Sei sulla pagina 1di 14

Applied Catalysis A: General 242 (2003) 1730

Aromatics hydrogenation on silicaalumina


supported palladiumnickel catalysts
V.L. Barrio a, , P.L. Arias a , J.F. Cambra a , M.B. Gemez a,
B. Pawelec b , J.L.G. Fierro b,1
a Department of Chemical and Environmental Engineering, School of Engineers, University of the Basque Country,
Alameda Urquijo s/n, 48013 Bilbao, Spain
b Institute of Catalysis and Petrochemistry, Spanish Council for Scientific Research (CSIC),

Campus de la UAM, Cantoblanco 28049 Madrid, Spain


Received 8 June 2002; received in revised form 29 August 2002; accepted 2 September 2002

Abstract
Nickel- and palladiumnickel catalysts supported on silicaalumina have been studied in the simultaneous hydrogena-
tion of naphthalene and toluene. These catalysts were characterised by means of X-ray diffraction, CO chemisorption,
temperature-programmed reduction, NH3 temperature-programmed desorption, diffuse reflectance spectroscopy, Fourier
transform infrared spectroscopy of adsorbed CO and X-ray photoelectron spectroscopy techniques. All the catalysts studied
showed higher initial intrinsic activity in the hydrogenation of toluene than of naphthalene. Regarding the hydrogenation of
toluene, the 1Pd8Ni/SA sample displayed the strongest resistance to deactivation by coke precursors as compared with the
Ni-free 1Pd/SA catalyst. For the 1Pd8Ni catalyst, the characterisation data pointed not only to a high degree of reducibility
of nickel but also the greatest exposure of Pd species. Both findings appear to be related to the development of nickel hy-
drosilicate species at the support interface. Indeed, a better resistance towards deactivation was obtained by Pd incorporation
and by increasing Ni-loading.
2002 Elsevier Science B.V. All rights reserved.
Keywords: Aromatics hydrogenation; Nickel and palladium catalysts; Silicaalumina support

1. Introduction at low reaction temperatures and moderate hydro-


gen pressures, although in general they show low
Owing to environmental and clean-fuel legislation resistance to sulphur poisoning [4]. Due to the low
[13], deep hydrodearomatisation of diesel fuel has cost and acceptable resistance to sulphur poisoning,
been the focus of many recent studies. Supported Ni-based systems are good alternatives to noble metal
noble metal catalysts are well known for their high hydrogenation catalysts [5,6]. These catalysts are par-
hydrogenation activity for deep hydrodearomatisation ticularly suited to the production of middle distillates
through the hydroconversion of hydrocarbons derived
through the FischerTropsch process.
Corresponding author. Tel.: +34-94-6017282;
Most studies on Ni-based catalysts have been car-
fax: +34-94-6014179.
ried out using silica or alumina supports. The use of
E-mail addresses: iapbacav@bi.ehu.es (V.L. Barrio),
jlgfierro@icp.csic.es (J.L.G. Fierro). silicaalumina as a support has received less atten-
1 Tel.: +34-91-5854769; fax: +34-91-5854760. tion. Recently, Guimon et al. [5] reported that the

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 4 8 9 - 1
18 V.L. Barrio et al. / Applied Catalysis A: General 242 (2003) 1730

addition of Al2 O3 to SiO2 has a positive effect on the 2. Experimental procedure


decrease in the coke formation. This was explained in
terms of the different interaction between the metal 2.1. Catalyst preparation
and the support, as shown by X-ray photoelectron
spectroscopy and temperature-programmed reduction A commercially available support, SiO2 Al2 O3
studies. Among Ni-based catalysts, the PdNi formu- (SMR 5-473, Grace Davison Chemical, hereafter re-
lation seems to be of special interest. One important ferred to as SA) containing 28 wt.% alumina was
aspect of PdNi systems is the very mild deactivation calcined in air at 773 K for 3 h prior to catalyst prepa-
they undergo along the successive reaction experi- ration. Ni/SA and Pd/SA catalysts were prepared by
ments [810]. The use of palladium-promoted amor- wet impregnation of the support with aqueous so-
phous silicaalumina supported nickel catalysts for lutions of Ni(NO3 )2 4H2 O (Carlo Erba, 99%) and
the hydroisomerisation of alkanes has been patented Pd(NO3 )2 2H2 O (Fluka, >98%), respectively. Once
[7]. The lower deactivation of silica-supported PdNi adsorption equilibrium had been reached, excess
alloys as compared to unsupported ones has been re- water was removed in a rotary evaporator. Then,
ported for 1,3-butene hydrogenation [8,9]. In the case the impregnates were dried at 383 K in air for 12 h
of this latter reaction, a strong decrease in deactiva- and calcined in air at 723 K for 2 h. The bimetallic
tion on PdNi/SiO2 catalysts was explained in terms PdNi/SA catalysts were prepared by impregnation
of a preferential migration of Pd atoms on low coor- of calcined Ni/SA catalysts with the Pd(NO3 )2 2H2 O
dination sites occupied by Ni atoms [8]. Additionally, (Fluka, >98%) solution, followed by drying and cal-
Hermann et al. [10] correlated the increased activity cination as above. The metal contents of the catalysts,
of the Pd layer on Ni with a 0.8 eV upward shift of Pd as determined by plasma analysis, are reported in
3d core-levels. By contrast, limited modifications in Table 1. For all the catalysts, the nominal palladium
electronic structure, which have no influence on reac- composition was 1 wt.%. Monometallic and bimetal-
tivity, were observed for the PdNi/SiO2 catalysts [8]. lic systems are referred to xNi and 1PdxNi, respec-
In keeping with the foregoing, it seemed to be of tively, where x denotes the nominal nickel percentage.
interest to study the catalytic behaviour of bimetallic It can be observed in Table 1, that x can be 4 or 8,
PdNi supported on silicaalumina. Accordingly, in and implies a Ni content of 4 or 8 wt.%.
this paper, nickel- and palladium-loaded nickel cata-
lysts were studied in the simultaneous hydrogenation 2.2. Catalyst characterisation
of naphthalene and toluene. The catalysts in their ox-
idic, reduced and used forms were characterised by The chemical compositions of the catalysts were de-
different techniques with a view to explaining the re- termined by inductively coupled plasma atomic emis-
lationship between activity and catalytic properties. sion spectroscopy (ICP-AES) using a Perkin-Elmer

Table 1
Chemical compositiona , textural propertiesb and metal dispersionc of Pd/SA, Ni/SA and NiPd/SA catalysts
Catalyst Metal contenta Dc (%) BETb (m2 /g) dp b (nm) VN2 b (cm3 (CN)g1 )
Ni (wt.%) Pd (wt.%)

SA 394 7.5 118.4


1Pd 0.9 372 6.9 102
4Ni 4.0 26.6 389 6.6 107
8Ni 8.3 14.6 375 5.5 103
1Pd4Ni 3.9 0.8 21.7 324 6.9 89
1Pd8Ni 8.2 0.8 16.1 326 5.7 89
aAs determined by chemical analysis.
bBET specific area, average pore diameter (dp ) and volume of nitrogen adsorbed at P/P 0 = 0.2 as measured by N2 adsorptiondesorption
isotherms at 77 K.
c As measured by CO chemisorption.
V.L. Barrio et al. / Applied Catalysis A: General 242 (2003) 1730 19

Optima 3300DV instrument. Samples were solu- is an easily reducible oxide, even at room temperature,
bilised in a mixture of HF, HCl and HNO3 at 363 K all temperature-programmed reduction experiments
and homogenised in a microwave oven. Diffuse re- were carried out after sample (0.05 g) conditioning (air
flectance UV-Vis spectroscopy was performed on treatment at 473 K for 0.5 h) and further exposure to
the calcined samples using a Shimadzu-UV-2100 the 10% H2 /Ar mixture (air liquide, 99.996% purity;
spectrophotometer. The calcined samples were also flow 60 ml/min) at temperatures from 263 to 1173 K.
characterised by X-ray diffractometry according to Temperature was increased at a rate of 15 K/min and
the step-scanning procedure (step size 0.02 ), using a the amount of H2 consumed was determined with a
computerised Seifert 3000 diffractometer, with Cu K thermal conductivity detector (TCD). A cooling trap
( = 0.15406 nm) radiation, and a PW 2200 Bragg placed between the sample and the detector retained
Brentano /2 goniometer equipped with a bent the water formed during the reduction process.
graphite monochromator and an automatic slit. Scan- Temperature-programmed desorption of ammonia
ning 2 angles ranged from 10 to 70 . Line-broadening measurements were carried out with the same appara-
measurements were carried out using the NiO peak tus described for temperature-programmed reduction.
(43.35 in 2). The full-width at half-maximum The sample (0.050 g) was pretreated in an He (air liq-
(FWHM) of these peaks was corrected for instrumen- uide) stream at 383 K for 1 h and then reduced in a 10%
tal broadening (b). Diffuse reflectance spectra (DRS) H2 /Ar flow at 673 K for 1 h. After cooling down to
of the finely ground calcined samples were recorded 473 K, the reduced sample was ammonia-saturated in
in the 240800 nm range with a Shimadzu UV-2100 a stream of 5% NH3 /He (air liquide) flow (50 ml/min)
spectrophotometer, using BaSO4 as a reference and for 1 h. Thereafter, after catalyst equilibration in a he-
converted to the SchusterKubelkaMunk function. lium flow at 400 K for 0.5 h, the ammonia was des-
The textural properties of the calcined catalysts orbed using a linear heating rate of 10 K/min from 400
were evaluated from the N2 (air liquide, 99.994%) to 873 K. The water evolved was trapped in a KOH
adsorptiondesorption isotherms obtained at 77 K trap placed just before the TCD. In order to determine
over the whole range of relative pressures, using a the total acidity of the catalyst from its NH3 desorp-
Micromeritics ASAP-2000 apparatus, for samples tion profile, the area under the curve was integrated.
previously outgassed at 413 K for 18 h. A value of A semiquantitative comparison of the strength distri-
0.1620 nm2 was taken for the cross-section of the bution was achieved by Gaussian deconvolution of the
physically adsorbed N2 molecule. BET specific areas peaks. Medium and strong acidities were defined as
were computed from these isotherms by applying the areas under the peaks at the lowest and highest
the BET method over the 0.0050.25 P/P0 range. In desorption temperatures, respectively.
all cases, correlation coefficients above 0.999 were The infrared spectra of chemisorbed CO were
obtained. recorded with a Nicolet 5ZDX Fourier Transform
Volumetric CO chemisorption isotherms at 303 K spectrophotometer, working with a resolution of
were obtained in order to estimate metal dispersion. 4 cm1 over the entire spectral range and averaged
Before measurements, the sample was reduced under over 100 scans. The samples, in the form of self-
H2 at 673 K for 1 h and then outgassed at 105 mbar. supporting wafers (thickness ca. 10 mg/cm2 ), were
In order to avoid sintering of the Pd0 particles, the reduced in flowing hydrogen at 673 K for 1 h, and
1Pd sample was reduced at 573 K. After cooling the then outgassed under vacuum at the same tempera-
sample to room temperature, CO was admitted and ture for 1 h. After admission of CO at room tempera-
the adsorption isotherm was measured. The monolayer ture (30 mbar), the fraction of physically adsorbed
adsorption capacity was obtained by extrapolation to molecules was removed by outgassing at room temper-
zero-pressure the linear part of the isotherm. In the ature for 15 min. Net infrared spectra of chemisorbed
calculation of the dispersion, the stoichiometry of CO CO were obtained after subtraction of the background
to either Pd or Ni was assumed to be unity. spectrum of the solid.
Temperature-programmed reduction profiles were The photoelectron spectra of the fresh, reduced and
obtained on a semiautomatic Micromeritics TPD/TPR used catalysts were recorded on a VG Escalab 200R
2900 apparatus interfaced with a computer. Since PdO electron spectrometer equipped with a hemispherical
20 V.L. Barrio et al. / Applied Catalysis A: General 242 (2003) 1730

electron analyser, using a Mg K (h = 1253.6 eV, 3. Results


1 eV = 1.603 1019 J) 120 W X-ray source. The
ECLIPSE software was used to record and anal- 3.1. Chemical and textural properties of oxidic
yse spectra. The reduction of calcined samples was catalysts
achieved in flowing hydrogen at 673 K for 1 h. The
samples used in the reaction were embedded in Although under hydrogenation conditions the cata-
i-octane in order to avoid exposure to air. All catalysts lysts are in reduced state, it is relevant to study their
were outgassed at 105 mbar, and then transferred oxidic precursors because there is a direct relationship
to the ion-pumped analysis chamber, where residual between the properties of calcined catalysts and the
pressure was kept below 7 109 mbar during data corresponding reduced samples.
acquisition. Pd 3d5/2 , Ni 2p3/2 , Si 2p and Al 2p The chemical composition and textural properties of
core-level spectra were recorded and the correspond- the NiPd/SA catalysts as revealed by chemical analy-
ing binding energy (BE) referenced to the C 1s line at sis and nitrogen adsorptiondesorption isotherms, re-
284.9 eV (accuracy within 0.1 eV). Peak intensities spectively, are given in Table 1. For all samples, the
were estimated by calculating the integral of each shape of the N2 adsorptiondesorption isotherms (not
peak after subtraction of an S-shaped background shown) and their hysteresis loops were similar to the
and fitting the experimental peak to a combination of bare SA. As expected, the sorption capacity of the
Lorentzian/Gaussian lines (20% L/G). SA decreased after Ni and/or Pd incorporation. The
drop in the adsorbed volumes of N2 at a relative pres-
2.3. Activity measurements sure of P/P 0 = 0.2, and consequently in the BET
area and porosity, indicate that some of the microp-
The catalysts were tested in the hydrogenation ores and/or the smaller mesopores become occupied
of aromatics. The feed composition was 20 wt.% or blocked by nickel and/or palladium species. Notice-
of naphthalene and 2 wt.% of toluene dissolved in ably, the 8Ni/SA catalyst shows a similar BET area to
n-hexadecane. The reaction was performed in a that of the 1Pd/SA sample. The similar BET specific
bench-scale unit equipped with a stainless steel fixed area of the 1Pd4Ni and 1Pd8Ni catalysts shows that
bed catalytic reactor (1.15 cm i.d. and 30 cm length). the decrease in the BET specific area is mainly gov-
In order to avoid hot spots, 0.5 g of catalyst was erned by the second impregnation step with the Pd
diluted with 0.5 g of SiC (both in the 0.420.5 mm salt. According to the X-ray profiles (not shown), all
particle size range) at a mass ratio of 1:1. Before the calcined catalysts exhibited only one peak com-
the reaction, the catalyst was activated in situ by re- ing from the support (2 = 24.16), but none from the
duction at atmospheric pressure under 166.6 ml/min Pd or Ni phases. Thus, it was inferred either the for-
of H2 /N2 mixture (9:1 v/v) at 673 K for 3 h. After mation of well-dispersed amorphous PdO/NiO phases
catalyst activation, the reactor was cooled to the reac- or solid state reaction between NiO and support to
tion temperature and the system was pressurised. The form an ill-crystallized surface compound (Ni alumi-
reaction conditions were P = 5 MPa; T = 548 K; nate/silicate) [12].
WHSV = 0.48 h1 ; and the hydrogen-to-liquid feed
ratio was 550840 Nm3 /m3 . The reaction was kept 3.2. CO chemisorption
running until steady-state was reached. Liquid sam-
ples were collected for 1 h and analysed by GC Data on the metal dispersion of catalysts obtained
(HP5890) equipped with a FID detector and for prod- by the chemisorption method are given in Table 1.
uct identification a mass selective detector (HPMSD Metal dispersion follows the trend: 4Ni > 1Pd4Ni >
5973) and an atomic emission detector (HPG2350A) 1Pd8Ni > 8Ni. As the metal dispersion of a given
were also used. Besides the high conversions ob- metal usually increases in consecutive impregnations
served for all catalysts, the reaction was free of mass [13], for 1Pd8Ni sample some re-dispersion of the
transport effects since, according to the Weisz param- Ni species after impregnation of 8Ni catalyst with
eter [11], the conversion was always proportional to the palladium solution could be expected. However,
the mass of the catalyst or to the inverse of the flow. for the1Pd4Ni sample, in which Ni-loading is much
V.L. Barrio et al. / Applied Catalysis A: General 242 (2003) 1730 21

lower and Ni species interact strongly with the sup- in NiO generates the latter). Considering the support
port surface, re-dispersion of the nickel phase during composition (28% of Al2 O3 only) and calcination
consecutive impregnation with the palladium salt did temperature (723 K), it is more likely that the nickel
not occur. hydrosilicate would be formed [14], although the
formation of the nickel aluminate cannot be excluded.
3.3. UV-Vis diffuse reflectance analysis
3.4. Temperature-programmed reduction
In order to confirm the identity of the nickel
The temperature-programmed reduction profiles
species and the formation of nickel hydrosilicate, the
of the Ni/SA and PdNi/SA catalysts are shown
UV-Vis diffuse reflectance spectra of Ni/SA cata-
in Fig. 2A and B, respectively. The temperature-
lysts are obtained. Fig. 1 shows the UV-Vis diffuse
programmed reduction profile of the bare SA, also
reflectance spectroscopy profiles for the representa-
tive 4Ni and 8Ni catalysts in the visible range. Both
catalysts showed two distinctive major bands at 370
and 650 nm, and a minor one at around 530 nm.
Since a similar band was observed for NiO obtained
by decomposition of nickel nitrate [12], the band at
370 nm was attributed to the octahedral coordinated
Ni2+ ions. Considering the diffuse reflectance spec-
troscopy study of Houalla and Delmon [12] on the
Ni/silicaalumina catalyst with several alumina con-
tents, we ascribed a band at 650 nm as being due to
Ni2+ in a tetrahedral environment, and the band at
530 nm as being due to Ni3+ (the excess of oxygen

Fig. 2. Temperature-programmed reduction profiles of Ni/SA (A)


and PdNi/SA (B) catalysts. For the sake of comparison, the
temperature-programmed reduction profiles of the Pd/SA and bare
Fig. 1. UV-Vis diffuse reflectance spectra of 4Ni/SA and 8Ni/SA silicaalumina support are given in figures (A) and (B), respec-
catalysts. tively.
22 V.L. Barrio et al. / Applied Catalysis A: General 242 (2003) 1730

included for comparison, exhibits two overlapping catalysts (typical of NiO) shows a broader reduction
peaks of about 947 and 1087 K, associated with the re- peak, which is consistent with a more heterogeneous
duction of oxidic impurities. Temperature-programmed distribution of oxidic nickel species.
reduction profiles of the 4Ni and 8Ni catalysts show
a broad peak with a maximum at around 800 K with 3.5. Temperature-programmed desorption of
a shoulder around 900 K. Considering the UV-Vis ammonia
diffuse reflectance spectra (Fig. 1), the peak at 800 K
and its shoulder at around 900 K can tentatively be Ammonia temperature-programmed desorption is a
assigned to the reduction of Ni2+ species in the oc- commonly used technique for the titration of surface
tahedral and tetrahedral coordination, respectively, acid sites. It is generally accepted that ammonia is an
the latter interacting with the support more strongly excellent probe molecule for testing the acidic prop-
than the former. As can be seen, the increase in erties of solid catalysts since its strong basicity and
Ni content leads to an increase in the intensity of small molecular size allow the detection of acidic sites
the peaks around 800 K and 900 K. Thus, the ex- located even in very narrow pores [17]. The strength
tent of formation of a nickel hydrosilicate/aluminate of an acid site can be related to the corresponding
increases upon raising the Ni content, and this in- desorption temperature, while the total amount of
crease is larger for binary samples. For the 1Pd8Ni ammonia desorption after saturation coverage permits
sample, it is suggested that NiO particles placed quantification of the number of acid sites at the surface.
on the nickel hydrosilicate/aluminate phase are The temperature-programmed desorption profiles are
dominant. given in Fig. 3. Experimental curves were analysed by
From temperature-programme reduction data, the a mathematical fitting program and deconvoluted into
formation of PdNi alloy is excluded since after Pd Gaussian functions. The area of the components, ex-
incorporation onto Ni catalysts no lowering in the NiO pressed as moles of ammonia per gram of catalyst, is
reduction temperature did occur. This can be expected given in Table 2. Thus, temperature-programmed des-
considering that palladium has a lower surface tension orption peaks at two temperature ranges of 529640
and a larger atomic radius than Ni as well as the Ni and 700900 K can be ascribed to ammonia adsorbed
is in large excess compared to Pd. As a consequence, on middle- and high-strength acid sites, respectively.
Pd tends to be expelled out of the Ni matrix leading Middle-temperature acid sites were split into two
to a strong Pd segregation [9]. groups of acid sites: the first from 529 to 576 K and
Temperature-programmed reduction profiles of the the second ones from 601 to 640 K.
Pd-containing catalysts show an H2 -consumption In comparison with the SA support, the total
peak at 327 K and a negative one at 355 K. The for- acidity of the catalysts after metal incorporation
mer comes from the reduction of the Pd oxide species diminishes and strong acidity almost completely
[15] while the latter is usually ascribed to the desorp- disappears, due to the exchange of H+ by Ni2+
tion of hydrogen from the decomposition of a bulk
palladium hydride formed through H-diffusion into Table 2
the Pd crystallites [16]. For the binary systems both Total acidity and distribution of the acidic sites for Pd/SA, Ni/SA
peaks are much more intense than in the 1Pd sample, and PdNi/SA catalystsa
suggesting that the structure and/or location of the Catalyst Acid amount (mmol NH3 /gcat )
PdO phase changes in the presence of nickel. For the
T536576 K T601650 K T702711 K
1Pd4Ni catalyst, the maximum of this peak appears
at almost the same temperature, indicating a similar Bare SA 5.3 5.8
1Pd 3.8
particle size of PdO particles. For the 1Pd8Ni cat-
4Ni 3.1 3.4
alyst, the peak of Pd reduction is also larger than in 8Ni 2.9 4.3
its 1Pd4Ni counterpart. Moreover, the shift in the 1Pd4Ni 1.0 1.0
temperature maximum (350 versus 327 K) points to a 1Pd8Ni 0.7 1.5
stronger interaction between Pd species and the sup- a
As measured by the temperature-programmed desorption of
port. In addition, the second peak of the bimetallic ammonia.
V.L. Barrio et al. / Applied Catalysis A: General 242 (2003) 1730 23

Fig. 3. Normalised NH3 temperature-programmed desorption profiles of reduced 1Pd/SA, Ni/ASA, 1PdNi/SA catalysts and bare support.
For the sake of comparison, profiles have been deconvoluted.
24 V.L. Barrio et al. / Applied Catalysis A: General 242 (2003) 1730

during the impregnation step or to some interaction respectively [19]. The bridged form is characteristic
between the metals and H+ along the calcination of the carbonyl (Pd2 CO) species on the (1 0 0) planes
step. It was noted that this decrease did not correlate of the metal crystal [20]. In comparison with the 1Pd
with the amount of metal incorporated. As a general sample, the IR spectra of the 1Pd8Ni catalyst show
trend, the proportion of weak middle-temperature the less intense band shifted from 2107 to 2097 cm1 .
acid sites decreases with increasing metal contents Weakly adsorbed CO, giving a band in the region
whereas the strong middle-temperature acidity fol- 20702090 cm1 , has been reported for Ni/Al2 O3 and
lows the opposite trend. According to these results, the attributed to CO held by isolated Ni0 atoms or at Ni
middle-temperature acidity in the 536676 K range sites adjacent to oxygen atoms or oxide support [21].
might well be associated to acid sites on the support Considering TPR data, the most likely explanation is
and the middle-strength acidity in the 601650 K that 1Pd8Ni catalyst contains NiO crystallites, which
range could be ascribed to Brnsted-type acidity on are stabilized against reduction in hydrogen. There-
Ni [18]. Compared with the monometallic catalysts, fore, its band at 2097 cm1 may be attributed to CO
the bimetallic ones show the peak maximum of Ni2+ adsorbed at nickel hydrosilicate species.
acidity shifted toward a higher temperature, and this For 1Pd and 1Pd8Ni samples, the most apparent
shift increases with the increase in Ni contents. This difference in region 20631700 cm1 is the general
indicates a higher strength of this type of acidic site increase in the intensity of bands assigned to bridging/
in the bimetallic systems. multisite adsorption. The bands at 1987 and 1880 cm1
could be indicative of the two- and three-fold, respec-
3.6. Fourier transform infrared spectroscopy tively, bridging sites of the CO bonded on the Ni0
of the CO probe [22]. Similarly, the strong band at 1842 cm1 and
much weaker band at 1731 cm1 may be assigned to
The infrared spectra of adsorbed CO on the reduced CO multisite adsorbed on Pd0 or Ni0 phases.
(673 K, 1 h) 1Pd and 1Pd8Ni samples are shown
in Fig. 4. At saturation coverage, the spectra of the 3.7. X-ray photoelectron spectroscopy
1Pd sample displays an intense absorption band at
2107 cm1 and a less intense one at 1959 cm1 , cor- Photoelectron spectra of Si 2p, Al 2p, Ni 2p and Pd
responding to carbon monoxide bonded to the surface 3d core-levels were recorded from reduced and used
of palladium atoms in linear and bridged structures, samples. The binding energies and peak percentages
are summarised in Table 3. The binding energies of
Si 2p and Al 2p were observed in all the samples
at 74.5 and 102.5 eV, respectively. For the reduced
sample, the Ni 2p3/2 profile exhibits three peaks: one
at 852.8 eV, due to metallic Ni0 species; the second
one at 855.4 eV, ascribed to unreduced Ni2+ species,
and the third oneat ca. 860 eVis the satellite of the
second one. This satellite has also been considered for
quantification [23]. Additionally, the Pd 3d5/2 profile
displays a single peak at 335.0 eV, characteristic of
Pd0 [24]. Noticeably, for used Pd-containing catalysts
this peak is shifted toward higher binding energies
(from 335.0 to 335.8 eV). By contrast, in comparison
with the 1Pd4Ni sample, the peak indicative of Ni0
species in the 1Pd8Ni sample is shifted toward lower
binding energies (from 852.7 to 851.5 eV), indicating
a higher electron density in the reduced Ni species. It
Fig. 4. Fourier Transform infrared spectra of adsorbed CO for can also be noted that, with palladium incorporation,
1Pd/SA and 1Pd8Ni/SA catalysts. the reducibility of the nickel species increases from
V.L. Barrio et al. / Applied Catalysis A: General 242 (2003) 1730 25

Table 3
Binding energies (eV) of core electrons and surface atomic ratios of reduced (fresh) and used Pd/SA, Ni/SA and PdNi/SA catalystsa
Catalyst Ni 2p3/2 Pd 3d5/2 Ni/(Al + Si) atomic ratio Pd/(Al + Si) atomic ratio

1Pd Fresh 335.0 0.010


Used 335.8 0.005
4Ni Fresh 852.7 (18) 0.029
855.3 (82)
Used 852.8 (40) 0.032
855.3 (60)
8Ni Fresh 852.7 (25) 0.040
855.3 (75)
Used 852.8 (37) 0.039
855.3 (63)
1Pd4Ni Fresh 852.6 (31) 335.0 0.029 0.015
855.1 (69)
Used 852.7 (54) 335.8 0.016 0.008
855.2 (46)
1Pd8Ni Fresh 852.6 (30) 335.2 0.049 0.023
855.2 (70)
Used 851.5 (41) 335.8 0.048 0.013
854.6 (59)
a From X-ray photoelectron spectroscopy.

1825 to 3031% (see peak percentages in parentheses


in Table 3).
For quantification, the Ni/(Al + Si) and Pd/(Si + Al)
atomic ratios are also offered in Table 3. The
Ni/(Al + Si) surface ratio did not change after
on-stream operation. The only exception is the
1Pd4Ni sample, in which this ratio decreases strongly
in the used sample. Since an intense carbon signal (C
1s peak) was observed in this sample, its Ni/(Al + Si)
ratio should be taken with caution. Thus, the nickel
phase remains virtually unchanged upon on-stream re-
action. The opposite occurs with palladium. Looking
at the Pd/(Al + Si) ratios in Table 3, it is evident that
Pd undergoes some sintering in all the used samples.

3.8. Simultaneous toluene and naphthalene


hydrogenation

Simultaneous hydrogenations of toluene and naph-


thalene were chosen as a test reaction. For the hy-
drogenation of toluene and naphthalene, the intrinsic
activity at 548 K (expressed as gram of toluene or
naphthalene per gram of catalyst and per hour) plot- Fig. 5. Effect of the reaction time on intrinsic activity in the
ted against the run time are shown in Figs. 5 and 6, hydrogenation of toluene. Reaction conditions were: T = 548 K;
respectively. As expected, the intrinsic activity of P = 50 bar; WHSV = 0.48 h1 .
26 V.L. Barrio et al. / Applied Catalysis A: General 242 (2003) 1730

catalysts decreased with time on stream. For the hy-


drogenation of toluene, the 8Ni/SA and 1Pd8Ni/SA
samples show higher activity as compared to the
1Pd/SA sample during the first 7 h on-stream. Con-
sidering the run time without deactivation, the
1Pd8Ni/SA sample shows an approximately two-fold
higher resistance to deactivation than its Pd-free
8Ni/SA catalyst. Similar behaviour has previously
been reported for PdNi alloys supported on silica;
these underwent less deactivation than the Pd-free Ni
catalyst [8]. According to the steady-state data shown
in Fig. 5, only the 1Pd/SA sample is active in the
hydrogenation of toluene whereas the activity of the
other catalysts resembles that of bare SA. Contrary to
toluene hydrogenation, the steady-states of the naph-
thalene hydrogenation activities of all the catalysts
are much more active than bare SA. The activities of
both 1Pd/SA and 1Pd8Ni/SA catalysts are highest
and are very close each other. Interestingly, unlike
Fig. 6. Effect of the reaction time on intrinsic activity in the the hydrogenation of toluene, these two catalysts
hydrogenation of naphthalene. Reaction conditions as in Fig. 5. do not undergo deactivation in the hydrogenation of
naphthalene.

Fig. 7. Effect of the reaction time on selectivity in the hydrogenation of naphthalene (tetralin and trans-decalin formation) and toluene
conversion on the 1Pd4Ni/SA catalyst. Reaction conditions as in Fig. 5.
V.L. Barrio et al. / Applied Catalysis A: General 242 (2003) 1730 27

Fig. 8. Effect of the reaction time on selectivity in the hydrogenation of naphthalene (tetralin and trans-decalin formation) and toluene
conversion on the 1Pd8Ni/SA catalyst. Reaction conditions as in Fig. 5.

For the hydrogenation of naphthalene at 548 K, the As seen in Fig. 7, the main product in the hydro-
selectivities toward tetralin and trans-decalin forma- genation of naphthalene over the 1Pd4Ni/SA cata-
tion for the most active 1Pd4Ni/SA, 1Pd8NiSA and lyst is tetralin. A similar type of behaviour is observed
1Pd/SA catalysts are plotted in Figs. 79, respectively. for the 4Ni/SA catalyst (data not shown). Under the

Fig. 9. Effect of the reaction time on selectivity in the hydrogenation of naphthalene (tetralin and trans-decalin formation) and toluene
conversion on the 1Pd/SA catalyst. Reaction conditions as in Fig. 5.
28 V.L. Barrio et al. / Applied Catalysis A: General 242 (2003) 1730

conditions used in this work, the hydrogenation of


naphthalene to tetralin can be considered as an irre-
versible reaction because at T = 548 K and a hydro-
gen pressure of 50 bar the equilibrium concentration of
naphthalene is negligible [2]. Contrary to the 4Ni/SA
and 1Pd4Ni/SA catalysts, the main reaction product
for the 1Pd/SA, 8Ni/SA and 1Pd8Ni/SA catalysts
is decalin. In particular, the 1Pd/SA catalyst shows
the highest selectivity toward decalin formation (over
90%) (Fig. 9). For the 1Pd4Ni/SA and 1Pd8Ni/SA
catalysts, tetralin and toluene conversion are compared
in Figs. 7 and 8, respectively. The decrease in toluene
conversion in both samples is accompanied by an op-
posite increase in selectivity toward tetralin formation
in the hydrogenation of naphthalene. A similar type
of behaviour is seen for the other catalysts.
All the catalysts studied show a preferential
trans-decalin formation during the whole run time.
The 1Pd/SA catalyst shows the greatest selectivity
toward trans-decalin formation (about 70%) (Fig. 9).
The preferential trans-decalin formation contrasts
with the general behaviour of Group VIII metals
[25,26], but is similar to the trend observed with metal
sulphide [26]. For metal sulphides, the edgewise ad-
sorption of the naphthalene, which allows H-addition
from opposite sites, has been proposed [26]. Such
addition is possible because hydrogen chemisorption
is heterolytic [27]. Interestingly, selectivity toward Fig. 10. Comparison of the intrinsic activity (a) and turnover fre-
cis-decalin formation increases with run time over quencies (TOFs) at zero-time conversion (s1 ) (b) for simultane-
Ni/SA and PdNi/SA catalysts, but not on the 1Pd/SA ous toluene (Tol) and naphthalene (NP) hydrogenation over Pd/SA,
Ni/SA and PdNi/SA catalysts. Reaction conditions as in Fig. 5.
one. Since the 1Pd/SA catalyst shows the greatest
stability during on-stream operation, it seems that
coke formation on the active sites might change the
adsorption mode of naphthalene, thus influencing the hydrogenation activity compared to palladium and the
steroselectivity of decalin formation. synergy between Pd and Ni is hardly expected [8]. The
other important point is the larger TOFt=0 of all cat-
alysts for naphthalene than for toluene. This is in line
4. Discussion with previous findings [28,29] and could be related to
a decrease in the resonance energy per aromatics ring
In order to avoid the influence of deactivation and as well as to differences in the -electron density in the
to establish some catalyst structureactivity correla- aromatics ring as a result of the inductive effect of the
tions, Fig. 10(a) and (b) compare the initial activ- methyl group [30]. As expected, for both toluene and
ity (calculated at zero-time conversion) expressed in naphthalene, there is no correlation between the metal
gToluene/Naphthalene /(gcat h) and the turnover frequen- dispersion determined by CO chemisorption (Table 1)
cies (TOFs), respectively. Considering the latter, the and TOFs values. These confirm that hydrogenation
major feature is that, irrespectively of the reacting of aromatics is a structure insensitive reaction, i.e. spe-
molecule, the 1Pd sample showed larger TOFt=0 than cific activity does not depend on the details of the sur-
other Ni catalysts. This is because Ni has a weaker face structure of the metal crystallites.
V.L. Barrio et al. / Applied Catalysis A: General 242 (2003) 1730 29

Contrary to TOFt=0 values, all catalysts showed Considering additional activity from acid sites
greater specific reaction rates in the hydrogenation of [3,3134], the differences in activity expressed as the
toluene than naphthalene (Fig. 10(a)). For toluene, the initial reaction rates and in TOFt=0 suggest that both
initial intrinsic activity ranking is: 1Pd8Ni/SA > metal and acid sites of silicaalumina may be in-
8Ni/SA 1Pd/SA
4Ni/SA > 1Pd4Ni/SA
volved in the rate-determining step of the hydrogena-
SA, whereas for naphthalene the reactivity of all the tion of toluene [37]. Indeed, contrary to naphthalene,
catalysts is very similar. On comparing the activities the bare SA was active in hydrogenation of toluene.
of both the 1Pd/SA and 8Ni/SA catalysts with the bi- However, in this study no exists correlation between
nary 1Pd8Ni/SA one, the absence of a synergy ef- the ammonia temperature-programmed desorption
fect is evident. A similar absence of a synergy effect data (Table 2) and the trend in catalytic activity
was observed for the silica-supported PdNi systems (Fig. 10(a)). Moreover, the reduction temperature of
tested for the hydrogenation of 1,3-butadiene [8]. the Ni species after Pd incorporation did not change
Considering Ni-containing catalysts, several expla- (Fig. 2). Thus, the additional activity from hydrogen
nations can be advanced to explain the greater reaction spilled over from metal to the acid sites where toluene
rate of the 1Pd8Ni/SA catalyst in the hydrogenation can be adsorbed is hardly possible. This is due to the
of toluene at steady-state: (i) electronic modification formation of large amount of a nickel hydrosilicate
of the metal particles [10]; (ii), the absence of sinter- phase, which modifies the environment of the acid
ing of the metal phases; (iii) additional activity from sites to a considerable extent.
acid sites [3134]; and (iv), increasing resistance to Finally, considering the greater resistance to deacti-
poisoning of the metal sites of the catalysts by coke vation of the 1Pd8Ni/SA catalyst, it can tentatively be
precursors [810]. assumed that as a consequence of the stronger metal-
Considering point (i), the IR spectra of adsorbed lic character of Ni atoms the hydrogenation of ad-
CO and X-ray photoelectron spectroscopy data of sorbed coke precursors may be favoured. Despite this,
the 1Pd8Ni/SA catalyst rule out the participation of the stronger metallic character of Ni0 is produced dur-
electronic modification of Pd atoms by surrounding ing on-stream operation, and this was not seen in the
Ni atoms. The limited modification of the electronic reduced samples. Since 1Pd8Ni/SA catalyst shows
structure of the PdNi/SiO2 catalyst, which have no resistance to deactivation only for 7 h, and after this
effect on the reactivity, were also observed by Re- run time a dramatic inhibition in activity occurs, some
nouprez et al. [8]. Moreover, the similar Ni/(Al + Si) changes in the structure of the catalyst after coke sat-
atomic ratio for the reduced and used 1Pd8Ni/SA uration must occur. This could be explained in terms
catalyst precludes its deactivation due to the sintering of the notion that a drastic decay in the conversion of
of Ni particles. Since the temperature-programmed toluene would be accompanied by a parallel decrease
reduction measurements point to the presence of in selectivity toward decalin formation. Noticeably,
nickel hydrosilicate phases in all the samples, their this decrease in hydrogenation capability was accom-
role could be similar to that of a NiAl2 O4 phase panied by parallel increase in the metallic character
[35]. This phase is considered to act as a support for of Ni0 . Since the X-ray photoelectron spectroscopy
Ni phases [35], providing a stabilising effect while data confirmed a decrease in the exposure of Pd atoms
maintains a very low reactivity [6,36]. Since for the during the reaction, in agreement with the findings of
reduced and used catalysts, the Ni surface exposure Renouprez et al. [8], we can tentatively propose that
measured correlates well with hydrogenation activity, during on-stream reaction the Pd atoms migrate to
the initial formation of nickel hydrosilicate/aluminate the sites where coke is preferentially adsorbed. Since
could inhibit the sintering of nickel species. The only coke formation starts on low coordination metal sites
exception is the 1Pd4Ni catalyst. This is because of [8] and their surroundings and then continues on the
the longer period needed for its stabilisation during support [38], the strong resistance to the deactivation
on-stream operation (100 h; the rest of the catalysts would be related to Pd incorporation and nickel hy-
only required 24 h). As a consequence, this sample drosilicate formation, which might inhibit the growth
simultaneously presents a large amount of coke on of coke precursors at the support interface. Since Pd
the surface. exhibits a greater hydrogenation ability than Ni, once
30 V.L. Barrio et al. / Applied Catalysis A: General 242 (2003) 1730

palladium becomes deactivated a drastic activity de- [6] J.A. Pea, J. Herguido, C. Guimon, A. Monzon, J.
cay is to be expected. Santamaria, J. Catal. 159 (1996) 313.
[7] C.M. van Ballegoy, W.H.J. Stork, J.P. Gilson, Eur. Patent
Appl. EP 587245 A1, 16 Mar 1994, p. 7.
[8] A. Renouprez, J.F. Faudon, J. Massardier, J.L. Rousset, P.
5. Conclusions
Delichre, G. Bergeret, J. Catal. 170 (1997) 181.
[9] P. Miegge, J.L. Rousset, B. Tardy, J.C. Bertolini, J. Catal.
The surface characteristics of SA-supported Ni 149 (1994) 404.
and PdNi catalysts and their catalytic performance [10] P. Hermann, B. Tardy, D. Simon, J.M. Guignier, B. Bigot,
for aromatics hydrogenation have been studied. The J.C. Bertolini, Surf. Sci. 307 (1994) 422.
following conclusions were obtained: (i) an increase [11] C.N. Satterfield, Heterogeneous Catalysis in Industrial
Practice, 2nd ed., Mc Graw-Hill, New York, 1980.
in activity in the hydrogenation of toluene might [12] M. Houalla, B. Delmon, J. Phys. Chem. 84 (1980) 2194.
be attained by Pd incorporation and by increasing [13] C.V. Cceres, J.L.G. Fierro, M.N. Blanco, H.J. Thomas, Appl.
Ni-loading. However, for the binary (PdNi) catalysts, Catal. A: Gen. 10 (1984) 333.
no synergy effect between both Pd and Ni metals was [14] J. van de Loosdrecht, A.M. van der Kraan, A.J. van Dillen,
observed; (ii) the greater activity of the 1Pd8Ni/SA J.W. Geus, J. Catal. 170 (1997) 217.
[15] T.C. Chang, J.J. Chebn, V.T.J. Yeh, J. Catal. 96 (1985) 51.
catalyst can be associated with its greater resistance to [16] M.A.Vannice, P. Chou, in: Proceedings of the 8th International
deactivation due to coke formation because the larger Congress on Catalysis, Tokyo, 1984, p. V-99.
amount of metal phase incorporated could inhibit the [17] M.C. Kung, H.H. Kung, Catal. Rev. Sci. Eng. 27 (3) (1985)
growth of coke precursors at the support interface. 425.
As a general conclusion, the incorporation of 1% Pd [18] S. Benbenek, E. Fedorynska, P. Winiarek, React. Kinet. Catal.
Lett. 51 (1) (1993) 189.
and a larger amount of Ni (8 wt.%) are beneficial for [19] J.K. Lee, H.K. Rhee, J. Catal. 177 (1998) 208.
aromatics hydrogenation at 548 K over silicaalumina [20] P. Marcot, A. Akhachane, J. Barbier, Catal. Lett. 36 (1996)
supported catalysts. These catalysts could be used 37.
in the final stages of refinery processes, where high [21] J.B. Peri, J. Catal. 86 (1984) 84.
quality and very low environmental impact diesel fuel [22] B. Pawelec, L. Daza, J.L.G. Fierro, J.A. Anderson, Appl.
Catal. A: Gen. 145 (1996) 307.
must be produced. [23] W. Dai, M. Qiao, J. Deng, Appl. Surf. Sci. 120 (1997) 119.
[24] Y. Yazawa, H. Yoshida, N. Takagi, S. Komai, A. Satsuma, T.
Hattori, J. Catal. 187 (1999) 15.
Acknowledgements [25] A.W. Weitkamp, Adv. Catal. 18 (1968) 1.
[26] J. Shabtai, N.K. Nag, F.E. Massoth, in: M.J. Phillips, M.
The University of the Basque Country, Repsol-YPF Ternan (Eds), Proceedings of the 9th International Congress
Foundation, and grants-in-aid from CICYT (Spain) on Catalysis, Calgary, vol. 1, Chem. Institute of Canada,
and the EU supported this research. One of us (B.P.) Ottawa, 1988, p. 1.
[27] R. Redey, W.K. Hall, J. Catal. 119 (1989) 534.
acknowledges financial support from the Spanish Min- [28] M.V. Rahaman, M.A. Vannice, J. Catal. 127 (1991) 251.
istry of Science and Technology through the Ramon [29] R.M. Navarro, B. Pawelec, J.M. Trejo, R. Mariscal, J.L.G.
y Cajal program. Fierro, J. Catal. 189 (2000) 184.
[30] C. Moreau, P. Geneste, in: J.B. Moffat (Ed.), Theoretical
Aspects of Heterogeneous Catalysis, Van Nostrand Reinhold,
References New York, 1990, p. 256.
[31] M.V. Rahaman, M.A. Vannice, J. Catal. 127 (1991) 251.
[1] A. Avidan, B. Klein, R. Ragsdale, Hydroc. Process 80 (2) [32] P. Chou, M.A. Vannice, J. Catal. 107 (1987) 129.
(2001) 47. [33] S.D. Lin, M.A. Vannice, J. Catal. 143 (1993) 539, 554, 563.
[2] A. Stanislaus, B.H. Cooper, Catal. Rev. Sci. Eng. 36 (1) [34] H. Liu, G.D. Lei, W.M.H. Sachtler, Appl. Catal. A: Gen. 137
(1994) 75. (1996) 167.
[3] B. Pawelec, R. Mariscal, R.M. Navarro, S. van Bokhorst, S. [35] A. Al-Ubaid, E.E. Wolf, Appl. Catal. 40 (1988) 73.
Rojas, J.L.G. Fierro, Appl. Catal. A: Gen. 225 (2002) 223. [36] R. Lamber, G. Schulz-Ekloff, J. Catal. 146 (1994) 601.
[4] C.H. Bartholomew, P.K. Agrawal, J.R. Katzer, Adv. Catal. 31 [37] S. Siffert, J.-L. Schmitt, J. Sommer, F. Garin, J. Catal. 184
(1982) 135. (1999) 19.
[5] C. Guimon, N. El Horr, E. Romero, A. Monzon, Stud. Surf. [38] C.L. Pieck, R.J. Verderone, E.J. Jablonski, J.M. Parera, Appl.
Sci. Catal. 130D (2000) 3345. Catal. 55 (1989) 1.

Potrebbero piacerti anche