Palladium-based catalysts for the synthesis of alcohols

Article abstract of DOI:10.1016/S1381-1169(03)00412-6

The effects of precursor and support composition on the structure and performance of Pd-based catalysts for the synthesis of methanol and higher oxygenated compounds were presented. CO hydrogenation over Pd-supported on ZnO, Zr(OH)₄, and ZrO₂ using PdCl₂ and Pd(NO₃)₂ as salt precursors was studied. The activity and selectivity were strongly dependent on the salt precursor and on the acid basic properties of the support. The order of acidity of the supports was Zr(OH)₄ > ZrO₂ > ZnO independently of the salts precursor used. The Pd(NO₃)₂/ZnO catalyst showed the highest methanol selectivity (98%), which was related to the presence of large particles of a PdZn alloy. Pd(C)/ZnO catalyst was very selective towards methanol formation (70% selectivity). For the Zr(OH)₄ and ZrO₂ catalysts, higher hydrocarbon selectivity was obtained and always a smaller particle size was observed when using chloride. On zirconia-supported catalysts, methanol formation was not very much affected with precursor salt. However, as a difference from the ZnO-supported catalysts, oxygenated compounds other than methanol (dimethylether, methyl formate, and isobutanol) were obtained.

Full text of DOI:10.1016/S1381-1169(03)00412-6

Journal of Molecular Catalysis A: Chemical 206 (2003) 339–351
Palladium-based catalysts for the synthesis of alcohols
M. Josefina Pérez-Zuritaa,∗, Michelle Cifarellib, M. Luisa Cubeiroa,
Juan Alvareza, Mireya Goldwassera, Egle Pietria, Luis Garciab,
Antoine Aboukaisc, Jean-François Lamonierc
a
Facultad de Ciencias, Escuela de Qu´ımica, Centro de Catálisis Petróleo y Petroqu´ımica, Universidad Central de Venezuela,
Apartado Postal 47102, Caracas DF 1040, Venezuela
Facultad de Ingenier´ıa, Escuela de Qu´ımica, Centro de Catálisis Petróleo y Petroqu´ımica, Universidad Central de Venezuela,
Apartado Postal 47102, Caracas DF 1040, Venezuela
b
c
Laboratoire de Catalyse et Environement, Université du Littoral Coˆte d’Opale, EA 2598, 59140 Dunkerque, France
Received 12 February 2003; received in revised form 15 May 2003; accepted 15 May 2003
Abstract
The present work intends to illustrate the effects of precursor and support composition on the structure and performance
of Pd-based catalysts for the synthesis of methanol and higher oxygenated compounds. The power of the XRD technique and
the reaction itself, as characterisation tools, was evidenced.
The hydrogenation of carbon monoxide has been studied over palladium supported on ZnO, Zr(OH)4 and ZrO2 using
PdCl2 and Pd(NO3)2 as salt precursors. Catalysts were characterised by XRD and Raman spectroscopy. The results show
that activity and selectivity are strongly dependent on the salt precursor and on the acid–basic properties of the support. As
seen from the CH3OH/DME ratio, the order of acidity of the supports is: Zr(OH)4 > ZrO2 > ZnO independently of the salt
precursor used. The Pd(NO3)2/ZnO catalyst shown the highest methanol selectivity which was related to the presence of large
particles of a PdZn alloy.
A double bifunctionality on the catalysts for the production of higher oxygenated seems evident. On the one hand, an
acid–base bifunctionality of the support seems to be needed for the formation, stabilisation, chain growth and further reactivity
of intermediates such as formate, as well as for the dissociative adsorption of CO, and on the other hand, a metal-support
bifunctionality, where the palladium metal particles would play the role of hydrogenating the intermediates and possibly
adsorb non-dissociatively the CO.
© 2003 Elsevier B.V. All rights reserved.
Keywords: Palladium; ZrO2; Zr(OH)4; ZnO; CO hydrogenation; Methanol; DRX; Syngas
1. Introduction
carbon monoxide over supported palladium catalysts
produced methanol in high selectivity. They suggested
that the ability of palladium to catalyse methanol for-
mation under elevated pressures results from its ex-
cellent hydrogenation ability coupled with its inherent
tendency to adsorb CO non-dissociatively even at re-
action temperatures.
Many papers have been written since Poutsma et al.
[1] reported for the first time, that hydrogenation of
∗
Corresponding author. Tel.: +5812-6051231;
fax: +5812-6051231.
The reason for the continuous interest on the study
of palladium-based catalysts on the CO hydrogenation
E-mail addresses: marperez@reacciun.ve,
marperez@strix.ciens.ucv.ve (M. Josefina Pe´rez-Zurita).
1381-1169/$ – see front matter © 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S1381-1169(03)00412-6

340
M. Josefina Pe´rez-Zurita et al. / Journal of Molecular Catalysis A: Chemical 206 (2003) 339–351
reaction comes from the fact that they provide an al-
ternative to the presently used copper-based catalysts
as they possessed tolerance towards sulphur poisoning
[2,3].
From all the work published after Poutsma et al. [1],
major advances have been made in our understanding
of palladium-based systems.
nearly three-fold more active than Pd(1 1 1). However,
for a fixed morphology, the specific methanol synthe-
sis activity of Pd/La2O3 was a factor of 7.5 greater
than that of Pd/SiO2. A metal-support interaction was
claimed to be responsible for these results.
Jackson et al. [6] reported on the effect of using
molybdenum and tungsten trioxides as supports for
the medium pressure CO hydrogenation over group
VIII metals. All metals tested, except iron, showed
activity enhanced by up to two-fold orders of mag-
nitude when either molybdenum or tungsten trioxide
replaced silica as support. They suggested that under
reaction conditions, a fast spillover/reverse spillover
of hydrogen occurred between the metal and the
support and they stated that this process allowed an
effective hydrogen concentration increase to the reac-
tive intermediates and hence, causes an enhancement
of the rates. No comments were made regarding the
product distribution but it can be seen from their
results that on Pd/SiO2 catalysts, methanol was the
main product while when WO3 was used as support,
selectivity shifted towards methane. Dimethylether
formation was obtained on Pd/WO3 but the authors
do not mention any acidity effect.
It is clear now that the catalytic properties of palla-
dium for CO hydrogenation are quite sensitive to the
nature of the support, the presence of promoters, the
palladium particle size and the metal precursor. How-
ever, the way in which each of these parameters affect
the behaviour of palladium-based catalysts remains a
subject of controversy.
From the pioneer work of Fajula et al. [4], it was
clear that the nature of the support was crucial to ob-
tain methanol. They studied palladium supported on
Y zeolite and on three different commercial SiO2.
Their first observation was that under their experimen-
tal conditions (270–290 ◦C and 15–16 atm) the selec-
tivity, the activity and the stability of their catalysts
were strongly dependent on the type of support used.
Depending on the source and grade of the silica used
a 20-fold variation and a nearly 100% variation in se-
lectivity towards methanol was observed.
CO hydrogenation on supported palladium cata-
lysts was investigated by Ali and Goodwin [7] us-
ing Steady-State Isotopic Transient Kinetic Analysis
(SSITKA) in order to explore the differences in the
catalytic behaviour resulting from the use of differ-
ent supports with different acidities (SiO2, Al2O3 and
SiO2·Al2O3). The nature of the support significantly
altered the activity and selectivity of palladium. The
order of activities at steady state for a given precursor
was Al2O3 > SiO2 · Al2O3 > SiO2. It was found that
Pd/SiO2 produced essentially only methanol. The use
of acidic supports (Al2O3 and SiO2·Al2O3) resulted in
the formation of dimethylether which was limited by
the amount of methanol formed. Palladium on acidic
supports produced significant amounts of methane.
The SSITKA results showed that the nature of the sup-
port altered only slightly the intrinsic activity of the
sites producing methane and methanol. However, the
number of surface intermediates leading to methane
and methanol were significantly affected by the sup-
port used. They concluded that the impact of the sup-
port on the reaction rate was determined by its effect
on the concentration of active sites/intermediates on
palladium.
On how the support was affecting the performance
of palladium catalysts was then the subject of major
debate. The explanation of Fajula et al. [4] in their
work was that methanol synthesis was favoured on
small metal crystallites where CO was weakly ad-
sorbed while methane formation was enhanced when
the density of acidic sites at the surface of the support
was high.
Hick and Bell [5] reported on the effect of sup-
port on palladium morphology as expressed by the
distribution of Pd(1 0 0) and Pd(1 1 1) planes deter-
mined from “in situ” infrared spectra of adsorbed CO.
They observed that the crystallite morphology of a
Pd/SiO2 catalyst was the same independently of pal-
ladium loading: 90% of the surface was comprised of
Pd(1 0 0) planes and 10% of the surface was comprised
of Pd(1 1 1) planes. By contrast, the crystallite mor-
phology of a Pd/La2O3 catalyst changed with palla-
dium loading. Primarily Pd(1 0 0) planes were exposed
at low-weight loading while Pd(1 1 1) planes were ex-
posed at high-weight loading. They observed that the
palladium morphology influences the specific activ-
ity of Pd/La2O3 for methanol synthesis: Pd(1 0 0) was

M. Josefina Pe´rez-Zurita et al. / Journal of Molecular Catalysis A: Chemical 206 (2003) 339–351
341
From the above discussion, it seems to be a con-
sensus on the fact that acidity on the support favours
the rate of methane formation and enhances overall
activity. The fact that a Pd/SiO2 catalyst (which is
rather a neutral support) produced almost exclusively
methanol was not related to any property of the sup-
port. A clearer picture on this respect arises when the
promoters/impurities effect is reviewed.
A more recent work of Gusovius et al. [13], where
the study of a series of calcium promoted Pd/SiO2
catalysts with different calcium loading was reported
suggested that (i) calcium locates itself preferentially
on the palladium particles (rather than on the support)
and (ii) that calcium species located on (and perhaps
also near) the palladium particles give rise to the cat-
alytic activity for C1-oxo. Moreover, the requirement
for the promoter to be located on the palladium par-
ticle means that an effective promoter must not only
have favourable chemical properties for the reaction
but must also have a high affinity for the metal particle
and its oxidic precursor.
Particle size have also an influence on the perfor-
mance of palladium catalysts [4,5,14–17]. When a
reaction is structure sensitive, formation of smaller
particles on supported catalysts is of considerable im-
portance as the catalytic surface area is enlarged and,
as a consequence, the catalytic activity is enhanced.
Kim et al. [14,15] reported the study of the CO hydro-
genation reaction over palladium catalysts prepared by
a water-in-oil micro emulsion method developed by
them, with the aim of producing well-dispersed cat-
alysts. They reported a size distribution of palladium
particles remarkably narrow with an average particle
size much smaller than that found when conventional
wet impregnation method was used. A drawback of
this method of preparation was that some of the pal-
ladium particles were wholly or partially embedded
in the support. However, the catalysts prepared by
the micro emulsion method were found to exhibit a
much higher activity (three times as high) for the CO
hydrogenation reaction than the catalysts prepared by
impregnation. Product selectivity was considered to
be independent of the palladium particle size. The
same conclusion was reported by Kirillov and Ryndin
[16]. Gotti and Prins [8] on the other hand reported a
greater promoter-metal contact area for catalysts with
a small noble metal particles which results in a more
efficient promotion of the methanol formation.
Gotti and Prins [8] prepared Pd and Rh catalysts
using an ultra pure SiO2 doped with Al, Fe, Na, K, or
Ca in order to investigate their activity and selectivity
on the CO hydrogenation reaction. They observed
that additives had a strong influence on the catalytic
properties. The doping with alkali and alkaline earth
oxides led to a strong suppression of the CO dissocia-
tion. Particularly, basic additives, such as calcium had
an strong promoting effect on the methanol forma-
tion. In another work, the same authors [9] stated that
molecular adsorption of CO or CO2 and the availabil-
ity of activated hydrogen are not sufficient to form
methanol: basic metal oxides are needed to give pal-
ladium a high methanol activity. In addition, the basic
oxide additives must be close to or in contact with the
palladium particle to be effective in methanol synthe-
sis. Furthermore, the clear correlation between cal-
cium amount, noble metal surface and C1-oxo activity
may confirm that the formation of methanol occurs
through formate intermediates formed by the reaction
of CO with the basic support or at the interface be-
tween the promoter oxide and the noble metal particle.
From their results, they concluded that the rate en-
hancement for methanol formation of the metal oxides
followed a volcano curve when plotted as a function
of the metal ion electro negativity, with a maximum
for metal oxides with a moderate basic nature.
In an earlier work, Nonneman et al. [10,11] shown
that, not only by adding alkaline promoters to the
metallic catalysts it was possible to enhance methanol
formation but that impurities on the support could play
an important role as they can be leached out during the
impregnation step and promote the metal sufficiently
to produce methanol and higher oxygenates.
The prevalent opinion is that the promoter should be
located on or close to the noble metal surface, probably
as oxide patches [12] and that the presence of basic
promoters like calcium, close or on the active metal
particle has a strong promoting effect on the methanol
production.
An earlier work of Rieck and Bell [17] has shown
that dispersion had an effect on the interaction of H2
and CO with Pd/SiO2. They reported that the distri-
bution of H2 ad-states changes with dispersion due
possibly to a change on the morphology of the palla-
dium crystallites or in the co-ordination of adsorbed
H atoms. For the CO adsorption, these authors state
that the ratio of linearly held CO to bridge-bonded

342
M. Josefina Pe´rez-Zurita et al. / Journal of Molecular Catalysis A: Chemical 206 (2003) 339–351
CO decreases with the dispersion, as does the ra-
tio of Pd(1 0 0) to Pd(1 1 1) planes on the surface of
the palladium crystallites. It was concluded that CO
dissociation occurs preferentially from bridged sites
and proceeds more readily with decreasing disper-
sion. Consistent with this, the turnover frequency for
methanation increases with decreasing dispersion.
Dispersion may be influenced by the salt precursor
used although there is not consensus on its real ef-
fect. Gotti and Prins [8] reported a dramatic influence
of the salt precursor for the Pd/SiO2 catalysts. Cata-
lysts prepared from PdCl2 and Pd(NH3)4(NO3)2 had a
higher hydrocarbon activity than when prepared from
Pd(NO3)2. This observation correlates with strong dif-
ferences in metal dispersion. Apparently, the CO dis-
sociation is only favoured on small palladium particles
and the metal salt anion has an indirect effect through
its influence on the final metal dispersion. In contrast,
Ali and Goodwin [7] stated that the palladium precur-
sor had no apparent impact on palladium dispersion.
However, these author reported that the total rate of
CO conversion was somewhat higher for the catalysts
prepared using PdCl2 rather than Pd(NO3)2 precursor
at steady state.
For the synthesis of ZnO, a solution of oxalic
acid [H2C2O4·2H2O, Riedel de Haen, 4, 5 ml de
H2O/g H2C2O4·2H2O] was added drop wise un-
der continuous stirring to a zinc nitrate solution
[Zn(NO3)2·6H2O, Riedel de Haen, 0, 18 ml H2O/g
Zn(NO3)2·6H2O] at room temperature. The result-
ing zinc oxalate (ZnC2O4), was left on digestion
for 12 h, filtered dried at 110 ◦C for 24 h and cal-
cined at 400 ◦C (ramp 2 ◦C/min) for 3 h to finally
obtain a ZnO with a surface area of 24 m2/g. All
supports were impregnated with both, an aqueous
solution of Pd(NO3)2·H2O [Strem Chemical-named
Pd(N)/support] and HCl acidified aqueous solution
of PdCl2 [Aldrich, 0, 3 g de PdCl2/ml HCl-named
Pd(C)/support] dried at 110 ◦C overnight and calcined
in air at 400 ◦C (ramp 2 ◦C/min) for 3 h to give a
nominal content of 3% Pd (w/w).
2.2. Catalysts characterisation
Surface area of the supports and catalysts were
obtained from nitrogen adsorption and desorption at
77 K in a Micromeritics Flow Sorb II 2300, using
a N2–Ar (30% N2) gas mixture. For the analysis,
20 mg of the samples were heated at 200 ◦C for 2 h
under flowing nitrogen. Palladium loadings were
determined by inductively-coupled plasma-atomic
emission spectrometry (ICP-AES).
“In situ” programmed temperature XRD data for
structural analysis and particle size determination were
collected with a Siemens D5000 diffractometer using
Cu K␣ radiation (λ = 1.5418 Å) by placing the solids
on a gold sample holder at 2θ = 20–60◦ with a 0.02◦
step size. Background correction, K␣2 stripping and
peak identification was accomplished with the Diffract
AT program from Siemens. The mean particle size was
derived from XRD line broadening, using the simple
Scherrer equation.
From the latest studies, one can safely said that the
major parameters which affect the performance of Pd
catalysts are
1. the acid–base properties of the support;
2. the addition of promoters/additives;
3. the nature of the salt precursor.
The present work intend to give further evidence of
the effects of the acid–base properties of the support
and the nature of the salt precursor by means of XRD,
Raman spectroscopy and the CO hydrogenation reac-
tion itself.
Raman spectra were recorded using a Dilor XY
monochromator with subtractive dispersion. The sam-
ples were excited using a 632.8 nm radiation of a
He–Ne laser. The samples were analysed under atmo-
spheric pressure after being calcined in air at 400 ◦C.
2. Experimental
2.1. Catalysts preparation
The catalysts were prepared by incipient wetness
impregnation using as supports a commercial Zr(OH)4
(MEL Chemicals, 310 m2/g), ZrO2 which was obtai-
ned by calcination of the former at 500 ◦C (ramp 3 ◦C/
min) for 15 h (43 m2/g) and ZnO synthesised follow-
ing the method described by Cubeiro and Fierro [18].
2.3. Activity testing
Catalysts were tested for hydrogenation of CO using
a 40 cm, 1/2 in a 316 stainless steel reactor. In order to

M. Josefina Pe´rez-Zurita et al. / Journal of Molecular Catalysis A: Chemical 206 (2003) 339–351
343
Table 1
avoid contamination of the catalysts with Fe carbonyls
Chemical analyses and surface area
coming from the gas cylinder, an activated carbon trap
was placed upstream of the catalyst to adsorb such
metal carbonyls. A mixture of H2:CO:N2 (63:32:5)
was fed from a mass flow controller (Brooks Instru-
ments 5850E). Nitrogen was used as internal standard.
The catalyst (600 mg) was placed in the centre of the
reactor between two beds of glass beams separated
from the catalyst bed by glass wool. Downstream of
the reactor the line was kept at 120 ◦C to avoid prod-
uct condensation. The catalysts were reduced in a flow
of H2 (50 ml/min) at 400 ◦C (ramp 2 ◦C/min) for 3 h
under atmospheric pressure. Pressure was increased to
working pressure (50 bar) using the H2:CO:N2 mix-
ture by means of a back pressure regulator. Reaction
temperature (280–350 ◦C) was reached as the pres-
sure was increased. In all cases, the reaction started at
280 ◦C and temperature was increased every 24 h to
reach 320 and 350 ◦C. The data reported in the present
paper are the corresponding to a reaction temperature
of 320 ◦C. Space velocity was 5.000 ml/(gcat h). Ex-
periments were carried out for 72 h on stream. The re-
actor effluents were analysed every hour using an on
line Varian 3300 gas chromatograph equipped with a
Carbosieve SII column (SS, 12 ft.×1/8 in.) and a TCD
detector for analysis of CO, N2, CH4 and CO2 and
a Chrompack CP 9100 gas chromatograph equipped
with a Quadrex Corp methyl silicon capillary column
(fused SiO2, 50 m × 0.32 mm) and a FID detector for
analysis of the hydrocarbons and oxygenated com-
pounds. CO conversion was determined using nitrogen
as an internal standard. Chromatograms were corre-
lated through methane and selectivities were obtained
based on converted carbon.
Catalyst
Pd (%)
Zr (%)
Zn (%)
Surface area
(m2/g)
Pd(N)/ZnOa
2.78
2.65
3.73
3.85
3.53
3.32
–
–
79.65
75.00
–
–
–
–
21
21
146
146
37
Pd(C)/ZnOa
b
b
Pd(N)/Zr(OH)4
Pd(C)/Zr(OH)4
65.53
65.40
69.50
69.00
b
Pd(N)/ZrO2
Pd(C)/ZrO2
b
35
a Dried.
b Calcined.
3.1. XRD results
Fig. 1 shows the evolution of the XRD patterns of
Pd(N)/ZnO catalysts under reduction conditions. The
characteristic signals of ZnO in an hexagonal struc-
ture (JCPDS 790205) are observed at room tempera-
ture which confirm that the synthesis method lead to a
well crystallised ZnO. The signal at 2θ = 38.44◦ is as-
signed to gold (JCPDS 040784) on the sample holder.
When the sample is reduced, two additional bands at
2θ = 41.62 and 44.10◦ are seen. These two signals are
characteristic of a PdZn alloy (JCPDS 6-620). These
results are in good agreement with those reported by
Iwasa et al. [19] and by Cubeiro and Fierro [18]. With
increasing temperature, better defined signals can be
observed with the mean particle size increasing from
8 nm (at 250 ◦C) to 23 nm (at 450 ◦C). The presence of
a signal at 40.0◦ assigned to palladium metal (JCPDS
5-681) can also be assessed from the diffractogram
at 400 ◦C but not that of the metallic zinc. It cannot
be ruled out, however, that some zinc metal could be
present in small particles.
Fig. 2 shows the XRD pattern of the catalyst after
reaction. No important changes are observed as com-
pared with the reduced catalyst. The presence of the
signals at 41.62 and 44.10◦, evidence the presence of
the PdZn alloy in the reacting catalyst.
Fig. 3 shows the evolution of the XRD patterns for
the Pd(C)/ZnO catalyst under reduction condition. As
in the Pd(N)/ZnO catalyst all the signals attributed to
ZnO are present on the diffraction patterns although
their intensity is lower. This could be due to the fact
that, since an acid solution was used for the prepa-
ration of this catalyst, it is possible that some of the
zinc oxide was partially dissolved, resulting in the
3. Results and discussion
Table 1 shows the chemical analyses and surface
area of all the prepared catalysts. In all cases, the real
content was very similar to the nominal one. The cata-
lysts maintain similar surface areas as compared with
that of the support, with the exception of the catalysts
supported on Zr(OH)4 where values are half of the ob-
served for the support, attributed to the fact that these
catalysts were calcined at 400 ◦C. Under this condi-
tion, the Zr(OH)4 partially crystallises leading to a re-
duction of the surface area.

344
M. Josefina Pe´rez-Zurita et al. / Journal of Molecular Catalysis A: Chemical 206 (2003) 339–351
Fig. 1. Evolution of the XRD patterns of Pd(N)/ZnO catalysts.
formation of a new oxide phase or of zinc oxychloride
after the thermal treatment with lower crystallinity. In
contrast with the results observed for the Pd(N)/ZnO
catalysts, the 41.62 and 44.10◦ signals attributed to
the PdZn alloy can be only seen when the temperature
reach 400 ◦C as compared with the apparition of the
same signals at 250 ◦C for the Pd(N)/ZnO.
Fig. 4 shows the XRD pattern of Pd(N)/ZnO and
Pd(C)/ZnO catalysts reduced at 400 ◦C (reduction
temperature before reaction). The mean particle sizes
as derived from the XRD line broadening are 21 nm
and 9 nm, respectively. It is clear that smaller metal
particles are obtained when PdCl2 is used as pal-
ladium precursor, compared to catalysts prepared
using Pd(NO3)2. Fig. 5 shows the Pd(C)/ZnO cata-
lysts after reaction. In this case, the reacted catalyst
did not show the presence of the signals assigned
to the PdZn alloy. This fact seems to be an indi-
cation that the alloy is present in small particles.
From the above discussion, it is possible to con-
clude that the PdZn alloy is formed regardless the
precursor salt but the particle size is greatly affected
by it.
The evolution of the XRD patterns of Pd(N)/
Zr(OH)4 catalyst with calcination temperature is
shown in Fig. 6. At room temperature the signals
attributed to ZrO2 (2θ = 30.3, 34.6 and 50.3◦) in its
tetragonal structure (JCPDS 79177) are present. It
has to be pointed out that this catalyst was calcined
at 400 ◦C before the analysis. When temperature is
increased up to 700 ◦C, the original signals remain,
becoming a little bit better defined.
Fig. 2. XRD pattern of the Pd(N)/ZnO catalyst after reaction.

M. Josefina Pe´rez-Zurita et al. / Journal of Molecular Catalysis A: Chemical 206 (2003) 339–351
345
Fig. 3. Evolution of the XRD patterns of Pd(C)/ZnO catalyst.
During the cooling down process (last diffractogram
at 27 ◦C), a new small signal appears at 2θ = 28.2◦
assigned to ZrO2 in its monoclinic structure (JCPDS
371484). Additionally, when the temperature is in-
creased up to 500 ◦C and upwards, three new signals
appeared attributed to the tetragonal phase of PdO
(JCPDS 431024): a shoulder at 2θ = 33.5◦, and other
signals at 2θ = 41.9 and 54.7◦. After reaction, in ad-
dition to the signals assigned to the tetragonal phase
of ZrO2, signals at 2θ = 28.0◦ attributed to the mon-
oclinic phase and at 2θ = 40.0◦ attributed to metallic
palladium (JCPDS 5-681) were observed.
For the Pd(C)/Zr(OH)4 catalyst, a poor crystallinity
of the tetragonal phase of ZrO2 (signals at 2θ = 30.3,
34.6 and 50.3◦) is observed (Fig. 7). In a previous
work [20], we reported using TGA/TDA analyses that
Fig. 4. XRD pattern of the Pd(N)/ZnO and Pd(C)/ZnO catalysts reduced at 400 ◦C.

346
M. Josefina Pe´rez-Zurita et al. / Journal of Molecular Catalysis A: Chemical 206 (2003) 339–351
Fig. 5. XRD pattern of the Pd(C)/ZnO catalyst after reaction.
when using Pd(NO3)2 as precursor salt crystallisation
starts at 425 ◦C while when using PdCl2, crystallisa-
tion starts at 484 ◦C. These results lead us to claim
that crystallisation is retarded by the presence of chlo-
ride. As a consequence, the lack of crystallinity for
ZrO2 is attributed to the need of a higher calcination
temperature for this sample.
ture signal at 2θ = 28.8◦ appears after cooling down.
After reaction, the presence of a signal at 2θ = 40◦
assigned to palladium metal in the catalysts prepared
using Pd(NO3)2 as precursor and its absence in the cat-
alyst prepared using PdCl2 provided further evidence
that the precursor salt has an effect on the palladium
particle size of the final catalyst.
Another important feature in Fig. 7 is the absence
of signals at 2θ = 41.9 and 54.7◦ attributed to PdO.
These results lead us to conclude that again, when us-
ing nitrate as precursor, the palladium particle size of
the final catalyst is larger than those observed when us-
ing chloride as precursor. Again the monoclinic struc-
Fig. 8 shows the XRD patterns of the fresh
Pd(N)/ZrO2 and Pd(C)/ZrO2 catalysts. For both
Pd/ZrO2 catalysts, no much information can be drawn
from the XRD results as the patterns are very rich in
signals. The overlapping of the different diffraction
peaks does not allow to evidence the presence of Pd
neither PdO phases by XRD. Nevertheless, it can be
said that for the Pd/ZrO2 catalysts the characteristic
signals assigned to the monoclinic phase of ZrO2
at 2θ = 28.2 and 31.0◦ are present along with a
tetragonal phase signal at 2θ = 30.5, 34.6 and 50.3◦.
In order to overcome this problem and assess a
possible difference in particle size or palladium crys-
tallinity on Pd/ZrO2 catalysts, Raman spectrometry
measurements were conducted as it has been reported
in the literature that PdO shows a sharp band at around
648–652 cm−1 [21–23].
Fig. 9 shows the Raman spectra for ZrO2, Pd(C)/
ZrO2 and Pd(N)/ZrO2 samples in the 250–750 cm−1
range. The spectrum of pure zirconia exhibits the
bands characteristics of ZrO2 in the monoclinic
phase [22]. When PdCl2 salt is used, no changes in
the spectrum are observed. This observation is an
Fig. 6. Evolution of the XRD patterns Pd(N)/Zr(OH)4 catalyst.

M. Josefina Pe´rez-Zurita et al. / Journal of Molecular Catalysis A: Chemical 206 (2003) 339–351
347
Fig. 7. Evolution of the XRD patterns Pd(C)/Zr(OH)4 catalysts.
indication of a well dispersed palladium phase [21].
3.2. CO hydrogenation reaction results
When Pd(NO3)2 salt is used, a clear increase in the
intensity of the band at 648 cm−1 is observed. In this
case, a contribution of the PdO phase seems to be
an evidence of the presence of a better crystallised
phase when the nitrate precursor is used. Again, the
effect of the precursor salt on the crystallisation of
palladium is assessed.
Under the reaction conditions used (Treaction
=
320 ◦C, SV = 5000 ml/(gcat h), Preaction = 5 MPa,
H2/CO = 2) conversions were lower than 10%.
Fig. 10 shows the catalytic activity (mol CO/
(kg Pd h) for all the catalysts studied. In general,
higher activities were obtained, when using PdCl2.
Additionally, Pd/ZrO2 catalysts were more active than
Pd/Zr(OH)4 and Pd/ZnO catalysts.
The fact that catalysts prepared with palladium chlo-
ride are more active can be related to particle size. As
seen by the Raman and XRD results, the particle size
of the palladium when using palladium chloride was
smaller than that obtained when palladium nitrate was
the precursor salt, so it is possible to conclude that dif-
ferences in activities with precursor salt can be related
to dispersion.
In addition of the salt precursor effect, the sup-
port seems to play an important role. The differ-
ence in activity showed when support was changed
could be related either to its acid–base proper-
ties [9], its ability to chemisorbs CO dissocia-
tively [8,24], to the presence of cationic palladium
species formed through a metal-support interaction
Fig. 8. XRD patterns of the Pd(N)/ZrO2 and Pd(C)/ZrO2 catalysts.

348
M. Josefina Pe´rez-Zurita et al. / Journal of Molecular Catalysis A: Chemical 206 (2003) 339–351
Fig. 9. Raman spectra for ZrO2, Pd(C)/ZrO2 and Pd(N)/ZrO2 samples.
as reported by Shen et al. [25] or again to metal
dispersion.
in its tetragonal phase after calcination, while in the
catalysts supported on ZrO2, the former is present in
its monoclinic phase. This can explain the differences
in activity among Pd/Zr(OH)4 and Pd/ZrO2 catalysts.
The lower activity observed for the Pd/ZnO catalysts
could be related to the basic character of the support.
Gotti and Prins [8] reported that basic sites led to a
strong suppression of CO dissociation which, as stated
above, does not contribute with gross activity. The
particle size effect cannot be ruled out, especially be-
cause the presence of surface hydroxyls groups in the
Pd/Zr(OH)4 catalysts could influence palladium depo-
sition, controlling particle size. Evidence of this was
reported elsewhere [20].
The more active catalysts were those supported on
zirconia. Two possible explanations could be given:
(i) the reported ability of ZrO2 to chemisorbs CO
dissociatively [24] or, as reported by Gotti and Prins
[9] and (ii) to its amphoteric character. On the other
hand, it has been recently stated [26] that the CO
adsorption capacity of ZrO2 depends on the crystal-
lographic phase. The adsorption capacity of mono-
clinic ZrO2 was reported to be significantly higher
than that of tetragonal ZrO2. As seen on our XRD
results, catalysts prepared on Zr(OH)4 had the ZrO2
Figs. 11 and 12 show the product selectivity for
the catalysts studied. The catalysts prepared us-
ing ZnO as support shows interesting results: while
the Pd(N)/ZnO catalyst was very selective towards
methanol formation (98% selectivity), Pd(C)/ZnO
catalyst produced mainly methane (70% selectivity).
The high methanol selectivity of the Pd(N)/ZnO could
safely be attributed to the presence of large particles
of the PdZn alloy as the XRD results evidenced the
presence of larger particles of a PdZn alloy when
Pd(NO3)2 was used as precursor salt. Similarly, the
high hydrocarbon formation obtained when using
Fig. 10. Catalytic activity (mol CO/(kg Pd h) of the catalysts as a
function of support and salt precursor.

M. Josefina Pe´rez-Zurita et al. / Journal of Molecular Catalysis A: Chemical 206 (2003) 339–351
349
Fig. 11. Methanol selectivity as a function of support and salt precursor.
PdCl2 could probably be explained relating it with
the presence of small particles [8]. In order to obtain
further evidence on the need of bigger PdZn alloy par-
ticles as a condition to have high methanol selectivity,
the Pd(C)/ZnO catalyst was reduced at 600 ◦C using
a higher heating rate with the aim of promote sinter-
ing (Table 2). The results obtained show a complete
change of the product distribution and do not leave
any doubt. A selectivity of 89% towards methanol
proved that indeed, large particles of the PdZn alloy
are needed to obtain methanol.
For the Zr(OH)4 and ZrO2 supported catalysts, a
similar behaviour was observed: when using chlo-
ride, higher hydrocarbon selectivity was obtained and
Fig. 12. Hydrocarbon selectivity as a function of support and salt
precursor.
Table 2
Catalytic results on the CO hydrogenation (Treaction: 320 ◦C)
Catalyst
Carbon selectivity (%)a
Activityb
CH4
C2+
CH3OH
DME
C2–C3-oxo
C4-oxo
Total oxo
CH3OH/DME
Pd(N)/ZnO
Pd(C)/ZnO
Pd(C)/ZnOd
Pd(N)/ZrOH4
Pd(C)/ZrOH4
Pd(N)/ZrO2
Pd(C)/ZrO2
79
65
41
1.4
62.1
7.5
22.7
29.9
30.5
39.1
0.4
7.9
3.9
2.7
4.2
6.4
3.1
97.5
29.9
88.6
58.8
53.0
51.2
52.0
0.2
0.0
0.0
12.5
10.1
7.6
0.5(0.4c)
0.0
0.0
1.2(0.6c)
1.3
0.0
0.0
0.0
2.0(1.4e)
1.5(1.2e)
3e
98.2
29.9
88.6
74.5
65.9
63.1
57.8
487.5
∞
∞
57
4.7
5.2
6.7
16.3
121
119
152
1.3
1.7(0.5c)
3.2
0.9
a CO2 free.
b Units: mol CO converted/(kg Pd h).
c Methyl formate.
d Sinterized.
e iso-C4.

350
M. Josefina Pe´rez-Zurita et al. / Journal of Molecular Catalysis A: Chemical 206 (2003) 339–351
always a smaller particle size were observed by RA-
MAN and XRD. On the other hand, as seen in Table 2,
selectivity towards total oxygenated compounds was
higher when using nitrate. The previous XRD results
indicated that when using Pd(NO3)2 as precursor, the
palladium particle size was larger. Again particle size
seems to play an important role on the catalytic per-
formance, as it can be said that large particles favours
oxygenated compounds and small particles favours hy-
drocarbon formation. It cannot be ruled out, however
an effect of conversion on selectivity. A relationship
between high conversion and hydrocarbon formation
and low conversion and oxygenated formation can also
be established.
On zirconia-supported catalysts, methanol forma-
tion was not very much affected with precursor salt.
However, as a difference from the ZnO-supported cat-
alysts, oxygenated compounds other than methanol
(i.e. dimethylether, methyl formate and isobutanol)
were obtained on zirconia-supported catalysts. All
these products can be formed by the presence of sites
introduced by the support.
The CH3OH/DME ratio is an indication of the
acidic properties of the catalysts. The presence of
DME is an evidence of the presence of acid sites as it is
well known that DME is produced via methanol dehy-
dration. On the other hand, the presence of iso-butanol
and methyl formate is evidence of the presence of ba-
sic sites. As seen from the CH3OH/DME ratio, the or-
der of acidity of the supports is: Zr(OH)4 > ZrO2 ≫
ZnO independently of the salt precursor used. Cata-
lysts with higher values of CH3OH/DME ratio have
little or none acidic sites and, as a consequence, for-
mation of mostly methanol and methyl formate is
observed. On the contrary, the catalysts with both type
of sites like Pd/Zr(OH)4 and Pd/ZrO2 (as evidenced
by the lower CH3OH/DME ratio and the presence
of methyl formate) produced the higher amount of
higher oxygenated. On the other hand, the presence
of iso-butanol indicates that the chain growth mech-
anism applying for oxygenated compounds seems to
be aldolic condensation which would take place on
basic sites. A double bifunctionality on the catalysts
for the production of higher oxygenated seems evi-
dent. On the one hand, an acid–base bifunctionality
of the support seems to be needed for the formation,
stabilisation, chain growth and further reactivity of
intermediates such as formate as well as for the dis-
sociative adsorption of CO, and on the other hand,
a metal-support bifunctionality, where the palladium
metal particles would play the role of hydrogenate the
intermediates and possibly adsorb non-dissociatively
the CO. An alternative explanation was given recently
by Bell who reported [26] that on Cu/ZrO2 cata-
lysts, methanol synthesis occurs via a bifunctional
mechanism, where the ZrO2 adsorbs CO which is hy-
drogenated in the metal to methoxide groups. These
methoxide groups are released from the ZrO2 surface
via either hydrolysis or reductive elimination. He de-
scribes this process assuming that Cu serves to adsorb
H2 dissociatively and provide atomic hydrogen to the
ZrO2 via spillover. The role of the ZrO2 support is
described in terms of its amphoteric characteristic.
The Lewis acid centres contribute to the formation of
carbonate species whereas the hydroxyl groups are
involved in the formation of bicarbonate and formate
species.
A straight forward interpretation of our results is
not easy, mainly because the many interacting ef-
fects involved. However, a clear influence of particle
size and the acid–basic properties of the support
were evidenced. The nature of the precursor salt has
an indirect effect on the catalytic properties of the
systems, as it seems to influence the particle size
which in turn is responsible for selectivity. ZnO and
ZrO2 are known to be components of the low and
high pressure methanol synthesis commercial cata-
lysts. Furthermore, ZrO2 is in itself an iso-synthesis
catalyst. The role of these supports in CO hydro-
genation does not seem to be that of an inert support
for palladium; instead they seem to play the role of
co-catalysts.
4. Conclusions
This investigation provides additional evidence that
the selectivity and activity of supported palladium cat-
alysts for the CO hydrogenation reaction depends on
the nature of the salt precursor, particle size and on the
acid–base properties of the support. From the analysis
of our experimental data, the following conclusions
may be drawn
1. In general, higher activities are obtained when us-
ing PdCl2 instead of Pd(NO3)2.

M. Josefina Pe´rez-Zurita et al. / Journal of Molecular Catalysis A: Chemical 206 (2003) 339–351
351
2. The Pd(NO3)2/ZnO catalyst shown the highest
References
methanol selectivity which was related to the
presence of large particles of a PdZn alloy.
3. The particle size was dramatically influenced by
the salt precursor, PdCl2 lead to smaller particles
than Pd(NO3)2.
4. Particle size seems to play an important role on
the catalytic performance: large particles favours
oxygenated compounds and small particles favours
hydrocarbon formation.
5. As seen from the CH3OH/DME ratio, the order of
acidity of the supports is: Zr(OH)4 > ZrO2 > ZnO
independently of the salt precursor used.
6. Pd/ZrO2 catalysts were more active than Pd/
Zr(OH)4 and Pd/ZnO catalysts.
7. A double bifunctionality on the catalysts for the
production of higher oxygenated seems evident. On
the one hand, an acid–base bifunctionality of the
support seems to be needed for the formation, sta-
bilisation, chain growth and further reactivity of
intermediates such as formate as well as for the
dissociative adsorption of CO, and on the other
hand, a metal-support bifunctionality, were the pal-
ladium metal particles would play the role of hy-
drogenating the intermediates and possibly adsorb
non-dissociatively the CO.
[1] M.L. Poutsma, L.F. Elek, P.A. Ibarbia, A.P. Risch, J.A. Rabo,
J. Catal. 52 (1978) 157.
[2] M.N. Berube, B. Sung, A. Vannice, Appl. Catal. 31 (1987)
133.
[3] L.R. Radovic, A. Vannice, Appl. Catal. 29 (1987) 1.
[4] F. Fajula, R. Anthony, J. Lunsford, J. Catal. 73 (1982) 237.
[5] R.F. Hick, A.T. Bell, J. Catal. 90 (1984) 205.
[6] S.D. Jackson, B.J. Brandreth, D. Winstanley, Appl. Catal. 27
(1986) 325.
[7] S.H. Ali, J.G. Goodwin Jr., J. Catal 176 (1998) 3.
[8] A. Gotti, R. Prins, Catal. Lett. 37 (1996) 43.
[9] A. Gotti, R. Prins, J. Catal. 175 (1998) 302.
[10] L.E.Y. Nonneman, V. Ponec, Catal. Lett. 7 (1990) 213.
[11] L.E.Y. Nonneman, A.G.T.M. Bastein, V. Ponec, R. Burch,
Appl. Catal. 62 (1990) L23.
[12] V. Ponec, Catal. Today 12 (1992) 227.
[13] A.F. Gusovius, T.C. Watling, R. Prins, Appl. Catal. 188 (1999)
187.
[14] W.Y. Kim, T. Hanaoka, M. Kishida, K. Wakabayashi, Appl.
Catal. 155 (1997) 283.
[15] W.Y. Kim, H. Hayashi, T.M. Kishida, H. Nagata, K.
Wakabayashi, Appl. Catal. 169 (1998) 157.
[16] V.L. Kirillov, Y.A. Ryndin, React. Kinet. Catal. Lett. 59 (2)
(1996) 351.
[17] J.S. Rieck, A.T. Bell, J. Catal. 106 (1987) 46.
[18] M.L. Cubeiro, J.L.G. Fierro, Appl. Catal. A 168 (1998) 307.
[19] N. Iwasa, T. Mayanagi, N. Ogawa, K. Sakata, N. Takezawa,
Catal. Lett. 54 (1998) 119.
[20] M.L. Cubeiro, M.J. Pérez-Zurita, V. Londoño, M. Goldwasser,
E. Pietri, L. Garc´ıa, Rev. Soc. Venezolana Catal., in press.
[21] H. Ohtsuka, T. Tabata, Appl. Catal. B.: Environ. 26 (2000)
275.
[22] M. Kilo, C. Schild, A. Wokaun, A. Baiker, J. Chem. Soc.
Faraday Trans. 88 (10) (1992) 1453.
Acknowledgements
[23] A. Sekulic´, K. Furic´, M. Stubicar, J. Mol. Struct. 410 (1997)
275.
[24] T. Iiazuka, Y. Tanaka, K. Tanabe, J. Catal. 76 (1982) 1.
[25] W.J. Shen, M. Okumura, Y. Matsumura, M. Haruta, Appl.
Catal. A 213 (2001) 225.
The authors wish to acknowledge the financial sup-
port received from CONICIT-CONIPET through the
Project No. 97-003739. The XRD analyses were con-
ducted at Université des Sciences et Techniques de
Lille.
[26] A.T. Bell, Stud. Surf. Sci. Catal. 136 (2001) 13.