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.
CATALYTIC COMBUSTION OF DIESEL SOOT ON PEROVSKITE TYPE OXIDES W. Sri Rahayu, L. Monceaux, B. Taouk and P. Courtine Ddpartement de Gdnie Chimique, Universitd de Technologic de CompiOgne, B.P. 649, 60206 CompiOgne C~dex, France
ABSTRACT The combustion of soot particulates is studied on perovskite type oxide catalysts having the general formula : La0.8Sr0.2Mnl.x_yB'x~yO 3 ( B ' = Pt, Ru or Pd, x < 0.01, and F represents a deficiency in B site, y < 0.09). In order to determine the relative activity of various catalytic compositions, TGA and DTA experiments are performed on a mixture of catalyst and soot under air flow. The ignition temperature (Ti), the temperature of the maximum of combustion rate (Tin) and the soot conversion are determined. Soot is simulated either by a mixture of carbon black, fuel and lube oil or by carbon black alone. After a screening, the best catalytic results are obtained for two compositions, i.e. Lao.sSro.2Mno.999Pdo.0010 3 and Lao.sSro.2Mno.91~0.090 3. In both cases, the combustion temperature (Tin) of carbon black is found to be lowered from 650~ to about 440~ The influence of several parameters on the catalytic properties of these latter formulations is studied : catalyst preparation method, specific area, ageing, carbon black/catalyst ratio.
Compared to gasoline engines, diesel powered automobiles emissions are low in carbon monoxide but, unfortunately, they contain a higher proportion of NOx and soot particulates which have a serious impact on urban air quality and human health. Therefore, legislation makes it a duty to siglfificantly reduce these emissions . Soot particulates may be separated from the exhaust stream by filtration through a porous trap , but this device causes the increase of exhaust back pressure and therefore affects fuel economy and vehicle performance, consequently traps must be periodically regenerated by oxidizing the collected particulates. Several methods are used in order to regenerate traps. These
methods can be divided into catalytic and non-catalytic ones. In the first case, fuel additives [3, 4] and catalytic trap coatings are necessary [5-13] whereas in the non-catalytic case, the trap is electrically heated or with the help of a burner. As fuel additives may cause engine disorders and extemal heating may lead to the trap destruction, then, it seems that catalytic trap coatings could be a satisfying alternative. Different kinds of catalysts have already been studied. Research works in this field proceeded through noble metals and transition metals or alkaline oxides based catalysts. Pt and Pd, though showing some activity [5-7], cause serious problems, as they are poisoned by sulphur. It has been shown that catalytic performance and durability are improved by using noble metals doped praseodymium, neodymium and samariuln oxides [ 13]. Comparison between oxides of copper, manganese, chromium and vanadium shows that V205 is one of the most active catalysts for the combustion of diesel soot . Catalyst made up of V205 and CuO with a molar fraction 0.9 in vanadium leads to an elflmncement of the activity . Another screening study on oxides of potassium, vanadiuln, copper, manganese and cobalt shows that the systems (K-V-O) and (K-Cu-O) can reduce the ignition temperature of carbon black of about 260~ . It is to be noticed that no study about the stability of these compounds is reported. Then, the purpose of this study was to investigate the possibility of lowering of combustion temperature of diesel soot particulates by using perovskite type oxides (ABO3) as catalysts. These phases are known to be good catalysts for total oxidation of hydrocarbons and CO. They are thermally and chemically resistant. It is possible to introduce in their structure little amount of noble metals or other ions [14, 15]. The basic formula of the studied phases is Lao.2Sr0.sMnO3 doped or not by platinum, palladium and n~thenium. To evaluate the catalytic activity of catalysts, thennogravimetric (TGA) and differential thermal analysis (DTA) are used.
2. EXPERIMENTAL PROCEDURE
2.1. Preparation of perovskite catalysts The studied catalysts have the following general formula : Lao.aSr0.8Mnl_x_yB'x~)yO3_x where B' is a noble metal (Pt, Pd and Ru), F cation vacancies and 1 refers to the oxygen non stoichiometry. The perovskite phases are obtained by decomposition of equimolar mixtures of the corresponding metal nitrates (La, Sr and Mn) and noble metal (Pt, Pd mad Ru) chlorides.
565 The precursors are synthesized by a sol-gel process. This method that was developed and accorded to Baythoun et al. , consists in dissolving, gelating, drying and activation steps. The salts are dissolved, in either water (hereafter first method), or ethylene glycol as solvent (second method) with citric acid. i) In the first case, a solution of mixed salts is evaporated to dryness at 80~ and the gel obtained is dehydrated at 60-70~ for 5 hours and calcined with a heating rate of 5~ for 6 hours at 600~ under air. ii) In the second case, the mixtures are stirred while heating around 80-110~ tmtil the sol is obtained. The as-prepared gels are dried and calcined by heating rate of 5~ under air, for 6 hours up to 600~ for crystallization.
2.2. Preparation of simulated diesel soot The simulated diesel soot is synthesized by a mechanically mixing of carbon black (70%), diesel fuel (15%) and lube oil (15%). The carbon black used (Regal 660 supplied by the Cabot France Co) has a specific area of 112 m2/gr., and the particle size is around 0.24 lam in diameter. 2.3. Activity test The combustion of diesel soot is conducted using TG-DTA Setaram 92. Sample to be analyzed is prepared by mixing catalysts with simulated diesel soot or carbon black alone and grinding in an agate mortar. Then it is loaded in a platinum crucible on initial weight around 66 mg. The experimental conditions are 6~ of heating rate from 20~ to 600~ 1.5 1/h of oxidizing atmosphere flow. The computer system allows to draw both TGA and DTA curves as shown in Figure 1. TGA curve gives the conversion of diesel soot mad the ignition temperature Ti. On the other hand, DTA curve provides the maximal rate of oxidation corresponding to the top of the exothermic signal (Tm). 2.4. Characterization of samples The specific areas of samples are measured using a triple point BET method of the nitrogen adsorption at 77K on a Quantasorb Jr. surface area analyser. The X-ray diffraction (XRD) analysis of the samples is carried out with an Xray powder diffractometer using CuKa radiation and curved sensitive detector. Temperature-progratmned desorption (TPD) is carried om under vacuum (~10-6mbar). Prior to each run the sample is treated in air stream at 600~ and then cooled to room temperature. Then, the temperature of the sample is raised at a constant rate of 10~ under vacuum. Oxygen, CO and COa are detected using a quadrupole mass spectrometer (QTMD Carlo Erba).
Ti TG ('.z)
I-"" i -:.~
TEMPERATURE (C) 7oo _~_._._._~_~
/ t / u,,"
X t \\
F~AT F'LOW ( ~ 4
Figure 1. Thermal analysis (TG and DTA) of" a) simulated soot " b) mixture of LMPds catalyst and soot (15 wt %)
3. RESULTS 3.1. Catalytic Test Reactivity of simulated soot 9Results obtained with the simulated soot without catalyst are shown in Figure 1.a. Two steps of weight loss are observed. In the temperature range 150~ to 400~ a 25 % weight loss is assigned to oxidation of hydrocarbons (filel and oil). hi the range 500~176 a second loss corresponds to the oxidation of carbon black. The DTA curve shows two exothermie peaks at 350~ and 670~ associated respectively with hydrocarbons and carbon black oxidation. With dry carbon black, no modification on ignition temperature of the particulates is observed.
567 Screening : TG and DTA experiments are carried out with mixtures of soot and catalysts.~Jn Table 1, ignition temperature T i and maximum rate of combustion Tm are noted and related to temperatures obtained for soot combustion without catalyst. An important catalytic effect is observed in all cases, and the maximum effect is reached for the sample containing palladium (in stoichiometrie substitution). From comparison of the various combustion temperatures, it is possible to deduce the following order of catalytic activity : "LMs" < "LMPts" < "LMPdn" < "LMRun" < "LMRus" < "LMn" < "LMPds". These results show that catalytic activity is influenced by two parameters : the presence of the substituted noble metal and the non stoichiometry. Soot/catalyst ratio : A series of experiments with different [soot/catalyst] ratios is carried out with stoichiometric substituted Pd doped perovskite as catalyst. Results show that the combustion temperature increases when the [soot/catalyst] ratio is higher than 15 wt%. This is probably due to the lack of contact between catalyst surface and all the soot particulates beyond this point. Catalyst ageing and specific area : Simulation tests on Pd containing catalyst in the presence of 5 wt % of soot are conducted in a furnace in the same conditions as for TG-DTA, from room temperature up to 800~ during 10 minutes, then an aliquot part of the sample is taken off, mixed with soot and tested in TG-DTA, whereas the remaining of the sample is kept in the fi~mace for further runs. The results (Table 2) show that catalytic reactivity is weakly influenced by ageing test despite the decrease of surface areas, the stabilization of which is reached after the third run. (Ethylene-glycol preparation gives the highest specific area mid activity.) Isothermal t e s t : H e a t i n g in TG-DTA up to the chosen temperature is performed under pure nitrogen which is then replaced by air for combustion. The sample is then heated at higher temperatures until the total combustion of soot is achieved. These experiments evidence that the combustion rate of soot increases with temperature, and that the carbon black can be totally and rapidly eliminated at a telnperature higher than 450~ whereas the liquid hydrocarbon part of the simulated soot is completely oxidized at 200~
568 Table 1. Screening test results. Catalysts are tested with 15 wt % soot
S~nbol Tcalc 1st DTA (~ %loss 64.0 Hydrocarbon Carbon Black (CB) 18.2 Stimulated Soot 600 25.0 Lao. 8Sro. 2MnO3 +~ LMs 600 20.0 La0.8Sro.2Mn0.91Fo.o903+~, LMn 20.0 Lao. 8Sr0.2Mno.999Pt0.0o 103+~ LMPts 600 23.3 Lao.8Sr0.2Mn0.9Pt0.0o8F0.o9203+~, LMPtn 600 16.7' Lao. 8Sro.2Mno. 9Ruo. 103+X LMRus 900 20.0 18.3' Lao.8Sro.2Mno.9Ruo. 103 +X LMRus 600 25.9 18.3' La0.8Sr0.2Mno.9Ru0.008F0.09203+~, LMRun 600 23.3 16.7" Lao.8Sro.2Mno.999Pdo.oolO3+~ LMPds 900 23.3 Sample
peak Tm 305 350 300 275 295 305 340 355 305 345 280 302 325 352
Lao.8Sro.2Mno.999Pdo.oolO3+k LMPds l_zo.8Sro.2Mno.9Pdo.oo8Fo.09203+k LMPdn
2 nd DTA % loss 90.2 59.0 73.7 56.7 60.2 56.7 56.7
peak Tm 350 675 670 480 460 485 460
50.0* 8O.0 60.0 56.7
525 605 450 490
Tcalc "temperature of calcination, *" two DTA peaks. "s" indicated on symbol means a stoichiometric formula, and "n" is non stoichiometry 3.2. Characterization 3.2.1. XRD Carbon black : The XRD pattern (Figure 2) shows four wide peaks due to poorly crystallized graphite. These peaks correspond to the stacks of parallel hexagonal layer planes. Catalysts : XRD patterns indicate the existence of one phase which crystallizes in rhombohedral structure with the following parameters : a~7.7 5A and a~90.27 ~ (R3m).
Table 2 "Preparation method, surface area and ageing influence on the catalytic activity of the catalyst containing Pd "LMPds" calcined at 600~ The soot to catalyst ratio is 5 wt %. Preparation method
spec. area m2/~.
1 2 3 4
23.4 17.78 16.17 15.67
1st DTA peak* % wt loss Tm ~ 225 28 230 28 . . . 260 22
2nd DTA peak** % wt loss Tm ~ 445 70 480 62 . 455 75 460 65 490 85 500 60 600 81 . 470 62
* 9hydrocarbons oxidation ; ** 9carbon black oxidation.
Figure 2. XRD pattern of carbon black 3.2.2. TPD Several TPD analysis are carried out conceming carbon black, Pd containing catalyst and mixture of both of them. In the first sample (Figure 3.a), desorption of CO and CO2 was observed. CO2 desorbs at low temperature (200-400~ while CO desorption begins at 200~ to become important at 550~ CO and CO2 arise from surface carbon-oxygen complexes. The TPD curves of catalyst (Figure 3.b) show two important desorbed species : CO2 and 02. CO2 certainly results from the decomposition of residual carbonated components remaining during the preparation. 02 may
correspond to adsorbed and absorbed oxygen species on the perovskite structure . When the catalyst is mixed with the carbon black, C02 and CO desorptions are observed while oxygen desorption disappears (Figure 3.c). (,o +
o ,. - '~~176
./- ":'":" " :'2 " 9
~ 0 2
II C 0 2
,4 ~ 1/
\.. j'~" ~ /
Temperature (~ C)
Figure 3. TPD profiles of 02, CO and C02 from 9a) carbon black ; b) Lao.8Sro.2Mno.999Pdo.o0103+2 ; c) mixture of catalyst and carbon black
571 4. DISCUSSION
i).Reactivity of carbon black alone According to the XRD pattern (Figure 2), the carbon black Regal 660 has a surface characteristic similar to graphite structure. Recent investigations in scanning tunnelling microscopy [26, 27] have shown that this type of carbon black is actually constituted by the superposition of the overlapping of many graphite planes, rolled up in an "icospiral" structure, whose exposed edges take the form of "fish scales". This led to models of soot formation. Figure 4 also gives our example of HRTEM characteristic image, similar to that obtained by Bourrat . On the particulate surface, carbon-oxygen complexes, which are previously formed, decompose and give rise to a formation of CO2 at low temperature and CO at higher temperature [18, 19]. TPD measurements on carbon black sample confirmed these phenomena (Figure 3.a).
Figure 4. H R T E M image o f the carbon black used as a reactant
The mechanism of carbon oxidation is very difficult to investigate for several reasons : i) the high exothennicity of the reaction provides a difference of temperature between the solid surface mad the gas ; ii) the existence of several varying parameters during reaction due to mass transfer effects (pore structure and particulate size modifications, swelling, cenosphere formation). In spite of these problems, many authors agree with the fact that the mechanism of the C-O2 reaction is the same for the different kinds of carbon (graphite, char, coal, carbon black, etc.) However, the reactivity varies with the structure, the morphology, the specific areas and the surface complexes. Generally, the mechanism comprises three steps : firstly, oxygen adsorption at the edges described above with C-O complexes formation, secondly, complex decomposition, CO2, CO formation and finally new active sites liberation.
572 ii) Reactivity of carbon black on catalyst. When the reaction is performed in presence of a catalyst, the mechanism is more complicated. In the case of perovskite type oxide, the catalytic effect is due to transition ion in B site. It is known that the octahedral environment of the B ions splits the d-orbitals into two levels ; the lower (t2g) level contains orbitals that are less repulsed by negative point charges (oxygen) thin1 those in the higher (eg) level. At the surface, dz 2 orbital is the lowest eg level. In regard to CO oxidation, it is observed that the maximum activity is reached in both cases for an occupation of the eg levels of less than one electron, the t2g levels being halffilled or totally filled [14, 15, 20, 21]. In our case, the B cation is Mn3+ and Mn 4+. The electronic configuration is respectively 3d 4 (t2g 3, eg 1) and 3d 3 (t2g 3, egO). In both configurations, Mn cation is in the maximum space interval of activity. Concerning CO/O2 reaction on perovskite oxide, the "suprafacial" mechanism is assumed, as well as for the carbon black/oxygen reaction. Carbon adsorption on the catalyst surface could be made through the C-C bond or the C-O surface complexes, assuming that C is bonded to the Mn ion with donation of carbon lone pair into the empty 3dz 2 orbital to form s bond accompanied by back donation of the t2g electrons of Mn ion to anti-bonding m-orbital of C-O or C-C. Moreover, the mechmlism begins by silnultaneous adsorption of carbon and oxygen, the interaction between adsorbed species causes the CO2 formation, the desorption of which releases the catalytic active sites. Finally, the screening investigation shows : 1) that the best performance is obtained with "LMn" m~d "LMPds". The enhancement of the activity is due to the non-stoichiometry of Mn cation in the former case and to Pd substitution in the latter. The Mn4+/Mn 3+ ratio is in turn 0.57 for LMs, 1.0 for LMPds and 1.2 for L1Vhl. Therefore, the improvement of activity may be correlated, first of all, to the Mn4+/Mn 3+ ratio, particularly when noble metals are absent. In this way, Vrieland suggested that Mn4+ and Mn3+ do not act as individual surface ions, but form a part of a large group which acts as either an electron donor or acceptor . 2) As far as precious metal dopes are concerned, Pd is the most active for the carbon oxidation. Inside the perovskite matrix its oxidation state is Pd 2+ as it was reported that the platinum in the same kind of structure is in the form of dissolved tetravalent ion . The role of palladium call be due to its known catalytic property : - its affinity to the rr-allylic bond favours the formation of activated surface complexes with the carbon black ;
573 in both Pd 2+ and Pd 0 forms, it is assumed that the palladium is a soft Pearson acid. Thus, it preferably reacts with a soft Pearson base such as the >C=C < bonds present in the carbon black. Among the four noble metals studied, Pt, Pd, Rh and Ir for the catalytic oxidation of graphite single crystals , the catalytic activity is indeed in the following order: Pd >> Pt > Ir > Rh. -
CONCLUSION In conclusion our results have shown that: the perovskite type oxides (Lao.8Sr0.2Mnl_x_yB'x~yO3) are highly active in the catalytic combustion of diesel soot at temperature above 320~ (Ti) ; the catalytic activity is significantly influenced by the noble metal substitution (Pd, Pt, Ru) in B site and by the non-stoichiometry. The best performance is obtained by two samples ; La0.8Sro.2Mno.999Pdo.oolO3 and -
Lao.8Sro.zMno.91 ~ 0.0903 . the enhancement of the activity is due to : i) the catalytic properties of Pd ions in the former catalyst and ii) the increase of Mn4+/Mn 3+ ratio in the latter. -
D.J. Ball and R.B. Stack, CAPOC II, Ed. A. Crucq, Elsevier, Amsterdam (1991) p. 337. 2 F. In&a, SAE Paper No. 885151 (1988). 3 V.D. Rao, J.B. White, W.R. Wade, M.G. Aimone and H.A. Cikanek, SAE Paper No. 850014 (1985). 4 M.R. Montierth, SAE Paper No. 840072 (1984). 5 R. Domesle, E. Koberstein, H.D. Pletka and H. Voelker, US Patent No.4,515,78 (1985). 6 B.J. Cooper and J.E. Thoss, SAE Paper No. 890404 (1989). 7. D.J. Ball and R.G. Stack, SAE Paper No. 902110 (1990). 8 Y. Watabe, K. Irako, T. Miyajima and T. Yoshimoto, SAE Paper No. 830083 (1983). 9 J. Widdershoven, F. Pischinger, G. Lepperhoff, K.P. Schicki, J. Strutz and S. Stahlhut, SAE Paper No. 860013 (1986). 10 A.F. Ahlstr6m and C.U.I. Obenbrand, Appl. Catal., 60 (1990) 143. 11 A.F. Ahlstr6m and C.U.I. Obenbrand, Appl. Catal., 60 (1990) 157.
12 P. Ciambelli, P. Parrella and S. Vaccaro, CAPOC II, Ed. A. Crucq, Elsevier, Amsterdam (1991) p. 323. 13 H. Makoto and S. Koichi, European Patent No 0397411 A2 (1990). 14 R.J.H. Voorhoeve, Adv. Materials in Catalysis, Eds. J.J. Burton and R.L.Garten, Academic Press, London (1977). 15 L.G. Tejuca, J.L.G. Fierro and J.M.D. Tascon, Adv. in Catalysis, 36 (1989) 237. 16 M.S.G. Baythoun and F.R. Sale, J. Mater. Sci., 17 (1982) 2757. 17 Y. Teraoka, M. Yoshimatsu, N. Yamazoe and T. Seiyama, Chem. Lett., (1984) 893. 18 D.L. Trimm, Catalysis-Special Report of Royal Society of Chemistry, No. 4, London (1980). 19 N.M. Laurendeau, Prog. Energy Combust., 4 (1978) 221. 20 R.J.H. Voorhoeve, J.P. Remeika and L.E. Trimble, Aim. N.Y. Acad. Sci., 272 (1976) 3. 21 J.M.D. Tascon and L.G. Tejuca, Reac. Kinet.Catal.Lett., 15 (1980) 185. 22 E.G. Vrieland, J. Catal., 32 (1974) 415. 23 D.W. Jolmson Jr., P.K. Gallagher, G.K. Werthem and E.M. Vogel, J. Catal., 8 (1977) 87. 24 R.T.K. Baker and R.D. Sherwood, J. Catal., 61 (1980) 378. 25 H.W. Kroto, A.W. Allafmad A.P. Bahn, Chem. Rev., 91 (1991) 1213. 26 J. Lahaye and G. Prado, Soot in combustion and its toxiproperties, NATO Series VI, Plenum (1983). 27 J.B. Dolmet and E. Custodero, Bull. Soc. Chim. Fr., 131 (1994) 115-117. 28 X. Bourat, Extended Abstracts 21th Biemlal Conf. Carbon, ACS, 229 (1993).
ACKNOWLEDGEMENTS The authors are indebted to Peugeot SA and STTS for financial support, and particularly thank Dr. Belot, Dr. Le Borga~e and Mr. Foucaud for their contribution and helpfid discussions.