Sei sulla pagina 1di 13

~

AA
PT
PA
LL
E
IY
DSS
C
I
A: GENERAL

ELSEVIER

Applied CatalysisA: General 165 (1997) 429-441

Hydrogenation of crotonaldehyde on Pt/TiO2 catalysts:


Influence of the phase composition of titania
on activity and intramolecular selectivity
Peter Claus

a*
'

Sabine Schimpf b, Rainer Sch6del c, Peter Kraak c,


Wolfgang M6rke d, Dieter H6nicke b

a lnstitutfiir Angewandte Chemie, Abt. Katalyse, Rudower Chaussee 5, D-12489 Berlin, Germany
b Lehrstuhlfiir Technische Chemie, Technische Universitiit Chemnitz, D-09107 Chemnitz, Germany
c KataLeuna GmbH, D-06236 Leuna, Germany
d lnstitutfiir Analytische und Umweltchemie, Martin-Luther-Universitiit Halle, D-06217 Merseburg, Germany

Received 10 August 1996; received in revised form 4 July 1997; accepted 7 July 1997

Abstract

The hydrogenation of crotonaldehyde in the gas phase at 413 K and 2 MPa over Pt/TiO2 catalysts reduced at 473 K (LTR)
or 773 K (HTR) was investigated in order to examine if the catalytic properties could be altered by the phase composition of
TiO2. The catalysts prepared by ion-exchange or sol-gel technique were characterized by physisorption, chemisorption, EPR,
TEM and XRD measurements. It was found that the phase composition of the support has a strong influence on the activity
which decreased with increasing anatase fraction (~ANA).Furthermore, the specific activities on a per g Pt basis were higher in
the case of the HTR catalysts than those of their LTR counterparts. The selectivity to crotyl alcohol ranged from 30 to 40% at
crotonaldehyde conversions up to 50% independent of the catalyst support used. However, at higher conversions up to 80% the
highest selectivity of 53% to crotyl alcohol was obtained over a Pt catalyst with ~ANA----65%. Both the observed behavior in
selectivity which corresponds to the TOF data for product formation on individual Pt catalysts as well as the catalytic
properties of physical mixtures of the catalysts suggest that the differences in selectivities are connected with different degrees
of competitive adsorption and readsorption of the products which depend on the TiO2 phase composition. In addition, the solgel derived Pt catalyst gave a selectivity of 51% for crotyl alcohol which shows that the SMSI effect could be more
pronounced by the close contact of Pt with the anatase matrix. 1997 Elsevier Science B.V.
Keywords: Anatase; Crotonaldehyde; EPR spectrum; Hydrogenation; Platinum catalyst; Rutile; Sol-gel technique; TiO2
phase composition

1, I n t r o d u c t i o n
The selective hydrogenation of c~,fl-unsaturated

to the corresponding allylic alcohols remains a goal of

aldehydes containing both a C=C and a C=O group

considerable importance for the industrial production


of fine chemicals and for fundamental research in
catalysis. The hydrogenation of an c~j~-unsaturated

*Corresponding author. Tel.: (+49-030) 6392 4322; fax: ( + 4 9 030-6392 4350); e-mail: claus@aca.fta-berlin.de,

aldehyde proceeds according to a formal reaction


network which consists of competitive and consecu-

0926-860x/97/$17.00 5;) 1997 Elsevier Science B.V. All rights reserved.


Pll S0926-860X(97)00224-X

P. Claus et al./Applied Catalysis A: General 165 (1997) 429-441

430

CH3_CH=CH_CaO
j - ' - . . . "H,,,

H 2 ~ 2

,r.~

CH3-CH=CH-CH20H

(5) ~

"'~3)

H2~

(2) ~
CH3-CH2-CH2-C~0-H
(4)/J..

,f/

H2

CH3_CH2.CH2.CH20H
Scheme 1. Reactionnetworkofcrotonaldehydehydrogenation[1].

tive reaction steps in both functional groups as shown


in Scheme 1 for the hydrogenation of crotonaldehyde
[1]. The formation of the desired product, crotyl
alcohol (reaction 1), is not the thermodynamically
favored one, and therefore, crotonaldehyde is preferably hydrogenated to butyraldehyde (reaction 2)in the
presence of most of the conventional hydrogenation
catalysts. Thus, crotyl alcohol selectivity must be
controlled by changes in the rate constants of the
competitive reactions and in the adsorption constants
of the components. Indeed, true bimetallic catalysts
prepared by controlled surface reaction (CSR), e.g. of
(n-CnH9)4Sn with supported group VIII metals (Rh,
Pt, Ru [1]) and reduced in hydrogen at 623 K, were
able to control the intramolecular selectivity by favoring the hydrogenation of the C=O group instead of the
C=C group. As shown in the crotonaldehyde hydrogenation on a series of bimetallic Rh-Sn/SiO2 catalysts, the selectivity to crotyl alcohol (trans and cis
isomers) increased strongly with increasing tin content, reaching values between 65 and 74% provided
that the Sn/(Sn + Rh) atomic ratio is more than 40%
[2]. It is important to note that the intramolecular
selectivity of the gas phase hydrogenation of c ~ , g unsaturated aldehydes can be controlled by the pressure. For the first time, a marked increase in selectivity
to the allylic alcohol from 12 to 66% with increasing
pressure from 0.1 to 2 MPa was observed in the
hydrogenation of crotonaldehyde over a bimetallic
Rh-Sn/SiO2 catalyst [1]. By varying the pressure at
413 K the turnover frequency for the formation of
crotyl alcohol was increased by a factor of 5.3 whereas
that for the formation of butyraldehyde was decreased
at about I/3 of specific activity at 0.1 MPa. A reaction
mechanism has been proposed which explains the

formation of the reaction products via 7r-allylic or


oxo-Tr-allylic surface species [2]. Furthermore, Vannice et al. showed that on monometallic Pt catalysts
the selectivity to crotyl alcohol was increased up to
37% using a partially reduced support as TiO2 which
can activate the C=O group of crotonaldehyde by the
creation of special active sites at the Pt metal-support
interface [3]. Moreover, metal ion additives in liquid
phase hydrogenations [4,5], electron donor ligand
effects [6], steric constraints on the metal surface or
in the metal environment [7,8], and the nature and
position of substituents on the C=C group [9] can
influence the selectivity to allylic alcohols.
The present study was undertaken in order to find
out if the intramolecular selectivity could be altered by
changes in the phase composition of the titania used as
support for monometallic Pt hydrogenation catalysts.
Therefore, self-prepared pure TiO2 modifications anatase and rutile as well as a commercial TiO2 (P25,
Degussa) were used to prepare supported Pt catalysts
by an ion-exchange method. Furthermore, the sol-gel
technique was also applied to prepare a Pt catalyst
using Ti-isopropylate as a precursor. The catalysts
were characterized by means of adsorption and chemisorption techniques, X-ray diffraction, electron
microscopy and EPR studies. Their catalytic properties were estimated in the gas phase hydrogenation of
crotonaldehyde at 423 K and 2 MPa total pressure.

2. Experimental
The TiO2 modifications anatase and mille used as
support materials were prepared as follows: anatase
was obtained by hydrolysis of titanium(IV)-isopropoxide (Ti(O-iC3H7)4, Aldrich). The precipitate
TiO(OH)2 was dried at 383 K and calcined at 783 K
for 5 h. Rutile was prepared from TIC14 (Aldrich) by
hydrolysis at pH 0-1 according to a modified procedure reported elsewhere [10]. The solid Ti(OH)4 thus
obtained was treated under a stream of air at 723 K for
3 h. In addition, P25 (Degussa) was used as commercial TiO2 support material. The titania supported
catalysts Pt/rutile, Pt/anatase and Pt/P25 were prepared by cation exchange using Pt(NH3)4(NO3)2
(Aldrich) as a precursor at pH 11 in an aqueous
solution. Subsequently, the catalysts were dried at
423 K and reduced by a standard treatment: the low

P. Claus et al./Applied Catalysis A: General 165 (1997) 429~141

temperature reduction in situ performed under flowing


hydrogen (90 ml min -1) at 473 K (4 h) is designated
as LTR, whereas the high temperature reduction at
773 K (4 h) is designated as HTR. The sol-gel catalyst
Pt/SG was prepared by the application of the sol-gel
method [11] using a precursor solution consisting of
Pt(NH3)2(NO3)2 (Aldrich) and Ti(O-iC3H7)4, dissolved in a 2-methyl-2,4-pentanediol/2-propanol
mixture. By this procedure the coordination sphere
was modified which ensured a homogeneous solution
of the metal alkoxide [12]. After adding water to
induce hydrolysis and heating at 373 K a white gel
was obtained. The solid material was dried under low
residual pressure (11,3 kPa) at 373 K (2 h) and then
again at 433 K for several hours. After calcination in
air at 673 K, the dried materials were reduced in
flowing hydrogen at 773 K. The metal contents of
the catalysts were determined by atomic emission
spectroscopy with inductive coupled plasma (AESICP, Perkin Elmer Optima 3000XL) after dissolving
the materials in a mixture of HF/HNO3 by means of a
MDS-2000 microwave unit (CEM).
The catalysts were characterized by nitrogen physisorption, hydrogen chemisorption, electron microscopy, X-ray diffraction (XRD) and electron
paramagnetic resonance (EPR). Nitrogen adsorption
isotherms and BET surface areas, respectively, were
obtained at 77 K using a Sorptomatic 1900 (Fisons).
The chemisorption studies were performed by a pulse
technique described earlier [13]. The platinum particle
size was determined by high resolution transmission
electron spectroscopy (HRTEM), using a JEOL 100C
microscope. XRD pattern were conducted on a URB 6
diffractometer (Seifert FPM) using monochromatic
CuK,~ radiation. EPR spectra of the reduced catalysts
(without air exposure) were obtained with a X-band
spectrometer (ERS 220) equipped with a dual sample
cavity. Signal intensity was evaluated using a cornputer program allowing baseline adjustments and
comparison with a reference (Fe 3+ in glass and ultramarine, respectively). These spectra were recorded at
123 K.
Gas phase hydrogenation of crotonaldehyde was
performed in a computer controlled fixed-bed microreactor system which has been described in detail
elsewhere [14]. This equipment allows the performance of high pressure gas phase hydrogenations of
unsaturated organic compounds which are usually

431

liquids with low vapor pressures at standard conditions (STP). Crotonaldehyde (CA, Aldrich, distilled
before use) and the hydrogenation products were
analyzed on-line by means of a gas chromatograph
(HP 5890) equipped with a flame ionization detector
and a 30 m J&W DB-WAX capillary column.
The selectivities of reaction products were calculated from moles of the product formed per moles of
crotonaldehyde converted, and the activities of the
catalysts were expressed as specific activities (on a
gram of Pt basis) or in the form of turnover-frequencies, TOF (based on hydrogen chemisorption data).
The experimental equipment allowed us to get the first
representative data point after 180 s time on stream. In
the case of the LTR catalysts some deactivation was
observed during a time on stream of 100 to 160 min.
The effect was more pronounced for the Pt/anatase
catalyst compared to the Pt/rutile and Pt/P25 catalysts. For their HTR counterparts only the anatasesupported Pt catalyst showed a deactivation with time
on stream up to 100 min. All reported activities are
steady-state data.

3. Results and discussion


The composition of the catalysts, their specific BET
surface areas and pore volumes from nitrogen physisorption, the Pt dispersion obtained from hydrogen
chemisorption, and the Pt particle size of the HTR
catalysts determined from HRTEM are reported in
Table 1. As the reduction temperature increased, a
strong decrease in hydrogen uptake was observed for
the Pt/TiO2 catalysts. This effect was more pronounced for the futile and anatase-supported Pt catalysts than for Pt/P25. The observed suppression in
hydrogen chemisorption after the high temperature
reduction is characteristic of strong metal-support
interactions (SMSI), e.g. see [15]. Although several
attempts were made to explain the SMSI effect via an
electron transfer from the support to the metal [ 16-19]
or by the formation of intermetallic phases [20-22], it
has been well accepted now that a partial reduction of
TiO2 gives rise to TiO~ suboxide species (x < 2)
which migrate onto the metallic surface and cause
physical blockage of dispersed surface Pt particles
without changes in the metallic particles size
[3,23,24]. Thus, the apparent discrepancy between

P. Claus et aL/Applied Catalysis A: General 165 (1997) 429-441

432

Table 1
Composition, specific BET surface areas, pore volumes, Pt dispersion and Pt particle size of the Pt/TiO2 catalysts used in this study
Catalyst

Pt [wt%]

Phase composition
of T i t 2 [wt%] a

Tred [K] b

SBET [m2 g J]

Vp [era 3 g J]

H/Pt

dpt [nm] c

Pt/rutile-LTR
Pt/rutile-HTR
Pt/P25-LTR
Pt/P25-HTR
Pt/anatase-LTR
Pt/anatase-HTR
Pt/SG

0.77
0.77
0.88
0.88
0.53
0.53
0.6

100% mtile
100% mtile
65% anatase, 35% rutile
65% anatase, 35% rutile
100% anatase
100% anatase
100% anatase

473
773
473
773
473
773
773

36
35
45
49
39
39
44

0.145
0.156
0.222
0.280
0.169
0.175
0.06

0.55
0.015
0.16
0.014
0.12
0.005
0.07 d

-3
-3
-2
4

a Determined by XRD
b Reduction temperature
c Pt particle sizes were determined by HRTEM only for HTR catalysts. The LTR counterparts were analyzed by XRD; see text.
d CO/Pt (by CO chemisorption)

the observed hydrogen uptake after HTR and the low


Pt particle size (as determined from HRTEM, Table 1)
is due to the partial coverage of the Pt surface by the
TiOx adspecies and is not due to a sintering of the
metallic phase. The preparation of Pt/TiO2 catalysts
by the cation-exchange method via Pt(NH3)4(NO3)2
ensures the generation of small Pt particles after LTR
and HTR with a mean particle size of 2-3 nm independent of the titania phase. In the case of the Pt/TiO2
catalyst obtained by the sol-gel method the fraction
exposed (CO/Pt) was 0.07 (Table 1). This estimate
represents only an apparent dispersion because the
HRTEM investigation of this catalyst gave a mean
particle size of 4 nm. Therefore, the low chemisorption capacity is probably due to an incorporation of Pt
into the sol-gel matrix in addition to the physical
blockage of dispersed surface Pt particles induced by
SMSI.
The X-ray measurements were conducted only with
the LTR catalysts and showed no Pt signals. This result
suggests either that the Pt particle size could be lower
than 3--4 nm taking into account the resolution of the
diffractometer or that the Pt content of the catalysts
(< 0.88 wt.%) is too low to give an XRD pattern,
Also, phase transformation of T i t 2 from anatase to
rutile was not observed by X-ray diffraction, neither
after calcination of the precipitate formed during
anatase preparation nor after the high temperature
reduction of both the Pt/anatase and the Pt/SG catalysts. The determination of the phase composition in
Pt/P25 by XRD gave 65% anatase and 35% rutile in
contrast to the usually published values of 80% ana-

tase and 20% rutile. Furthermore, the specific BET


surface areas did not change when the catalysts were
treated by LTR or HTR. The use of the sol-gel method
for preparing a Pt/TiO2 catalyst led to a titania which
was analyzed by XRD as the anatase phase (100%).
The calcined and reduced Pt/SG catalyst had a BET
surface area of 44 m 2 g - ~, and an average pore diameter of 3 nm. The pore volume (0.06 cm 3 g - t ) is
significantly lower compared to that in Pt/TiO2 catalysts prepared by ion-exchange. It is also noticeable
that the BET surface area of the Pt/SG powder is
negligible (2 m 2 g - l ) if it was calcined at 473 K. The
results demonstrate that the textural properties of the
titania-based catalysts can be controlled by the method
of preparation and the temperature treatment during
calcination [11,12].
A deeper characterization of the SMSI state
between Pt and TiO2 in the catalysts of the present
study was possible by the investigation of EPR active
species. The supports and the catalysts were investigated after HTR and LTR by EPR [32]. The EPR
spectra of the catalysts recorded after the HTR step
without air exposure are shown in Fig. 1. Those of the
catalysts Pt/P25 and Pt/SG consist of the well
described distinctive signal of Ti 3+ ions at g ~ 1.97
[25-27] and the more or less resolved broad signal of
Pt + i o n s (d 9, S----1/2) at g < 2.35 [28,29]. Pt/anatase
revealed a narrow signal at geff ---- 4.21 except that of
Pt + ions in the region of g < 2.3 (spectrum 1, Fig. 1).
The former corresponds to paramagnetic dimers of
Ti 3+ ions arising during the formation of very small
seeds of Magneli phases under HTR conditions. The

433

P. Claus et al./Applied Catalysis A: General 165 (1997) 429-441

"

Sl/

signal at g ~ 1.96 whose line width increased with


increasing both the concentration of Ti 3+ ions and the
temperature of the EPR measurement [31 ]. Moreover,
tails of Ti 3+ signals were observed in a field region of
g > 2 [31]. After high temperature reduction of both
the supports and the catalysts at 773 K small portions
of the TiO 2 were transformed into very small seeds of
Magneli phases, and thus, the Ti 3+ part of their EPR
spectra consists of a superposition of the narrow signal
of Ti 3+ ions of the support and the broader one of the
Ti 3+ ions of the Magneli phase. These circumstances
mainly determine the EPR spectrum of Pt/rutile
(spectrum 3, Fig. 1) and Pt/anatase (spectrum 2,
Fig. 1). The EPR signal of Pt + ions appears at
g ~ 2.3 as a broad bump and is hardly perceivable
.3+
due to overlay w i t h TIMagneli
in the case of Pt/anatase
and Pt/rutile. After low temperature reduction at
473 K Pt/rutile and Pt/anatase showed Ti 3+ and
Pt + signals, the latter are characterized by g values
at gl = 2.75, g2 = 2.30, g3 "~ 2.1 and gl = 2.28,

g = 2.1s A
~_
1
-

'
~
~
" "7-2-- ~
.07

!/
. f
/ ~ J '.:
! , . .=
: :~ ~v /

4
'.

10o

, ~
15o

~ , ~
2oo 2so

3oo

,
3~0

, . ,
4oo ~
Bo [mr]

g2 = 2.15, g3 ~ 2.1 for Pt/rutile and Pt/anatase,


respectively, [32].
As the high-field part of the Pt + spectrum
(gxx = 2.07) is superimposed with that of the Ti 3+
signal the components of the g tensor of the Pt + ions in
the P t / S G H T R catalyst were revealed as an example
by spectral calculation as described previously [33].
The best fit of the entire spectrum is shown in Fig. 2. It
was obtained if the following four different signals
were allowed to overlap: (i) 195pt+ species (I = 1/2,

Fig. 1. EPR spectra of the Pt/TiO2 catalysts: Pt/anatase ( 1 ) ,


Pt/anatase after an additional HTR treatment (2), Pt/rutile (3),
Pt/P25 (4), Pt/SG (5).

signal disappeared after an additional H T R (spectrum


2, Fig. 1), because the formation of Ti,O2,_t phases is
connected with diamagnetically coupled Ti 3+ pairs
[30]. The non-diamagnetically coupled Ti 3+ ions of
the Magneli phases gave rise to a broad symmetrical

Pt

pt0

130

230

330

430

-Bo [m'l'J

Fig. 2. Experimental (1) and calculated (2) EPR powder spectrum of the Pt/SG catalyst. The parameter set is given in the text.

434

P. Claus et al./Applied Catalysis A: General 165 (1997) 429-441

using the parameter set: gzz = 2.33, gvv = 2.15 and


gxx = 2.07, ABx ----ABy = 2.0 and ABz = 10.0 mT,
Axx = 10 10 -4 cm -1,
Ayy = 20 10 -4 cm -1,
Azz = 75 x 10 -4 cm-l), Pt + (I = 0, using the same
set without hyperfine structure, HFS); (ii) Tlanatas
e'3+
(gxx = gvv = 1.970, gzz = 1.906, ABxx = z3Bvv =
5.0 mT, ABzz = 7.0 mT); (iii) T1Magneli
.3+
(giso =
1.960, ABiso = 50.0 mT) and (iv) a symmetrical
signal (Pt ) at g = 2.33, AB = 100.0 roT. The four
"3+
'3+
signals of Pt +, Tlanatase,
T1Magneli
and p t 0 have proportions of about 20%, 65%, 10% and 5%, respectively,
The Pt signal can be attributed to nanocrystalline Pt
with a particle size of dpt _< 1.5 nm [34,35]. Such
small metallic particles show CESR (conduction
electron spin resonance) signals if the condition
6 < kBT is fulfilled (6 is the mean level separation
and kB the Boltzmann constant) [35-38]. A symmetrical signal of nanocrystalline Pt with g = 2.28 and a
temperature-dependent line width (25 mT at 77 K and
142 mT at 298 K) was mentioned by Katzer et al. [29].
They assumed that the signal was due to the odd
electron effect [39] in Pt/A1203. The observed g factor
is in good agreement with the value predicted by
theoretical considerations [44]. A similar CESR signal
was observed on Pd crystallites in PdNaY zeolites
after hydrogen or oxygen adsorption [40]. In this case
the odd electron effect was caused by the interaction of
nanocrystalline Pd with H2 or O2; the samples showed
a Curie-like behavior. The parameters of the Pd CESR
signal at 120 K (g = 2.28 and AB = 80 roT) [40] and
those of Katzer et al. [29] are in good agreement
with the results of the present study. We presume
that the odd electron effect in the Pt/SG-HTR
sample is caused by the intimate contact between
the Pt and Pt + ions which is not limited to the metal
semiconductor interface. A further consequence of
this quantum size effect is the better resolution of
the Pt + signals in the case of the Pt/SG catalyst
(Fig. 1), since the relaxation of the Pt + ion in the
vicinity of the metallic platinum is influenced by the
metallic particle size [37,41].
The EPR spectra of Pt + ions in the HTR catalysts
Pt/anatase and Pt/P25 were fitted using the same
components of the g tensor as for Pt/SG but without
HFS and greater line widths (gxx = 2.35, gyy = 2.15
and gzz = 2.07, ABx = 9.0, ABy = 9.0 and
ABz = 12.0 mT). From the same g tensor it can be
concluded that the Pt + coordination is nearly the same

in these samples. The greater line widths can be


interpreted by a broader variety of the coordination
surrounding the Pt + ions, e.g. surface species or the
presence of larger Pt and TiO2 crystals compared with
the Pt/SG. This conclusion is supported by the results
of EPR measurements after exposing the samples to
air at room temperature. While both the Pt + and the
Ti 3+ signals disappear in the case of Pt catalysts
prepared by ion exchange, these signals survived at
treatment temperatures up to 423 K with the Pt/SG
catalyst. After hydrogen treatment of the TiO2-supported Pt catalysts at room temperature the Pt + and
Ti 3+ signals appeared again together with signals of
O~- radicals. This behavior was mostly pronounced in
the case of Pt/P25. From the EPR studies it can be
concluded that the Pt + ions in Pt/rutile, Pt/anatase
and Pt/P25 are surface species whereas in the case of
the sol-gel derived catalyst Pt/SG they are embedded
in the nanosized matrix. The double integration of the
EPR spectra gave the following Ti 3+ contents in the
high temperature reduced Pt/rutile, Pt/P25, Pt/SG
and Pt/anatase catalysts: 7.5 1015, 8.3 1015,
2.1 1016 and 1 1017 spins mg 1, respectively.
The Pt + content of the HTR catalysts increased in
the sequence Pt/rutile, Pt/P25, Pt/anatase, Pt/SG
corresponding tO 8.7 1013, 2.1 x 1014, 6 1014
and 7 1015 spins mg -~, respectively. Thus, values
between 0.04 and 1% for the Ti3+/Ti 4+ ratio and
between 0.3 and 3.6% for the Pt+/Pt ratio can be
derived. Both ratios are in fair agreement with the
literature [26,29,42] but the Ti3+/pt + ratio (Pt/rutile:
85, Pt/P25: 40, Pt/anatase: 25, Pt/SG: 3) demands
further explanation. We assume that the distinct differences in the Ti 3+ content are connected with the
formation of very small seeds of the Magneli phase
which is not observable by XRD. Note that in Pt-free
titania samples the Magneli phase can be formed from
rutile [49]. In samples under investigations, i.e. in the
presence of Pt, the formation of Magneli phase over
the anatase route cannot be excluded (see Fig. 2).
Nevertheless, the phase transformation starts in both
modifications (rutile and anatase) at different Ti 3+
contents. Diamagnetically and non-coupled Ti 3+ ions
arise during the transformation when concentrations
are different in rutile and anatase. The differences in
the Pt + contents are connected with the phase transformation also. It seems that the mobility of the lattice
components during the rutile transformation [30] in

P. Claus et al./Applied Catalysis A: General 165 (1997) 429~141

hydrogen atmosphere impedes the electron transfer


from Pt to Ti 4+ ions and oxygen vacancies. Moreover,
the Ti3+/Pt + ratio points to the fact that more than one
way exists to generate Ti 3+ ions: (i) reduction of the
support by hydrogen; this mechanism prevails in the
case of catalysts prepared by ion exchange, and (ii)
electron transfer from Pt to Ti 4+ and oxygen vacancies [26,43]; this mechanism predominates in the
Pt/SG catalyst. According to the rigid band model,
Pt with an increasingly negative charge results from
the raising excess of Ti 3+ ions compared with the
number of Pt + ions [42]. For that reason, considerable
consequences for the catalytic behavior can be
expected if the phase composition of the support is
altered. Thus, from the above-described sequences of
the Pt + and Ti 3+ ion contents it is suggested that, not
only the number of these species at the interfacial
surface depends on both the different structures of the
individual TiO2 phases and their stability against
phase transformation, but also the extent of the electron transfer process is increased if the sites at the
Pt-TiO2 interface are in intimate contact,
In the gas phase hydrogenation of crotonaldehyde
over these titania-supported Pt catalysts the main

=
O

products were crotyl alcohol (trans and cis, CyOH),


n-butyraldehyde (BA) and n-butanol (BuOH). Moreover, small amounts of other products corresponding
to the proposed reaction mechanism [2] were detected
(selectivities: allylcarbinol < 1%; ethyl methyl ketone
< 0.5%; 2-butanol < 0.5% and CwC4-hydrocarbons
< 4%). The results of the hydrogenation experiments
showed marked differences between the low temperature reduced (LTR) and the high temperature reduced
(HTR) Pt/TiO2 catalysts on the one hand and the use
of catalysts with pure anatase or rutile or a commercial
P25 support material on the other. From Fig. 3 it can
be seen that the phase composition of TiO2 (expressed
as (ANA) used as support for Pt catalysts for crotonaldehyde hydrogenation has a strong influence on the
conversion, which was never observed before. In the
case of both the HTR and LTR catalysts the conversion
decreased with increasing ~ANA- Furthermore, the
HTR catalysts showed a marked higher conversion
than their LTR counterparts. The observed changes in
activity by varying the phase composition as well as
the reduction temperature cannot be ascribed to a
variation of the Pt particle size because all the titaniasupported Pt catalysts used in the present study exhibit

100

11

sI

80

60

>
t-'

435

.....
"~

!
~

60

~.,
\

40~

~-,

-40

20

20
! RUTILE

P25

ANATASE '

l
0.0

0.1

0.2

03

0,4

0.5

0.6

--1
0.7

0.8

0.9

*-0

1,0

~ANA

Fig. 3. Conversion of crotonaldehyde (CA) vs. anatase content of titania used as support for Pt/TiO2 catalysts (0.77% Pt/rutile: 0.88%
Pt/P25; 0.53% Pt/anatase) reduced at high temperature (773 K, HTR) or reduced at low temperature (473 K, LTR) (T 413 K. p = 2 MPa.
H2/CA
20, W/I~CA = 19 gh mol 1).

436

P. Claus et al./Applied Catalysis A: General 165 (1997) 429-441


1200-T

1000t Pt/RUTILE
800

Pt/P25

Pt/ANATASE

o oO,oo ,io
o-

!
LTR

HTR
BB CyOH

II

LTR
[~

HTR
BA

[~

LTR

HTR

BuOH

Fig. 4. Specific activities (per g Pt basis) for the formation of (trans + cis) crotyl alcohol (CyOH), n-butyraldehyde (BA) and n-butanol
(BuOH) during crotonaldehyde hydrogenation over the catalysts 0.77% Pt/rutile; 0.88% Pt/P25; 0.53% Pt/anatase (HTR or LTR) at
T = 413 K, p = 2 MPa, H2/CA = 20, W/F~cA = 19 gh mo1-1.

similar Pt dispersions (2-3 nm observed by HRTEM,


Table 1). Since the catalysts had different Pt contents
(Table 1), specific activities for product formation
were calculated on a per g Pt basis. Fig. 4 shows that
the specific activities towards the hydrogenation of the
C=O bond are higher in the case of the titania-supported HTR catalysts than those of their LTR counterparts. This is a remarkable result because the data
available in the literature show that the activities (per g
Pt) did not vary significantly or were even lower
compared to LTR catalysts in crotonaldehyde hydrogenation [3]. Furthermore, as shown in Fig. 4, the
specific activities (per g Pt) for the formation of the
C=O hydrogenation products (CyOH, BuOH) were
markedly increased by high temperature reduction for
each of the three HTR catalysts, whereas those for the
formation of the C=C hydrogenation product BAwere
decreased only for Pt/rutile and Pt/P25. Although this
catalytic behavior is due to the SMSI effect which has
a strong influence on the corresponding product selectivities [3], the observed product distribution corresponds to a complex reaction network involving
competitive and consecutive reactions in both of the
functional groups (see Scheme 1, [1,3]).

The product selectivities of the three catalysts were


compared with respect to their phase composition at
constant conversion level (Fig. 5). The conversion of
crotonaldehyde at constant temperature, pressure and
molar educt ratio was controlled over a large extent
from about 10% up to 99% by variation of the space
time ( W / F A ) , and the observed selectivities were
plotted versus conversion for the three HTR catalysts.
At lower crotonaldehyde conversions the selectivity to
CyOH ranged from about 30 to 40% independent of
the catalyst support used. Nevertheless, this is a
marked increase in selectivity to the allylic alcohol
compared to A1203 and SiO2-supported Pt or Rh
catalysts [ 1-3,45]. However, crotyl alcohol selectivity
decreased significantly to 25% at higher conversions
(XcA) in the case of Pt/rutile (XcA = 98%), increased
up to 53% using the Pt/P25 catalyst (XcA = 84%),
and remained unchanged on Pt/anatase (up to
XCA = 72%). In addition, a decrease in selectivity
to BA and an increase in selectivity to BuOH with
increasing XCA was observed for the Pt/TiO2 catalysts
whatever the support used. These observations can be
explained by the assumption that readsorptions of the
products (BA, CyOH) and/or competitive adsorptions

P. Claus et al./Applied Catalysis A: General 165 (1997) 429--441

100 (a)
eyOH
9- Pt/RUTILE-HTR

~,
a0
. ~OH
-- ro
60~
1
"~ 50~
40
o~ ao
20~
10
~
o.-. , ....
, , . ....
, . , . , .
lo 20 ao 40 50 60 7o o0 90 loo
Conversion [%]
100- (b)
~/OH
Pt/P26-HTR

ao-

ao-

auOH

TO-

--60~ 5040ao20t
100t

,_.--~*
, . . . . . . . . . . . . . . . . .
10 20 ao 40 50 60 T0 ao o0 loo
C o n v e m i o n [%]

10090-

80

(C)

CyOH
BA
* BuOH

Pt/ANATASE-HTR

70

~
>

]3

6050-

4o

c~ 30
a0

-~

_____,,,_/-~--------*

lO
o-.

lO' ' 20' ' 3o' ' 4o' . 5'0. 8o


. . 7'0. 8'0
. ~

1oo

Conversion [%1

Fig. 5. Selectivity vs. conversion of crotonaldehyde for CA


hydrogenation over Pt/TiO2-HTR catalysts ((a) 0.77% Pt/rutile;
(b) 0.88% Pt/P25; (c) 0.53% Pt/anatase) at T = 4 1 3 K ,
p = 2 MPa, H2/CA = 20.

between the hydrogenation products occur which


seems to depend on the TiOz phase composition: In
the case of the Pt/anatase catalyst (Fig. 5(c)) a strong
adsorption of butyraldehyde could lead to the satu-

437

rated alcohol by a consecutive hydrogenation whereas


CyOH selectivity remains constant. The data of the
Pt/rutile catalyst (Fig. 5(a)) imply that the decrease of
both the crotyl alcohol and the butyraldehyde selectivity is due to a stronger adsorption compared to
Pt/anatase. Further hydrogenation of both products
gave the highest selectivity to butanol. The increase in
CyOH selectivity with increasing conversion of CA
(Fig. 5(b)), which corresponds to high concentrations
of CyOH and low concentrations of CA, suggests a
strong effect due to competitive adsorption in the case
of the Pt/P25 catalyst. Beside a stronger adsorption of
the ~,3-unsaturated aldehyde compared to the allylic
alcohol, a readsorption of BA and its subsequent
hydrogenation to BuOH seems to be possible.
Based on the obtained selectivities at 20%, 40% and
80% conversion, the rates of formation of the different
hydrogenation products, expressed as turnover-frequencies (TOF obtained using the data of H2 chemisorption), were now calculated, and they are shown in
Fig. 6. For the catalysts Pt/rutile ((ANA ~ 0) and
P t / P 2 5 (~ANA = 0 . 6 5 ) , TOF values for the formation
of CyOH and BA are similar and range from 5 to
20 s -I. However, in the case of Pt/anatase ((ANA = 1)
they increased significantly, suggesting a change in the
nature of the active sites capable of hydrogenating the
conjugated carbonyl group with increasing anatase
content, starting at (ANA = 0.65 in the catalyst support, Despite the higher TOF values for both CyOH
and BA at (ANA ---- 1 (Fig. 6(a-c)), BA was the most
predominant product with the Pt/anatase catalyst at
each conversion level (Fig. 5(c) and Fig. 6(a-c) at
(ANA = 1). Considering the formation of BA proceeds
on metallic Pt particles, whereas the formation of
CyOH is due to special sites involving both the Pt
and TiOx species created at the Pt-TiO2 interface by
SMSI [3], the observed behavior suggests that the
fraction of the latter does not increase relative t o P t
sites by a variation of the titania phase in the three
TiO2-supported catalysts of this study. The order of the
TOF values of the hydrogenation products at the same
conversion level and anatase content of the catalyst
(Fig. 6) corresponds clearly to the observed sequence
in product selectivities, as indicated in Fig. 7 for
CyOH. Neither pure futile nor pure anatase can be
used as support to alter the selectivity to CyOH
significantly within conversion levels of 20 to 80%.
Selectivity to CyOH was clearly enhanced by the use

P. Claus et al./Applied Catalysis A: General 165 (1997) 429-441

438

6o
" CyOH
60-. BuoHBA

ta)/re(a)
~
/ ~

40-

. conversion = 20 %
u. a00
~- 2- tF10

~
~
o
o.o o11 o12 o.3 o14 0.5 o18 o17 o.a o.a 1.o

~ANA

60

u.
O

CyOH

&

50-

(b)

BA
BuOH

40~ conversion = 40 %
30-

a0
100

o14o15o16o)

o.o o.1 ola o13

ola o.9 1.o

~ANA

s]i!~-I
~

(C)

50
401

conversion

u. 3o~

0i

--80'/o
/=

1--

2o
1
,
....
.
. ,
,
o.o o.t o.2 o.a 0.4 0.5 o.s 0.7 o.a 0.9 1.o
~ANA
,

Fig. 6. Dependence of turnover-frequencies (TOE based on


hydrogen chernisorption) for the formation of (trans + cis) crotyl
alcohol (CyOH), n-butyraldehyde (BA) and n-butanol (BuOH) on
anatase content (~ANA) at constant CA conversion of (a) 20%, (b)
40% and (c) 80% (HTR catalysts, T = 413 K, p = 2 MPa,

H2/CA = 20).

of mixed TiO2 phase in the Pt/P25 catalyst provided


that a high conversion (XcA = 80%) was reached.
Only in the latter was the TOF for the formation of
CyOH higher than those for BA and BuOH formation
(Fig. 6(c), (ANA z 0.65) leading to allylic alcohol as
the main product. As mentioned previously, the
increase of the concentration of crotyl alcohol with
increasing CA conversion, i.e. at low concentrations of
CA, suggests that competitive adsorption effects are
responsible for the marked change in CyOH selectivity which depends on the TiO2 phase composition as
already discussed. Nevertheless, the increase in the
specific activities of CyOH formation (per g Pt) when
using Pt/TiO2-HTR catalysts instead of their low
temperature counterparts (Fig. 4) implies strong support for the model of interfacial active sites [3].
However, in spite of the lower chemisorption capacities of the HTR catalysts the spreading of TiOx
species over the Pt surface under HTR conditions
cannot be interpreted by a simple decoration of the
metal surface because conversions and specific activities obtained during CA hydrogenation in the present
study were even higher compared to the LTR catalysts.
Such catalytic behavior of catalysts initiated by SMSI
was also found in the hydrogenation of carbon monoxide [46,47] and acetone [48].
Finally, the use of physical mixtures of Pt/rutile and
Pt/anatase (Fig. 7) supports the assumption that the
observed differences in selectivies between rutile,
anatase and P25-supported Pt catalysts are connected
with a different competitive adsorption, rather than
due to the creation of different active sites by variation
of the TiO2 phase composition. Again, the selectivity
to CyOH was clearly enhanced by higher crotonaldehyde conversions at (ANA = 0.56, leading to almost
the same value (ScyoH = 58%) at 80% conversion as
in the case of the P25-supported Pt catalyst. Since no
transformation of titania from anatase to rutile was
observed by XRD, neither after calcination of the
catalyst precursornoraftertheHTRstep, the catalytic
results suggest that the respective amounts of the
active sites of Pt/anatase and Pt/rutile in the physical
mixture at an anatase content similar to the Pt/P25
catalyst give rise to similar competitive adsorption
properties.

Before considering the catalytic behavior of the


sol-gel derived Pt catalyst (Pt/SG), it should be
remembered that the formed titania was a pure anatase

P. Claus et al./Applied Catalysis A: General 165 (1997) 429-441

I O0

439

I O0

l
t

HTR

P = 2 MPa, T = 413 K, H2/CA = 20

801
k

-8o

o
u 60-

XCA = 80 %

--'~
O

XCA =40%

I ~

40

Pt/SG
~i~- _ ~ -

- - '~ . . . .

- -

..

.....

"-- ...

-40

-~ 20

-60

N,~

....

o~

RUTILE

XCA = 20 %

025

04 , , , , ,
0,0

0.1

- 20

0.2

0.3

: 1,
0.4

0.5

0.6

0.7

ANATASE

~,
0.8

0.9

~-0
1.0

~ANA
physical mixtures at XCA = 20 % (~ANA = 0.20, 0.56, 0.80)

- sT- - Pt/TiO2 catalysts at XCA= 40 %

- ~ - PffTiO 2 catalysts at XCA = 80 %

physical mixture at XCA = 80 % (~ANA = 0.56)

[]

Pt/TiO 2 catalysts at XCA = 20 %

physicalmixtureat XCA= 40 % (~ANA = 0.56)

Fig. 7. Comparision of selectivity to crotyl alcohol vs. anatase content (~ANA) at constant conversion level for CA hydrogenation over
Pt/TiO2-HTR catalysts (0.77% Pt/rutile; 0.88% Pt/P25; 0.53% Pt/anatase): open symbols. Additional points (black symbols) are included for
physical mixtures of Pt/rutile and Pt/anatase ((ANA = 0.2, 0.56 and 0,8) at XCA = 20% and for a physical mixture with (ANA = 0.56 at
XCA = 40% and 80%, respectively, as well as for the sol-gel derived titania-supported Pt catalyst (Pt/SG: O).

phase. The use of the same reaction conditions


(T = 413 K, p = 2 MPa, molar ratio H2/CA = 20)
as in the case of Pt/anatase prepared by ion exchange
gave a different conversion of CA (< 5%). In Fig. 8
the specific activities (on a per g Pt basis) for the
formation of CyOH, BA and BuOH and for the
disappearance of crotonaldehyde are shown. It can
be seen that the specific activity obtained with Pt/SG
for CyOH formation is higher than that of BA formation which gave a selectivity towards the allylic
alcohol of 51% (Fig. 7). This selectivity is quite
similar to that of the P25-supported Pt catalyst at
80% of CA conversion, but significantly higher cornpared with Pt/anatase prepared by ion exchange. In
the latter case the obtained CyOH selectivity was
about 32% independent of the conversion (Fig. 6(c)
and Fig. 7). This is a noticeable difference and suggests that the SMSI effect which may alter the selectivity towards the unsaturated alcohol is more

pronounced by the close contact of Pt with the anatase


matrix which is ensured by the sol-gel method.

4. Conclusions
The activity of titania-supported Pt catalysts prepared by ion exchange and used for the hydrogenation
of crotonaldehyde was altered by the phase composition of titania. Furthermore, at each point of the TiO2
phase composition the high temperature reduced
Pt/TiO2 catalysts in the SMSI state were intrinsically
more active than their low temperature reduced counterparts. This is an important result because the activity on a gram of metal basis determines the volume of
the catalyst in the hydrogenation reactor and can be
explained by the creation of special sites in the metalsupport interfacial region. Therefore, the selectivity
for the desired product, i.e. crotyl alcohol, increases by

440

P. Claus et al./Applied Catalysis A: General 165 (1997) 429-441

10o
80

mental work has been greatly appreciated. The authors


are grateful to Prof. K6hler (Miinchen) for his supply
of the EPR calculation program.

Pt/SG

,r-

',,.,,

References

~. 60
-5
40
.>__
"~

~
20

0
CyOH

RA

BUOH

CA

Fig. 8. Specific activities (per g Pt basis) lbr the formation of


(trans + cis) crotyl alcohol (CyOH), n-butyraldehyde (BA) and nbutanol (BuOH), and for the disappearance of crotonaldehyde (CA)
during CA hydrogenation over Pt/SG (T = 413 K, p = 2 MPa,
He/CA = 20).

using high temperature reduced Pt/TiO2 catalysts.


However, for each of those catalysts the product
selectivities depend on the conversion of crotonaldehyde to a different extent which demonstrates the
importance of competitive adsorption and readsorption processes during the hydrogenation of c~,/3-unsaturated aldehydes. The results clearly suggest that
the extent of such processes depends on the anatase/
rutileratio. In comparison with the Pt/anatase catalyst
prepared by ion exchange, the selectivity to crotyl
alcohol is enhanced over the sol-gel-derived
Pt/anatase catalyst which shows (i) that the preparative technique could be the decisive factor for
controlling the creation of interfacial sites and (ii)
the important impact of the sol-gel method for the
preparation of new hydrogenation catalysts.

Acknowledgements
This study was supported by the Bundesminister ftir
Bildung, Wissenschaft, Forschung und Technik
(BMBF) during the 'catalysis project' under grant
03D0028. Assistance by Mrs. H. Mtinzner (Chemnitz), Mrs. G. Mertsching and M r s . B. N e u s t a d t
(Halle) as well as Mr. M. Lucas (Berlin) in experi-

[1] P. Claus, D. H6nicke, in: M.G. Scaros, M.L. Prunier (Eds.),


Catalysis of organic reactions, Chemical Industries Series,
Vol. 62 (15th Conference on Catalysis of Organic Reactions,
ORCS, May 02-05, 1994, Phoenix, Arizona/USA), Marcel
Dekker, New York, 1995, p. 431.
[2] P. Claus, Chem.-Ing.-Tech. 67 (1995) 1340.
[3] M.A. Vannice, B. Sen, J. Catal. 115 (1989) 65.
[4] S. Galvagno, A. Donato, G. Neri, R. Pietropaolo, D.
Pietropaolo, J. Mol. Catal. 49 (1989) 223.
[5] S. Galvagno, C. Milone, G. Neff, A. Donato, R. Pietropaolo,
Stud. Surf. Sci. Catal. 78 (1993) 163.
[6] A. Giroir-Fendler, D. Richard, P. Gallezot, Stud. Surf. Sci.
Catal. 41 (1988) 171.
[7] P. Gallezot, B. Blanc, D. Barthomeuf, M. Pais da Silva, Stud.
Surf. Sci. Catal. 84 (1994) 1433.
[8] A. Giroir-Fendler, D. Richard, E Gallezot, Catal. Lett. 5
(1990) 175.
[9] T.B.L.W. Marinelli, S. Nabuus, V. Ponec, J. Catal. 151 (1995)
431.
[10] F. Cavani, E. Foresti, F. Parrinello, Appl. Catal. 38 (1983)
311.
[11] S. Niwa, F. Mizukami, S. Isoyama, T. Tsuchiya, K. Shimizu,
S. Imai, J. Imamura, J. Chem. Tech. Biotechnol. 36 (1986)
236.
[12] G. Zehl, S. Bischoff, F. Mizukami, H. Isutzu, M. Bartoszek,
H. Jancke, B. Lticke, K. Maeda, J. Mater. Chem. 5 (1995)
1893.
[13] R. SchtJdel, M. Keck, H.-D. Neubauer, Chem. Techn. 44
(1992) 93.
[14] M. Lucas, P. Claus, Chem.-lng.-Tech. 67 (1995)773.
[15] s.J. Tauster, S.C. Fung, R.L. Garten, J. Am. Chem. Soc. 100
(1978) 170.
[16] B.H. Chen, J.M. White, J. Phys. Chem. 86 (1982) 3534.
[17] C. Ocal, S. Ferrer, J. Chem. Phys. 84 (1986) 6474.
[181 s. Sakellson, M. McMillan, G.L. Haller, J. Phys. Chem. 90
(1986) 1733.
[19] D.E. Resasco, R.S. Weber, S. Sakellson, M. McMillan, G.L.
Haller, J. Phys. Chem. 92 (1988) 189.
[20] B.C. Beard, P.N. Ross, J. Phys. Chem. 90 (1986) 6811.
[21] H.C. zur Loye, A.M. Stacy, Langmuir 4 (1988) 1261.
[22] T. Sheng, X. Guoxing, W. Hongli, J. Catal. 11! (1988)
136.
[23] S. Engels, B. Freitag, W. M6rke, W. Roschke, M. Wilde, Z.
anorg, allg. Chem. 474 (198l) 209.
[24] P. Meriaudeau, J.E Dutel, M. Dutaux, C. Naccache, Stud.
Surf. Sci. Catal. 11 (1982) 95.
[25] L, Bonneviot, G.L. Hailer, J. Catal. 113 (1988) 96.
[26] T.M. Salama, H. Hattori, H. Kita, K. Ebitani, T. Tanaka, J.
Chem. Soc., Faraday Trans. 89 (1993)2067.

P. Claus et al./Applied Catalysis A: General 165 (1997) 429-441

[27] J.E Houlihan, L.N. Mulay, Phys. Stat. Sol. b 61 (1974)


647.
[28] T. Hulzinger, R. Prins, J. Phys. Chem. 87 (1983) 173.
[29] J.R. Katzer, G.C. Schmidt, J.H.C. van Hoff, J. Catal. 59
(1979) 278.
[30] J.B. Goodenough, Progr. Solid State Chem. 5 (1971) 149.
[31] C. H. Perry, D. L. Kinser, L. K. Wilson, in R.A. Levy, R.
Hasegava (Eds.), Amorphous Magn., Proc. 2nd Int. Symp.,
1976, Plenum Press, New York, 1977.
[32] W. M6rke, P. Claus, in preparation.
[331 L.K. White, R.L. Belford, J. Am. Chem. Soc. 98 (1976) 4428.
[34] W.P. Halperin, Rev. Mod. Phys. 58 (1988) 533.
[35] D.A. Gordon, R.E Marzke, W.S. Glausinger, J. Physique C2
suppl. 7 38 (1977) C2.
[36] R. Kubo, J. Phys. Soc. Jpn. 17 (1962) 976.
[37] A. Kawabata, J. Phys. Soc. Jpn. 29 (1970) 902.
[38] R. Kubo, A. Kawabata, S. Kobayashi, Ann. Rev. Mater. Sci.
14 (1984) 49.

441

[39] S. Bandow, K. Kimura, Solid State Commun. 75 (1990)


473.
[40] F. Blatter, K.W. Blazey, J. Phys. Chem. Solids 52 (1991) 629.
[41] J.-L. Millet, J.-P. Borel, Solid State Commun. 43 (1982)
217.
[42] P. Meriaudeau, O.H. Ellestad, M. Dufaux, C. Naccache, J.
Catal. 75 (1982) 243.
[43] J.C. Canosa, P. Malet, G. Munuera, J. Sanz, J. Sofia, J. Phys.
Chem. 88 (1984) 2986.
[44] C. Schober, G. Kurz, H. Wonn, V.V. Nemoshkalenko, V.N.
Antonov, Phys. Star. Sol. b 136 (1986) 233.
[45] H. Berndt, H. Mehner, P. Claus, Chem.-lng.-Tech. 67 (1995)
1332.
[46] M.A. Vannice, C.C. Twu, J. Catal. 82 (1983) 213.
[47] M.A. Vannice, C. Sudhakar, J. Phys. Chem. 88 (1984) 2429.
[48] B. Sen, M.A. Vannice, J. Catal. 113 (1988) 52.
[49] A.R. West, Grundlagen der Festk6rperchemie, VCH, Weinhelm, 1992.

Potrebbero piacerti anche