perature responses of the fiber catalyst bed are found to be always better than those of beds containing catalyst beads.
Literature Cited Acres, G. J. K., Cooper, E. J.. Platinum Met. Rev., 16, 74 (1972). Benrey. R. M., Search, 3 (1973). Bowdiich, F . W., "A General Motors Publication," 1973. Dwyer, F. G.,Morgan, C. R., Mobil Research and Development Corporation, Research Department, Paulsboro. N.Y.. 1973. Fogel, A,, Kinet. Katal., 5, 496 (1964). Hum. P. W., Dozois, C. L., Chase, J. O., Ellis, C. F.,Ferrin, P. E., Division of Refining, 42, 657, (1962). Innes, W. D., Tau, K., "The Kirk-Othmer Encyclopedia of Chemical Technology," p 814, 1963.
Klimisch, R. L., Schlatter, J. C., presented to American Ceramic Society, Flint, Mich., 1972. Kuo, J. C. W., Morgan, C. R., Lassen, H. G.. SAE Paper No. 710289, 1971. Levenspiel, O., "Chemical Reaction Engineering," 2d ed, Wiley. New York, N.Y., 1972. Leventhal, B., Statement to EPA Hearing on Application for One-Year suspension of Auto Emission Standards, 1972. Nicholas, D.M., Ph.D. Thesis, University of Pittsburgh, 1975. Snyder, P. W., Stover, W. A.. Laseen. G. G., SAE Paper No. 720479, 1972. Schlatter, J. C., Klimisch, R. L., Taylor, K. C., Science, 179, 798 (1973). Shishu, R. C.. Kowalcyzyk, L. S.,Platinum Met. Rev., 18, 58 (1974). Sourirajan. S..Accomazzo, M. A,, Can. J. Chem., 38, 1990 (1960). Voltz, S. E., Morgan, C. R., Liederman. D.. Jacob, S. M., Ind. Eng. Chem.. Prod. Res. Dev., 12, 294 (1973).
Receioed for reuiew February 28,1975 Accepted October 9,1975
Carbon Monoxide Oxidation over a Platinum-Porous Fiber Glass Supported Catalyst D. M. Nicholas and Y. 1.Shah' Department of Chemical Engineering, University of Pinsburgh, Pittsburgh, Pennsylvania 1526 1
This paper presents the kinetics of carbon monoxide oxidation over a platinum-porous fiber glass supported catalyst. In a recent article by Nicholas et al. (1976), the advantages of this catalyst for an automobile exhaust converter were illustrated. We examined the kinetic data illustrating the basic activity of fiber glass supported platinum catalyst for carbon monoxide oxidation and compare the results with the similar results with other platinum catalysts published in the literature. The kinetic data were correlated by three types of model: (a) a powerlaw model, (b) Langmuir-Hinshelwood single site model, and (c) Langmuir-Hinshelwood dual site model. The Langmuir-Hinshelwood dual site model was found to correlate the experimental data best under the entire range of reaction conditions. The experimental data indicated the inhibition effect of carbon monoxide on the rate of oxidation. At low temperatures fiber catalyst appears to be more active than other platinum catalysts investigated in the literature.
Introduction The kinetics of carbon monoxide oxidation have been studied as early as 1922 (Langmuir, 1922). A considerable amount of attention has been, however, given to this reaction in recent years because of its importance in automotive emission control. The kinetics of this reaction has been examined for a variety of catalysts and cat,alyst supports. The most widely used catalysts for the carbon monoxide oxidation have been platinum-supported type catalysts. Langmuir (1922) found that carbon monoxide oxidation over platinum wire could involve the reaction of gaseous carbon monoxide with adsorbed atomic oxygen. Sklyarov (1969), however, found that the oxidation of carbon monoxide involved molecular oxygen and carbon monoxide. With his dual site Langmuir-Hinshelwood model, Shishu and Kowalczyk (1974) also indicated that carbon monoxide reacted with adsorbed molecular oxygen. It is generally agreed that the oxidation of carbon monoxide over a platinum catalyst is inhibited by carbon monoxide a t low conversion and a t low temperatures. Harned (1971) found that a t conversions less than 80%, the rate of carbon monoxide oxidation was inversely proportional to carbon monoxide concentration and directly proportional to oxygen concentration. At conversions greater than 80%, rates were directly proportional to carbon monoxide to the half power. Voltz et al. (1973), in a kinetic study of carbon
monoxide oxidation on a platinum-alumina catalyst, also found the rate of the reaction to be inhibited by carbon monoxide. They proposed a Langmuir-Hinshelwood type of rate model to take into account the inhibition effect of carbon monoxide. Very recently, Shishu and Kowalczyk (1974) also obtained a similar type of Langmuir-Hinshelwood rate model. They also reported a power law-type rate model which indicated the rate of oxidation to be directly proportional to oxygen partial pressure and inversely proportional to the half power of the carbon monoxide partial pressure. All of Shishu's data were, however, for carbon monoxide conversion less than 10%. In the preceding paper, Nicholas et al. (1976) showed that a porous fiber glass-supported platinum catalyst presents some advantages over conventional pellet catalysts for use in an automobile exhaust catalytic converter. Because of the lower thermal mass of fiber catalysts, a converter packed with the fibers has a quicker temperature response than one packed with the conventional pellets. For a simulated automobile exhaust mixture, the fiber catalyst appears to perform better than the conventional pellets a t low temperature. The purpose of the present paper is to examine intrinsic activity of a fiber-supported platinum catalyst for the oxidation of carbon monoxide. The results are compared with the similar results reported in the literature for other platiInd. Eng. Chem., Prod. Res. Dev.. Vol. 15, No. 1, 1976
35
num catalysts. Three kinetic models for the reaction are presented.
Experimental Section The equipment used in the present investigation consisted of an inlet metering system, a reactor system, and an analytical system (see Figure 1). The inlet metering system consisted of two feed gas lines and a helium purge line. The first feed line consisted of a mixture of carbon monoxide, carbon dioxide, and nitrogen and a water saturator, which introduced 10 mol % water vapor into the feed. The second feed line consisted of pure oxygen. The amount of carbon monoxide used in the feed was varied by using cylinders of premixed gases containing varying amounts of carbon monoxide. The amount of oxygen was varied by adjusting the flow of oxygen into the system. The reactor system consisted of the reactor itself, a heating furnace, a manometer, and a reactor by-pass line. The reactor was constructed of quartz and served both as a preheater and catalyst bed. The preheat section of the reactor, approximately the first 30 cm, was packed with 2-3-mm diameter spheres of high silica glass. The catalyst bed was packed with the catalyst in the next section of the reactor and the remaining length of the reactor was also packed with glass beads. The reactor, with an i.d. of 2 cm, had a thermocouple well passing down the middle of the reactor enabling measurement of temperatures in the bed. Three thermocouples were positioned as follows: one at the inlet of the bed, the second a t the middle of the bed, and the third a t the exit of the bed. A thermocouple was also placed on the outside wall of the reactor in the middle of the catalyst bed. The analytical system consisted of a Hewlett-Packard Research Chromatograph with a thermal conductivity cell used as a detector. Separation of oxygen, nitrogen, and carbon monoxide was obtained on a Molecular Sieve 13X column while separation of carbon dioxide and water was obtained on an 8-ft Porapak Q and an 8-ft Porapak R column. The system conditions investigated were catalyst bed temperatures varying from 160 to 305 "C, total flow varying from 30 000 to 70 000 GHSV, and oxygen and carbon monoxide concentrations both varying from approximately 0.2 to 2.0 mol %. Carbon dioxide and water were both maintained a t approximately 10 mol % while the amount of nitrogen was approximately 78-80 mol %. The catalyst used was a porous fiber glass catalyst containing 0.27 wt % platinum. The fiber glass catalyst had a BET surface area of 74.7 m2/g, a bulk density of 0.37 g/cm3, and a porosity of 0.044; these properties were obtained after impregnation of the platinum and before the fiber catalyst was used for any experiments. Some other properties of the catalyst obtained from a BET analysis are summarized in Table I. More detailed descriptions of the catalyst properties and its preparation are given by Nicholas (1975). The composition of the feed gas was first determined by by-passing the reactor system and allowing the feed gas to enter the gas chromatograph system. Once the feed gas composition was determined, the feed was allowed to enter the reactor which was set a t a predetermined temperature. After a steady-state condition for the bed temperature was achieved, the exit stream of the reactor was analyzed in the gas chromatograph. Data were taken only when the catalyst bed was essentially isothermal. The experimental data were correlated by three kinetic models: (I) a power law model, (11) a Langmuir-Hinshelwood single site model, and (111) a Langmuir-Hinshelwood dual site model. The power law model was used to correlate only the low conversion data. 36
Ind. Eng. Chem., Prod. Res. Dev., Vol. 15,No. 1, 1976
I. Power Law Model The experimental data with CO conversions less than approximately 13% were correlated by the following power law rate expression
Rco = Ae-E/R*T(C0)a(02)b
(1)
where Rco = rate of disappearance of CO, (CO) = mol % carbon monoxide, ( 0 2 ) = mol % oxygen, A = frequency factor based on catalyst volume, cc/sec/cc of catalyst, E = activation energy for the reaction, cal/g-mol, R* = universal gas constant, 1.987 g cal/g mol K, and T = temperature in K. From the experimental data, rates of reaction were calculated following the differential reactor type of analysis presented by Shishu and Kowalczyk (1974). This analysis is reasonable for low conversions and makes use of the following expression to calculate the rate of disappearance of carbon monoxide
where Ax = conversion of carbon monoxide, V = volume of catalyst, cc, and F = volumetric flow rate of carbon monoxide cc/sec. The values of (CO) and ( 0 2 ) used in eq 1 are the reactor inlet concentrations of the reactants; it is assumed these concentrations are the same at the inlet and outlet of the catalyst bed because of the low conversions. The experimental data were fitted to eq 1 using both linear and nonlinear least-square techniques. The values of power coefficients a and b in eq 1 obtained in this manner were approximately -0.28 and 1.07, respectively. The value of a = -0.28 demonstrates the inhibition effect of carbon monoxide on the rate of reaction. The values of a and b obtained here are comparable to the ones obtained by Shishu and Kowalczyk (1974) as a = -0.5, b = 1.0 with a honeycombplatinum catalyst. The frequency factor A and the activation energy E obtained in the present experiments were 3.8 X 10' (sec-') and 6400 cal/g-mol, respectively. Based on these values of the rate parameters, a parity plot of model predicted carbon monoxide conversion vs. the ones obtained experimentally is shown in Figure 2. The activation energy and the frequency factor obtained in this study differ considerably from the similar values of 16 000 cal/g-mol and 9.1 X lo4 (sec-l) obtained by Shishu and Kowalczyk (1974) for a honeycomb-type monolith supported platinum catalyst. For the range of temperatures investigated here, a comparison between the rate constants obtained for the present fiber catalyst and the monolith supported catalyst of Shi. shu and Kowalczyk (1974) is shown in Figure 3. These results indicate that the intrinsic rate constant obtained in the present study was larger than the one obtained by Shishu and Kowalczyk (1974) a t low temperature levels. Unfortunately, no data on the platinum surface area for the catalyst used by Shishu and Kowalczyk (1974) have been reported to permit a comparison of the intrinsic activity of the present catalyst and the one for their catalyst. It should be noted, however, that the w t % platinum on the catalyst used by Shishu and Kowalczyk (1974) is approximately one-half (0.14%) of that used in the present study, but the total weight per unit volume of catalyst (wt % Pt X bulk density of catalyst) is similar for both catalyst. The surface area per unit volume of catalyst (BET surface area X bulk density) used in the present study is approximately 3.5 times (27.8 m2/cc of catalyst) that used by Shishu and Kowalczyk (1974). Assuming all platinum sites are available for reaction, a comparison of the rate constants based on the surface area of the catalysts (denoted as K") is
F
ROTAMETEH
WATER
SATURATOR
TO VENT
C O . C 0 2 . & N2 T A N K
REACTOR & PREHEATER
WEST
OR
TEST METER
c
SAMPLING VALVE U
02 TANK
He T A N K
Figure 1. Equipment for the present study.
Table I. Properties of Porous Glass Fiber Catalyst
____
Wt % Pt BET area, m'/g Average pore radius, ?i Total pore volume, cc/g Pore radius distribution Volume % pore, radius, ?i
0.27 74.7 32.0
12
0.12
15.8/7-20 12.2/20-30 11.9/30-40 1 2 . 1/40-50 48.1/50-100 7.71100-200 1.1/200-300 34
Platinum crystallite size, A
8 OBSERVED CARBON M O N O X I D E CONVERSION 1911 4
shown in Figure 3. Comparison of these two plots shows that the kinetic constants based on surface area for the fiber Catalyst are larger than the ones for the honeycomb catalyst a t temperatures below approximately 275 "C.
0
PREDICTED CARBON M O N O X I D E CONVEqSION !' I
Figure 2. Parity plot of power law model predicted carbon monoxide conversions vs. observed carbon monoxide conversions.
11. Langmuir-Hinshelwood Single Site Model This model (henceforth denoted as L-H-S model) was attempted to fit the data obtained a t all conversions in the present study. The model was of the following form
3 l!01 1
(3) 102
where (CO) and
K
are defined as before and K , = Aae.-Ea/R*T(Co)-1
(02)
K, = Are-Er/R*Tsec-'
CCiSEC CCOF CATALYST 3110"
(02)-]
SHISHU A N D
here K, is an adsorption rate constant for carbon monoxide and K, is the intrinsic rate constant with corresponding activation energies and frequency factors. Substituting the expression for the rate of carbon monoxide oxidation (eq 3) into the following equation for a plug flow reactor
.=-I
(CO) d(C0)
(COIO
Rco
(4)
where 7 = residence time, in seconds, based on volume of catalyst/volumetric flow rate. (C0)O = initial concentration of carbon monoxide in mole %. One obtains
KOWALCLYK S
103
31101'
HONEYCOMB C A T A L Y S T 119741
!6
1 7
13
19
20
21
22
23
1000lT 1%
Figure 3. Rate constants for power law model vs. temperature for fiber and Shishu and Kowalczyk's honeycomb catalysts. Ind. Eng. Chem., Prod. Res. Dev.. Vol. 15, No. 1, 1976
37
Table 11. Frequency Factors and Activation Energies for the L-H-SModel at All CO Conversions The mol % oxygen at any time, (021, was replaced by the expression based on the stoichiometric amount of oxygen by (02)
=
(02)o
- 0.5 [(CO)o - (CO)]
(6)
Fiber catalyst 1.79 -1000 4.4 (1015 7 . 1 (10)3 Voltz et al. alumina 0.655 -1910 1.8 (10)" 2.5 beads a (CO)-l. b (sec)-'(02)-'. c cal/gmol.
where ( 0 2 ) o = initial concentration of oxygen in mol %. Combining eq 5 and 6 one obtains Kr7 =
(C0)o [w2
ice,
IW
+ Pwz(C0) + z ~ ( C O ) ~ ] ~ ( C O( 7))
[a(CO) + Y(co)21 here a = ( 0 2 ) o - 0.5(cO)0, y = 0.5, z = Ka, and w = 1. An integration of (7) results as -W2
K,T = -In
a
[ G -(C0)o 1
+ *In y
[GI +
80
60 OBSERVED CARBON MONOXIDE CONVERSION
40
($1
+
+
where G = ( a y(CO)o)/(a y(C0)) and (CO) = (CO)o(l - x ) and x = conversion of carbon monoxide in mole fraction. Equation 8 was used to obtain values of A,, E , , A,, and E , by use of both a nonlinear least-squares technique and a patterned search technique. Table I1 gives the values obtained for the present model. Voltz et al. (1973) also obtained a rate expression similar to (3), but in his model, he accounted for effects of propylene and nitric oxide on carbon monoxide oxidation. If one sets the concentrations of propylene and nitric oxide equal to zero, values of A,, E,, A,, and E, can be obtained for the comparison purposes, and these are also given in Table 11. Figure 4 shows a parity plot of the L-H-S model predicted carbon monoxide conversions vs. the ones obtained experimentally. Values of the various rate constants for the present L-H-S model and the model of Voltz et al. (1973) are plotted vs. temperature in Figure 5 . The values for the adsorption rate constant, K,, for the fibers and alumina beads of Voltz et al. (1973) are close over the entire temperature range studied and are decreased as temperature is increased (1000/T decreased). The value of K, for the fibers does not increase sharply with temperature, but at low temperatures these values are considerably higher than the ones for the alumina beads. Low activation energy for the present catalyst indicates that pore size for the fibers may not be large enough to completely eliminate the effects of internal diffusion on the rates of reaction. Unfortunately, no data on platinum w t 96, or platinum surface area for Voltz's catalyst have been reported to permit a comparison of intrinsic activity of the present catalyst to his.
20
0 20
0
40
60
100
PREDICTED CARBON M O N O X I D E CONVERSIONI%I
Figure 4. Parity plot of predicted L-H-S model carbon monoxide conversions vs. observed carbon monoxide conversions.
lt
I' 1WOil i°KIS1
111. Langmuir-Hinshelwood Dual Site Model
The model (henceforth denoted as L-H-D model) which produced the best fit of our data was of the form (9)
where K , is defined as before
here Kb is an adsorption rate constant for oxygen and K,' is the pseudo intrinsic rate constant with corresponding activation energies and frequency factors. Experimental data with carbon monoxide conversions less than approximately 13% were first correlated by a dif38
Ind. Eng. Chem., Prod. Res. Dev., Vol. 15, No. 1, 1976
Figure 5. Kinetic and adsorption constants for fiber and Voltz et al. alumina catalysts as functions of temperature for the L-H-S model. ferential bed analysis in order to compare the kinetic constants with the ones obtained by Shishu and Kowalczyk's (1974) model. Values of the frequency factors and activation energies obtained by a nonlinear least-squares technique are shown in Table I11 along with the similar values reported by Shishu and Kowalczyk (1974). Using these values, the rate constants as functions of temperature for both catalysts are shown in Figure 6. Over the temperature range considered here, the values for the adsorption rate constant for carbon monoxide, K,, for both catalysts are about the same. Also, for both catalysts, K, decreases as temperature increases. The adsorp-
Table 111. Frequency Factors and Activation Energies for L-H-D Model at Low CO Conversions Fibers catalyst 6.30(10)-' Honeycomb catalyst of Shishu and Kowalczyk (1974) 2.20(10)-2 a (CO)-'. b ( 0 2 ) -c' . (CO)(sec)-'. d cal/g-mol.
-1.0(10l3
i.3(10)-~
-1000
3.1(10)6
8.2(10)3
-5.9( 10)3
1.3(10 ) - 6
-1.4(10)4
9.4(10)9
2.5(10)4
Table IV. Frequency Factors and Activation Energies for L-H-D Model for All CO Conversions 3 11013
2.63 -1000 1.009 -1000 .2.0(10)5 (CO)-'. b ( 0 2 ) -c' (CO)(sec)-'. , dcal/g-mol.
1101~
8.5(10)3
/'
3 11012
K, SHISHU A N 0
/'
12
KOWALCZY K'S HONEYCOMB C A T A L Y S T
/\
10 K a HONEYCOMB
A
CATALYST
/A
8 OBSERVED CARBON M O N O X I D E
3
n
A
d
PERFECT FIT L I N E
CONVERSION
I%! 4
1
311Oj.l
/
/16
17
18
19
20
21
22
I
4
23
12
8
PREDICTED C A R B O N M O N O X I D E CONVERSION 1%)
1000 T I'Kl
Figure 6. Kinetic and adsorption constants for L-H-D model vs. temperature for fiber and Shishu and Kowalczyk honeycomb catalysts.
Figure 8. Parity plot of L-H-D model predicted carbon monoxide
conversion vs. observed carbon monoxide conversion for low carbon monoxide conversions.
1101
3 1101
RCO !lo1 2
CCI1SEC CC OF C A T A L Y S T
311013
0 0
20
40
60
80
100
PREDICTED C A R B O N M O N O X I D E CONVERSIOY 1% !lOl
Figure 9. Parity plot of L-H-D model predicted carbon monoxide conversion vs. observed carbon monoxide conversion for all con-
versions. 6
17
l B
19
20
21
22
23
1000 T l ° K l
Figure 7. Rates of carbon monoxide oxidation for L-H-D model vs. temperature for fiber and Shishu and Kowalczyk's honeycomb catalysts.
tion constants for oxygen, Kb, is found to be considerably smaller for the fiber catalyst than for the honeycomb catalyst of Shishu and Kowalczyk (1974). This implies that adsorption of oxygen on the fiber catalyst is less important than adsorption on the honeycomb catalyst for the rates of oxidation a t the temperatures shown in Figure 6. As ex-
pected, the adsorption constants for oxygen, Kb, are smaller than that of carbon monoxide, K,, for both catalysts. T h e values of K,' for the fiber catalyst are much higher than those for the honeycomb catalyst. T h e results for the overall rates of disappearance of carbon monoxide are shown in Figure 7 . These rates are based on typical reactant compositions of 2 mol % carbon monoxide and 1 mol % oxygen. As can be seen in Figure 7, at the low temperatures, Rfiber are much higher than those of Rhoneycomb, but as the temperature increases, these two values begin to approach each other. Figure 8 shows a parity plot of L-H-D model predicted carbon monoxide conversion versus the Ind. Eng. Chem., Prod. Res. Dev., Vol. 15,No. 1, 1976
39
SYMBOL
MOLE%
MOLE%
o 28
0
'%I
1 I
CONVERSION
40
21
0 46
22
0 20
/
I
/
MODEL
/ /-.
j>& 2 0 1 ,
"
0
455
495
535
575
615
BED TEMPERATURE (OK1
Figure 10. Observed and L-H-D model predicted carbon monoxide conversion vs. bed temperature a t GHSV = 50 000 h-l.
%.
I_
The reasons for high activity at low temperatures and low temperature sensitivity of the fiber catalysts compare to other conventional catalysts are not as yet clearly understood. A preliminary electron microscope study of the platinum distribution over the fiber support indicates that platinum is reasonably well distributed on the surface (Nicholas, 1975). Similar distributions for other catalysts reported in the literature are not known. An extensive electron microscope study of the fiber catalyst is planned to understand the reasons for its high intrinsic activity at low temperatures.
0 55 11
80
0.56
0.19
22
60
-
CARBON MONOXIDE CONVERSION
(%I LINESARE PREDICTED CONVERSIONS FOR L W D MODEL
455
495
535 BED TEMPERATURE
575
615
(OK1
Figure 11. Observed and L-H-D model predicted carbon monoxide conversion vs. bed temperature a t GHSV = 70 000 h-l.
ones obtained experimentally at low carbon monoxide conversions. To correlate the data over all the carbon monoxide conversions, an expression similar to eq 8 was obtained upon substitution of eq 9 into eq 4 except that now K, = K,Kfi,' z = K, W
+ 0.5Kb
= Kb(02)o - 0.5(CO)o
(11)
+1
(12)
Again values of activation energies and frequency factors were obtained by use of a nonlinear least-squares technique and patterned search technique on eq 8 with the above values of K,, z , and w. These values as summarized in Table IV provided the best fit of all the data obtained in the present study. Figure 9 shows a plot of predicted carbon monoxide conversion vs. observed conversions for all the data obtained in the present study using the L-H-D model. I t is clear from this plot that the model fits the data best a t low conversions. The experimental data are best modeled a t low conversions because under this condition the reactor was isothermal within only a couple of degrees. At high conversions, it was very difficult to maintain the reactor isother-
40
mal. In some high conversion runs, the temperature along the reactor varied by 5-10'. Under these situations, an arithmetic average temperature was used for model fitting of the data. Figure 10 shows the predicted and observed conversions of carbon monoxide vs. bed temperature a t various feed concentrations of carbon monoxide and oxygen and a GHSV of 50000-1. The enhancement effect of oxygen on the oxidation of carbon monoxide can be seen in Figure 10 by comparing the curves a t 1.1 mol % CO and 0.28 mol % 0 2 , or by comparing the curves for approximately 2.15 mol % CO and the oxygen concentrations of 0.46 and 0.20 mol %. The inhibition effect of carbon monoxide can be seen by comparing the curves for 1.1 and 2.1 mol % CO a t an oxygen concentration of 0.46 mol %. Figure 11 also shows predicted and observed carbon monoxide conversions vs. bed temperature a t various feed concentrations but a t a GHSV = 70 000 h-l. The inhibition effect of carbon monoxide is again seen by comparing the curves for 0.55 and 1.1 mol % carbon monoxide and at the oxygen concentration of 1 mol
Ind. Eng. Chem., Prod. Res. Dev., Vol. 15, No. 1, 1976
Conclusions As a result of the present study, the following conclusions are made. (1) Kinetic expressions for carbon monoxide oxidation over a fiber glass supported platinum catalyst show the inhibition effect of carbon monoxide on the rate of oxidation. (2) At low temperatures fiber catalyst appears to be more active than other platinum catalysts investigated in the literature. (3) The experimental kinetic data were best fitted by the Langmuir-Hinshelwood dual site kinetic model. Acknowledgment One of the authors (D. M. Nicholas) gratefully acknowledges financial support of PPG industries. The help of Gulf Research and Development Company, Harmarville, Pa., for the catalyst property evaluation is gratefully acknowledged. Literature Cited Harned. J. L., Paper No. 720520, SAE National Automotive Engineering Meeting, Detroit, Mich., May 1972. Langmuir, I., Trans. Faraday Soc., 17, 621 (1922). Nicholas, D. M.,Ph.D. Dissertation, University of Pittsburgh, 1975. Nicholas, D. M., Shah, Y. T., Zlochower, I. A.. Ind. Eng. Chem., Prod, Res. Dev., 15, 29 (1976). Shishu, R. C., Kowalczyk, L. S., Platinum Met. Rev., 18, 58 (1974). Sklyarov. A. V., Tret'yakov, I. I., Shab. B. R.. Roginski, S. Z., Dokl. Phys. Chem. 189,829 (1969). Voltz, S. E., Morgan, C. R. Liederman, D., Jacob, S. M.. Ind. Eng. Chem.. Prod. Res. Dev., 12,294 (1973).
Received for review April 15, 1975 Accepted October 9,1975
