Você está na página 1de 12

Applied Catalysis A: General 232 (2002) 147158

Co/SiO2 catalysts for selective hydrogenation of crotonaldehyde II:


influence of the Co surface structure on selectivity
E.L. Rodrigues, J.M.C. Bueno
Departament of Chemical Engineering, Universidade Federal de So Carlos, Caixa Postal 676, 13565-905 So Carlos (SP), Brazil
Received 10 September 2001; received in revised form 29 January 2002; accepted 1 February 2002

Abstract
The effect of interaction of silica-supported CoOx on catalytic properties of Co/SiO2 catalysts was investigated. The CoOx
precursors supported on silica were obtained by impregnation of the support with cobalt nitrate solution. Cobalt precursors,
with different interactions with support, were obtained by changes of impregnation solvent, drying time and temperature; and
calcination temperature. The Co surface structure of the catalysts was characterized by diffuse reflectance FTIR spectroscopy
(DRIFTS) of adsorbed CO and temperature-programmed desorption of hydrogen (TPD-H2 ). The TPD-H2 and DRIFT of
CO spectra indicate the presence of at least four different Co surface sites, labeled , , , and . The relative amount of
these species varied depending on the degree of CoOx interaction with SiO2 in precursors. For crotonaldehyde (CROALD)
hydrogenation in gas phase, selectivity to crotyl alcohol and butanol depended strongly on the /( + ) ratio. Interestingly,
selectivity to butyraldehyde does not depend on , , or sites present. 2002 Elsevier Science B.V. All rights reserved.
Keywords: Co catalysts supported; Crotonaldehyde hydrogenation on supported Co catalysts; Desorption of hydrogen adsorbed on Co catalysts

1. Introduction
Unsaturated alcohols (UOLs) obtained by selective
catalytic hydrogenation of ,-unsaturated aldehydes
are important intermediates in perfumes, flavoring,
and pharmaceutical production. Selective hydrogenation of these aldehydes to the corresponding UOL is
a rather difficult process, especially under heterogeneous catalysis conditions. Within the last decade, extensive efforts have been made to understand the different effects that control the intramolecular selectivity
to C=O groups hydrogenation [1,2]. It is known that
a properly designed catalyst should strongly polarize
the carbonyl group in an adsorbed aldehyde molecule,
enhancing its reactivity to hydrogen and diminishing
carboncarbon double bond activity.
Corresponding author.
E-mail address: jmcb@power.ufscar.br (J.M.C. Bueno).

The selective hydrogenation of the C=O group on


the noble metals Pt, Ru, and Rh is very difficult because the C=C group hydrogenation is thermodynamically and kinetically favored [3,4]. However, Englisch
et al. [5], in studying the hydrogenation of crotonaldehyde (CROALD) in gas phase over Pt/SiO2 , observed
an increase in selectivity toward the UOL from 8 to
43% with increases of metal particle size. It was observed that the high fraction of Pt(1 1 1) surfaces favors adsorption of the ,-unsaturated aldehydes via
C=O group. Theoretical calculations and studies on
single crystal catalytic surfaces showed that the adsorption structure of ,-unsaturated aldehydes depends on the metal surface planes and the constituents
of C=C group [6,7].
Among no noble metal, Co shows higher hydrogenation selectivity to UOL [810]. The Co shows
higher selectivity than the noble metals as Pt, Rh and
Ru. A strong selectivity dependence was observed by

0926-860X/02/$ see front matter 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 6 - 8 6 0 X ( 0 2 ) 0 0 0 9 0 - X

148

E.L. Rodrigues, J.M.C. Bueno / Applied Catalysis A: General 232 (2002) 147158

Nitta et al. [8,9] on Co particle size in the liquid phase


hydrogenation of cinnamaldehyde and CROALD over
monometallic cobalt catalyst that was prepared by the
reduction of cobalt chrysolite, Co3 (OH)4 SiO5 . Selectivity to UOLs was enhanced with increase of cobalt
particle size. Hindered adsorption of cinnamaldehyde
by olefinic bond, e.g. the steric repulsion between
flat metal surfaces in large metal particles and aromatic ring, was responsible for high selectivity toward C=O group hydrogenation [11]. Through the
study of the liquid phase hydrogenation of acrolein
on Co/silica-alumina catalyst prepared from cobalt
nitrate, Nitta et al. [9] observed that the higher Co
loading resulted in large particle size of Co and
higher selectivity to allyl alcohol. This result suggests
that structure sensitivity in this reaction cannot be
explained only in terms of the steric effect of bulky
substituent on substrate -carbon.
Even though satisfactory selectivity towards UOL
has been obtained with monometallic Co catalysts
through selective hydrogenation of ,-unsaturated
aldehydes [8,9], studies involving the influence of the
catalyst surface structures in this reaction are scarce.
In a previous study [12], we showed that CROALD
hydrogenation selectivity depends on the degree of interaction of CoOx precursors species and silica support.
In the present work, we attempt to find correlation
of different CoOx precursor species over silica, and
the Co superficial structure and its catalytic properties
using as a reaction test hydrogenation of CROALD in
gas phase.

2. Experimental
2.1. Catalyst preparations
Degussa silica, aerosil-200 (200 m2 /g), was used as
the support. Two Co catalysts series were prepared by
the impregnation method, using different treatments
to obtain different metalsupport interactions [13].
The first precursor series of Co catalysts was prepared from aqueous solutions of Co(NO3 )2 (99.999%,
Aldrich) with different cobalt loadings. To obtain
low CoOx SiO2 interaction, resulting gel was dried
at 110 C for 48 h and heated to 200 C at 2 C/min
in synthetic air for 6 h [13]. This samples series was

labeled X-CoSi-LI, where X is the Co loading wt.%


and LI means low interaction.
The second precursor series of Co catalysts was obtained by SiO2 impregnation with Co(NO3 )2 ethanol
solution. To produce high CoOx SiO2 interaction,
the gel was dried at room temperature and heated to
400 C for 1 h in synthetic air. This samples series
was labeled X-CoSi-HI, where X is the Co loading
%w/w and HI means high interaction series.
The precursors of X-CoSi-LI and X-CoSi-HI catalysts series were reduced before reaction in situ
by heating in H2 from room temperature to 480 C
at 10 C/min, and continuing for 2 h in H2 flow of
60 cm3 /min.
2.2. Catalyst characterization
Catalysts were characterized by inductively coupled plasma atomic emission spectroscopy (ICP),
temperature-programmed reduction (TPR), X-ray
powder diffraction (XRD), temperature-programmed
desorption of hydrogen (TPD-H2 ), and infrared spectroscopy of adsorbed CO.
The TPR analyses were performed on a Micromeritics Pulse Chemisorb 2705. The precursor samples
(about 30 mg) were heated in a U-shaped fused silica
cell with 5% H2 /N2 in a total flow of 40 cm3 /min.
The samples were heated from room temperature to
1000 C at 10 C/min. Gas flow from the U-shaped
fused silica cell passed through a trap at 60 C; H2
consumption was monitored with a thermal conductivity detector (TCD) connected to a PC data station.
XRD patterns were collected with a Siemens
D-5000 diffractometer using Cu K radiation with a
graphite monochromator.
The TPD-H2 profiles were obtained with a
Micromeritics Pulse Chemisorb 2705. All gas streams
(UHP-5.0, AGA, Brazil) were purified by a zeolite
13X filter and a manganese oxide filter. The precursor samples (about 100 mg) in the U-shaped fused
silica cell were reduced by heating in H2 flow from
room temperature to 480 C at 10 C/min and staying
in this temperature for 2 h. This reduction condition
was similar to that utilized for activation in reaction
tests. The sample was cooled to room temperature
in H2 flow and exposed for 1 h to H2 to ensure
the maximum adsorption [14]. Next, samples were
cooled to 30 C and purged in Ar for 1 h. Hydrogen

E.L. Rodrigues, J.M.C. Bueno / Applied Catalysis A: General 232 (2002) 147158

desorption was obtained by heating at 10 C/min in


Ar at 30 cm3 /min flow. The H2 desorbed was monitored with a TCD. The amount of HCo species was
calculated from desorption peaks area obtained by
Gaussian deconvolution of TPD-H2 spectra.
The infrared spectra of CO adsorbed were collected
with a Nicolet Magna 750 infrared spectrometer with
diffuse reflectance, Fourier transform, and a Spectra
Tech cell (DRIFTS). For CO adsorption, the following
procedure was used: (i) the undiluted oxide sample
was placed in a DRIFTS cell with ZnSe windows;
(ii) the DRIFTS cell was purged with dry N2 free of
O2 for 15 min; (iii) the sample was reduced in H2 /N2
(25/75 v/v) by heating to 480 C at 5 C/min under a
total flow of 25 ml/min. The sample was kept at 480 C
for 2 h; (iv) after reduction, the sample was cooled in
two ways: (a) from 480 to 25 C in H2 /N2 mixture;
(b) from 480 to 300 C in H2 /N2 mixture, proceeding
from 300 to 25 C in a N2 stream; (v) back-ground acquisition was done at 25 C and used as a reference for
the spectra of adsorbed CO; (vi) CO was introduced in
the DRIFTS cell by two procedures: (a) CO pulse of
1 l in N2 flow; (b) continuous flow of CO/N2 (with
4% CO v/v) in a total flow of 25 cm3 /min. The sample
in the DRIFT cell was kept on stream until the spectra
of adsorbed species began to reach steady-state. The
cell was then purged with N2 at 25 C and spectra
were recorded together with desorption time.
2.3. Reaction
Vapor-phase hydrogenation of CROALD was carried out in a tubular fused silica microreactor (9 mm
i.d.) at 120 C and under atmospheric pressure. The
precursors were reduced in situ by heating in H2 flow
from room temperature to 480 C at 10 C/min, staying
at this temperature for 2 h. The reactor was then cooled
to the desired reaction temperature. The gases (H2 and
N2 , 99.999% purity, AGA) were purified by passing
over MnO/SiO2 and, 4 A molecular sieve. A constant
flow of vapor-phase CROALD was established by
using a bubbling H2 /N2 (1/1 v/v) gas mixture, as the
carrier gas, through a thermostabilized saturator containing liquid CROALD (>99% purity, Aldrich). Reactions were carried out with a CROALD/H2 /N2 feed
of 1/45/45 at total flow rate of 60 ml/min. The feed
reactor was bypassed the reactor until steady-state
composition was reached. Reactor effluents were

149

analyzed by on-line gas chromatography, employing a


13 ft 20 M carbowax on a chromosorb-W column. The
initial reaction data were obtained by extrapolation to
t = 0.
The reaction data of activity and selectivity for different samples were normalized to isoconversion of
CROALD. For different catalysts, reaction data were
scaled to a similar molar flow rate and catalyst amounts
were adjusted to obtain similar conversion values. Selectivities were scaled to isoconversion, defined as the
number of mols of product i formed by the number
of mols of converted CROALD in the hydrogenation
0
was calculated from the rate of
reaction. The YCarbon
carbon deposited divided by 4rate of total CROALD
converted.
3. Results and discussion
3.1. Catalyst characterization
3.1.1. TPR
The hydrogen consumption profiles of the precursor
series X-CoSi-HI and X-CoSi-LI after calcinations are
shown in Fig. 1. As previously shown [12], the degree
of CoOx interaction with SiO2 support depends on Co
loading and preparation conditions of samples. Two
regions of H2 consumption can be identified according to reduction temperature of CoOx species present:
(a) lower temperatures region (<550 C) and (b)
higher temperatures region (>550 C).
X-CoSi-LI series samples show predominant CoOx
species that are reduced at temperatures below 550 C
(Fig. 1a). The first peak, with a maximum at 200 C,
can be assigned to the reductive decomposition of
residual nitrate present in X-CoSi-LI precursor series [13,15,16]. Rosynek and Polansky [15] verified,
through TPR monitored with mass spectrometric analysis, that the reduction of cobalt salt, Co(NO3 )2 6H2 O,
occurred in two steps. The most intense hydrogen consumption peak occurred with maximum at 280 C. It
was due to reductive decomposition of the nitrate ion:
Co(NO3 )2 +3H2 CoO+2NO+3H2 O. The second
peak occurred at 370 C and was attributed to subsequent reduction of CoO to metallic cobalt. The reduction of cobalt salt is influenced by the SiO2 support,
the peak due to reductive decomposition of the nitrate
ion in Co(NO3 )2 /SiO2 is shifted downward slightly to
250 C [15]. Kogelbauer et al. [16] described similar

150

E.L. Rodrigues, J.M.C. Bueno / Applied Catalysis A: General 232 (2002) 147158

Fig. 1. TPR profiles of sample series: (a) X-CoSi-LI and (b) X-CoSi-HI.

results for a Co(NO3 )2 /Al2 O3 precursor, which exhibit


a TPR peak at 230 C.
Unsupported Co3 O4 exhibits two reduction peaks
at 300 and 360 C, respectively, according to some
authors [13,17,18]. XPS results [18] show that theses peaks were due to reduction of Co3 O4 phase in
a two-step process (Co3 O4 3CoO 3Co). This
happened due to diffusional limitation during a TPR
experiment. For precursors with small particles of supported Co3 O4 phase, reduction of this phase in TPR
experiments has been observed to occur in one-step
[13,1517]. The XRD pattern for X-CoSi-LI precursor
series exhibits broad peaks at the position characteristic of Co3 O4 , indicating the formation of small particles of Co3 O4 . The ratio of H2 consumption between
peaks at 300 and 360 C increases with Co loading
decrease. Okamoto et al. [19] ascribed the first peak
to small Co3 O4 particles and the subsequent reduction peak maximum to larger particles. From these evidence, the TPR peaks with maxima at 300 and 360 C
for X-CoSi-LI samples, observed in Fig. 1 a, can be
assigned to Co3 O4 -like phase reduction [13,17]. However, the H2 consumption at 360 C can overlap with
the reduction of CoO, formed by decomposition of
residual nitrate [15].
The shoulder at 430 C could be related to a reduction of larger Co3 O4 crystallites and/or species
of Con+ having an intermediate interaction with
silica support [13,17]. Ratios between area of the
peak at 430 C, and sum of areas of peaks at 300
and 360 C (ACon+ /ACo3 O4 ) for X-CoSi-LI precursor

series are listed in Table 1. The samples show increase of ACon+ /ACo3 O4 ratio values with Co loading
decrease. Since the samples with low loading show
higher fractions of cobalt reducible above 550 C, it
is reasonable to suppose that the shoulder at 430 C
is more likely due to the reduction of Con+ species
interacting with silica surface than to the reduction of
larger Co3 O4 crystallites. Although the reduction of
CoO bulk should occur at 370 C [15], the species of
CoO cluster interacting with support is reducible in
the high temperature region of 400 to 500 C [13,15].
For X-CoSi-LI series, the TPR in region (b) shows
two small peaks at 650 and 800 C. Theses H2 consumption are characteristic of Co+ species interacting strongly with silica (Fig. 1a). Strong interaction
between cobalt species and silica can cause the formation of silicates and/or hydrosilicates [13,15,1921].
Silica contains many silanol groups, which may react
with cobalt ions in solution forming silicates and/or
Table 1
Influence Co loading in the ratio between the Con+ and Co3 O4
species
Sample

ACon+ /ACo3 O4 a

8-CoSi-LI
14-CoSi-LI
25-CoSi-LI
13-CoSi-HI
35-CoSi-HI

0.76
0.64
0.58
1.47
0.24

Area peaks (ACon+ ; ACo3 O4 ) were calculated by Gaussian


deconvolution.
a

E.L. Rodrigues, J.M.C. Bueno / Applied Catalysis A: General 232 (2002) 147158

cobalt hydrosilicate. van Steen et al. [13] obtained a


TPR peak at 900 C of a precursor of Co/SiO2 catalyst
calcined at 1000 C and the H2 consumption indicated
reduction of divalent cobalt. This result was in accordance with XPS measurements reported by Ming and
Baker [22]. Under these conditions, the formation of
orthocobalt silicate by a solid-state reaction between
CoO and SiO2 is known [23]. From this evidence, hydrogen consumption between 550 and 850 C could be
ascribed to the reduction of cobalt hydroxysilicate-like
species or surface cobalt silicate species.
In the X-CoSi-HI series, different CoOx species
can be formed, depending strongly on the Co loading (Fig. 1b). For sample with highest Co loading (35-CoSi-HI), the main species are reduced at
T < 550 C (region a). The H2 consumption profile
for 35-CoSi-HI sample is similar to that obtained
for X-CoSi-LI series at the temperature region from
300 to 500 C. However, for this sample, the first
two peaks at 375 and 417 C are characteristics of
the reduction of Co3 O4 -like particles in two steps
(Co3 O4 CoO Co). This interpretation can be
supported by XRD results that show the presence of
large particle size of Co3 O4 phase (Fig. 2). In addition, TPR results show that the H2 consumption of the
first peak area is approximately one-fourth of the sum
of areas of the two peaks, suggesting the reduction of
large Co3 O4 -like particles in two steps.
The sample with intermediate Co loading (13-CoSiHI) shows CoOx species reducible at regions (a) of
lower temperatures region (<550 C) and (b) higher
temperatures region (>550 C) (Fig. 1b). The profiles
of H2 consumption in region (a) suggest the presence
of CoOx species similar to those obtained for sample
with high Co loading (35-CoSi-HI). The H2 consumption at higher temperature region (b), the TPR decom-

151

Fig. 2. XRD patterns of samples: (a) 8.6-CoSi-HI, (b) 13-CoSi-HI,


(c) 35-CoSi-HI, (): Co3 O4 .

position spectra suggest the presence of two species,


with maximum reduction temperature of about 770
and 850 C. These species correspond supposedly to
cobalt hydroxysilicate-like species or of surface cobalt
silicate species [13,15,1921].
For sample with lowest Co loading (8.6-CoSi-HI),
the main Co species present are reduced at region (b)
T > 550 C (Fig. 1b), shows two species with maximum reduction temperatures of about 770 and 850 C.
The results show that low Co loading and higher calcinations temperature for this series samples favored
high interaction between CoOx species and silica,
similar to that previously described by van Steen et al.
[13].
Table 2 shows the amount percentage of reduced
Co, obtained from H2 consumption by X-CoSi-LI
and X-CoSi-HI series during the reduction procedure.

Table 2
TPD-H2 results and percentage of Co reduced of some studied catalysts
Sample

8-CoSi-LI
14-CoSi-LI
25-CoSi-LI
8.6-CoSi-HI
13-CoSi-HI
35-CoSi-HI

Co reduced (%)

100
100
100
2025
3545
100

H/Co (102 )

8.8
8.2
8.9

4.0
4.6

Hydrogen species (%)

/( + )

/( + )

59
39
35

32
42

30
36
35

56
30

11
16
17

12
16

0
10
13

0
12

5.4
1.5
1.2

2.7
1.5

2.7
1.4
1.2

4.7
1.1

152

E.L. Rodrigues, J.M.C. Bueno / Applied Catalysis A: General 232 (2002) 147158

Fig. 4. Correlation among the ratio /( + ) and ratio of


the species Con+ /Co3 O4 , and with the ratio of the species
(Con+ + Co+ )/Co3 O4 , (): Con+ /Co3 O4 and (): (Con+
+ Co+ )/Co3 O4 .

Fig. 3. TPD-H2 profiles of sample series X-CoSi-LI and X-CoSi-HI.

X-CoSi-LI sample series shows that 100% of cobalt


ions are reduced after activation, independently of
Co loading. However, for the X-CoSi-HI series, the
amount percentage of Co reduced after activation
depends strongly on the cobalt loading.
3.1.2. TPD-H2
The TPD-H2 profiles for different X-CoSi-LI
and X-CoSi-HI series are shown on Fig. 3. These
TPD-H2 profiles show peaks at 50, 95, 120160,
and 160300 C, assigned to different HCo species,
which are labeled , , , and , respectively. Different desorption temperature of hydrogen due to
different superficial metallic structures was previously
observed for Pd [24], Co [12,25], and Pt [26]. Our
results suggest at least four different types of superficial metallic sites on Co surface samples. The relative
amounts of HCo species are summarized in Table 2.
The hydrogen species, , is not observed in
8-CoSi-LI and 13-CoSi-HI samples. However, the
species concentration increases with increase of
Co loading, and consequently with the decrease
of CoOx SiO2 interaction in precursor samples.

Correlations between the ratios of the hydrogen


species /( + ) and the different precursors species
of CoOx interacting with the silica are presented in
Fig. 4. From this figure, it can be seen that there are:
(i) a correlation exists between ratio /( + ) and
the ratio of species Con+ /Co3 O4 and (ii) a correlation
exists between ratio /( + ) and the ratio of the
species (Con+ +Co+ )/Co3 O4 , where Co+ are cobalt
species strongly interacting with silica, with reduction
temperatures above 500 C. It suggests that the sites
adsorbing hydrogen species dominate in relation to
those sites adsorbing hydrogen species and when
the interaction increases between the cobalt oxide
species and silica support. No correlations are obvious
between the concentration of sites adsorbing hydrogen
species and the different species of CoOx interacting with silica. The results suggest that adsorbing hydrogen and species are predominantly formed by
reduction of Co3 O4 phase and the adsorbing hydrogen
species are predominantly generated from reduction of two precursors Con+ and Co+ species, which
have intermediate and strong interaction with silica,
respectively.
It is worth noting that independent of the presence of Co3 O4 in samples with lower Co loading
(X-CoSi-LI and X-CoSi-HI), sites adsorbing hydrogen
were not observed. The results suggest that a decrease in loading of the Co3 O4 phase in the precursor
favors larger uniformity of Co sites generated in the
catalysts.

E.L. Rodrigues, J.M.C. Bueno / Applied Catalysis A: General 232 (2002) 147158

153

Fig. 5. DRIFT spectra of CO adsorbed on the 14-CoSi-LI sample: in the presence of preadsorbed hydrogen obtained (a) by CO pulse on
N2 stream; (b) by CO/N2 flow; (c) in CO/N2 (4% Co v/v) flow; (d) after desorption in N2 flow for 3 min at 25 C; with time of desorption
when CO adsorption was done; (e) in the presence of preadsorbed hydrogen and (f) in the absence of preadsorbed hydrogen.

3.1.3. DRIFTS-CO
Fig. 5a shows DRIFTS spectra of CO adsorbed on
the 14-CoSi-LI sample (cooled in a H2 /N2 gaseous
mixture stream after reduction), obtained for CO pulse
on N2 stream. The spectra show a continuous shift
of the position band of 25 cm1 to higher frequencies (up to 2005 cm1 ) with increasing CO coverage. In the spectra of sample exposed to a continuous flow of mixture CO/N2 = 4/96 by volume, new
bands at 1950, 2030, 2050, 2115, and 2170 cm1 are
observed, whose intensities increase with the contact
time between CO and catalyst (Fig. 5b and c). During

evacuation the bands at 2170, 2115, and 2050 cm1


disappeared, while the band at 2030 cm1 decrease,
shifting to 2000 cm1 (Fig. 5d and e).
As shown in Fig. 5c, the DRIFTS spectra of adsorbed CO on reduced 14-CoSi-LI sample was cooled
to 300 C in H2 /N2 mixture followed by cooling in
N2 stream from 300 to 25 C at 5 C/min. TPD results
suggest that there is no hydrogen adsorbed on Co surface (see Fig. 3). Bands at 2005, 2030, 2070, 2115,
and 2170 cm1 are revealed in the spectra (Fig. 5c).
During evacuation, the bands at 2115 and 2170 cm1
disappeared, while the band at 2070 cm1 initially

154

E.L. Rodrigues, J.M.C. Bueno / Applied Catalysis A: General 232 (2002) 147158

shifted to 2060 cm1 (Fig. 5d). This band decreases


in intensity and shifts to 2005 cm1 with increasing
evacuation time (Fig. 5f).
A similar result was reported previously by
Heal et al. [27] and Choi et al. [28] for Co/SiO2
and Co/Al2 O3 , respectively. At a CO pressure of
10 Torr, the authors [27,28] reported a band at
21602180 cm1 and a doublet band between 2030
and 2060 cm1 . Kadinov et al. [29] also observed
that the spectrum of Co/Al2 O3 at 70 Torr displayed a
band at 2055 cm1 with intensity comparable to that
at 2025 cm1 , a weak band around 2110 cm1 , and a
strong band at 2175 cm1 .
The small band observed at 2170 cm1 does not
contain a substantial contribution from the gaseous
CO stretching vibration at 2143 cm1 according to
Lapidus et al. [30]. The band in the 21602180 cm1
regions has been assigned to linearly adsorbed CO on
Co2+ or Co3+ ions [27,2931].
According to Kadinov et al. [29], the band at
2115 cm1 can be assigned to CO adsorbed on Co+
ions and this band would also be attributed to CO
complexes with partially reduced cations, which are
an intermediate species during reduction of Co3 O4
to Co metal [31]. Yu Khodakov et al. [31] observed
that the band at 2181 cm1 is likely due to the difficulty to reduce Co oxidized species. The authors
demonstrated that this band was the most intense for
the sample, where crystalline Co3 O4 concentration
is rather low and most of Co is in an amorphous
phase.
Several studies report one band in the 20002050
cm1 region after CO adsorption onto reduced cobalt
catalysts. This band has been attributed to linearly
bonded molecules with surface cobalt atoms of relatively large metal particles [27,2932].
When the interaction between CO and the
14-CoSi-LI sample surface takes place in the presence of preadsorbed hydrogen, the bands observed
show higher intensity than in absence of hydrogen
(Fig. 5c and d).
During adsorption of CO onto the 14-CoSi-LI sample in the presence of preadsorbed hydrogen, the intensity of band at 2050 cm1 increases with an increase
in contact time between CO and the catalyst. This
band disappears during the evacuation of CO from gas
phase at room temperature. However, in the absence
of preadsorbed hydrogen, a new band at 2070 cm1

is observed. On evacuation, the band shifted to


2060 cm1 and its intensity decreased with time
(Fig. 5e and f).
The different behavior of the band in the region
20702050 cm1 could be explained by the presence
of other species. In the absence of preadsorbed hydrogen, the band at 2070 cm1 should be due to the
overlapping bands caused by CO species on metallic
cobalt with weak electron-donor properties (with partial positive charge Co+ ) [29,30] and polycarbonyl
species [29]. After evacuation at room temperature,
the considerable decrease in relative intensity of the
band at 2060 cm1 , in relation to the bands at 2005
and 2030 cm1 could be explained by the lower stability of the COCo+ structure compared to that of the
polycarbonyl species. In the presence of preadsorbed
hydrogen, the band at 2050 cm1 has been attributed
to a structure of the hydrocarbonyl type [27].
The bands at 2030 and 2005 cm1 can be assigned
to CO adsorbed in a linear form on Co metals with
different surface structure [2731,33]. Both band positions are independent of the preadsorbed hydrogen
presence.
The bands in the 20001800 cm1 region have been
attributed to the CO adsorbed in a bridged form on
metal particles [30,31,34,35]. In the presence of preadsorbed hydrogen, a band at 1950 cm1 is observed on
the spectrum of 14-CoSi-LI sample (Fig. 5c and d).
The DRIFTS spectra of CO adsorbed on X-CoSi-LI
and X-CoSi-HI samples series are displayed in Fig. 6.
The band at 2170 cm1 is assigned to linearly adsorbed CO on Co2+ or Co3+ ions from oxide species.
The relative amount of these species is favored with
decrease in Co loading. The DRIFTS-CO results show
that these species dominate in the 13-CoSi-HI sample
which is in agreement with TPR results. On the other
hand, the relative intensity of the band at 2115 cm1
attributed to adsorbed CO on Co+ ions is practically
unchanged in the samples studied.
The bands observed in the region 20502000 cm1
(Fig. 6) are assigned to CO adsorbed in linear form
on Co metal sites and the broad band in region
20001900 cm1 is related to CO adsorbed in bridge
form on Co metal sites.
The TPD-H2 and DRIFTS of CO adsorption results
showed four different identifiable metallic superficial
structures. Based on these results and supposing that
the site reacting more strongly with hydrogen also

E.L. Rodrigues, J.M.C. Bueno / Applied Catalysis A: General 232 (2002) 147158

155

Fig. 6. DRIFT spectra of CO adsorbed on the X-CoSi-LI and X-CoSi-HI samples in the presence of preadsorbed hydrogen: (a) in CO/N2
flow and (b) after desorption in N2 flow for 3 min at 25 C.

interacts strongly with CO, the following correlation


can be suggested: sites adsorbing hydrogen species
may be responsible for CO adsorption characterized
by IR band at 2050 cm1 . Consequently, the bands
at 2030, 2005, and 19851930 cm1 may be assigned
to the adsorbed CO at sites which adsorbed hydrogen species , , and , respectively. These suppositions are supported by the good correlation between
the ratio of the hydrogen species /( + ) and the
ratio of intensity of the three bands I2030 /I(2005+1950) ,
shown in Fig. 7 (where I2030 , I2005 , and I1950 correspond to the intensity of bands at 2030, 2005, and
19851930 cm1 , respectively).
Therefore, it can be concluded that the samples
show four surface sites, which were denoted as , ,
, and . The generation of these sites is a function
of the CoOx species structures formed on silica. Precursors with larger Co loading and prepared in such a
way as to obtain less metalsupport interaction are derived mainly by Co3 O4 particles that are responsible
for generating of the and sites. The Co loading
and preparation methods influence the distributions of
Co sites in catalysts. The increase of CoOx SiO2 interaction in the precursor favors the generation of

sites to the detriment of and sites. Besides, increased amounts of CoOx species that are strongly interacting with the support in the oxide precursor favor
the presence of Co2+ and Co3+ ions that are difficult
to decrease in the reduced sample. No correlation of
site generation with structure of the precursor CoOx
species was found.

Fig. 7. Correlation between ratio of intensity of CO bands adsorbed


I2030 /I(2005+1950) and the ratio of the hydrogen species /( + ).

156

E.L. Rodrigues, J.M.C. Bueno / Applied Catalysis A: General 232 (2002) 147158

Table 3
Products formation rate, ri (mol/g Co s), in different time on stream
Sample

rT0

rH0

X0 (%) ri0

X1 (%) ri0

CrOH BAL BOL rCrOH /


rBAL
8-CoSi-LI
14-CoSi-LI
25-CoSi-LI
8.6-CoSi-HI
13-CoSi-HI
35-CoSi-HI

52.5
62.5
45.8
19.6
24.7.
41.1

43.9
51.7
40.3
9.06
15.4
35.6

30
30
28
30
34
35

5.80
6.25
2.97
1.31
3.05
3.39

31.7
36.1
29.4
6.44
10.4
26.4

6.44
9.31
7.92
1.31
1.94
5.83

0.18
0.17
0.10
0.20
0.29
0.13

X2 (%) ri0

CrOH BAL BOL rCrOH /


rBAL
19
14
17
7
12
22

5.53
4.52
3.17
1.05
1.85
3.20

25.0
20.8
20.5
4.13
6.26
19.7

3.83
3.89
4.41
0.39
0.62
3.33

0.22
0.22
0.15
0.26
0.29
0.16

CrOH BAL BOL rCrOH /


rBAL
14
12
12
5
6
15

3.88
4.20
2.25
0.83
1.12
2.45

17.0
18.0
14.4
2.95
2.94
14.0

2.33
3.2
1.67
0.31
0.32
2.0

0.23
0.23
0.16
0.28
0.37
0.18

Xi : conversion of CROALD: i = 0 correspond at t = 0, i = 1 correspond at after period initial of deactivation, i = 2 correspond at


steady-state; rT0 = initial total rate of CROALD converted, rH0 = initial rate of CROALD hydrogenation.

3.2. Catalytic activity

residue deposition could be responsible for such deactivation. The catalyst deactivation has been related
to irreversibly adsorbed organic molecules on the
catalyst surface [36,37], and decarbonylation of the
reactant molecule yielding carbon monoxide, which
is adsorbed on metal under reaction conditions [36].
The products formation rates (ri ), at different times
on stream (Tables 3 and 4) show that for products of
partial hydrogenation, the butyraldehyde (rBAL ) formation showed stronger deactivation than crotyl alcohol (rCrOH ) formation. The results obtained through
isoconversion at different times on stream demonstrate
similar behavior such as increase of ratio rCrOH /rBAL
and decrease of the rate of butanol formation (rBOL )
with time on stream. From these results, it is reasonable to suppose that carbon deposition on Co surface
can block higher hydrogenation sites and cause geometrical effects that decrease the hydrogenation of
C=C bond. To minimize the effect of surface modifi-

Reaction data for CROALD hydrogenation over


silica-supported Co catalysts are summarized in
Table 3. The products observed during CROALD hydrogenation were butyraldehyde (BAL), crotyl alcohol
(CrOH), and butanol (BOL). The products formation
rates (ri ) and total activity of the samples depend
on the catalysts preparation conditions and time on
stream. For all samples, two periods of activity and
selectivity with time on stream were observed. The
first period is characterized by strong deactivation
and occurred during the first 100 min on stream. The
second period is marked by a slight deactivation; the
steady-state was reached after 500 min on stream. The
blank run showed that steady-state of composition in
the effluent reactor can be reached in 10 min on stream.
Catalyst deactivation was followed by carbon balance
values lower than 100%, suggesting that carbonaceous

Table 4
Initial total rate of CROALD converted (rT0 )
Sample

8-CoSi-LI
14-CoSi-LI
25-CoSi-LI
8.6-CoSi-HI
13-CoSi-HI
35-CoSi-HI

rT0 (mol/g Co s)

52
62
46
20
25
41

0
YCarbon

Selectivity, % (Si0 )
CrOH

BAL

BOL

13
12
7
14
20
10

72
70
73
72
68
74

15
18
20
14
12
16

0.063
0.042
0.030
0.135
0.095
0.053

ri0 (mmol/g Co s)

0
0
rCrOH
/rBAL

0
rCrOH

0
rBAL

0
rBOL

5.8
6.2
3.0
1.3
3.0
3.4

31.7
36.1
29.4
6.4
10.4
26.4

6.4
9.3
7.9
1.3
1.9
5.8

0.18
0.17
0.10
0.20
0.29
0.13

At X CROALD = 30%; rate formation of products (ri0 ), selectivities to hydrogenation products (Si0 ) and yield in carbon deposition over Co
0
catalysts (YCarbon
).

E.L. Rodrigues, J.M.C. Bueno / Applied Catalysis A: General 232 (2002) 147158

Fig. 8. Correlation between the initial selectivities to products (Si0 )


and the ratio of the sites /( + ). Initial reaction data obtained
0
0
0
, (
): SCrOH
, (): SBOL
.
at X CROALD = 30%, (): SBAL

cation by carbon deposition, the reaction data analyses were done from data obtained at the initial time
on stream.
For correlations of the initial reaction, data are expressed as initial selectivities to different hydrogenation products (Si0 ). The correlation between Si0 and
relative amount of sites (/( + )) is shown in Fig. 8.
0 ) was not influenced
Selectivity to butyraldehyde (SBAL
by ratio of sites /( + ). Good correlations are found
0
0
between the SCrOH
and SBAL
with the ratio of sites
0
/( + ). The SCrOH is favored by increase of the
relative amount of sites . The selectivity to butanol
0 ) was favored by increase of the relative amount
(SBAL
of the sum of sites and .
Different adsorptions of CROALD on metal surfaces can be considered [6]: adsorption 1,2 as an
interaction through the carbonyl groups; adsorption
3,4 as an interaction through the ethylene group; and
adsorption 1,4 as an interaction through the both
groups. Thus, the obtained results can be interpreted
taking in to consideration the different CROALD adsorption modes on the different (, , or ) Co sites.
Increased ratio of sites /( + ) can means a relative increase of CROALD adsorption by adsorption
1,2 mode rather than by adsorption 1,4 and 3,4
modes. The presence of type sites favors a CROALD
adsorption type 1,2 and consequently increases the
crotyl alcohol formation. The presence of sites type
and favors CROALD adsorption type 1,4 and
leads to preferential simultaneous hydrogenation of
C=C and C=O bonds, and then to butanol formation.

157

Fig. 9. Correlation of ratio of CO adsorbed intensity bands at


0
I2170 /I2050 with the initial yield side reaction products (YCarbon
);
data obtained at XCROALD = 30%.

Interestingly, the results suggest that the CROALD adsorption by 3,4 mode to hydrogenation of C=C bond
has no preferential sites , or .
The correlation of the initial yields of carbonaceous
0
product deposition (YCarbon
) with the ratio of bands
intensity of CO adsorbed at 2170 and 2050 cm1
(I2170 /I2050 ) is shown in Fig. 9. The bands at 2170 and
2050 cm1 correspond to the CO adsorbed on Co ions
and metal, respectively. This correlation suggests that
the increase of initial yields to carbonaceous prod0
ucts deposition (YCarbon
) is related to the presence of
cobalt ions on the surface.
4. Conclusions
The results show a dependence of interaction CoOx
precursor, with silica on type of Co surface sites
formed by reduction. The TPD of H2 and DRIFTS
of CO spectra suggested the presence of at least four
different Co surface sites, labeled , , , and . The
relative amount of these species varied depending
on the degree of CoOx interaction with SiO2 in precursors. Catalysts obtained from precursors mainly
constituted by large Co3 O4 -like species, exhibited
lower /( + ) site ratio than catalysts obtained from
precursors containing CoOx species, which strongly
interacted with SiO2 support.
For hydrogenation of CROALD in gas phase, the
increase of the relative amount of type sites favors
hydrogenation of C=O bond to crotyl alcohol formation. Increase of the relative amount of type and

158

E.L. Rodrigues, J.M.C. Bueno / Applied Catalysis A: General 232 (2002) 147158

sites favors simultaneous hydrogenation of the C=C


and C=O bonds to butanol formation. Hydrogenation
of C=C bond to butyraldehyde formation does not depend on the , or type sites.

Acknowledgements
This work was supported by the FAPESP (Fundao
para o Amparo a Pesquisa do Estado de So Paulo)
and CNPq (Conselho Nacional de Desenvolvimento
Cientifico e Tecnolgico, Brasil).
References
[1] P. Gallezot, D. Richard, Catal. Rev. Sci. Eng. 40 (1998) 81.
[2] P. Claus, Topics Catal. 5 (1998) 51.
[3] P. Rylander, Catalytic Hydrogenation Over Platinum Metals,
Academic Press, New York, 1967.
[4] C.L. Thomas, Catalytic Processes and Proven Catalysts,
Academic Press, New York, 1990.
[5] M. Englisch, A. Jentys, J.A. Lercher, J. Catal. 166 (1997) 25.
[6] F. Delbecq, P. Sautet, J. Catal. 152 (1995) 217.
[7] P. Beccat, J.C. Bertolini, Y. Gauthier, J. Massardier, P. Ruiz,
J. Catal. 126 (1990) 451.
[8] Y. Nitta, K. Ueno, T. Imanaka, Appl. Catal. 56 (1989) 9.
[9] Y. Nitta, T. Kato, T. Imanaka, Stud. Surf. Sci. Catal. 78
(1993) 83.
[10] C. Ando, H. Kurokawa, H. Miura, Appl. Catal. A 185 (1999)
L181.
[11] A. Giroir-Fendler, P. Gallezot, D. Richard, Catal. Lett. 5
(1990) 175.
[12] E.L. Rodrigues, A.J. Marchi, C.R. Apestequia, J.M.C. Bueno,
Stud. Surf. Sci. Catal. 130 (2000) 2087.
[13] E. van Steen, G.S. Sewell, R.A. Makhothe, C. Micklethwaite,
H. Manstein, M. de Lange, C.T. OConnor, J. Catal. 162
(1996) 220.

[14] J.A. Amelse, L.M. Schwartz, J.B. Butt, J. Catal. 72 (1981) 95.
[15] M.P. Rosynek, C.A. Polansky, Appl. Catal. 73 (1991) 97.
[16] A. Kogelbauer, J.G. Goodwin Jr., R.R. Oukaci, J. Catal. 160
(1996) 125.
[17] A. Sexton, A.E. Hughes, T.W. Turney, J. Catal. 97 (1986)
390.
[18] B. Viswanathan, R. Gopalkrishnan, J. Catal. 99 (1986) 342.
[19] Y. Okamoto, K. Nagata, T. Adachi, T. Imanaka, K. Imanaka,
T. Takyu, J. Phys. Chem. 95 (1991) 310.
[20] L.B. Backman, A. Rautiainen, A.O.I. Krause, M. Lindblad,
Catal. Today 43 (1998) 11.
[21] L.B. Backman, A. Rautiainen, M. Lindblad, O. Jylh, A.O.I.
Krause, Appl. Catal. A 208 (2001) 223.
[22] H. Ming, B.G. Baker, Appl. Catal. A 123 (1995) 23.
[23] M. Ruger, Chem. Abstr. 18 (1924) 156.
[24] J.S. Rieck, A.T. Bell, J. Catal. 103 (1987) 46.
[25] L.B. Backman, A. Rautiainen, M. Lindblad, A.O.I. Krause,
Appl. Catal. 191 (2000) 55.
[26] S. Tsuchiya, Y. Amenomiya, R. Cvetanovic, J. Catal. 19
(1970) 245.
[27] M.J. Heal, E.C. Leisegang, R.G. Torington, J. Catal. 51 (1978)
314.
[28] J.-G. Choi, H.-K. Rhee, S.H. Moon, Appl. Catal. 13 (1985)
269.
[29] G. Kadinov, Ch. Bonev, S. Todorova, A. Palazov, J. Chem.
Soc., Faraday Trans. 94 (1998) 3027.
[30] A. Lapidus, A. Krylova, V. Kazanskii, V. Borovkov, A.
Zaitseu, J. Rathousky, A. Zukal, M. Janclkov, Appl. Catal.
73 (1991) 65.
[31] A. Yu Khodakov, J. Lynch, D. Bazin, B. Rebours, N. Zanier,
B. Moisson, P. Chaumette, J. Catal. 168 (1997) 16.
[32] K. Sato, Y. Inoue, I. Kojima, E. Miyazaki, I. Yasumori, J.
Chem. Soc., Faraday Trans. 1 (80) (1984) 841.
[33] S. Todorova, V. Zhelyazkov, G. Kadinov, React. Kinet. Catal.
Lett. 57 (1) (1996) 105.
[34] J. Ansorge, H. Frster, J. Catal. 68 (1981) 182.
[35] R. Queau, R. Poilblanc, J. Catal. 113 (1972) 200.
[36] M. English, A. Jentys, J.A. Lercher, J. Catal. 166 (1997) 25.
[37] F. Coloma, J. Llorca, N. Homs, P.R. Piscina, F.
Rodriguez-Reinoso, A. Seplveda-Escribano, Phys. Chem.
Chem. Phys. 2 (2000) 3063.

Você também pode gostar