J. Phys. Chem. 1993,97, 5703-5712
5703
Characterization of Ni/SiOz Catalysts during Impregnation and Further Thermal Activation Treatment Leading to Metal Particles Catherine Louis,’ Zbeng Xing Cheng, and Michel Che Laboratoire de RPactivitP de Surface et Structure - URA 1106 CNRS, UniversitP P. et M.Curie, 4 place Jussieu, 75252 Paris Cedex 05, France Received: January 13, 1993
Nickel nitrate impregnated onto silica, identified as such after drying at 25 OC, is gradually transformed into basic nickel nitrate, Ni(N0&*2Ni(OH)2, and then into nickel phyllosilicate (1: 1 nickel phyllosilicate) upon drying at 90 OC. At this temperature, the amount of phyllosilicates which depends on the drying time, can reach 20% of the overall Ni. The temperature programmed reduction (TPR) profiles of impregnated nickel exhibit three peaks of hydrogen consumption at 300, 380, and about 500 O C , attributed to the decomposition of nickel nitrate or basic nitrate into NiO, the reduction of NiO, and the reduction of 1:l nickel phyllosilicates, respectively. The last peak corresponds to the reduction of 20% of the overall Ni content regardless of the Ni loading and the conditions of drying at 25 or 90 OC. This fixed percentage proves that phyllosilicate may be still formed at the beginning of the TPR. After TPR up to 700 OC, the nickel is totally reduced and the metal nickel particles have a constant average size of about 65 %I while the concentration of particles per surface unit increases with the Ni loading. Before complete reduction, the nickel phyllosilicates are probably located at the interface between silica and the remaining nickel, for which they act as anchoring sites. During TPR, after nitrate decomposition, the NiO particles located on the phyllosilicates are reduced into Nio at 400 OC without any migration. Between 400 and 700 OC, the increase in metal particle size is due not only to the reduction of the phyllosilicates but also to NiO migration induced by thermal effect and probably also by the weakening of the anchoring strength because of the reduction of phyllosilicates. Calcination prior to TPR leads to the formation of larger metal particles and to their aggregation, owing to the partial decomposition of phyllosilicates and leading to the formation of larger oxide particles during calcination.
Introduction Supported nickel catalysts find widespread applicationsin many important industrial hydrogenation and hydrogenolysisprocesses such as the steam reforming of methane and higher paraffins or the methanation of coal synthesis gas.Iv2 They are commonly obtained by impregnation of a high surface area oxide carrier with a metal salt precursor followed by drying, activation via calcination, and reduction steps. Silica-impregnated nickel catalysts have been the subject of many studies.3-l1 The nickel nitrate precursor deposited by impregnation is weakly held by the carrier and therefore is easily removed by washing.I2 According to Houalla et al.,5 migration and aggregate formation may readily take place during drying and activation steps. Calcination of the nickel precursor prior to reduction was shown to be detrimental to preparing a high metallic surface area, i.e., small metal particle^.'^-^^ However, the reducibility of the uncalcined catalysts has not been studied in detail,l”I* and, for example, the attribution of the TPR peaks has not been fully elucidated. This point is developed in this work. Nickel silicateswith layer structure, also called phyllosilicates, hydrosilicates, hydroxysilicates, or surface silicates depending on the authors, are supposed to be formed in Ni/SiO2 catalysts prepared by impregnati0n,3,~*~~,~9,20 but to our knowledge, there has been no experimental proof. However, their existence was evidenced on Ni/SiOz catalysts prepared by deposition-precipitation2I-22 and by nickel hexaammine exchange and incipient wetness impregnation at controlled pH.23 Hence, one topic of this study is an attempt to check whether phyllosilicates are formed in the case of preparation by impregnation. Another topic is to investigate their possible influence on the formation of nickel metal particles. They may have an influence; a previous study24has shown that nickel phyllosilicates supported on silica and arising from the water washing of impregnated
nickel or preparation by nickel hexaammine exchange can act as nuclei. In other words, they act as anchoring sites for the growth of particles when nickel is impregnated in a second step, and they allow control of the metal nickel particle size. Therefore, the goal of this study is to get a better insight into the physicochemical phenomena occurring during the different steps of preparation leading to the formation of nickel metal particles supported on silica and to determine the parameters which control the metal particle size.
Experimental Section Catalyst Preparation. The Ni/Si02 catalysts were prepared by incipient wetness impregnation, Le., with an impregnation volume of 1.5 mL/g of silica. Aqueous solutions with different concentrations of nickel nitrate (0.17-3.4 mol/L; beyond this concentration, the solution is saturated) were put into contact with silica. The samples were kept at room temperature for 2 h and then dried either at room temperature or at 90 OC for various time intervals. Samples with 1.5-35 wt % of nickel (expressed per gram of dehydrated silica) were prepared. They were stored in air. They are hereafter referred to as INh, where I stands for impregnated and x for the Ni loading. In a further step, some samples were washed with water at room temperature and filtered on a Buchner. All the samples were prepared with a silica spherosil XOA400 (Rh6ne Poulenc, France, SBET = 400 m2/g, pore volume = 1.25 cm3/g, average pore size = 80 A). Some impregnated samples, hereafter referred to as INix(AD), were prepared witha nonporous silica Aerosil Degussa 200 (SBET = 200 m2/g, average primary particle size = 120 A). Techniques. Chemicalanalyses of nickel and nitrogen elements were performed in the chemical analysis center of CNRS (Vernaison, France). The X-ray diffraction (XRD) patterns were carried out on a
0022-3654/93/2097-5703$04.00/0 0 1993 American Chemical Society
Louis et al.
5704 The Journal of Physical Chemistry, Vol. 97, No. 21. 1993
TABLE I: Structural Parameters of the Ni Reference Samples Detennined by EXAFS at the Ni K-Edge of the Second Shell. samples shell atom number of neighbors distance (A) 0 (A) Q references Ni(OH)2 Ni:Mg(OH)z 1:l phyllosilicate 2: 1 phyllosilicate 0
Ni Mg Ni Si Ni Si
6.0 6.0 6.0 2.5 6.0 5.0
3.13 3.14 3.09 3.30 3.01 3.32
0.09 0.09 0.09 0.08 0.08 0.09
The best fits are obtained by minimizing the agreement factor Q. The electron mean free path,
1.9 x i t 3
21 21 this work
2.1 x 10-3
this work
r, is equal to 1.0 A-2 for all the samples.
TABLE II: Nickel Compounds Identified by XRD, after h y i n g and Storage in Air, of Impregnated Ni/Si02 Catalysts and of the Ni(N03)r6H20Impregnation Solution samples
treatments
Ni compounds identified by XRD
impregnation solution of Ni(N03)2,6H20
drying at 25 OC 90 OC/18 h 90 OC/42-137 h 25 OC/4 months drying at 25 OC 90 OC12-42 h 90 OC/12 h 90 OC/137 h 25 OC/4 months 90 OC/18-137 h 90 OC/351 h
no diffraction pattern no diffraction pattern Ni(N03)~2Ni(OH)2 Ni(N03)yNi(OH)y2H20 and trace of Ni(N03)~2Ni(OH)2 Ni(NO3)~6H20 Ni(N03)~6H20 Ni(N09)2.6HzO and trace of Ni(N03)2.2Ni(OH)2 Ni(N03)2-2Ni(OH)z N ~ ( N O J ) ~ . N ~ ( O H ) ~and -~H trace ~ Oof Ni(N03)2*2Ni(OH)z Ni(N03)2.6H20, Ni(N03)~4HzO,and N i ( N 0 3 ) ~ 2 H 2 0 Ni(N03)2.6H20, Ni(N03)2.4H20, Ni(N03)~2HzO, Ni(N03)2.Ni(OH)2.2H20, and Ni(N03)~2Ni(OH)2 N ~ ( N O ~ ) ~ . N I ( O H ) T ~and H ~trace O , of Ni(N03)2-2Ni(OH)2 no change
INil6
INi35
90 OC/4 months 25 OC/4 months
Siemens diffractometer (D5OO). The phase identifications were performed by comparison with the tabulated JCPDS d-spacing files. Extended X-ray absorption fine structure (EXAFS) measurements at the absorption edge of Ni were performed at the LURE radiation synchrotron facilities (EXAFS I) using the X-ray beam emitted by the DCI storage ring, according to conditions described earlier.z5v26The energy was scanned with 2-eV steps starting from 8200 to 9200 eV (Kabsorption edge of Ni: 8331 eV), using a channel-cut single crystal of silicon as monochromator. About 100 mg of ground samples was put into aluminum slides of different thicknesses. Amplitude and phase functions for 0, Ni, and Si neighbors were experimentallyobtained from the spectra of well-crystallized Ni(OH)2 and Ni-doped Mg(OH)2 (Ni:Mg(OH)z). The latter compound is considered as a reference for Si neighbors because the masses of Si and Mg are very close.I* Thedistances, numbers of neighbors, and DebyeWaller factors are gathered in Table I. The analyses of the EXAFS spectra were performed following standard procedure for background removal and normalization to the edge absorption. Two synthetic nickel silicates with layer structure, a 1:l phyllosilicate (Ni&Os(OH)4) and a 2:l phyllosilicate (Ni3SbO1~(OH)2), kindly,supplied by A. Decarreau (University of Poitiers, France), were used for DebyeWaller factor calibration (Figure 1, Table I). The layers in 1:l phyllosilicate consist of a brucite-type sheet containing Nil1 in octahedral coordinationand a sheet containing linked tetrahedral Si04units. In 2:l phyllosilicate, two sheets of linked Si04 units sandwich the brucite-type sheets.2s The samples were reduced by temperature programmed reduction (TPR), from room temperature to 700 OC, at a heating rate of 7.5 OC/min, in a flow of 5% H2 in argon (25 cm3/min) at atmospheric pressure. The intensities of the TPR profiles are expressed in arbitrary units. The reduction state of nickel after TPR was determined by the chemical analysis described by Coenen.20 This method is based on the measurement of the hydrogen gas generated on dissolution of the reduced nickel in 9 M sulfuricacid. After TPR, the catalysts are purged with argon at 450 OC (25 cm3/min, 1 h) in order to eliminate adsorbed H2, The dissolution of the reduced nickel in 9 N sulfuric acid generates hydrogen gas. Its volume is measured,
Figure 1. @weighted Fourier transform ( k 3 x ( k ) of ) EXAFS spectra of (a) 2:l phyllosilicate and (b) 1:l phyllosilicate.
-
and the reduction state is deduced from the equation
+
H2S04 Nio
NiSO,
+ H,
The nickel metal particle sizes were measured from electron micrographs obtained with a high-resolution electron microscope (HREM, JEOL l00CXII UHR). The average particle diameter d was calculated from the formula d = hidi/& (with En, 300), where ni is the number of particles of diameter di. The detection limit is 10and 15A for metal and oxide nickel particles, respectively. A measurement of the surface concentration of particles (N,)was obtained from counting the number of particles per surface area unit of silica on the micrographs. The measured surface area does not correspond to the real silica surface since ground particles of silica always possess a given thickness. However, the N,values enable the comparison of the particle concentration between different samples.
-
Results and Discussion
Changes in the Chemical Nature of the Catalysts upon Drying. After drying at room temperature, only the INi35 catalyst with the highest Ni loading exhibits a diffraction pattern characteristic of nickel nitrate, Ni(N03)2.6H20. For samples dried at 90 OC in air, the diffraction patterns are only observed for Ni loadings exceeding 16 wt %. They show that nickel nitrate is gradually transformed into the basic nitrate Ni(N03)~.2Ni(OH)2at 90 OC (Table 11). The latter is itself almost completely converted into another one, Ni(NO3)2qNi(OH)2.2H20, on standing in air at room temperature after drying for periods longer than 1 week. Such changes were not observed for catalysts dried at 25 OC. The time required for the transformation of the initial nickel nitrate at 90 "C increases
The Journal of Physical Chemistry, Vol. 97, No. 21, 1993 5705
Characterization of Ni/Si02 Catalysts
TABLE IIk Changes in the N/Ni Atomic Ratio of Impregnrted Ni/Si02 Catalysts and Impregnation Solution during Dryii at 90 O C * samples drying time at 90 OC (h) N/Ni INi 1.7
18 18 18 42 18 42 72 137 25 "C 18 72 137 357
INi3.5 INi6.1 INil6
INi35
impregnation solution of Ni(N03)~6H20
46
1.o 1.2
1.7
00 W
N l 33 3
1.3 2
1.6 1.4 0.74 2 2
1.9 1.6
1.6
m b
0.8
All the samples were stored in air before the determination of the N:Ni ratio by chemical analysis. Months. (I
with the Ni loading. This is confirmed by the measurement of the N/Ni atomic ratio, obtained by chemical analysis for several catalysts after different drying times at 90 OC (Table 111): for a given Ni loading, N/Ni decreases with the drying time and the evolution is less drastic for the higher Ni loadings. Therefore, two types of samples can be distinguished: (i) those dried at 25 or 90 OC for a short time, whose nickel nitrate is not transformed (N/Ni = 2), and (ii) those dried at 90 OC for a long time, whose nickel nitrate is at least partially transformed into basic nickel nitrate (N/Ni < 2). Short and long times depend on the sample Ni loading; for example, short is typically below 18 h for samples containing less than 6 wt 5% Ni and long above 72 h for the INi35 sample (Table 111). For unsupported nickel obtained by evaporation of a solution of nickel nitrate (0.34 and 3.4 mol/L), the transformations upon drying at 90 OC are not the same as those observed for nickel supported on silica (Table 11). After evaporation of the solution, Ni(N03)2*6H20 is slowly dehydrated into Ni(N03)2.4H20 and then into Ni(N03)2*2HzO. These three compounds are detected by XRD after more than 350 h at 90 OC. Finally, they are progressively transformed into two types of basic nitrates, Ni(NO3)yNi(OH)2.2H20and traces of Ni(N03)y2Ni(OH)z,which are the only species detected after 4 months at 90 OC. These results show that nickel nitrate is transformed faster into basic nitrate and via different steps of transformation when it is deposited on a support. In addition, nickel nitrate is transformed all the more slowly when the nickel loading is high. Its stability varies as follows: nickel nitrate from impregnation solution > INi35 > INil6 > > INil.7, indicating that silica favorsthe nitrate transformation. The stabilityof the basic nitrate during storage in air at 25 OC is also lower for supported nickel than from unsupported nickel (Table 11): silica dried at 90 ' C can adsorb water during air storage, which may react with Ni(N03)2.2Ni(OH)2and formNi(NO3)2*Ni(OH)y2H20. The role of water in the transformation of one basic nitrate into the other is supported by the results obtained by Gallezot and Prettre:29the fraction of Ni(NO&.Ni(OH)2,2H20 in the mixture with Ni(NO3)2*2Ni(OH)2disappearsuponheatingat 160'Chair (Table
...
IV). Several papers have dealt with the thermal transformation in air of bulk Ni(NO3)*-6H20 by thermogravimetric analysis (TGA), differential thermal analysis (DTA), and differential scanningcalorimetry (DSC) (refs 29-3 1 and references therein). First, the compound melts in crystal water if the heating rate is fast (Table IV,melting point = 57 and undergoes a stepwise dehydration, leading to the formation of various compounds such as Ni(NO3)2*4H20and Ni(NO3)2*2H20. At higher temperature, the compound is converted into basicnickel nitrates, Ni(NO&.Ni(OH)2.2H20 and Ni(N03)2*2Ni(OH)2,and then into NiO. Our
-
m VI
h
m 0 L
s W
h
e
v
c d
\ N
Y % a 8 v1
5706 The Journal of Physical Chemistry, Vol. 97, No. 21, 1993
Louis et al.
TABLE V: Characteristics of the Impregnated Ni/SiOz Catalysts Containing Ni(N0&.6HzO after Water Washing before washing
after TPR
after washing
samples
drying
washing n(IW)O
samples
Ni(wt %)
N/NiC
INi3.5 INi35 INi47 INi47 INi47
25 O C 18 h/90 OC 18 h/90 O C 18 h/90 O C 18 h/90 OC
1 1 1 3 6
IWNi0.2 IWNi0.4 IWNi0.6 IWNi2.3 IWNi5.1
0.18 0.44 0.64 2.34 5.1
7 2.6 0.8 0.3 0.08
Number of cycles of impregnation-washing.
(A)
Ns(10"/cm2)
aNiO
-
b b 23 23 17
0.4 6.7 10.2
Nonmeasurable. Atomic ratio.
results mainly differ from the published data29-3'by the fact that NiO was not formed, probably because our samples were maintained at too low a temperature. The transformation of supported nickel nitrate upon drying has been also previously studied by Burch and Flambard.I6 An XRD pattern was performed on a Ni/SiOZ catalyst (9.8 wt % Ni) prepared by impregnation with nickel nitrate and then dried at 120 OC in air for 16 h. It shows that nickel formed a mixture of Ni(NO3)2.Ni(OH)2-2H20 (70%) and Ni(OH)2 ( 5 % ) as well as a phase that resembled NiO (25%). The NiO phase was not detected in our samples, probably also because of the low drying temperature. It may be noted that the N/Ni atomic ratios of the INi35 and INil6 samples measured by chemical analysis after 137 h of drying at 90 "C and air storage are 1.6 and 0.74, respectively (Table 111). They are different from that of Ni(N03)2.Ni(OH)2.2H20 (N/Ni = l), which is the main species identified by XRD in these samples (Table 11). These results suggest that other nickel phases coexist with Ni(NO3)2.Ni(OH)2-2H20 and do not diffract: on one hand, nickel nitrate in INi35 was probably not completely transformed into Ni(N03)2.Ni(OH)2*2H20, and on the other hand, other nickel compounds which contain less or no nitrogen may coexist with Ni(N03)yNi(OH)2-2H20 in the INil6 sample. Since Ni(N03)~6H20and Ni(NO3)2*Ni(OH)2*2H20are water soluble,29 it is expected that washing the Ni/SiO2 samples with water will permit the elimination of these phases and leave the unknown ones, which can be studied separately. Characterizationof the Ni/SiOt Catalystsafter Water Washing. Samples ContainingNi(N03)76HzO. Table V shows that when the samples containing Ni(N03)2-6H20 are water washed (hereafter referred to as IWNix), their Ni loading is less than 1%of the initial one, which makes the characterization of nickel difficult. Nevertheless, it is possible to increase the nickel content by performingseveralcycles: impregnationwith high Ni loadingsdrying for a short time (90 OC/18 h)-water washingdrying (90 OC/18 h) (Tablev). Althoughnickelofthewaterwashedsamples did not diffract, it has been possible to identify it by EXAFS. The k3-weighted Fourier transform of the EXAFS spectrum of the IWNi2.3 sampleexhibitsa peak for thesecond coordination shell more intense than that of the first one (Figure 2). It looks similar to that of nickel phyllosilicates (Figure l).25 The simulation of the second peak, which gives a good fit for 6 Ni neighbors and 2.5 Si neighbors with Ni-*.Si distance of 3.31 A and Ni-Ni distanceof 3.08 A (Table VI) indicates that thenickel compound is probably a 1:l phyllosilicate (Table I). However, the catalysts still contain nitrates because of the presence of nitrogen (Table
V)* Samples ContainingNi(NO3)~Ni(OH),2H~O. The INi samples containing Ni(NO3)2.Ni(OH)2.2H20, washed with water as described above, retain 4040% of the initial Ni amount (Table VII). The N/Ni atomic ratios indicate that all the nickel involved in nitrate species was not eliminated. The water washed INil6 sample exhibits a diffraction pattern characteristic of Ni(N03)2*2Ni(OH)~, which was barelyvisiblebefore washing (Table 11). This result agrees with the fact that Ni(N03)2-Ni(OH)2-2H20 is water soluble whereas Ni(N03)2.2Ni(OH)2 is not.29 The &weighted Fourier transformof the EXAFS spectrum of the water washed INil6 sample gives a second peak slightly
+
h
.10 ".'P --Ch
.05
-.05
-.lo
4.000
6.000
8.000
k (A-S Figure2. EXAFS spectra of the impregnated Ni/Si02 catalyst containing N i ( N 0 ~ ) ~ . 6 H *after 0 water washing (IWNi2.3): (a, top) k3-weighted Fourier transform (k3x(k)) and (b, bottom) Fourier filtered inverse spectrum (dots) and simulated spectrum (solid line) for the second shell.
TABLE VI: Structural Parameters of the Water-Washed Impregnated NVSi02 Catalysts Determined by EXAFS at the Ni K-Edge of the Second Shell. shell no. of distance atom neighbors (A) .(A) IWNi2.3 Ni 6 3.08 0.09 Si 2.5 3.31 0.09 water-washed Ni 6.0 3.07 0.1 INi16/90 OC/90 h Si 1.3 3.30 0.08 samples
Q 5.2 X 10-' 6.2 X lo-'
The best fits are obtained by minimizing the agreement factor Q. The electron mean free path,,'l is equal to 1.0 A-2for all the samples.
r
TABLE W: Characteristics of the Kmpr Catalysts Containing Ni(NO&.Ni(OH)r2 WaSning before washing samples INi3.5 JNi6.1 INi9.2 INil6
ted Ni/Si02 after Water
2 0
after TPR
after washing
drying time at 90 OC (h)
Ni (wt %)
N/Nib
66 66 90 90
1.7 2.6 5.1 6.6
0.26 0.63 0.42 0.16
aNi0
Ns
(A)
(10"/cm2)
a
a a
a a
37
a
4.7
Not measured. Atomic ratio.
less intense than that of the first neighbors (Figure 3), indicating that if it corresponds to nickel phyllosilicate, the phase is not pure. A good fit of the second peak was obtained for 6 Ni and 1.3 Si neighbors (Table VI). The Ni-Si (3.30 A) and Ni-Ni distances (3.14A) areclosetothoseofnickelphyllodicates(3.30 and 3.09 A, respectively, Table I). Assuming that the Ni remaining after washing is only composed of a mixture of 1:l phyllosilicate and Ni(N03)2.2Ni(OH)2, their respective propor-
Characterization of Ni/SiO2 Catalysts
The Journal of Physical Chemistry, Vol. 97, No. 21, 1993 5707
k (A-1
Figure 3. EXAFS spectra of the INi16/90 OC/90 h impregnated Ni/ Si02 catalyst containing Ni(N03)yNi(OH)~2H20after water washing: (a, top) &'-weighted Fourier transform (k3x(k))and (b, bottom) Fourier filtered inverse spectrum (dots) and simulated spectrum (solid line) for the second shell. tion, x and y , may be deduced from the EXAFS data of the second shell: 6 Ni and 1.3 Si for the studied sample, 6 Ni for the Ni(N03)2*2Ni(OH)~ compound, and 6 Ni and 2.5 Si for the 1:l phyllosilicate. Therefore, x(2.5Si 6Ni) y(6Ni) = 1.3Si 6Ni, which together with x y = 100% leads to x = y = 50%. From the N/Ni ratio of 0.67 for Ni(N03)2-2Ni(OH)2and 0 for 1:l phyllosilicate,the theoreticalN/Niratioofthestudiedsample was found equal to 0.34. It is larger than the value 0.16 given by chemical analysis (Table VII). Therefore, the presence of Ni species containing no N, such as Ni(OH)2, cannot be excluded. The hypothesis of the presence of Ni(OH)2 remains consistent with the EXAFS results. Indeed, it is not possible to distinguish Ni ions as second neighbors in Ni(OH)* and Ni(N03)~2Ni(OH)z since the Ni-Ni distances are almost the same.29.30 In addition,Ni(OH)2 does not diffract when it is not well crystallized. It may be concluded that 50%of the Ni remaining after washing corresponds to a 1:l phyllosilicate and 50% to a mixture of Ni(N03)2.2Ni(OH)2 and Ni(OH)2. In summary, when the impregnated IN1 catalysts contain Ni(N03)2-6H20,i.e., are dried for a short time, water washing leaves on silica less than 1% of the initial nickel content as phyllosilicates. When they contain Ni(NO&.Ni(OH)2-2HzO, Le., are dried for a long time, water washing leaves 40-508 of the initial nickel content, half as a mixture of Ni(N03)202Ni(OH)*and Ni(OH)2 and half as phyllosilicates. In other words, about 20% of the initial nickel content remains on silica as phyllosilicates. Therefore, long drying times favor the transformation of nickel nitrate into basic nitrate and the formation of nickel phyllosilicates. Phyllosilicates present in impregnated samples are formed during the drying step. They cannot be formed during the impregnation step because the solution is acidic, and they are known to be formed in basic medium, for example, in solution during preparation of Ni/SiO2 by cation exchange at basic pHI7Js or deposition-precipitation.21.26 Characterization of the limpreputted Catalysts During TPR. The TPR profiles of the impregnatedcatalysts, performed without
+
+
+
+
200
300
400
500
600
10
T(OC1
Figure 4. TPR profiles of (a) impregnated INi6.1 catalyst, (b) INi6.1 catalyst calcined at 600 O C under 02, (c) INi6.1 catalyst calcined at 600 O C under 0 2 and H20, (d) Ni/SiOz catalyst impregnated with NiCl2, (e) bulk Ni(N0&.6H20 or basic nitrate, (f) impregnated Ni/SiO2 catalysts containing Ni(N0&.6H20 after water washing (IWNi0.6), and (g) impregnated Ni/SiO2 catalysts containing Ni(NO&.Ni(OH)y2H20 (INi16/90 OC/90 h) after water washing. any previous calcination, exhibit three peaks at 300, 380, and about 500 OC regardless of the Ni loading, the conditions of drying, and storage in air (Figure 4a). The main difference in the profiles is the relativeintensity of the fist peakand theamount of Hz consumed per Ni. The values of these two parameters are higher for samples containing nickel nitrate (HZ/Ni = 2) than for those containing basic nitrate (H2/Ni between 1 and 2, Table VIII). The oxidation of the reduced nickel by H2S04 confirms that nickel is completely reduced after TPR. In order to interpret the TPR profile, further experimentswere performed: After calcination in an O2 flow at 600 OC for 2 h, the INi6.1 sample gives only two reduction peaks at 370 and 450-500 OC (Figure 4b,c) with H2/Ni close to 1. The TPR profile of a sample prepared by impregnation with NiClz (3.2 wt %Ni) exhibitsafterdryingonlyonepeakat 385 OC withashoulder at 430 OC (Figure 4d); H2/Ni is close to 1. The TPR profiles of bulk Ni(N03)2*6H20and basic nitrate arisiig from evaporation of nickel nitrate solution and drying at 90 OC for 4 months (Table 11) exhibits a unique TPR peak at 360 OC (Figure 4e). H2/Ni is equal to 2 and 1.4, respectively. The TPR profiles of both types of water washed samples mainly exhibit a broad hightemperature peak at 450-500 OC (Figure 4f.g). The TPR experiment of the INi6.1 catalyst was interrupted after each H2 consumption peak, Le., at 320,400, and 700 OC. The analysis by XRD or electron microdiffraction shows that nickel is present as NiO after TPR up to 320 O C and as metallic nickel after TPR up to 400 or 700 OC. These TPR experiments lead us to conclude that the first peak at 300 OC arises from the decomposition of the nickel nitrate or
5708 The Journal of Physical Chemistry, Vol. 97, No. 21, 1993
Louis et al.
TABLE VIII: TPR Characteristics of Impregnated Ni/SiOz Catalysts peak 1 samples INi3.5 INi6.1 INi6. 1 INi6. 1 INi9.2 INil6 INi 16 INi35
drying 90 OCJ42 h 25 OC 90 OC142 h 90 OCJ170 h 90 "C 166 h 25 OC 90 OC166 h 90 OCJ72 h
("C) 305 300 290 310 301 290 303 303
Tmax
area (%) 31 48 33 25 35 48 33 29
peak 2 Tmax ("C)
385 375 375 380 382 380 380 385
peak 3
area (%) 56 42 54 61 53 42 54 58
basic nitrate into NiO: The H2/Ni consumption is smaller for basic nitrate than for nickel nitrate, and the peak at 300 OC is the only one to decrease with this ratio. The peak at 300 OC is not observed in theabsenceof nitrate, i.e., for thecatalyst prepared with NiC12 or for the calcined one. After TPR up to 320 OC, NiO is evidenced by XRD. The XRD and electron microdiffraction results allow attribution of the TPR peak at 380 OC to the reduction of NiO into NiO, which requires the nucleation of the metallic phase followed by propagation of the metal oxide reaction boundary.21 Roman and Delmon3 have indicated that the high degree of dispersion of supported nickel oxide may make nucleation more difficult and inhibit the propagation of the reduction reaction boundaries. As a consequence, the high-temperature peak at about 500 OC might be assigned to the reduction of NiO particles smaller than those reducible at 380 OC. However, werather suggest this peak to be due to the reduction of nickel phyllosilicates.. Indeed, the water washed samples containing phyllosilicates also exhibit a high-temperature peak (Figure 4f,g). In addition, phyllosilicates are known to be reducible at high temperature only: Wendt and May33 have reported that 1:1 phyllosilicate exhibits one TPR peak at 567 OC. More recently, Carriat et al.34 have shown that the temperature of the TPR peak increaseswith the crystallinity of phyllosilicates. It variesbetween 500 and 750 OC for 2:l phyllosilicateandbetween 420 and 660 OC for 1:l phyllosilicate. Litvin et al.35have also attributed the high-temperature peak (500-600 "C) of Ni/Si02 samples prepared by precipitation and drying to the reduction of nickel phyllosilicates. More recently, Clause et al.17@have shown that for Ni/SiO2 catalysts prepared by incipient wetness impregnation at different pH (5.7,8.3, and 10.7) and dried at 100 OC the TPR profile consists of three peaks, at 305,400, and about 500 OC. The third peak was also attributed to the reduction of 1:l nickel phyllosilicate, as determined by EXAFS, because both its intensity and the amount of 1:1 phyllosilicate were maximum at pH 8.3. Unsupported nickel nitrate or basic nitrate exhibits only one TPR peak at 360 O G , indicating that both nitrate decomposition and nickel reduction occur simultaneous(Figure 4d). In contrast, these two processes take place separately on supported nickel nitrate or basic nitrate (two peaks at 300 and 380 "C). The supported nitrate transformation occurs at about 60 OC lower than the bulk one. This result is in agreement with the fact that during drying at 90 OC nickel nitrate is transformed faster into basic nitrate. These results prove that the silica support favors the transformation of nickel nitrates. One may believe that the support acidity favorsthe nickel nitrate transformation. However, Burch and FlambardI6 observed the same TPR profiles for impregnated Ni/SiQ2 and Ni/TiOz, indicating that this interpretation is not correct. Our results agree with the DTA-TGA results obtained by Mile et d.I9They show that the temperature for each transformation step is lower for silica-supported nickel nitrate than for bulkcompound (Table IV). They haveattributed this effect to two possible reasons: "An increase in lattice defects and nucleation site densities for the NiO crystal growth in the blocks Of Nil1on the surface," and 'A reaction of the silica surface,
Tmax("C) 500 48 2 485 48 5 500 500 500 520
area (%) 13 10 10 14 12 10 13 13
H*/Ni 1.6 1.9 1.6 1.3 1.5 2.0 1.6 1.4
area 3/(area 2 + area 3) (%) 19 19 19 19 19 19 19 18
probably via surface OH groups, with nickel nitrate or basic nitrate to form surface nickel silicates or hydroxysilicates." As shown in Table VIII, the peak areas corresponding to the reduction of NiO and phyllosilicates represent 80% and 20% of the overall nickel content, respectively, no matter the Ni loading, the conditions of drying, and storage in air. NiO is therefore the main nickel compound formed after nitrate decomposition. The same proportion of 20% of nickel as phyllosilicate was found above in impregnated catalysts dried at 90 OC for a long time, Le., containingbasic nickelnitrate. It is higher than that obtained for samples dried at 25 or 90 "C for a short time (