J. Chem. Soc., Faraday Trans. I, 1987, 83 (lo), 3139-3148 Kinetics and Mechanism of Oxidative Dehydrogenation of Ethane and Small Alkanes with Nitrous Oxide over Cobalt-doped Magnesium Oxide Ken-ichi Aika,* Makoto Isobe, Kazuhiro Kido, Takeshi Moriyama and Takaharu Onishi Research Laboratory of Resources Utilization, Tokyo Institute of Technology, 4259 Nagatsuta, Midori-ku, Yokohama 227, Japan The kinetics of the decomposition of N,O and the partial oxidation of C,H, and other alkanes with N,O have been studied-over Co2+-doped MgO catalysts mainly at 473 K and below atmospheric pressure. The initial rate of N,O decomposition is proportional to the estimated amount of surface Co2+ ion, which suggests the Co2+ ion to be the active centre. Since the amounts of adsorbed oxygen are much greater than the estimated amounts of surface Co2+ ion, the adsorbed oxygen is inferred to migrate from the Co2+ ion centre to the MgO surface. No 0, was produced during the oxidation of C,H, and the initial rate of the oxidation reaction was almost identical to those of N,O decomposition.Thus, it was concluded that the oxidation of C,H, was controlled by the decomposition of N,O. Ethoxide is inferred to be a common intermediate, giving C,H, and surface oxidized species. Pure alkanes from C, to C, react at much the same rate with N,O, because the decomposition of N,O is the rate-determining step. If, however, mixed alkanes are used, differences in inherent reactivities appear. The reactivity sequence (n-C,H,, > C,H, z C,H, > C,H, > CH,) obtained is almost identical to that with oxygen atoms in the gas phase.There are several on the partial oxidation of C,H,. Some of the recent advances in this field are a finding of strong reactivity of surface 0- toward alkanes including C2H6,2 and the use of N,O as a source of 0- in order to obtain C,H,.3,4 However, 0- observable by e.s.r. spectroscopy is not always necessary for oxidative dehydrogenations of C,H, with N,O over CO~+-M~O.~,, In these reports the relation between C,H, selectivities and the nature of the catalyst, such as the bulk structure’ or surface-active have been discussed. However, all of these studies, including ours,5’ are insufficient to establish the reaction mechanism. Co2+-MgO has advantages for the study of the kinetics of the title reaction because it is an active and more selective catalyst than CoO/SiO, or C O O / A ~ , ~ , .~ ~ ~ While Co2+ ions make a solid solution with MgO,’** Co oxides are assumed to form an aggregated phase on SiO, and Al,03. Experimental MgO (Soekawa Rikagaku, 99.75 % purity) was impregnated with Co(NO,), - 6H,O in aqueous solution and a partly dried sample was extruded through a syringe. Pure MgO was also prepared by the treatment in water and extrusion. The dried pelleted catalyst thus made was evacuated at 773 K or at 1173 K for 3 h. The weight of the sample used decreased from 2.0 to 1.3 g during evacuation at 1173 K. B.E.T. surface areas measured after the evacuation were 64 m2 8-l for MgO, 162 m2 g-l for 0.02 YO Co2+-MgO, 220 m2 g-‘for 0.2 O/O Co2+-MgO, and 150 m2 g-l (for the weight measured before the evacuation) for 2 O/O Co2+-MgO, respectively.The surface area of MgO increases when Co2+ ions are 31393 140 Ox id at ive Dehydrogenation of Ethane added with a maximum above 0.2 YO of added cations. Co2+ contents are represented by molar percentages against MgO. The apparatus used is a closed circulation system, which is almost the same as that reported elsewhere.2 Initial amounts of the reactants (N,O and C2H') were ca. 6.67 kPa (1 kPa = 7.50 Torr) or 387 pmol. Most of the reactions were carried out at 473 K. 1.r. spectra were recorded using a JASCO A-3 spectrometer at room temperature and the reactions were carried out at 473 K in the i.r. cell. 1.r. spectra of pure chemicals adsorbed on 2% Co2+-MgO were recorded as references.CO, adsorbed at 473 K gave peaks at 1666(s), 1524(m) and 1314(s) cm-' (carbonates). HC0,H adsorbed at 473 K then evacuated at 723 K gave peaks at 1600(s) and 1382(m) cm-' (formates). CH,CO,H adsorbed at 473 K then evacuated at 723 K gave peaks at 1585(s), 1432(s), 1365(m), 1310(s), 1163(w), 947(w) and 832(w) cm-' (acetates). C,H,OH adsorbed at 298 K gave peaks at 1447(m), 1380(s), 1139(s), 11 lO(w), 1068(s) and 890(m) cm-l (ethoxides). Results Kinetic Analysis of N20 Decomposition A number of runs of N,O decomposition were carried out at 473 K over a 2 g sample of 0.2% Co2+-MgO catalyst. An example of the time course of reaction was shown in an earlier paper.' Since the amount of 0, produced is less than half the N, produced, some of the oxygen must be adsorbed, the amount being defined here as O(a) = N,(g) - 20,(g).The decomposition rate, R (N, pmol h-'), is taken from the slope of the curve for N, increase vs. time. Rate constants (RP;io) are plotted as a function of the amount of O(a) in fig. 1 for several runs over 0.2% Co2'-MgO samples at 473 K. The data can almost be reproduced as a linear relationship, where k is the initial rate constant (N, pmol h-l kPa-l) and [O,] is the amount of adsorbed oxygen when the reaction completely stops. The same relationships are observed for reactions at different temperatures (298473 K) or over other samples (0.02% and 2% Co2+). These results suggest that the reaction proceeds with the following mechanism which has been proposed by Tanaka and Ozakig~'O for the Cr,O, and Mn,O, catalyst: N,O + * + N,O(a) N,O(a) -+ N, + O(a) 20(a) -+ 0, + 2*.(2) (3) (4) Here, steps (2) and (4) are fast and step (3) is slow. Kinetic Analysis of Reaction between N20 and C2H, Reactions of C,H, with N,O were carried out over various Co2+-MgO catalysts at various temperatures. An example of the time course has been shown for the reaction over 2 g of 0.2% Co2+-MgO at 473 K in an earlier paper.6 0, is not released, but C,H, is produced, which suggests that the N,O decomposition is rate-determining. Rates of N, formation, expressed as RP;;,, are plotted as a function of amount of consumed 0 (= N,) in fig. 1. The initial rates of N,O decomposition with and without C2H6 are almost identical to each other and RP;;, decreases linearly with the amount of consumed 0, although the rate of decrease does not change compared with the reaction without C,H,.The same relationships are observed for other samples or for the runs at different temperatures (298-473 K).K. Aika et al. 3141 0 100 200 300 Fig. 1. Plots of rate constants (RP;;,) of N,O decomposition at 473 K us. adsorbed or consumed amount of oxygen atom over 2 g of 0.2% Co2+-MgO. Open circles denote pure N,O decomposition, whereas closed circles denote N,O decomposition under C,H6 oxidation. Duplicate runs were carried out over different batches of the sample for checking the reproducibility. The adsorbed or consumed amount of 0 is calculated from N,(g) - 202(g). adsorbed or consumed O/pmol Fig. 2. Relation between the amount of C2H6 reacted and amount of consumed 0 over 2 g of catalyst at 473 K.Runs over 0.2% Co2+-MgO (closed circles), runs over 2% Co2+-MgO (open circles). Each symbol corresponds to a separate run. (- - -) Slope 1 / 1 , (-) slope 1/4. The reaction is analysed in other ways. The amount of C2H, reacted (- AC2H,) is plotted as a function of the amount of consumed oxygen ( - A 0 = AN2) in fig. 2. Although the reaction rates are different over samples with different Co2+ content, plots of (-L\C,H,) us. (-do) coincide as is seen in fig. 2. The reaction stoichiometry of C2H6 against 0 was obtained from the slope (-dC2H6)/( -do), which changes from unity to ca. 0.25. The amount of C,H, produced is plotted as a function of the amount of C2H, reacted in fig. 3. The data also coincide irrespective of the Co2+ contents.The selectivity for3142 40 - 0 \ 5 5 20 x a 0 Oxidative Dehydrogenation of Ethane / /;* 50 100 C2H, reacted/pmol r" , , , , , ' , , , Fig. 3. Relation between the amount of C,H4 produced and the amount of C,H6 reacted at 473 K. Runs over 0.2 YO Co2+-MgO (closed circles), runs over 2 YO Co2+--MgO (open circles). Each symbol corresponds to a separate run. (---) Slope 0.6, (-) slope 0.25. Table 1. Competitive reactions of alkanes with N,O over Co2+-MgO at 473 K" decreasing rate of reactants in each run relative reactivity alkane initial rate constant RP,t,/pmol h-I kPa-' ratio 0.15 - - - - 1 .o 6.0 0.6 6.9 - 10 1.2 1 - 23b 7 - 9.7 27' 8 16 - CH4 C3*8 N,O" 13 8.1 9.0 13 12 45b - - - - C2H6 C2H4 - - - - - - - - 9.3 - n-C4H10 a N,O (6.7 kPa or 390 pmol) and mixed alkanes (each 6.7 kPa), 2 g of 0.2 YO Co2+-MgO evacuated at 1173 K.This sample was somewhat reactive compared with the other. Rate measured by N, production rate. Table 2. Relative reactivity of alkanes substrate alkane O(a)/MgO O-/MgO" O-/gasu *OHb -CH3b 0" 0.15 0.50 0.14 0.025 0.03 0.01 3 1 1 1 1 1 (1.67) (1.10) 8 1.27 1.33 7.7 (7) 12 16 0.67 1.7 - 4.5 19 473 298 298 295 455 307 - a Ref. (1 1). Ref. (12). Ref. (13).K. Aika et al. 3143 110% ~~ ~ 1600 1400 1200 wavenumberlcm-' Fig. 4. 1.r. spectra of the surface species formed during the reaction between N,O and C,H, at 473 K for 3 h over various MgO catalysts : (a) MgO, (6) 0.02 % Co2+-MgO, (c) 0.2 O/O Co2+-MgO, ( d ) 2 % Co2+-Mg0. ethene (dC2H4)/( -dC,H,) decreases from 0.6 to 0.25 with the progress of the reaction.For most cases the ratio of the pressure of C,H, to that of C,H, increases up to ca. 0.1 in ca. 3 h, when the reaction almost ceases. Reactions of Other Alkanes with N,O Although inherent reactivities of hydrocarbons with different carbon numbers should be different in an oxidation reaction, all pure alkanes from C, to C, react at much the same rate with N,O because the reactions of alkanes with N,O are controlled by N,O decomposition. No 0, is evolved in the gas phase during the oxidation reaction of any alkane by N20, which also means N,O decomposition is rate-determining. Some of the initial rates are shown in table 1 . If, however, mixed hydrocarbons are used, differences in inherent reactivities appear and the initial decrease of each reactant can be used to measure the relative rates of reaction of the different hydrocarbons.The two alkanes must compete with each other in their inherent reactivity against adsorbed oxygen produced via N20. The initial rates of mixed hydrocarbons are shown in table 1 and relative reactivities determined are listed in the last column. The reactivity increases with increasing carbon number of the alkane. Physical adsorption of C2H,, C2H4 and C,H, over MgO is negligibly small above 423 K2 Thus, the relative reactivity obtained here does not contain any adsorption terms. Observation of Surface Species during C,H,-N,O Reaction by I.R. Spectroscopy Since C2H, selectivity is far below 100 % without yielding other gaseous products, carbonaceous products should be left on the surface.Thus, i.r. spectra of various MgO catalysts were recorded after reaction for 3 h at 473 K. The results are shown in fig. 4. With an increase of Co2+ content several peaks become observable. These are identified as carbonate anions, both bidentate (1665 and 1314 ern-,) and unidentate (151 1 cm-l), formate (1600 and 1380 crn-l) and acetates (1585 and 1425 cm-'). Oxygen-containing species are not observed in the gas phase for reaction times > 3 h at 473 K.3144 Oxidative Dehydrogenation of Ethane co2+ content (%) 0.02 0.2 2 0.5 1 5 10 50 100 Fig. 5. Relation between the initial rate constant of N,O decomposition and the number of surface Co2+ ions. The initial rate, RP&, was measured over 2 g of catalyst at 473 K. 0, Rate of pure N,O decomposition; 0, rate of N,O decomposition under C,H, oxidation.Duplicate runs were carried out to check the reproducibility. surface Coz+ ions/pmol Discussion Active Centre for N20 Decomposition Since the kinetic data are represented by eqn (l), the corresponding elemental reaction steps [eqn (2)-(4)] are proposed, where eqn (3) is the rate-determining step. The initial rate constants (k = RP;;, at t = 0) obtained in the preceding results are plotted in fig. 5 as a function of the number of surface Co2+ ions (estimated by assuming no surface segregation of Co2+ and by using the B.E.T. surface area). The initial rate constant k is almost proportional to the number of Co2+ ions, which suggests that step (3) is catalysed by the Co2+ ion (active centre). Even if some surface segregation of Co2+ is assumed, the tendency would not be changed by much.An Arrhenius plot of the initial rate constant for 0.2% Co2+-MgO is shown in fig. 6 and an apparent activation energy of 37 kJ mol-1 was obtained. The observed activation energy is thought to be the value of the activation energy of step (3) less the heat of adsorption in step (2), which may be the reason why the value is so low. The value is lower than the reported one, 92f4 kJ mol-l, for 0.1 % Co2+-MgO calcined at 1273 or 1473 K.? Spillover of Adsorbed Oxygen from Co2+ to MgO Surface The amount of oxygen adsorbed during N,O decomposition (dotted line with arrow) and that of oxygen consumed during C2H, oxidation (solid line with arrow) are plotted in fig. 7 for three samples with different Co2+ contents. Maximum values of O(a) obtained from fig.1, which correspond to [O,] in eqn (l), are also plotted as open and closed circles. When all these values are compared with the estimated number of theK. Aika et al. 3145 1000 100 4 I cd a &i - 10 k: 5 O 1 hZ n P( - \ 5 -on 0.1 '. \ 1 I I I I 1.5 2 .o 2.5 3.0 103 KIT Fig. 6. Arrhenius plot of initial rate constant of N,O decomposition (RPNio) over 2 g of 0.2% Co2+-Mg0. 0, Rate of pure N,O decomposition; 0, rate of N,O decomposition under C,H, oxidation. The ordinate is scaled logarithmically. E, = 37 kJ mo1-I (9 kcal mol-I). Co2+ content (%) 0.02 0.2 2 500 1 / I / I I v I 1 I I / ; I / I I ; 101 IT I 1 I I I I / I Q 5 1 I / I I I / I / I I / I I - I / I I I I I . , . I I . , .. I 0.5 1 5 10 50 100 Fig. 7. Adsorbed or consumed oxygen atoms during the reaction at 473 K as a function of the number of surface Co2+ ions estimated. 0, Amount of saturated adsorbed oxygen (0,) in N,O decomposition (estimated) ; 0, amount of saturated consumed oxygen in N,O-C,H, reaction (estimated). Dotted and filled arrows show the actual amount of O(a) during N,O decomposition and N,O-C,H, reaction, respectively. surface Co2+ ions/pmoi3146 Oxidative Dehydrogenation of Ethane surface Co2+ ions, the former exceed the latter. This fact suggests that a part of the adsorbed oxygen spills over from the Co2+ centres to the MgO surface79*.f- if the estimation of surface Co2+ is correct: ( 5 ) In the case of C,H6 oxidation, most of the surface oxygen must be turned .into oxygenated species on MgO, as will be discussed below [step (6)].This would decrease the concentration of O(a)--.Co2+ further. This is why the rate of N20 decomposition during C2H6 oxidation is higher than that of pure N,O decomposition, as is seen in O(a) - . - Co2+ + O( a) - - - MgO. fig. 1. Initial Step of the Reaction between N20 and C2H, Since the stoichiometric ratio of C,H6 reacted and 0 consumed is unity in the initial stage (fig. 2 and table l), step (6) is reasonably assumed: O(a) + C2H6 -+ X(a). (6) This step is considered to be faster than step (4) because no gaseous 0, is obtained during the oxidation reaction. Thus the rates of the reaction are in the order: step (6) > step (4) > step (3). During the decomposition of N,O over 0.2% Co2+-MgO, C,H6 was added at 473 K.Small amounts of gaseous 0, formed by that time were consumed quite rapidly, even when the pressure was extremely low (1/200) compared to that of N,O. This means that the reverse reaction of step (4) (or any reaction of 0, with carbonaceous species) is also faster than step (3). Relative Reactivities of Small Alkanes with Adsorbed Oxygen The relative reactivities of alkanes with adsorbed oxygen observed here are compared with those for other substrates in table 2. The reactivity sequences, n-C,H,, > C,H, = C2H4 > C2H6 > CH, in this work is similar to those with 0- (in the gas phase or on MgO);ll however, it is much the same as other radicals such as .OH, .CH, or atomic 0.l2* l3 The reported reactivities with radicals are closely related to the dissociation energy of the C-H bond except for ethylene; H-CH, (104+ 1 kcal mol-l), H-C2H5 (98 & l), H-C,H, (2 108 f 2), H-i-C,H, (95 & l).14 These radicals or atoms react with alkanes to abstract hydrogen.12* 1 3 7 l5 The oxidation of alkanes over Co2+-MgO is also considered to start with hydrogen abstraction by O(a) on MgO: O(a) + RH + R + OH(a) O(1attice) + - R + OR(a).We cannot deduce the electronic state of O(a) from this work; however, the reactivity of O(a) somewhat resembles atomic 0 rather than the 0- radical when C,, C, and C, alkanes are compared in table 2. Further Step of the Reaction including Surface Products using N,O and C,H, C,H,OH(a) or C,H50(a) are assumed to partake in the ethane reaction for step (6b), although no i.r.information of this species has been obtained, probably because of the low concentration : C2H, + O(a) + C,H,OH(a) (6 4 C,H,OH(a) + C,H, + H,O(a) C,H,OH(a) + O(a) -+ CH,CHO(a) + H20(a). (7) (8) t Eqn (1) should be valid only for a case where amounts of the two kinds of adsorbed oxygen in eqn ( 5 ) are linearly correlated in the equilibration. Other cases might deform the straight line in fig. 1 ; however, we neglect such discussions here.K. Aika et al. 3 147 Surface ethoxide may decompose to give C2H4 [step (7)] or may be gradually oxidized to acetate [step @)I, formate and finally carbonates at 473 K as is observed by i.r. in fig. 4. Methoxide on MgO, which probably absorbs oxygen, has been reported to be oxidized to formate above 438 K,16 whereas it is not oxidized up to 673 K over clean MgO evacuated at 1003 K.17 Thus, oxygen adsorbed on MgO is thought to react with ethoxide at 473 K [step (S)].Along with the oxidation reaction of ethoxide to acetate, formate and carbonates, several times more oxygen atoms than C2H6 molecules must be consumed. This is why (-dC2H6)/(-do) decreases from 1 to 0.25 in fig. 2. It is considered that O(a) in step (3) is adsorbed on Co2+ ion, whereas the O(a) in step (8) and further steps are those adsorbed on MgO. The state of O(a) in step (6a) cannot be decided here, but it is suggested that it is located on Mg0.5 From the data at the initial stage in fig. 3, ratio of the rate of step (7) to the rate of step (8) is estimated to be 3/2. At the final stage of the reaction this ratio decreases to 1/3 (fig.3). At this stage PCaH4/PCaH is almost constant, ca. 0.1. This is probably because C,H4 is oxidized through the back-reaction of step (7) and the forward reaction of step (8). The steady-state condition for the ethoxide intermediate gives the relative reactivity of C2H, : C2H6 to be ca. 9. This means that C2H4 is much more reactive than C2H6. The relative reactivity of C2H4 : C2H6 is also ca. 7 from the direct method in table 1. Partial oxidation of alkanes must start by hydrogen abstraction; however, oxidation of alkenes may be different. The relative reactivity of C2H4 is almost the same as that of C,H,, whereas the dissociation energy of C-H in C2H4 (2 108 kcal mol-l) is much higher than that in C,H, (95 kcal mol-l). Oxygen addition is inferred to be the initial step for C2H4, as in the case for the 0 atom in the gas phase.', Conclusions Co2+-MgO is considered to have dual functions ; Co2+ catalyses N20 decomposition and MgO catalyses an oxidation of C2H6.The former reaction is the rate-determining step. Oxygen spills over from Co2+ to the MgO surface. The reaction mechanism is proposed for C2H4 formation. Surface intermediates (ethoxides) are either decomposed to give C2H4 with a selectivity of 0.6 or oxidized further. C2H4 is several times more reactive than C2H6, which depresses the ethylene selectivity along with the time of reaction. The nature of active oxygen on MgO which reacts with alkane seems to be different from 0-, but similar to the 0 atom in the gas phase. We thank Mr T. Nishiyama for his contribution. References 1 E. M. Thorsteinson, T. P. Wilson, F. G. Young and P. H. Kasai, J. Catal., 1978, 52, 116. 2 K. Aika and J. H. Lunsford, J . Phys. Chem., 1977, 81, 1393. 3 M. B. Ward, M. J. Lin and J. H. Lunsford, J. Catal., 1977, 50, 306.; T-J. Yang and J. H. Lunsford, J. 4 M. Iwamoto, T. Taga and S . Kagawa, Chem. Lett., 1982, 1469. 5 K. Aika, M. Tajima and T. Onishi, Chem. Lett., 1983, 1783. 6 K. Aika, M. Tajima, M. Isobe and T. Onishi, Proc. 8th Jnt. Congr. Catal., Berlin, 1984, (Dechema, Verlag Chemie, Weinheim, 1984), vol. 3, p. 335. 7 A. Cimino and F. Pepe, J . Catal., 1972, 25, 362. 8 V. Indovina, D. Cordischi, M. Occhiuzzi and A. Arieti, J . Chem. SOC., Faraday Trans. I , 1979, 75, 9 K. Tanaka and A. Ozaki, J . Catal., 1967, 8, 307. Catal., 1980, 63, 505. 2177. 10 K. Tanaka and A. Ozaki, Bull. Chem. SOC. Jpn, 1967, 40,420. 11 K. Aika and J. H. Lunsford, J . Phys. Chem., 1978, 82, 1794. 12 R. P. Overend and R. J. Cvetanovic, Can. J . Chem., 1975, 53, 3374. 13 J. T. Herron and R. E. Huie, J . Phys. Chem., 1969, 73, 3327. 14 D. M. Golden and S . W. Benson, Chem. Rev., 1969, 69, 125.3148 Oxidative Dehydrogenation of Ethane 15 N. N. Semenov, On Some Problems of Chemical Kinetics and Reaction Ability (Acad. Nauk. SSSR, 16 R. D. Kagel and R. G. Greenler, J. Chem. Phys., 1968, 49, lb38. 17 J. Kondoh, Y. Sakata, K. Muruya, K. Tamaru and T. Onishi, Appl. Surf Sci., 1987, in press. 18 S. Sat0 and R. J. Cvetanovic, Can. J. Chem., 1959, 37, 953. Moscow 1958). Paper 611933; Received 30th September, 1986