- Email: [email protected]
Structure sensitivity of the hydrocarbon combustion reaction over alumina-supported platinum catalysts
Studies in Surface Science and Catalysis 130 A. Corma, F.V. Melo, S. Mendioroz and J.L.G. Fierro (Editors) 9 2000 Elsevier Science B.V. All rights reserved.
575
Structure sensitivity of the hydrocarbon combustion reaction over aluminasupported platinum catalysts T.F. Garetto and C.R. Apesteguia* INCAPE (UNL-CONICET), Santiago del Estero 2654, (3000) Santa Fe, Argentina The reaction kinetics, structure sensitivity, and in-situ activation of cyclopentane and methane combustion on Pt/A1203 catalysts of different metallic dispersion were studied. The reaction orders in oxygen were 1 (cyclopentane) and zero (CH4). Methane oxidation turnover rates did not change significantly by changing the metallic dispersion but the cyclopentane combustion activity increased dramatically with increasing Pt crystallite size. On both reactions, the activation energies did not change by changing the Pt dispersion. Results are interpreted in basis of two different reaction mechanisms over the metallic Pt active sites. I. INTRODUCTION Platinum-based catalysts are highly active for oxidative removal of small amounts of hydrocarbon from gaseous or liquid streams. The effect of varying the platinum particle size on the catalytic combustion of different hydrocarbons has been extensively studied [1-4], but the results obtained are conflicting, probably because correlation between catalytic activity and metallic dispersion depends on the type of hydrocarbon to be abated. Several papers on light alkanes combustion, namely methane [ 1], propane [5], and butane [4] have reported that alkane oxidation turnover rates increase with increasing platinum particle size. In contrast, in a recent study on the C2H4 combustion over platinum-supported catalysts Pliangos et al. [6] proposed that turnover frequency changes, which cannot be explained by structure sensitivity considerations, are caused by interactions between the metal crystallites and the carrier. Similarly, Papaefihimiou et al. [7] reported that the benzene oxidation turnover rate on Pt/A1203 strongly increases with increasing Pt particle size but does not change by changing the Pt dispersion on Pt/SiO2 and Pt/TiO2 catalysts. Palladium and platinum catalysts are often activated on stream, ab-initio of the hydrocarbon combustion reaction [5,8,9]. This phenomenon has been widely studied on palladium-based catalysts; in the case of methane oxidation, several authors have proposed that the initial activation period is caused by reoxidation from Pd metal or oxygen-deficient PdOl.x to more active steady state PdO species [8]. In contrast, very few papers have been published using platinum-based catalysts [5] and the causes of induction periods on platinum remain unclear. In a recent paper [9], we studied the structure and reactivity of Pt/A1203 catalysts for benzene oxidation at low temperatures. In this work we report results on the oxidation of cyclopentane and methane over a set of Pt/AI203 catalysts of different metallic dispersion and chlorine concentration. Our goal was to obtain further information on the catalyst activation phenomenon and on the sensitivity of hydrocarbon oxidation turnover rates to Pt crystallite size. "Corresponding author. Email: [email protected],fax: 54-342-4555279
576 2. EXPERIMENTAL A Pt(0.3%)/AI203 catalyst (catalyst A) was prepared by impregnation at 303 K of a highpurity 7-A1203 powder (Cyanamid Ketjen CK303) with an aqueous solution of H2PtCI6.6H20 and HCI. After impregnation, samples were dried 12 h at 393 K and heated in air stream to 773 K. Then the chlorine content was regulated using a gaseous mixture of HC1, water and air. Finally, the sample was purged with N2 and reduced in flowing H2 for 4 h at 773 K. A set of three catalysts with different Do (Pt dispersion) was prepared by treating catalyst A in a 2% O2~2 atmosphere at 848, 873 and 903 K for 2 h (catalysts B, C, and D, respectively). The Pt dispersion was measured by 1-12 chemisorption by using the double isotherm method and a stoichiometric atomic ratio H/Pt~=I, where Pts implies a Pt atom on surface. The characteristics of catalysts A, B, C, and D are shown in Table 1. Tabla 1 Characteristics of the catalysts used in this work Catalyst Pt loading CI concentration (wt.-%) (wt.-%) A 0.30 0.95 B 0.30 0.61 C 0.30 0.58 D 0.30 0.60
Pt dispersion Do (%) 65 38 24 15
Hydrocarbon oxidation reactions were carried out at 1 atm in a fixed-bed tubular reactor. Cyclopentane (0.65%) or methane (2%) were fed in a 10% O2/N2 mixture. On-line chromatographic analysis was performed using a gas chromatograph equipped with a flame ionization detector and Bentone 34 or Porapak Q packed columns. Before gas chromatographic analysis, the reaction products were separated and carbon dioxide converted to methane by means of a methanation catalyst (Ni/Kieselghur) operating at 673 K. Two experimental procedures were used for catalyst testing. The complete oxidation of hydrocarbons was studied by obtaining curves of hydrocarbon conversion (X) as a function of temperature (light-off curves). The temperature was raised by steps of about 23 K, from 25 to 673 K (cyclopentane) or 913 K (methane). More fundamental differential reactor experiments (less than 10% conversion) were performed at constant temperature. 3. RESULTS AND DISCUSSION 3.1. Catalytic tests: Light-off curves Fig. 1 shows the X vs T curves obtained on catalyst A in two consecutive catalytic tests. The cyclopentane combustion started at about 473 K in the first run and the conversion increased then dramatically at ca. 553 K reaching a value of X ~ 100% between 663 and 673 K. The reaction was maintained at 673 K for 2 h and then the catalyst was purged and cooled down in nitrogen to 373 K. Subsequently, a second catalytic test was carried out. As shown in Fig. 1, the X vs T curve corresponding to the second run was clearly shifted to lower temperatures as compared to that obtained in the first run. Such a displacement of the light-off curves typically illustrates the catalyst activation phenomenon in hydrocarbon combustion reactions. To compare catalyst activities, we measured from light-off curves the value of the
577 temperature at X = 50 %, T i,j, 5~ where i identifies the catalyst and j indicates first (1) or second (2) runs. The difference ATSO = TSO 50 is a measure of the i,l - Ti,2
100 80
.
Methane
60
activation phenomenon on catalyst i. Table 2 shows that the AT5~ value for cyclopentane combustion was about 80 K. The CH4 combustion on Pt occurs at temperatures (T5~ = 823 K) significantly
-~ 4o
r
20 9
n 9
o =
!
I
1st. run 2nd. run
higher as compared with cyclopentane combustion (TS~ = 593 K). The consecutive
=
light-off curves for CH4 combustion were similar (Fig. 1), thereby suggesting that catalyst A is not activated ab-initio of this Fig. 1" Light-off curves on catalyst A reaction. On the other hand, we measured the Pt dispersion on catalyst A atter the second runs (D2, Table 2). By comparing the Do and D2 values in Table 2 it is inferred that the metal was severely sintered in both reactions after two consecutive catalytic tests. Similar experiments were carried out on catalyst C. For cyclopentane combustion, the light-off temperature in the first run (T c5~ = 518 K) was clearly lower than that obtained on 400
600
Temperature
8O0
1000
(K)
catalyst A; the initial catalyst activation was negligible (ATc5~ 5 K) as well as the metal sintering at~er two consecutive runs (Table 2). In contrast, for methane combustion the T~5~ and AT5~ values obtained on catalyst C were similar to those found on catalyst A. Tabla 2 Catalytic activity and Pt dispersion in two consecutive catalytic runs Catalyst Reactant Temperatures at X = 50% (K) T~~ T25~ AT 5~ A A C C
Cyclopentane Methane Cyclopentane Methane
593 823 518 823
513 818 513 823
80 5 5 0
Pt dispersion (%) Do D2 65 65 24 24
15 18 20 17
3.2. Low-conversion catalytic tests In order to establish the effect of the Pt crystallite size on catalyst activity, additional kinetically-controlled catalytic tests were performed In all the cases, the initial conversion was lower than 10%. The oxidation reactions were performed over catalysts A, B, C, and D at 443 K (cyclopentane) and 713 K (methane). For cyclopentane combustion, the activity of welldispersed catalyst A slowly increased with time on stream, but the turnover frequency (TOF, sl) on sintered catalyst D was constant along the 20 hour run. Over all the catalysts, methane oxidation rates did not change with time on stream. In Fig. 2 we have plotted the initial turnover frequencies as a function of the metallic dispersion. Cyclopentane combustion turnover rates
578
~
n
e
increased drastically with the Pt particle size; the TOF value on catalyst D (Do = 15%) was about 40 times higher than that measured on catalyst A (Do = 65%). This result shows that cyclopentane combustion on Pt/Al203 catalysts is a structure-sensitive reaction preferentially promoted on larger Pt crystallites. On the contrary, for methane combustion the effect of Pt dispersion on the catalytic activity was rather weak and the turnover rate does not change significantly with Do (Fig. 2). All these results suggest that the existence of initial activation periods is related to the sensitivity of
.L_._
0.1 v
I.i_
o
0.01
I
6O
o0 (%) Fig. 2: Turnover frequency (TOF) vs Do the combustion turnover rate to the Pt crystallite size. 3.3. Kinetic studies
Kinetic data were interpreted by considering a power-law rate equation: #
r o = k(P~ )~ (P~2~, where r0 (mot HC/hg Pt)is the initial reaction rate. In Figs. 3 and 4 the ro values obtained on catalyst A for the oxidation of cyclopentane and methane were represented in logarithmic plots as a function of P~c and p00 2 respectively. Reaction orders cx and 13 were determined graphically from Figs. 3 and 4. The reaction orders for cyclopentane Pg= 0.126 atm Methane
~ o
o
,_o r -1
o
Cyclopentane
-2
Cyclopentane~._~
443 K
-3 -4
o
-3
_~
'
713 K
Methane
e-
-2
PHc = 0.02 atm,
_15
'
_'4
'
_~3
In Pnc Fig. 3" Reaction orders in the hydrocarbon
I
I
-2.4
PHC= 6.5 10 .3 atm, 443 K i
In P0 2
I
-2.0
Fig. 4: Reaction orders in oxygen
combustion were ot _-- 0 and J3 -= 1 while values of cz _=_1 and J3 --- 0 were determined for CH4 combustion. Similar values for ct and 13 were measured on catalyst C. On the other hand, we plotted the In TOF values as a function of 1/T for calculating the apparent activation energy (Ea) and the preexponential factor A of both reactions on catalysts A, C and D via an Arrhenius-type function. The apparent activation energies were 11 + 1 kcal/mol (cyclopentane) and 17 + 1 kcal/mol (CH4), irrespective of the mean Pt crystallite size of the sample. For
579 cyclopentane combustion, we measured a A D/A A ratio of about 60. These results suggested that increasing the Pt particle size increases the density of active sites available for the ratedetermining step but does not modify the cyclopentane oxidation mechanism. The kinetic results show that CH4 and cyclopentane are oxidized by different mechanisms. The reaction orders obtained for cyclopentane combustion are well interpreted by considering that the reaction occurs via a Mars-Van Krevelen type mechanism [ 10], being the dissociative adsorption of oxygen on Pt the rate determining step. For cyclopentane oxidation on Pt, this mechanism may be represented by the following elementary steps: 02(g) + 2 L CP(g) + L
CP.L + O.L (CP...O).L
k~ > k2 >
20.L
k3 >
(CP...O).L
O.L
>
CP.L
CO2(g)+ H20(g)
where L represents the vacant active sites. The expression of initial rate r0 results: pO o klk3 o~PcP r0 = k,po ' + vik3PO p w h e r e v i is the stoichiometric coefficient of oxygen in the overall reaction.
(1) If k I < < k 3
Eq. (1) reduces to: klP~ 2 ro = - Vi
(2)
and the orders with respect to cyclopentane and oxygen predicted by Eq. (2) are 0 and 1, respectively, which are the approximate orders determined from our experiments. According to Eq. (2), any increase in rate constant k 1 accelerates the cyclopentane oxidation rate. The observed turnover rate increase with increasing Pt particle size would reflect therefore an increase in the density of reactive Pt-O species resulting from higher Pt oxidation rates. This assumption is consistent with previous work which showed that the number of Pt-O bonds of lower binding energy, i.e. the site density of more reactive surface oxygen, increases on larger Pt particles [2]. The initial activation of well-dispersed Pt catalysts in cyclopentane combustion would be caused by sintering of the metallic phase, which occurs in reaction conditions even if the cyclopentane the combustion reaction is performed at low-temperature and low-conversion regimes. The reaction is highly exhotermic and the Pt crystallite temperature is significantly increased in reaction conditions. Hot-spots on the metallic particles together with the presence of gaseous water cause the metal phase sintering at mild reaction conditions and the formation of larger, more active, Pt particles. The methane combustion has been interpreted by considering a Langmuir-Hinshelwood mechanism, where the rate-determining step is the abstraction of the first hydrogen on the
580
adsorbed methane molecule and oxygen chemisorption steps are not kinetically significants [8]. The proposed reaction pathways" 202+4.L
". K, > 4 0 . L
CI-I4+L
< K2 )
k
CH4.L + O.L CH3.L + 3 0 . L
>
CH4.L CH3.L+ OH.L
". K3 > C.L+3OH.L K4 > C O2.L+2L < K5 < > 2 H20(g) + 20.L + 2 L K6 < )' C02(g)+ L
C.L+20.L 40H.L
C02.L
leads to a complex kinetic rate expression: k K34K,K2 Pc., [Po~1,2
r= ,
+4K,K
,
+I
+2 K6
qKKK6 [p02]1/2Pc0212
When hydroxyl groups are the most abundant species, the initial rate expression becomes:
~ r0 -
(3)
PH2o
which is consistent with the observed experimental rate equation.
REFERENCES 1. K. Otto, Langmuir, 5 (1989) 1364. 2. P. Briot, A. Auroux, D. Jones and M. Primet, Appl. Catal., 59 (1990) 141. 3. M. Kobayashi, T. Kanno, A. Konishi and H. Takeda, React. Kinet. Catal. Lett., 37 (1988) 89. 4. R.F. Hicks, H. Qi, M.L. Young and R.G. Lee, J. Catal., 122 (1990) 280. 5. P. Mar6cot, A. Fakche, B. Kellali, G. Mabilon, M. Prigent and J. Barbier, Appl. Catal. B: Environmental, 3 (1994) 283. 6. C. Pliangos, I.V. Yentekakis, V.G. Papadakis, C.G. Vayenas and X.E. Verykios, Appl. Catal. B: Environmental, 14 (1997) 161. 7. P. Papaefihimiou, T. Ioannides and X.E. Verykios, Appl. Catal B: Environmental, 15 (1998) 75. 8. K. Fujimoto, F. Ribeiro, H.M. Avalos Borja and E. Iglesia, J. Catal., 179 (1998) 431. 9. T.F. Garetto and C.R. Apesteguia, J. Catal., in press. 10. P. Mars and D.W. van Kravelen, Chem. Eng. Sci., 3 (1954) 41.
575
Structure sensitivity of the hydrocarbon combustion reaction over aluminasupported platinum catalysts T.F. Garetto and C.R. Apesteguia* INCAPE (UNL-CONICET), Santiago del Estero 2654, (3000) Santa Fe, Argentina The reaction kinetics, structure sensitivity, and in-situ activation of cyclopentane and methane combustion on Pt/A1203 catalysts of different metallic dispersion were studied. The reaction orders in oxygen were 1 (cyclopentane) and zero (CH4). Methane oxidation turnover rates did not change significantly by changing the metallic dispersion but the cyclopentane combustion activity increased dramatically with increasing Pt crystallite size. On both reactions, the activation energies did not change by changing the Pt dispersion. Results are interpreted in basis of two different reaction mechanisms over the metallic Pt active sites. I. INTRODUCTION Platinum-based catalysts are highly active for oxidative removal of small amounts of hydrocarbon from gaseous or liquid streams. The effect of varying the platinum particle size on the catalytic combustion of different hydrocarbons has been extensively studied [1-4], but the results obtained are conflicting, probably because correlation between catalytic activity and metallic dispersion depends on the type of hydrocarbon to be abated. Several papers on light alkanes combustion, namely methane [ 1], propane [5], and butane [4] have reported that alkane oxidation turnover rates increase with increasing platinum particle size. In contrast, in a recent study on the C2H4 combustion over platinum-supported catalysts Pliangos et al. [6] proposed that turnover frequency changes, which cannot be explained by structure sensitivity considerations, are caused by interactions between the metal crystallites and the carrier. Similarly, Papaefihimiou et al. [7] reported that the benzene oxidation turnover rate on Pt/A1203 strongly increases with increasing Pt particle size but does not change by changing the Pt dispersion on Pt/SiO2 and Pt/TiO2 catalysts. Palladium and platinum catalysts are often activated on stream, ab-initio of the hydrocarbon combustion reaction [5,8,9]. This phenomenon has been widely studied on palladium-based catalysts; in the case of methane oxidation, several authors have proposed that the initial activation period is caused by reoxidation from Pd metal or oxygen-deficient PdOl.x to more active steady state PdO species [8]. In contrast, very few papers have been published using platinum-based catalysts [5] and the causes of induction periods on platinum remain unclear. In a recent paper [9], we studied the structure and reactivity of Pt/A1203 catalysts for benzene oxidation at low temperatures. In this work we report results on the oxidation of cyclopentane and methane over a set of Pt/AI203 catalysts of different metallic dispersion and chlorine concentration. Our goal was to obtain further information on the catalyst activation phenomenon and on the sensitivity of hydrocarbon oxidation turnover rates to Pt crystallite size. "Corresponding author. Email: [email protected],fax: 54-342-4555279
576 2. EXPERIMENTAL A Pt(0.3%)/AI203 catalyst (catalyst A) was prepared by impregnation at 303 K of a highpurity 7-A1203 powder (Cyanamid Ketjen CK303) with an aqueous solution of H2PtCI6.6H20 and HCI. After impregnation, samples were dried 12 h at 393 K and heated in air stream to 773 K. Then the chlorine content was regulated using a gaseous mixture of HC1, water and air. Finally, the sample was purged with N2 and reduced in flowing H2 for 4 h at 773 K. A set of three catalysts with different Do (Pt dispersion) was prepared by treating catalyst A in a 2% O2~2 atmosphere at 848, 873 and 903 K for 2 h (catalysts B, C, and D, respectively). The Pt dispersion was measured by 1-12 chemisorption by using the double isotherm method and a stoichiometric atomic ratio H/Pt~=I, where Pts implies a Pt atom on surface. The characteristics of catalysts A, B, C, and D are shown in Table 1. Tabla 1 Characteristics of the catalysts used in this work Catalyst Pt loading CI concentration (wt.-%) (wt.-%) A 0.30 0.95 B 0.30 0.61 C 0.30 0.58 D 0.30 0.60
Pt dispersion Do (%) 65 38 24 15
Hydrocarbon oxidation reactions were carried out at 1 atm in a fixed-bed tubular reactor. Cyclopentane (0.65%) or methane (2%) were fed in a 10% O2/N2 mixture. On-line chromatographic analysis was performed using a gas chromatograph equipped with a flame ionization detector and Bentone 34 or Porapak Q packed columns. Before gas chromatographic analysis, the reaction products were separated and carbon dioxide converted to methane by means of a methanation catalyst (Ni/Kieselghur) operating at 673 K. Two experimental procedures were used for catalyst testing. The complete oxidation of hydrocarbons was studied by obtaining curves of hydrocarbon conversion (X) as a function of temperature (light-off curves). The temperature was raised by steps of about 23 K, from 25 to 673 K (cyclopentane) or 913 K (methane). More fundamental differential reactor experiments (less than 10% conversion) were performed at constant temperature. 3. RESULTS AND DISCUSSION 3.1. Catalytic tests: Light-off curves Fig. 1 shows the X vs T curves obtained on catalyst A in two consecutive catalytic tests. The cyclopentane combustion started at about 473 K in the first run and the conversion increased then dramatically at ca. 553 K reaching a value of X ~ 100% between 663 and 673 K. The reaction was maintained at 673 K for 2 h and then the catalyst was purged and cooled down in nitrogen to 373 K. Subsequently, a second catalytic test was carried out. As shown in Fig. 1, the X vs T curve corresponding to the second run was clearly shifted to lower temperatures as compared to that obtained in the first run. Such a displacement of the light-off curves typically illustrates the catalyst activation phenomenon in hydrocarbon combustion reactions. To compare catalyst activities, we measured from light-off curves the value of the
577 temperature at X = 50 %, T i,j, 5~ where i identifies the catalyst and j indicates first (1) or second (2) runs. The difference ATSO = TSO 50 is a measure of the i,l - Ti,2
100 80
.
Methane
60
activation phenomenon on catalyst i. Table 2 shows that the AT5~ value for cyclopentane combustion was about 80 K. The CH4 combustion on Pt occurs at temperatures (T5~ = 823 K) significantly
-~ 4o
r
20 9
n 9
o =
!
I
1st. run 2nd. run
higher as compared with cyclopentane combustion (TS~ = 593 K). The consecutive
=
light-off curves for CH4 combustion were similar (Fig. 1), thereby suggesting that catalyst A is not activated ab-initio of this Fig. 1" Light-off curves on catalyst A reaction. On the other hand, we measured the Pt dispersion on catalyst A atter the second runs (D2, Table 2). By comparing the Do and D2 values in Table 2 it is inferred that the metal was severely sintered in both reactions after two consecutive catalytic tests. Similar experiments were carried out on catalyst C. For cyclopentane combustion, the light-off temperature in the first run (T c5~ = 518 K) was clearly lower than that obtained on 400
600
Temperature
8O0
1000
(K)
catalyst A; the initial catalyst activation was negligible (ATc5~ 5 K) as well as the metal sintering at~er two consecutive runs (Table 2). In contrast, for methane combustion the T~5~ and AT5~ values obtained on catalyst C were similar to those found on catalyst A. Tabla 2 Catalytic activity and Pt dispersion in two consecutive catalytic runs Catalyst Reactant Temperatures at X = 50% (K) T~~ T25~ AT 5~ A A C C
Cyclopentane Methane Cyclopentane Methane
593 823 518 823
513 818 513 823
80 5 5 0
Pt dispersion (%) Do D2 65 65 24 24
15 18 20 17
3.2. Low-conversion catalytic tests In order to establish the effect of the Pt crystallite size on catalyst activity, additional kinetically-controlled catalytic tests were performed In all the cases, the initial conversion was lower than 10%. The oxidation reactions were performed over catalysts A, B, C, and D at 443 K (cyclopentane) and 713 K (methane). For cyclopentane combustion, the activity of welldispersed catalyst A slowly increased with time on stream, but the turnover frequency (TOF, sl) on sintered catalyst D was constant along the 20 hour run. Over all the catalysts, methane oxidation rates did not change with time on stream. In Fig. 2 we have plotted the initial turnover frequencies as a function of the metallic dispersion. Cyclopentane combustion turnover rates
578
~
n
e
increased drastically with the Pt particle size; the TOF value on catalyst D (Do = 15%) was about 40 times higher than that measured on catalyst A (Do = 65%). This result shows that cyclopentane combustion on Pt/Al203 catalysts is a structure-sensitive reaction preferentially promoted on larger Pt crystallites. On the contrary, for methane combustion the effect of Pt dispersion on the catalytic activity was rather weak and the turnover rate does not change significantly with Do (Fig. 2). All these results suggest that the existence of initial activation periods is related to the sensitivity of
.L_._
0.1 v
I.i_
o
0.01
I
6O
o0 (%) Fig. 2: Turnover frequency (TOF) vs Do the combustion turnover rate to the Pt crystallite size. 3.3. Kinetic studies
Kinetic data were interpreted by considering a power-law rate equation: #
r o = k(P~ )~ (P~2~, where r0 (mot HC/hg Pt)is the initial reaction rate. In Figs. 3 and 4 the ro values obtained on catalyst A for the oxidation of cyclopentane and methane were represented in logarithmic plots as a function of P~c and p00 2 respectively. Reaction orders cx and 13 were determined graphically from Figs. 3 and 4. The reaction orders for cyclopentane Pg= 0.126 atm Methane
~ o
o
,_o r -1
o
Cyclopentane
-2
Cyclopentane~._~
443 K
-3 -4
o
-3
_~
'
713 K
Methane
e-
-2
PHc = 0.02 atm,
_15
'
_'4
'
_~3
In Pnc Fig. 3" Reaction orders in the hydrocarbon
I
I
-2.4
PHC= 6.5 10 .3 atm, 443 K i
In P0 2
I
-2.0
Fig. 4: Reaction orders in oxygen
combustion were ot _-- 0 and J3 -= 1 while values of cz _=_1 and J3 --- 0 were determined for CH4 combustion. Similar values for ct and 13 were measured on catalyst C. On the other hand, we plotted the In TOF values as a function of 1/T for calculating the apparent activation energy (Ea) and the preexponential factor A of both reactions on catalysts A, C and D via an Arrhenius-type function. The apparent activation energies were 11 + 1 kcal/mol (cyclopentane) and 17 + 1 kcal/mol (CH4), irrespective of the mean Pt crystallite size of the sample. For
579 cyclopentane combustion, we measured a A D/A A ratio of about 60. These results suggested that increasing the Pt particle size increases the density of active sites available for the ratedetermining step but does not modify the cyclopentane oxidation mechanism. The kinetic results show that CH4 and cyclopentane are oxidized by different mechanisms. The reaction orders obtained for cyclopentane combustion are well interpreted by considering that the reaction occurs via a Mars-Van Krevelen type mechanism [ 10], being the dissociative adsorption of oxygen on Pt the rate determining step. For cyclopentane oxidation on Pt, this mechanism may be represented by the following elementary steps: 02(g) + 2 L CP(g) + L
CP.L + O.L (CP...O).L
k~ > k2 >
20.L
k3 >
(CP...O).L
O.L
>
CP.L
CO2(g)+ H20(g)
where L represents the vacant active sites. The expression of initial rate r0 results: pO o klk3 o~PcP r0 = k,po ' + vik3PO p w h e r e v i is the stoichiometric coefficient of oxygen in the overall reaction.
(1) If k I < < k 3
Eq. (1) reduces to: klP~ 2 ro = - Vi
(2)
and the orders with respect to cyclopentane and oxygen predicted by Eq. (2) are 0 and 1, respectively, which are the approximate orders determined from our experiments. According to Eq. (2), any increase in rate constant k 1 accelerates the cyclopentane oxidation rate. The observed turnover rate increase with increasing Pt particle size would reflect therefore an increase in the density of reactive Pt-O species resulting from higher Pt oxidation rates. This assumption is consistent with previous work which showed that the number of Pt-O bonds of lower binding energy, i.e. the site density of more reactive surface oxygen, increases on larger Pt particles [2]. The initial activation of well-dispersed Pt catalysts in cyclopentane combustion would be caused by sintering of the metallic phase, which occurs in reaction conditions even if the cyclopentane the combustion reaction is performed at low-temperature and low-conversion regimes. The reaction is highly exhotermic and the Pt crystallite temperature is significantly increased in reaction conditions. Hot-spots on the metallic particles together with the presence of gaseous water cause the metal phase sintering at mild reaction conditions and the formation of larger, more active, Pt particles. The methane combustion has been interpreted by considering a Langmuir-Hinshelwood mechanism, where the rate-determining step is the abstraction of the first hydrogen on the
580
adsorbed methane molecule and oxygen chemisorption steps are not kinetically significants [8]. The proposed reaction pathways" 202+4.L
". K, > 4 0 . L
CI-I4+L
< K2 )
k
CH4.L + O.L CH3.L + 3 0 . L
>
CH4.L CH3.L+ OH.L
". K3 > C.L+3OH.L K4 > C O2.L+2L < K5 < > 2 H20(g) + 20.L + 2 L K6 < )' C02(g)+ L
C.L+20.L 40H.L
C02.L
leads to a complex kinetic rate expression: k K34K,K2 Pc., [Po~1,2
r= ,
+4K,K
,
+I
+2 K6
qKKK6 [p02]1/2Pc0212
When hydroxyl groups are the most abundant species, the initial rate expression becomes:
~ r0 -
(3)
PH2o
which is consistent with the observed experimental rate equation.
REFERENCES 1. K. Otto, Langmuir, 5 (1989) 1364. 2. P. Briot, A. Auroux, D. Jones and M. Primet, Appl. Catal., 59 (1990) 141. 3. M. Kobayashi, T. Kanno, A. Konishi and H. Takeda, React. Kinet. Catal. Lett., 37 (1988) 89. 4. R.F. Hicks, H. Qi, M.L. Young and R.G. Lee, J. Catal., 122 (1990) 280. 5. P. Mar6cot, A. Fakche, B. Kellali, G. Mabilon, M. Prigent and J. Barbier, Appl. Catal. B: Environmental, 3 (1994) 283. 6. C. Pliangos, I.V. Yentekakis, V.G. Papadakis, C.G. Vayenas and X.E. Verykios, Appl. Catal. B: Environmental, 14 (1997) 161. 7. P. Papaefihimiou, T. Ioannides and X.E. Verykios, Appl. Catal B: Environmental, 15 (1998) 75. 8. K. Fujimoto, F. Ribeiro, H.M. Avalos Borja and E. Iglesia, J. Catal., 179 (1998) 431. 9. T.F. Garetto and C.R. Apesteguia, J. Catal., in press. 10. P. Mars and D.W. van Kravelen, Chem. Eng. Sci., 3 (1954) 41.