR
proÐles are obtained showing two discrete peaks characterizing a Pt promoted and an unpromoted reduction of Fe O respec-
tively. No Pt migration occurs in mixtures of prereduced Pt/NaMor with Fe O ; this shows that surface migration of Pt0
clusters is negligible, but transport of PtO either through the gas phase or via the surface is likely. Pt migration is also detectable
at room temperature in mixtures stored for weeks in a moist atmosphere; in this case the data suggest surface migration of
2
3
2
3
2
3
2
3
2
hydrated Pt2` ions; the TPR proÐles are distinctly di†erent from those of the mixtures calcined in O . TPR also permits
2
discrimination between the promotion of oxide reduction by migrating Pt and “trueÏ hydrogen spillover. The latter phenomenon
requires transport of H atoms via protons and electrons and is realized with powder mixtures containing a semiconducting oxide,
such as TiO . Its TPR signature is a broad peak located between those for unpromoted and Pt promoted reduction. Physical
2
mixtures of Fe O and Pt/NaMor catalyze the reduction of acetic acid vapor to acetaldehyde via a MarsÈvan Krevelen mecha-
2
3
nism. In this case Pt migration helps to regenerate oxygen vacancies in the Fe O surface, whereas direct contact of CH CO H
3
4
3
2
vapor with Pt results in the formation of methane and higher hydrocarbons. The promoting e†ect of Pt is not observed after
prereduction of Pt/NaMor, because Pt0 does not migrate e†ectively under the conditions used.
Many catalytic and solid state reactions require surface trans-
port of intermediates across phase boundaries. For instance,
the reduction of an oxide such as WO or Fe O with hydro-
gen is signiÐcantly accelerated by the presence of Pt particles
that are deposited on the oxide. For such systems it is gener-
presence of a metal such a platinum or nickel, even if this
metal or its precursors is separated from the reducible oxide
by materials for which criteria (i) and (ii) appear to apply.8h12
Although the term hydrogen spillover is often used to describe
the observed acceleration of oxide reduction, this term is not
always justiÐed and a better deÐnition of the migrating species
appears desirable. In a previous study we showed that the
reduction of sulfate groups on zirconia is markedly enhanced
when this material is present in a physical mixture with zeolite
supported Pt clusters, but there remained uncertainty to what
extent this phenomenon could be ascribed to true spillover in
the rigorous deÐnition given above.13 We have also shown
that reduction of metal ions in zeolite cavities is enhanced by
the presence of ions of a second, more easily reducible
metal,14h16 but the precise nature of the atomic steps has not
yet been unravelled.
3
2 3
ally assumed that H is dissociatively chemisorbed on the Pt
2
surface, H atoms subsequently cross the phase boundary
between Pt and the oxide and attach themselves to O2~ ions,
while an electron is added to the transition metal cation:
H
] O2~wMn` ] H`wO2~wM(n~1)`
(1)
ads
Early observations of this phenomenon were reported by
Sinfelt and Lucchesi,1 Khoobiar2 and Kuriacose.3 Boudart
introduced the term “hydrogen spilloverÏ 4 which is now gen-
erally used. The literature on this subject is reviewed in several
papers.5h7
In some cases, clear evidence exists that the migrating
species is actually the transition metal (TM) or its precursor.
For instance, work of this group showed that co-reduction of
Al O supported Re-oxide and Pt-oxide is entirely controlled
This simple scheme breaks down, however, when the source
of the H atoms and the reducible oxide are separated by a
material characterized by (i) strongly endothermic adsorption
of hydrogen atoms and (ii) negligible semiconductivity.
2
3
by the surface mobility of the TM ions. If they are immobile,
TPR gives two separate peaks, one for Pt at low and one for
Re at higher temperature. However, if both ions can migrate
over the surface, only one TPR peak is observed, and is
located at low temperature. In this case, and only in this case,
are the reduced particles identiÐed as PtRe alloy clusters.17
Recently, hydrogen spillover was assumed by Pestman et
al.18,19 to be crucial for the catalytic reduction of acetic acid
to acetaldehyde over physical mixtures of, for instance, unsup-
ported Fe O and SiO supported platinum. In this case,
Materials for which these criteria are valid include SiO ,
2
Al O and zeolites. The Ðrst criterion prevents transport of H
2
3
atoms as chemisorbed but mobile H , because their surface
concentration is prohibitively low at low temperature. An
ads
alternative path transports H atoms via protons and electrons.
Indeed, the mobility of protons is high over surfaces exposing
both O2~ ions and OH~ groups, but the transport of elec-
trons requires semiconductivity which can be translated, for
pure oxides, as thermodynamic coexistence of cations in two
valence states Mn` and M(n~1)`. In the absence of impurities,
SiO , Al O and zeolites are thought to be electric insulators;
2
3
2
spiltover H atoms are assumed to partially reduce the surface
of the iron oxide; the acetic acid molecules re-oxidize it, in a
Mars-van Krevelen mechanism, by depositing one of its
oxygen atoms into an oxide vacancy:
2
2 3
consequently the second criterion seems to exclude them from
the group of prospective candidates for hydrogen spillover at
low temperature.
Still, there is a large amount of literature evidence showing
that chemical reduction processes are often accelerated by the
2 H
spiltover
] O2~ ] K
] H O
(2a)
surface
surface
2
J. Chem. Soc., Faraday T rans., 1998, 94(9), 1339È1346
1339

View Article Online
CH CO H ] K
] CH CHO ] O2~
surface
(2b)
(2c)
with a ramp of 1 K min~1 up to 723 K. During this treatment,
3
2
surface
3
Fe O was reduced to Fe O , the actual catalyst as shown by
2
3
2 4
H ] 2 H
Pestman et al.18 Subsequently, the temperature was decreased
with the same ramp to 473 K. The reaction proÐle was record-
ed during the entire temperature cycle.
2
spiltover
which adds up to:
The kinetics of Fe O reduction with a 5% H in Ar gas
CH CO H ] H ] CH CHO ] H O
(2d)
2
3
2
3
2
2
3
2
mixture was studied as a temperature programmed reduction
(TPR) proÐle with a Ñow rate of 30 ml min~1 and a tem-
perature ramp of 8 K min~1. The hydrogen uptake was moni-
tored by a thermal conductivity detector. In the mixtures the
content of Fe O was always kept constant at 50 mg, while
This observation potentially opens challenging perspectives
for selective catalysis, as geometric arrangements are conceiv-
able in which Pt acts exclusively as a source of hydrogen spill-
over, whereas the acetic acid is prevented from contacting the
Pt particles directly and thus undergoing undesired reactions.
In the present work we Ðrst ascertain whether physical mix-
tures of the type Fe O ] Pt/zeolite have the same catalytic
2
3
the total amount of the sample varied. Since the pretreatment
of the samples is an important issue in this research, these
conditions will be speciÐed when the results are presented.
For transmission electron microscopy (TEM) of the
samples, a Hitachi HF-2000 analytic electron microscope
operated with 200 keV and equipped with an energy disper-
sive X-ray (EDX) analyser was used.
2
3
propensity for the selective reduction of acetic acid to acetal-
dehyde as the system Fe O ] Pt/SiO described by Pestman
2
3
2
et al.18 It is further attempted to di†erentiate between (i) true
H spillover, i.e. migration of H atoms over the zeolite and the
iron oxide surfaces and (ii) migration of the transition metal.
In the latter case it appears desirable to identify the
migrating species; are these Pt metal clusters Pt oxide clusters
or Pt ions?
Results
E†ect of platinum on Fe O in the selective reduction of acetic
3
4
acid
In Fig. 1 the catalytic reduction of acetic acid is compared for
Fe O and Pt/Fe O as the catalyst. The partial pressures of
acetic acid and the reaction products are plotted vs. tem-
Experimental
3
4
3 4
Pt/NaMor (0.63 wt.% Pt-load; Mor \ Mordenite) was pre-
pared by ion exchange of a diluted aqueous solution of
Pt(NH ) (NO ) (Alfa Aesar) with a NaMor slurry [200 ml (g
perature. Over Fe O [Fig. 1(a)] the main products at high
3
4
temperature are hydrocarbons, especially methane, and
carbon monoxide. Acetaldehyde is produced with a selectivity
of only 20% at a maximum which is reached at 670 K.
Impregnation of Fe O with Pt [Fig. 1(b)] leads to an overall
3 4
3 2
zeolite)~1]. The zeolite used was LZ-M-5 (Si : Al \ 5.6) from
UOP. The ion exchanged catalyst was dried in air and subse-
quently calcined in pure O (Ñow rate: 500 ml min~1; heating
3
4
2
increase in activity and also an increase in selectivity towards
acetaldehyde. Additionally, the formation of ethanol is
rate 0.1 K min~1 to 783 K). At this stage it was stored over
NH Cl solution and used for further experiments.
4
Fe O (hematite, J. T. Baker Chemicals; 99.5% purity) and
2
3
TiO (P25, Degussa) were used as supplied. Pt/Fe O and
2
2 3
2 3
Pt/TiO were prepared by wet impregnation of Fe O and
2
TiO with an aqueous solution of Pt(NH ) (NO ) (Alfa
3 4 3 2
Aesar) and subsequent drying in air and calcination in O
2
2
(783 K). Five samples of Pt/Fe O with 1%, 0.1%, 0.03%,
0.01% and 0.001% Pt-loading were prepared using consecu-
tively diluted solutions. Pt/TiO contained 1 wt.% Pt. Fe O
2
3
2
3 4
and Pt/Fe O (1 wt.% Pt-loading) were synthesized by iso-
3
4
thermal reduction of Fe O and Pt/Fe O in H at 623 K and
2
3
2
3
2
573 K, respectively.
The used Fe O has a BET surface of 8.1 m2 g~1, which
2
3
leads to an average particle size of 140 nm. The size distribu-
tion of Fe O crystallites observed by TEM were 50 nm to
2
3
250 nm, which is in good agreement with the value obtained
from BET. It follows that microporosity can be neglected for
these particles. However, crystallites will agglomerate to form
larger particles of 1È2 lm diameter with mesopores.
Experiments were done using either one component cata-
lysts or physical mixtures of two or three components. The
two component mixtures were prepared by mixing and thor-
oughly grinding amounts of equal weight. When desired, a
third component was added afterwards and the entire mixture
ground again.
The catalytic studies were carried out in a standard plug-
Ñow microreactor, comprising a gas supply with an impurity
trap, a saturator to admit liquid, a temperature controlled
plug-Ñow microreactor and a two column gas chromatograph
(50 m HP-PONA capillary column with FID; Alltech-
Chemipack C18 packed column with TCD). The feed gas of
H
(60 ml min~1) contained 20È25 mbar CH CO H after
2
3
2
saturation. The amount of catalyst was 200 mg if using a
single component and 400 mg in the case of physical mixtures.
The reaction was studied in a temperature programmed way.
Fig. 1 Catalytic conversion of acetic acid vapor: (=) CH CO H,
3
2
(+) CH CHO, (…) CH CH OH, (L) CH , (|) CO, (È) C . (a)
3
3
2
4
2`
After an initial treatment at 473 K for 30 min in H and 30
Catalyst: unsupported Fe O ; (b) catalyst: Pt/Fe O produced by Pt
2
2 3
impregnation onto Fe O .
2 3
min in the reaction mixture the temperature was increased
2
3
1340
J. Chem. Soc., Faraday T rans., 1998, V ol. 94

View Article Online
observed. These results conÐrm the earlier observations of
Pestman et al.18
In a second experiment, a physical mixture of Fe O with
3
4
Pt/NaMor was probed as the catalyst. It is seen from Fig. 2(a)
that the conversion of acetic acid at low temperature is signiÐ-
cantly higher than with Fe O or Fe O impregnated with Pt.
3
4
3 4
Acetaldehyde is a major product, its yield peaks at 600 K.
Above 650 K the predominant reaction products are CO and
methane, other hydrocarbons are also formed. Fig. 2(b) shows
the reaction proÐle for Pt/NaMor. Over this catalyst,
methane, CO and small amounts of higher hydrocarbons are
the only products. It is evident that the catalytic proÐle over
the mixture of Fe O and Pt/NaMor is not a superposition of
3
4
the proÐles over the single components. A signiÐcant inter-
action between the two components has to be considered.
Understanding the nature of this interaction will be the main
objective of the research described below.
Fig. 3 TPR proÐles of Fe O reduction to Fe O . Curve A: pure
2
3
3 4
Fe O ; curve B: Pt/Fe O with 1 wt.% Pt deposited by impregna-
2 3
tion.
2 3
TPR proÐles of Fe O : the e†ect of added platinum
2
3
If some chemical interaction between Pt in Pt/NaMor and the
iron oxide is responsible for the non-additive behavior of that
mixture in the catalytic test, such interaction should also be
instrumental in the reduction of the Fe O with H to Fe O .
At Ðrst, the reduction behavior of pure Fe O under TPR
2
3
conditions was studied. The proÐle (Fig. 3 curve A) shows an
asymmetric reduction signal with a maximum at 670 K. In
this and the following TPR experiments the area under this
reduction peak corresponds within 10% error to the hydrogen
consumption for the reaction:
2
3
2
3 4
In general, a mere reduction is less complicated than a cata-
lytic network that includes other steps than local reduction of
the catalyzing oxide. Reduction of Fe O starts at the surface
2
3
of this solid and requires sites where H molecules can disso-
2
ciate. Platinum might be helpful in this step, if it is in direct
3 Fe O ] H ] 2 Fe O ] H O
(3)
2
3
2
3
4
2
physical contact, so that classical spillover to the oxide is pos-
The increasing hydrogen consumption above 700 K indicates
sible. This propensity was already shown for Pd/Fe O .20
2
3
further reduction to Fe O and Fe0. This reduction behavior of
Whether Pt clusters that are encaged inside an insulating
material, such as mordenite, can exert an e†ect on the
reduction of Fe O is not obvious a priori.
x
Fe O was observed several times before.20,21 In curve B of
2
3
Fig. 3 the reduction proÐle of impregnated 1% Pt/Fe O is
2
3
2
3
presented. The presence of Pt at the surface leads to a marked
lowering of the temperature of the maximum in the reduction
proÐle to 560 K. This indicates that the ability of metallic Pt
to dissociatively adsorb hydrogen is helpful; as both solids are
in direct physical contact the result can be explained in terms
of the classical hydrogen spillover model.
Fig. 4 shows reduction experiments of physical mixtures of
Fe O with Pt/NaMor. Curve A represents the proÐle of a
2
3
freshly prepared mixture with no further initial pretreatment.
In this mixture the Pt inside the zeolite is in its oxidized state,
Fig. 4 TPR proÐles of physical mixtures of Fe O with Pt/NaMor.
2
3
Fig. 2 Catalytic conversion of acetic acid vapor: (=) CH CO H,
Curve A: freshly ground mixture; curve B: mixture after calcination
3
2
(+) CH CHO, (…) CH CH OH, (L) CH , (|) CO, (È) C . (a)
2`
Catalyst: physical mixture of Fe O ] Pt/NaMor; (b) catalyst: Pt/
on O at 673 K for 10 min; curve C: physical mixture that was not
3
3
2
4
2
ground but merely shaken; curve D: mixture after storage for several
2
3
NaMor.
weeks.
J. Chem. Soc., Faraday T rans., 1998, V ol. 94
1341

View Article Online
as the Pt/NaMor had been calcined to 783 K in pure O . The
2
physical mixture is reduced under the same TPR conditions as
the samples mentioned above. A very small signal indicating
the reduction of Pt is hardly visible because of the very low Pt
loading. ProÐle A shows two reduction peaks of Fe O at
2
3
550 K and 660 K. The second peak, which corresponds nicely
to the signal of pure Fe O in Fig. 3, curve A, is much
2
3
more prominent than the Ðrst one, indicating that the major-
ity of Fe O does not take any kinetic advantage of the pres-
2
3
ence of Pt inside the zeolite.
This situation changes completely when the physical
mixture is calcined in pure O at 673 K for 10 min prior to
2
the TPR run, as can be seen in Fig. 4, curve B. Here, the most
prominent peak is the low temperature reduction signal with
its maximum at 560 K corresponding to curve B in Fig. 3.
Only a tiny signal remains at 660 K. Now the system behaves
like Pt in direct contact with Fe O .
2
3
To clarify the e†ect of the grinding procedure during prep-
aration of the physical mixture, the next experiment was
carried out after gentle mixing of the sample. The components
were ground separately and mixed inside the reactor by
Fig. 5 TPR proÐles of Pt/Fe O with various amounts of Pt. Curve
2
3
merely shaking it for 5 min. This mixture was calcined in O
A: 1 wt.% Pt; curve B: 0.1 wt.% Pt; curve C: 0.03 wt.% Pt; curve D:
0.01 wt.% Pt; curve E: 0.001 wt.% Pt.
2
in the same way as described above. Curve C in Fig. 4 shows
its TPR proÐle. Clearly, the main reduction takes place at
high temperatures (660 K), only a small reduction signal is
is located at 660 K. A possible model is that some fraction of
the Fe O is in efficient contact with Pt, while the other frac-
tion is not. With decreasing Pt content the former fraction
decreases and the latter increases.
visible at 550 K. It follows that calcining in O at 673 K
2
2
3
though necessary, is not sufficient to develop the strong
reducibility enhancement that was manifest in curve B. As
gently shaking will lead to less intimate contact between the
two components, and on average a much lower contact area
will be established, this experiment indicates that close prox-
imity between Fe O and Pt/NaMor is essential for the Pt
Even with 0.001 wt.% Pt there is still low temperature
reduction visible. The reducibility of Fe O appears to be a
2
3
highly sensitive phenomenon, registering even very small
amounts of Pt on its surface.
2
3
induced reduction enhancement. Therefore all further samples
with physical mixtures were prepared by grinding.
Diluting experiments
The most surprising observation was made when a physical
mixture was examined which had been stored after mixing for
several weeks at room temperature under atmosphere. The
TPR of this sample was recorded without previous calcina-
tion; the result is presented in curve D of Fig. 4. Here, the
proÐle shows two maxima at 560 K (sharp) and at 610 K
(broad). Comparing this to the initial state immediately after
mixing (Fig. 4, curve A) indicates that even at room tem-
perature an interaction between the two components takes
place when they are in close contact. Since this interaction is
not immediately observable after mixing, a slow chemical
reaction has to be considered. One possibility is the migration
of Pt ions or Pt-oxide clusters out of the zeolite onto the iron
oxide. This is, indeed, strongly suggested by the identical
reduction proÐle of the mixture after calcination (Fig. 4, curve
B) and the Pt/Fe O sample (Fig. 3, curve B) where Pt had
Diluting a physical mixture with an inert third solid provides
a statistical means to increase the average distance between
the component particles. If Pt migrates only in the stage
before the third component is added, one would expect that
the position of the TPR signals is una†ected by the diluent.
Therefore, experiments were designed in which the samples
were diluted with a tenfold excess (with respect to the Fe O
component) with NaMor. Preliminary tests indicated that if
using mordenite as the third component such ternary mixtures
had to be prepared in a dry atmosphere using precalcined
2
3
NaMor (673 K in O ) in order to minimize the e†ects of
2
physisorbed water. The Ðrst TPR proÐle (Fig. 6, curve A)
shows the behavior of 0.1% Pt/Fe O diluted with NaMor.
2
3
Since in this case Pt is known to be in contact with Pt/Fe O
2
3
we expect the reduction signal to be near 550 K. This is con-
Ðrmed by experiment, the TPR peak is located at 560 K, only
10 K higher than in the undiluted sample (Fig. 5, curve B).
This small di†erence can be attributed to the e†ect of physi-
sorbed water in NaMor or to a di†erent temperature distribu-
tion in the reactor due to the larger volume of the sample.
In curve B of Fig. 6 the TPR proÐle is presented of a diluted
Pt/NaMor ] Fe O mixture, which was calcined inO at 673 K
2
3
been deposited onto Fe O .
2
3
To detect a possible Pt migration onto iron oxide, TEM
studies with EDX were performed, however, no evidence was
found for the presence of Pt on Fe O . If Pt has indeed
2
3
migrated out of the zeolite onto the iron oxide, its presence
there is below the detection limit of these methods.
2
3
2
for 10 min prior to dilution. Again, the maximum of the
reduction signal is at 560 K, thus only slightly higher than
that of the undiluted mixture (Fig. 4, curve B). This is, again,
E†ect of the amount of platinum in Pt/Fe O
2
3
To get an idea as to the amount of Pt necessary to give
enhanced reduction of Fe O , series of impregnated
Pt/Fe O samples were prepared, having Pt contents of 1 wt.%,
a
strong evidence that during calcination in O some Pt
2
3
2
migrates to Fe O . Moreover, the proÐles of curve A and B
2
3
2 3
0.1 wt.%, 0.03 wt.%, 0.01 wt.% and 0.001 wt.%, respectively.
The TPR proÐles of these Ðve samples are shown in Fig.
5AÈE. It is obvious that there are two distinct reduction
domains, those which are inÑuenced by the presence of plati-
num at the surface (540È560 K) and those which are not (650È
660 K). Remarkably, there is no intermediate reduction
domain. Reduction is either enhanced by Pt and then gives a
TPR peak at 550 K, or it is not enhanced and the TPR peak
are very similar, suggesting that the amount of migrated Pt is
close to 0.1 wt.% with respect to Fe O .
The situation is more complicated in the case of physical
mixture that had been stored for extensive periods. Curve C
shows the proÐle of a diluted physical mixture after storage
for several months. In this case the reduction signal becomes
2
3
very broad, with
a smooth transition into the further
reduction of iron oxide. The low temperature maximum was
1342
J. Chem. Soc., Faraday T rans., 1998, V ol. 94

View Article Online
Fig. 6 TPR proÐles of ternary mixtures Fe O ] Pt/NaMor
Fig. 7 TPR proÐles of physical mixtures of Fe O ] Pt/TiO .
2
3
2
3
2
] NaMor. Curve A: 0.1% Pt/Fe O ] NaMor; curve B: Fe O
Curve A: Fe O ] Pt/TiO ; curve B: Fe O ] Pt/TiO ] tenfold
2
3
2
3
2
3
2
2
3
2
] Pt/NaMor calcined at 673 K prior to dilution with NaMor; curve
C: Fe O ] Pt/NaMor after storage for three months prior to dilu-
excess of TiO ; curve C: Fe O ] TiO .
2
2
3
2
2
3
tion with NaMor.
the TPR proÐle of a physical mixture of Pt/Fe O with
3
4
Fe O . Again the maximum reduction peak is observed at
2
3
near 620 K, which is 50 K higher than for the undiluted mix-
tures presented in Fig. 4, curve D.
540 K, due to the presence of platinum. In contrast, pure
Fe O gives no enhanced reduction of Fe O , discernible by
the high temperature reduction maximum at 660 K (Fig. 8,
curve C).
3
4
2 3
E†ect of “trueÏ hydrogen spillover on TPR proÐles
Again, the decisive clue is obtained from the diluted
mixture. Fig. 8, curve B shows the TPR proÐle of a physical
mixture of Pt/Fe O and Fe O with a tenfold excess of pure
All results described thus far can be rationalized in terms of a
model assuming migration of platinum. So far, it is not clear if
also migration of hydrogen atoms plays a role as in the clas-
sical model of hydrogen spillover. Therefore the following
experiments were carried out that focus on this possibility.
2
3
2 3
Fe O . As already observed in the experiments with TiO the
3
4
2
initial reduction signal at 540 K decreases and a new signal at
610 K shows up. Since Fe O is also a semiconducting
There is ample evidence in the literature22,23 that TiO , a
3
4
2
material, this can be interpreted again as the e†ect of hydro-
gen spillover.
semiconducting material, has the ability to transport spiltover
hydrogen atoms, i.e. protons and electrons. If Pt is supported
on TiO , it can be expected that a “trueÏ hydrogen spillover
2
Experiments with prereduced Pt/NaMor
e†ect should occur in physical mixtures with Fe O . The
2
3
question is, how would this be manifest phenomenologically
in TPR proÐles? Therefore, TPR experiments were designed
with physical mixtures of Fe O with Pt/TiO .
In previous TPR experiments concerning physical mixtures of
Pt/NaMor and Fe O there are indications that there is
2
3
2
3
2
migration of Pt, whether as oxide or as ions or as metal par-
ticles, from the zeolite to the iron oxide. To examine the last
possibility, the Pt/NaMor compound has to be reduced.
Fig. 7, curve A shows the proÐle of this mixture. The main
reduction peak is observed at 560 K, which corresponds
exactly to that observed for Pt/Fe O . To make sure that
2
3
TiO itself does not enhance the reduction of Fe O , a physi-
cal mixture of Fe O with pure TiO has been investigated.
Fig. 7, curve C shows the result. Now the reduction maximum
is observed at 690 K, which can be assigned to the reduction
of pure Fe O , not inÑuenced by TiO .
2
2 3
2
3
2
2
3
2
Since the Pt loading of TiO could be partially located on
the outer surface, this could lead to a direct contact of Pt and
2
Fe O , resulting in the behavior observed in Fig. 7, curve A.
2
3
To reduce this e†ect, again a dilution experiment was per-
formed where the Fe O and Pt/TiO physical mixture was
2
3
2
diluted with a tenfold excess of pure TiO . This should lead
2
to a statistical separation of the two interacting sites. Curve B
in Fig. 7 shows the result of this experiment. As expected, the
initial reduction signal at 560 K is strongly depressed. But
now the main reduction signal shows its maximum to be 620 K
and very broad. This is a completely new behavior. Since
the diluting compound is still able to spillover hydrogen, this
new reduction e†ect has to be assigned to the “trueÏ spillover
of hydrogen. The shift and broadening of the signal can be
explained by the larger distance over which the hydrogen
must be transported.
This e†ect can be used as a tool to investigate other sub-
stances for their spillover abilities. In the next experiment this
strategy was applied to study Fe O . Fig. 8, curve A shows
Fig. 8 TPR proÐles of physical mixtures with Fe O . Curve A:
3
4
Fe O ] Pt/Fe O ; curve B: Fe O ] Pt/Fe O ] tenfold excess
2
3
3
4
2
3
3 4
Fe O ; curve C: Fe O ] Fe O .
3
4
2
3
3 4
3
4
J. Chem. Soc., Faraday T rans., 1998, V ol. 94
1343

View Article Online
Therefore, the catalytic behavior of a physical mixture con-
taining prereduced Pt/NaMor and Fe O was investigated.
3
4
Fig. 10 shows that the catalytic signature of this mixture is
identical with that of pure Fe O ; i.e. in this mixture no pro-
3
4
moting e†ect of Pt is detected. This result proves that the pro-
moting e†ect shown in Fig. 2(a) is entirely caused by Pt which
migrated, as an oxidic Pt precursor, from the zeolite onto the
iron oxide.
Discussion
This discussion will focus on the enhanced reducibility of
Fe O to Fe O in physical mixtures with Pt/NaMor. It is
2
3
3 4
also related to the enhanced activity and selectivity in the
catalytic reduction of acetic acid to acetaldehyde on Fe O .
3
4
In a superÐcial way, this e†ect might be described in terms of
hydrogen spillover from Pt, located in the zeolite, to iron
oxide. However, a thorough analysis of the reduction kinetics
under di†erent circumstances leads to a more di†erentiated
point of view.
First, the temperature of maximum reduction rate of
Fe O , 660 K in pure samples, shifts to lower temperatures if
Fig. 9 TPR proÐles of physical mixtures Fe O ] Pt/NaMor. E†ect
2
3
of pretreatments: curve A: Fe O ] prereduced Pt/NaMor; Curve B:
2
3
2
3
the oxide is mixed with Pt/NaMor and treated at elevated
temperatures. Even storage at room temperature for days or
weeks results in a downward shift of the TPR peak. The
absence of this e†ect immediately after mixing is inconsistent
with the primitive model of hydrogen spillover. This is the Ðrst
reason for considering the presumably much slower process of
a migration of Pt out of the zeolite onto the iron oxide. Also,
the observation that grinding of the mixtures in a mortar is
crucial for the reducibility enhancement suggests a solid state
same as A after calcination at 673 K for 10 min; curve C: Fe O
2
3
] prereduced Pt/NaMor, after storage for 6 weeks.
Therefore, the following TPR experiments were carried out,
using physical mixtures of Fe O with prereduced (at 823 K
2
3
in 5% H ÈAr) Pt/NaMor. Since the initial mixture with no
2
further pretreatment (Fig. 4, curve A) did not show any indica-
tions of Pt migration, this is also to be expected for those
mixtures containing prereduced Pt/NaMor (Fig. 9, curve A).
It becomes more interesting when the di†erent pretreat-
ments are applied to this mixture. Fig. 9, curve B shows the
interaction. Grinding provides
a most intimate contact
between the two solids; virtually no reducibility enhancement
is observed for gently mixed systems which behave almost as
pure Fe O .
proÐle after calcination in O at 673 K for 10 min. The
2
2
3
maximum is obtained at 540 K, indicating that there is the
The most direct indication for Pt migration onto the Fe O
2
3
same behavior as after calcination of the initial mixture (Fig.
particles is provided by the fact that the TPR maximum at
550 K is identical for a physical mixture after calcination and
for samples in which Pt was directly supported on the Fe O .
The experiments in which the Pt loading on iron oxide was
varied clearly show that this TPR peak position is indepen-
dent of the quantity of the platinum. Only the ratio of the
fraction of Fe O reduced at this low temperature and that
which is reduced at the high temperature varies with the
amount of Pt. All these data suggest that the rate limiting step
is di†erent for the reduction of pure Fe O and Fe O in
4, curve B). Since pretreatment took place in an O atmo-
2
sphere at elevated temperature, this can be explained by reoxi-
2
3
dation of the small Pt metal particles leading to the same
behavior as observed with the initial mixture.
However, in the TPR proÐle of a mixture containing prere-
duced Pt/NaMor, which was stored for 6 weeks, the main
reduction peak remains at 660 K (Fig. 9, curve C). This is a
big di†erence to that of the initial mixture, presented in Fig. 4,
curve D. This last experiment indicates clearly that the migra-
tion of Pt0-metal particles is not responsible for the e†ects
observed at low temperature. Rather, the mixture remains
unaltered when using prereduced Pt/NaMor.
2
3
2
3
2 3
direct physical contact with Pt particles.
“TrueÏ hydrogen spillover requires that H atoms move from
the H donor to the H acceptor. When adsorption of H atoms
on an oxide is vastly endothermic with respect to gaseous H
molecules, this migration is likely to take place on two parallel
2
“railsÏ: protons jump from one surface O2~ ion to the next
while the underlying TM cations change their electrochemical
valency by one unit. In other words: the oxide has to be a
semiconductor with migrating n or p charge carriers. TiO is
2
an oxide fulÐlling these conditions and has been shown to
enable classical hydrogen spillover to take place. The present
data in Fig. 7 show that Pt/TiO does indeed enhance the
2
reduction of Fe O , as expected; but the response to dilution
2
3
with a tenfold excess of TiO creates a broad TPR peak. This
reduction rate maximum is located between the two extreme
2
cases found on Pt/Fe O . Since in this case hydrogen has to
2
3
travel over a considerable distance over the TiO surface and
2
cross grain boundaries until it reaches a Fe O particle, this
2
3
process needs a higher activation energy and a wider energy
distribution leading to the observed behavior. The phenom-
enological evidence is quite di†erent from that shown in Fig. 5,
where two discrete peaks exist which discriminate between
the Pt enhanced and the not enhanced reduction of Fe O .
Fig. 10 Catalytic conversion of acetic acid vapor: (=) CH CO H,
3
2
(+) CH CHO, (…) CH CH OH, (L) CH , (|) CO, (È) C . Cata-
2 3
Since Fe O , like TiO , is also a semiconductor, it should
3
3
2
4
2`
lyst: physical mixture of Fe O ] prereduced Pt/NaMor.
2
3
3
4
2
1344
J. Chem. Soc., Faraday T rans., 1998, V ol. 94

View Article Online
also be capable of true hydrogen spillover. Indeed, the data in
Fig. 8 reveal a very similar response to dilution. We wonder
whether this broad TPR signal of diluted samples located
between the temperature limits of the enhanced and not
enhanced reduction, might be used as an experimental cri-
terion for “trueÏ hydrogen spillover, distinguishing it from the
reduction enhancement caused by metal migration.
of course be reduced to Pt0 clusters capable of dissociating H
2
molecules. From previous work with Pt and Pd ions in zeo-
lites it is known28,29 that their mobility is vastly enhanced by
ligands in complexes such as Pt(NH ) 2` or Pd(NH ) 2`. We
3 4
3 4
therefore assume tentatively that the migrating individuals at
room temperature during extensive storage times are hydrated
ions, e.g. Pt(H O) 2`.
2
4
The remarkable e†ect on Fe O reduction that is caused by
Migration of ions out of the zeolite has, however, to be
compensated by a migration of other charge carriers in the
opposite direction. Possible candidates for that are H` or
2
3
minute amounts of platinum shows that a small number of Pt
atoms on an Fe O particle suffices to enhance the reduction
2
3
of that particle. This explains why Pt amounts below the
detection limit of EDX suffice to induce a signiÐcant reduction
enhancement. The result also shows that TPR is much more
sensitive for the migration of very small amounts of Pt than
EDX and other physical techniques.
Fe(OH) ` according to the following equation:
2
Fe O ] Pt2` ] 2 H O H 2 Fe(OH) ` ] PtO
(5)
2
3
2
2
Since up to now there is no evidence for Fe in the zeolite,
this mechanism remains speculative. It should be noted that a
zeolite containing Fe ions might behave as a semiconductor.
Indeed, the reduction proÐle of the mixture after storage at
room temperature, has some resemblance to those which were
obtained with mixtures with known hydrogen spillover.
In previous work of this group on physical mixtures of
Pt/NaY and sulfated zirconia, migration of Pt out of a zeolite
onto an oxide was proven directly. In that work more rigor-
ous calcination conditions were applied: the mixture was
Rather than expanding in vague speculations, we conclude
that there is evidence for Pt migration in the systems studied
here in agreement with our earlier results.13 Mere migration
of adsorbed H atoms over zeolite walls is ruled out as a major
cause of the strong enhancement of the Fe O reduction.
treated in pure O at 773 K for 2 h. After that treatment the
presence of Pt crystallites on the surface of sulfated zirconia
2
could be detected unambiguously by electron microscopy with
EDX and by electron di†raction.13
Since Pt migration is reasonably proven the question arises,
which Pt species is actually migrating, Pt0 clusters, Pt2` ions
or Pt oxide clusters?
2
3
Migrating platinum, not spiltover hydrogen, is responsible for
the non-additive catalytic signature of mixtures of Fe O with
2
3
either Pt/SiO or Pt/NaMor.
The present data show that migration is pronounced after
2
calcination in O but non-existent with prereduced Pt/
2
Conclusions
NaMor. These results strongly indicate that metallic Pt0 clus-
ters are not the migrating species. Accordingly, the e†ect of Pt
on the catalytic reduction of acetic acid is absent when Pt is
present as Pt0 inside the zeolite. For the migration after cal-
cination Pt ions or oxide clusters remain candidates. It is
TPR discriminates between Pt migration and “trueÏ hydrogen
spillover. The latter phenomenon is negligible for electrically
insulating oxides. In mixtures with semiconducting oxides
such as TiO , hydrogen spillover leads to broad TPR peaks
2
known from the literature that PtO is volatile at elevated
located between those characteristic of unpromoted and Pt-
2
temperature. The equilibrium constant24,25 for the reaction:
promoted reduction of Fe O . In contrast, deposition of very
2
3
small amounts of Pt onto Fe O results in TPR proÐles with
Pt(s) ] O ¢ PtO (g)
(4)
2
3
2
2
two discrete peaks, typical for unpromoted and Pt-promoted
reduction. The intensity ratio of these TPR peaks correlates
with the Pt content. Migration of Pt from mordenite cavities
onto Fe O requires close contact, obtained by grinding the
powder mixture, and oxidation. Migration of Pt0 is negligible.
PtO migration via the vapor phase is conceivable, but via the
gives a vapor pressure of PtO at 673 K in 1000 mbar O of
2
2
about 2 ] 10~10 mbar. Using as an assumption a Pt particle
size of 1 nm, the amount of platinum evaporated during the
calcination amounts to 5 lg or ca. 1% of the Pt loading. This
estimate disregards a potentially higher vapor pressure of
small particles. Since the activation energy for surface migra-
tion is certainly lower than the desorption energy, it is likely
2
3
2
surface more probable. At room temperature and in moist air,
migration of hydrated Pt2` ions is the most likely cause for
that PtO easily di†uses out of the zeolite. Therefore, it is rea-
the enhancement of Fe O reduction observed after storage of
2
2
3
sonable to assume that during calcination in O the main
the uncalcined powder mixture for weeks.
2
migrating species is PtO , either via surface di†usion or over
2
The e†ect of Pt in mordenite on the catalytic reduction of
acetic acid over Fe O is entirely due to Pt migration onto
the gas phase. The volatility of metal oxides is known to be
3
4
increased by water vapor,26 which could be produced by
the oxide. A true H spillover e†ect is negligible as follows from
the absence of this e†ect in mixtures where Pt was pre-reduced
and hence unable to migrate.
desorption of H O from the zeolite during calcination. Migra-
2
tion of PtO also provides a plausible rationalization of
2
results by Roessner et al.27 who observed bifunctional cataly-
sis of n-hexane conversion over the acid form of erionite that
Financial Support of this research by the National Science
had been in physical contact with Pt/Al O during calcination
in oxygen at 723 K.
Foundation,
acknowledged.
Contract
CTS-9629963
is
gratefully
2
3
Transport of Pt as PtO seems less plausible to explain the
present Ðnding that Pt precursors migrate even at room tem-
2
References
perature, when given sufficient time. If Pt is reduced prior to
mixing, even after six weeks storage no e†ect is visible on the
reducibility of Fe O . Thus Pt0 clusters are obviously not
1
2
3
4
5
J. M. Sinfelt and P. J. Lucchesi, J. Am. Chem. Soc., 1963, 85, 3365.
S. Khoobiar, J. Phys. Chem., 1964, 68, 411.
J. Kuriacose, Ind. J. Chem., 1967, 5, 646.
2
3
M. Boudart, Adv. Catal., 1969, 20, 153.
W. C. Conner, S. J. Teichner and G. M. Pajonk, Adv. Catal.,
1986, 34, 1.
W. C. Conner and J. L. Falconer, Chem. Rev., 1995, 95, 759.
G. M. Pajonk, in Handbook of Heterogeneous Catalysis, ed. G.
Ertl, H. Knozinger and J. Weitkamp, Wiley-Verlag Chemie,
Weinheim, 1997, vol. 3, p. 1064.
R. V. Dimitriev, K.-H. Steinberg, A. N. Detjuk, F. Hofmann, H.
Bremer and K. M. Minachev, J. Catal., 1980, 65, 105.
D. H. Lenz and W. C. Conner, J. Catal., 1987, 104, 288.
able to migrate at room temperature, even over a very long
time. The TPR proÐle D in Fig. 4 shows that after several
weeks at room temperature a substantial part of the Fe O is
6
7
2
3
reducible at the same low temperature as the oxide which was
calcined at 673 K (proÐle B). As it is difficult to imagine that
the extremely low vapor pressure of PtO at room tem-
2
8
9
perature can induce this Pt doping of part of the Fe O , it
2
3
appears more likely that migration of Pt-ions is responsible
for this low temperature transport of Pt from the zeolite to
part of the Fe O . Under the TPR conditions these ions will,
10 E. Baumgarten, R. Wagner and C. Lentes-Wagner, J. Catal.,
1987, 104, 307.
2
3
J. Chem. Soc., Faraday T rans., 1998, V ol. 94
1345

View Article Online
11 U. Roland, H. G. Karge and H. Winkler, in Zeolites and Related
Microporous Materials: State of the Art 1994, Stud. Surf. Sci.
Catal., ed. H. G. K. J. Weitkamp, H. Pfeifer and W. Holderlich,
Elsevier Science, Amsterdam, 1994, vol. 84, p. 1239.
12 F. Roessner and U. Roland, J. Mol. Catal., 1996, 112, 401.
13 B. Q. Xu and W. M. H. Sachtler, in Spillover and Migration of
Surface Species on Catalysis, Stud. Surf. Sci. Catal., ed. C. Li and
Q. Xin, Elsevier Science, Amsterdam, 1997, vol. 112, p. 229.
14 J. S. Feeley and W. M. H. Sachtler, Zeolites, 1990, 10, 738.
15 J. S. Feeley and W. M. H. Sachtler, J. Catal., 1991, 131, 573.
16 W. M. H. Sachtler, Z. Zhang, A. Y. Stakheev and J. S. Feeley, in
New Frontiers in Catalysis, Proc. 10th Int. Congr. Catalysis, ed.
F. S. A. L. Guczi and P. Tetenyi, Elsevier Science, Budapest, 1993,
part A, p. 271.
Li and Q. Xin, Elsevier Science, Amsterdam, 1997, vol. 112, p.
333.
21 O. J. Wimmers, P. Arnoldy and J. A. Moulijn, J. Phys. Chem.,
1986, 90, 1331.
22 U. Roland, R. Salzer, T. Braunschweig, F. Roessner and H.
Winkler, J. Chem. Soc., Faraday T rans., 1995, 91, 1091.
23 U. Roland, R. Salzer and L. Summchen, in Spillover and Migra-
tion of Surface Species on Catalysts, Stud. Surf. Sci. Catal., ed. C.
Li and Q. Xin, Elsevier Science, Amsterdam, 1997, vol. 112, p.
339.
24 C. B. Alcock and G. W. Hooper, Proc. R. Soc. L ondon, Ser. A,
1960, 254, 551.
25 H. Schafer and A. Tebben, Z. Anorg. Chem., 1960, 304, 317.
26 D. Hildenbrand, K.-H. Lau and J. G. McCarthy, in Abstracts
13th North American Meeting of the Catalysis Society May 2È6
1993, paper PA38, ed. G. L. Haller, A. Varmice, J. Goodwin and
T. Mazanec, The Catalysis Society.
17 S. M. Augustine, M. S. Nache†, C. M. Tsang, J. B. Butt and
W. M. H. Sachtler, Proc. 9th Int. Congr. Catal., 1988, ed. M. J.
Phillips and M. Ternan, Chemical Institute of Canada, Calgary,
1988, vol. 3, p. 1190.
27 F. Roessner, U. Roland and T. Braunschweig, J. Chem. Soc.,
18 R. Pestman, R. M. Koster, J. A. Z. Pieterse and V. Ponec, J.
Faraday T rans., 1995, 91, 1539.
Catal., 1997, 168, 255.
28 S. T. Homeyer and W. M. H. Sachtler, J. Catal., 1989, 117, 91.
29 S. T. Homeyer and W. M. H. Sachtler, J. Catal., 1989, 118, 266.
19 R. Pestman, Thesis, National University of Leiden, Leiden, 1995.
20 G. Xu, Y. Zhu, J. Ma, H. Yan and Y. Xie, in Spillover and Migra-
tion of Surface Species on Catalysts, Stud. Surf. Sci. Catal., ed. C.
Paper 8/00200B; Received 6th January, 1998
1346
J. Chem. Soc., Faraday T rans., 1998, V ol. 94