A. Frennet and J.-M. Bastin (Eds.) Catalysis and Automotive Pollution Control 111
Studies in Surface Science and Catalysis, Vol. 96 9 1995 Elsevier Science B.V. All rights reserved.
S T R U C T U R E AND C A T A L Y T I C A C T I V I T Y OF MIXED OXIDES OF P E R O V S K I T E S T R U C T U R E V. Mathieu - Deremincel, J.B. Nagy and J.J. Verbist Groupe de Chimie Physique, Facultds Universitaires Notre-Dame de la Paix, 61 Rue de Bruxelles, B-5000 Namur, Belgium
ABSTRACT The Lal-xCexBO3 (B = Ti, Cr, Mn, Fe, Ni, Co) mixed oxides of perovskite structure present a catalytic activity for CO oxidation and NO reduction which increases with increasing Ce content. LaCoO3 and LaMnO3 are the more active compounds for both CO oxidation and NO reduction. The oxidation of CO by 02 follows a suprafacial type mechanism where the adsorbed oxygen is the active species. The reduction of NO by CO is best explained by a redox mechanism. Ce is only incorporated in a limited amount in the perovskite structure and a concomitant formation of CeO2 is also observed. For a low Ce content, CeO2 is highly dispersed on the surface of the catalyst particles, while larger particles are obtained at higher Ce content. The presence of Ce leads to an increase in both CO oxidation and NO reduction activities. The CO oxidation is dependent on the dispersion of CeO2 on the mixed oxides particles. On the other hand, the NO reduction does depend on the Ce content in the perovskite structure.
Despite the fact that perovskite-type oxides have been suggested as substitutes for noble metals in automotive exhaust catalysis , relatively few studies were devoted to the synthesis, characterization and catalytic activity of these catalysts. These compounds, of general formula ABO3, generally include a lanthanide element A and a transition metal B. Partial substitution of A by Ce is also interesting, as Ce is a well known additive of exhaust catalyst . Its oxygen storage ability is used to broaden the air/fuel ratio window, improving the activity of three way catalysts (CO, hydrocarbon oxidation, NO reduction).
1 Present address : Solvay & Cie, Rue de Ransbeek 310, B-1120 Bruxelles, Belgium
394 It seems that the lattice oxygen plays a direct role in the oxidation of CO. Indeed, it was found that the catalytic activity is maximum, if the bond energy of the lattice oxygen is minimum . Hence, the oxygen vacancy is also an important factor for the catalytic activity. The partial substitution of the A element can monitor the oxygen vacancy. If Ce(IV) is introduced in the perovskite structure A site vacancies are produced . The partial substitution of B element by Pt (IV) increases greatly the CO oxidation activity of Lao.7Pbo.3MnO3 perovskite . For the NO reduction by CO AMnO3 perovskites were the most studied catalysts . The importance of anion vacancies was emphasized for the good catalytic activity. The ACoO3 was also found a good candidate for both CO + 02 and CO + NO reactions . Very recently perovskite-related oxides were also proposed as catalysts for the direct decomposition of NO into N2 and 02 [8, 9]. This paper deals with the characterization and catalytic activity of cerium substituted lanthanum perovskites Lal-xCexBO3 (B = Ti, Cr, Mn, Fe, Ni, Co),x ,varying from 0 to 0.6 .
Lal_xCexBO3 compounds were synthesized by evaporation of a solution of the respective metal nitrates or oxalates and calcination of the precursors either at 1000~ or at 1200~ for 5 h . X-ray powder diffraction patterns were recorded on a Philips PW 1730 diffractometer using Ni filtered Cu Ka radiation. XPS spectra were obtained on a Hewlett Packard 5950 A spectrometer with A1 Ka radiation. The catalysts were tested for the CO + 02 and CO + NO reactions in a flow reactor system. Before the tests, the compounds (1.0 g) mixed with small glass beads (weight ratio 1:1) were pretreated overnight in N2 at 400~ and then in a mixture of CO, 02, N2 and Ar or CO, NO, N2, Ar, respectively for 30 min at the same temperature to attain steady- state conversion. The same flow rate (Q = 150 ml/min), the same mixture of reagents (20 % CO, 10% 02, 70 % N2 or 20% CO, 20% NO, 20% Ar, 40% N2, respectively, in mol %) and the same reaction temperature (T = 100-400~ or 300-500~ respectively) were used to compare the activity of the different catalysts. For the determination of the kinetic schemes, the differential mode of the reactor was used (% conversion < 15%).
395 3. R E S U L T S AND DISCUSSION 3.1. X-ray Diffraction
Since no traces of La2 03 or BOx appear in the diffraction patterns of LaBO3 (B = Cr, Mn, Fe, Co), the perovskite is completely formed at 1000~ in the systems not containing Ce. With B = Ni or Ti, additional phases appear which are La2 NiO4 or La2 Ti2 07, respectively. Moreover, the stoechiometry of the latter perovskite is Lao.7TiO3. The Ce introduced in the reagents is not totally incorporated in the perovskite lattice. The excess crystallizes as CeO2 [4, 12]. The latter is influenced by the nature of metal B (Table 1). For compositions x < 0,05, no trace of CeO2 is detected. This can be due either to a total incorporation of Ce into the perovskite lattice or to a very highly dispersed (and/or amorphous) CeO2 not detected by XRD. For higher Ce content, the relative intensity of CeO2 increases with increasing x, the slope being in the following order as a function of B : Ni > Co > Mn > Cr > Fe > Ti.
Table 1 Relative heights of X-ray diffraction peaks characterizing the different catalysts calcined at 1000~ ICeO2 flLaBO3
0.01 0.03 0.05 0.07 0.10 0.15 0.20 0.40 0.60
0 0 0.05 0.13 -
0 0.06 0.22 0.73 -
0 0.06 0.14 0.31 0.77 1.37
0 0.04 0.08 0.08 0.17 0.21 0.46 0.98
0 0.12 0.17 0.22 0.26 0.35 0.88 1.41
0 0 0.13 0.21 0.33 0.28 0 58 1 20 1 79
ICeO2-d=3.16A;ILao.7TiO3-d=2.74A;ILaCrO3 - d = 2.74 A; ILaMnO3 - d = 2.72 A; ILaFeO3 - d - 2.78 A; ILaCoO3 - d = 2.67 A; ILaNiO3 - d = 2.74 A b :x is defined as Lal_xCexBO3
Table 2 Amount of adsorbed oxygen (in %) and Olattice/La atomic ratios of perovskites calcined at 1000~ Perovskites
La0.7TiO3 LaCrO3 LaMnO3 LaFeO3 LaCoO3 LaNiO3
35.9 42.5 49.2 61.9 84.9 71.0
4.67 3.04 2.53 1.75 0.59 1.09
a : Theoretical atomic ratios : 3.0, except for Ti(4.3) The formation of extra perovskite lattice CeO2 gives rise to three different results: - appearance of A vacancy sites (O) following Lal-xOxBO3 +xCeO2, - migration of (excess) B ions outside the perovskite lattice to yield BOx : Lal-xB 1-xO3-x + xBOx + xCeO2. - the occurrence of both phenomena. For low Ce content, the first process is predominant, while for higher Ce content BOx phase also appears. The relative importance of the two processes is highly dependent on the nature of B. Neither TiO2 nor Cr203 are detected for any initial x value. Only traces of Fe203 are shown for x = 0.40 and 0.60. Mn203 starts to appear for x = 0.20 and CoO for x = 0.10. NiO is detected for any x values studied. 3.2. X - r a y p h o t o e l e c t r o n s p e c t r o s c o p y
The valence state of La and B ions in LaBO3 perovskites is equal to (III) in all compounds without Ce, but for Ti, where Ti is in (IV) oxidation state, in agreement with the stoechiometry ofLao.7TiO3 perovskite. The atomic ratio computed form the B2p and La3d core peak intensities is close to the theoretical value of 1 in LaBO3 for B=Cr, Mn, Fe and Co. For Ti, this ratio is equal to 1.44, corresponding to the above mentioned stoechiometry. The XPS spectra of O ls are more interesting to characterize the nonstoechiometric composition of the perovskites. Indeed, their catalytic activity is
397 dependent on the excess or default of oxygen in the lattice. The higher energy O ls peak (Eb = 530.5 eV with respect to Eb of carbon, 284.7 eV) is attributed to adsorbed oxygen and the lower energy peak (Eb = 529.0 eV) to lattice oxygen . The amount of adsorbed oxygen increases as a function of the atomic number of B ions, exception for Ni (Table 2). From the comparison of the theoretical OlatticefLa ratio (equal to 3) with the experimental values, oxygen deficiency is detected for all B ions, excepting Ti and Cr (Table 2). The importance of oxygen deficiency increases with the atomic number of B ions. Hence, it seems that the oxygen is essentially adsorbed at anion vacancies. The perovskites synthesized in presence of Ce do not exhibit any bond energy difference for La or B ions. On the oflaer hand, despite of the complexity of the Ce XPS spectra, the Ce3d core peak intensities and the Ce(IV) satellite intensity allowed us to estimate the relative amotmt of Ce(III) and Ce(IV) on the surface [10,11]. The amount of Ce(III) is considered as being incorporated in the perovskite lattice, Ce(III) replacing La(III) in Lal-xCexBO3. For x<0.20, the incorporation of Ce(III) decreases with increasing x values for all B ions (Figure 1). The amount of Ce(III) is a fimction of the nature of B ions. The following sequence is obtained : Ti > Fe > Mn > Co > Cr. A rather high amount of Ce(III) accompanies Ti, while it is difficult to incorporate it in presence of Cr. Fe, Mn and Co represent intermediate cases. 100
9 9 o 9
Ti Cr Mn Fe Co
Figure 1. Variation of the amount of Ce(llI) (in %) as a function of x in the perovskites Lal-xCexB03.
Table 3 Atomic ratios Ce/La from XPS La3d and Ce3d peak intensities for the catalysts Lal_xCexB03 calcined at 1000~ Ce/La (XPS) x
0.03 0.05 0.07 0.10 0.15 0.20 0.40 0.60
0.031 0.052 0.075 0.11 0.18 0.25 0.67 1.50
0.12 0.23 0.45 -
0.12 0.23 0.36 0.79 -
0.20 0.25 0.23 0.28 0.51 0.56 1.15
0.22 0.21 0.30 0.25 0.23 0.56 0.68 0.90
0.13 0.19 0.17 0.28 0.43 0.54 0.89
Following these values, two behaviours can clearly be distinguished. For x ~ 0.40, the surface atomic ratios are higher than the bulk values for all B ions, showing a surface excess Ce. Oppositely, for x > 0.40, due to the sintering of CeO2 particles, the surface atomic ratios are lower than the corresponding bulk values. For low x values, the surface CeO2 is highly dispersed and not detected by XRD (Table 1). For high x values, the sintering of CeO2 particles is also confirmed by XRD. Note, that the surface Ce(III)/La values are generally independent on the Ce/La ratio in the synthesis mixture. This amount of Ce is considered being incorporated in the perovskite lattice. As a conclusion, it can be underlined that Ce(III) is incorporated in the lattice, while CeO2 is either in a highly dispersed form or is found as larger particles. The O 1 s bond energies are not influenced by the Ce substitution. Two types of oxygens - lattice and adsorbed - are also distinguished as for the pure perovskites. However, the quantitative analysis is more difficult in this case, because both perovskite and CeO2 contain the two types of oxygen. The Olattice/La ratios are all smaller than the corresponding theoretical values, indicating that all the catalysts are oxygen deficient. For Fe, Co and Ni, the decrease of Oads as a fimction of x shows the same variation as Ce(III) surface (Figure 2a). Hence, it can be concluded that the
399 incorporation of Ce(III) in the lattice has a negative effect on the adsorption of oxygen. 100t 80
O Fe *
Figure 2. Variation of the amount of adsorbed oxygen (in %) as a function of x in the perovskites Lal-xCexB03. For Cr and Mn, a continuous decrease of Oads is observed, while for Ti, Oads increases greatly as a function of x (Figure 2b). The latter could be explained by the non-stoechiometry of the initial perovskite Lao.7TiO3, by the high dispersion of the CeO2 formed which also contribute to the amount of Oads. 3.3. Catalytic activity The pure perovskites are all more active for CO + O2 than CO + NO reactions. The best catalysts for both reactions are LaMnO3 and LaCoO3. The activity of the different pervoskites for CO oxydation can be linked semiquantitatively with the ease of anionic vacancy formation in the lattice, described by the B-O bond energy (Figure 3). The formal kinetics were determined only on the less active LaCrO3 catalyst. The Langmuir-Hinshelwood model explains quite well the experimental rate equation 0.5 R=kPcoPo 2
Figure 3. Comparison between the catalytic activity and the B-O bond energy in the perovskites LAB03 ~ = 77, Cr, Mn, Fe, Co, Ni). No CO conversion was observed in absence of adsorbed oxygen, confirming the reaction mechanism involving both adsorbed species. The apparent energy of activation is equal to 12.2 kcal mo1-1. The CO + NO reaction is better explained by a redox mechanism, where CO is oxidized by the catalyst which is regenerated by the reduction of NO. Moreover, a dissociative adsorption of NO better explains the experimental results on LaCoO3 and LaMnO3, while a molecular adsorption has to be supposed on LaCrO3 adn LaFeO3. Simultaneous adsorption of CO and NO shows a competiton for the same active sites . The latter are generally surface oxygens leading to the formation of nitrates, nitrosyls, dinitrosyls, and carbonates, carbonyls, respectively. On the other hand, the dissociative adsorption of NO leads to the formation of N2 at higher temperature. The following reaction scheme is adequate for the LaCoO3 and LaMnO3 catalysts: CO + *
N O + 2"
<:,a N--* + O--*
CO--* + O--*r
C O 2 + 2*
1/2N2 + *
401 Leading to R = kp~op,o o.5 1 + k'p~o
The computation of the activation energy allows one to determine Ea3 + 0.5 AHNO : 14.4 kcal mo1-1 for LaMnO3 and 12.5 kcal mo1-1 for LaCoO3, respectively. Ea3 is the activation energy for the CO oxidation (step 3) and AHNO is the adsorption energy of NO in a dissociated form. The heat of adsorption of CO is determined to be equal to -7 kcal mo1-1, value to be compared to the literature value of- 6 kcal mol-1 [ 15]. The competition between CO and NO is detected in the reaction scheme CO + NO on LaCrO3 and LaFeO3. Indeed, a negative partial order equal to -1 is obtained for NO on both catalysts. The reaction scheme allows one to explain the experimental partial orders and the rate equation R - kpco I + k'p~o
From the apparent activation energies, the heat of adsorption for NO adsorption is determined to be equal to - 9 kcal mol-1. The addition of Ce to the perovskites leads to different effects depending on the nature of B ions and on the relative amount of Ce. It has to be emphasized, that CeO2 itself is also a good catalyst for CO oxidation with 02 (97 mol % CO conversion at 300~ This activity is equal to that of LaMnO3, but it is inferior to the activity of LaCoO3. The following reaction scheme is adequate for the LaCoO3 and LaMnO3 catalysts 9 CO + * NO + 2*
C O ~ * + O~*r N~* r
CO~* N~* + O~*
CO2 + 2* 1/2N2 + *
402 Leading to R = kp~~176 0.5 1 + k'p~o
The computation of the activation energy allows one to determine Ea3 + 0.5 AHNO : 14.4 kcal mo1-1 for LaMnO3 and 12.5 kcal mol-1 for LaCoO3, respectively. Ea3 is the activation energy for the CO oxidation (step 3) and AHNO is the adsorption energy of NO in a dissociated form. The heat of adsorption of CO is determined to be equal to -7 kcal mol-1, value to be compared to the literature value of- 6 kcal mol-1 . The competition between CO and NO is detected in the reaction scheme CO + NO on LaCrO3 and LaFeO3. Indeed, a negative partial order equal to -1 is obtained for NO on both catalysts. The reaction scheme allows one to explain the experimental partial orders and the rate equation R = kpco I + k'p~o
From the apparent activation energies, the heat of adsorption for NO adsorption is detennined to be equal to - 9 kcal mo1-1. The addition of Ce to the perovskites leads to different effects depending on the nature of B ions and on the relative amount of Ce. It has to be emphasized, that CeO2 itself is also a good catalyst for CO oxidation with O2 (97 mol % CO conversion at 300~ This activity is equal to that of LaMnO3, but it is inferior to the activity of LaCoO3. The activity of LaCoO3 first increases as a function of x in Lal-xCexCoO3. This increase is explained by the beneficial effect of dispersed CeO2 on the surface of the catalyst. However, the concomitant increase of Ce(III) incorporation into the perovskite lattice leading to A site vacancies could suggest that these cation vacancies could also have a positive effect on CO oxidation, as it was suggested earlier . Note, that a mechanical mixture of CeO2 (5 mol%) and LaCoO3 is less active than the corresponding Ce containing Lao.95Ceo.05CoO3 catalyst. For higher Ce content, the activity decreases and is levelled off, showing that the larger CeO2 particles are less active. Similar conclusions can be drawn for the Mn, Fe and Cr containing catalysts.
403 The activity of Lal-xCexTiO3 catalysts is systematically lower than that of Lao.7TiO3. A first large decrease of the activity for x = 0.05 is followed by a slight increase for x > 0.05 due to the formation of CeO2 on the surface. Finally the Lal-xCexNiO3 catalysts are more active than LaNiO3 for x > 0.2. For lower x values the influence of Ce is negligible. CeO2 is a relatively good catalyst for NO + CO reaction. Its activity (equal to 10 mol% CO conversion at 300~ is higher than those of Ti, Cr, Mn, Fe and Ni perovskites (1.7 %, 2.5 %, 9.0 %, 5.9 % and 2 . 1 % , respectively). Only Co perovskite shows a higher activity (equal to 40.8 % conversion). Note, that the activities are always lower for the NO reduction than the CO oxidation. For all perovskites the increase in activity as a function of x can be linked to the increase of Ce(III) content in the lattice. For Ti and Ni perovskites, after an initial increase of the activity a plateau is reached. For Fe and Cr perovskites, the activity is decreasing after the initial increase, showing a maximum in the activity vs x curves. The Mn perovskite shows an initial small decrease of activity followed by a larger increase as a function of x. The behaviour of Lal-xCexCoO3 perovskites is more complex. A quite large initial decrease of activity is observed for x < 0.05. After that value a maximum is reached at x = 0.20 (activity = 65 mol % CO conversion at 300~ followed by a decrease in activity (ca 40 mol% CO conversion at x = 0.60). For the initial addition of Ce, the activity decreases because the CeO2 fonned has a lower activity. At higher Ce content, larger CeO2 crystallites are formed freeing the more active surface of the perovskite catalyst. Finally, the activity decreases because the amount of Ce(III) incorporated in the lattice decreases. Note, that for small x values, the influence of Ce(III) concentration is hidden by the effect of CeO2 on the surface. A kinetic study for the CO + NO reaction on La0.95Ceo.05CoO3 revealed a decrease of the partial order with respect to NO : it is equal to 0.2, in comparison with the value of 0.5 obtained for the pure perovskite. Further studies are necessary to ascertain the specific role of both Ce(IV) and Ce(III) ions in the CO + NO reaction.
404 4. CONCLUSION
Ce(m-) is incorporated in the perovskite lattice for low x values in Lal-xCexBO3 perovskites.The remaining Ce is in a highly dispersed CeO2 form. For higher x values, larger CeO2 crystals are formed. The different phases formed are active in both CO + 02 and CO + NO reactions. The activity for the former reactions is always higher than that for CO + NO. The best perovskites are obtained with B = Mn and Co.
10 11 12 13 14 15 16
W.F. Libby, Science, 171 (1971) 499. F. Le Normant, P. Bernhardt, L. Hilaire, K. Kili, G. Krill and G. Maire, Stud. Surf. Sci. Catal., 30 (1987) 221. T. Shimizu, Chem. Lett., (1980) 1. T. Nitadori, S. Kurihara and M. Misono, J. Catal., 98 (1986) 221. D.W. Jolmson, P.K. Gallagher, G.M. Wertheim and E.M. Vogel, J. Catal., 48 (1977) 87. R.J.H. Voorhoeve, J.P. Rameika, L.E. Trimble, A.S. Cooper, F.S. Disalvo and P.K. Gallagher, J. Solid State Chem., 14 (1975) 395. G.L. Bauerle, N.T. Thomas and K. Nobe, Chem. Eng. J., 4 (1972) 199. Y. Teraoka, T. Harada, H. Fund~awa and S. Kagawa, Guczi L. et al. (eds), New Frontiers in Catalysis, Stud. Surf. Sci. Catal., 75 (1993) 2649. H. Shimada, S. Miyama and H. Kuroda, Chem. Lett. (1990) 1. V. Deremince-Mathieu, H. Collette, J.B. Nagy, E.G. Derouane and J.J. Verbist, Inorg. Chim. Acta, 140 (1987) 41. V. Deremince-Mathieu, PhD Thesis, FUNDP, Namur, 1989. H. Arai, T. Yamada, K. Eguchi and T. Seyiama, Appl. Catal., 26 (1986) 265. J.L.G. Fierro and L.G. Tejuca, Appl. Surf. Sci., 27 (1987) 453. see e.g.M.A. Pefia, J.M.D. Tascon, J.L.G. Fierro and L. Gonzalez Tejuca, J. Colloid Interface Sci., 119 (1987) 100 and references therein. J.M.D. Tascon and L.G. Tejuca, React. Kinet. Catal. Lett., 15 (1980) 185. R.J.H. Voorhoeve, in 'The Catalytic Chemistry of Nitrogen Oxides', J.C. Larson, ed., New York, 1975, p 215.