Ind. Eng. Chem. Res. 1987,26, 1419-1424 Mair, A. D.; Soroczak, M. M. U S . Patent 4588498, 1986. Moudgil, B. M.; Chanchani, R. Miner. Metall. Process. 1985, 2(1), 13-25. Ratobylskaya, L. D.; Klassen, V. I.; Boiko, N. N.; Baskakova, M. I.; Smirnov, Yu. M. “Development and Industrial Introduction of New Concentration Processes for Phosphorites of Complex Mineral Composition”, Proceedings of the 11th International Mineral Processing Congress Seminar on Beneficiation on Lean Phosphates with Carbonate Gangue, Cagliari, Sardinia, April 1975; pp 17-39. Rule, A. R.; Clark, C. W.; Butler, M. 0. “Flotation of Carbonate Minerals From Unaltered Phosphate Ores of the Phosphoria Formation”, Report of Investigation No. 7864,1974; U.S.Bureau of Mines, Washington, DC.
1419
Rule, A. R.; Clark, C. W.; Butler, M. 0. “Flotation of Carbonate Minerals From Unaltered Phosphate Ores of the Phosphoria Formation”, Proceedings of the 11th International Mineral Processing Congress Seminar on Beneficiation on Lean Phosphates with Carbonate Gangue, Cagliari, Sardinia, April 1975; pp 167-186. Rule, A. R.; Kirby, D. E.; Dahlin, D. C. Min. Eng. 1978,30(1),37-40. Smani, M. S.; Blazy, P.; Cases, J. M. AIME Trans. 1975, 258, 168-182. Snow, R. E. U.S. Patent 4 144 969, 1979. Snow, R. E. U S . Patent 4 364 824, 1982.
Received for review August 14, 1986 Accepted April 10,1987
Selective Oxidation of Propene on a Mo-Pr-Bi Catalyst Gojko Kremeni6,* Jose M. L. Nieto, Juan M. D. TascGn, and Luis G. Tejuca Instituto de Catdlisis y Petroleoqulmica, C.S.I.C.,28006 Madrid, Spain
Sol W . Weller Department of Chemical Engineering, State University of New York at Buffalo, Buffalo,New York 14260
The adsorption of oxygen and propene and the kinetics of formation of acrolein during the oxidation of propene over a Mo-Pr-Bi catalyst were studied. Preadsorption of propene a t 373 K does not appreciably affect the total adsorption of oxygen. The total adsorption of propene on a surface with preadsorbed oxygen a t 373 K is about 25% lower than that on a clean surface. This decrease is assumed to result from steric hindrance. The oxygen isobar exhibits an ascending branch, indicating a change in the nature of the adsorbed species, probably associated with the appearance of dissociatively adsorbed oxygen above 550 K. The kinetic data are best fitted by equations derived from the redox Mars-van Krevelen mechanism and from the stationary state of the adsorption model assuming in both cases nondissociative adsorption of oxygen. Reasonably good fits are also obtained with the power rate law. It is concluded that the data- and model-fitting procedures are useful in arriving a t rate equations that best describe the kinetic results, but it is hazardous to extend this analysis to conclusions concerning the detailed mechanism. Propene can be oxidized to acrolein in the gas phase over a number of catalysts. In recent years a large number of patents has been reported for this reaction (Hucknall, 1974). However, only in a limited number of cases have the kinetic aspects been studied, mainly due to the complexity of this catalytic reaction. The dependence of acrolein formation with propene or oxygen partial pressure gives reaction orders ranging from 1to 0. Activation energy values have been found to be from 50 to 210 kJ mol-l (Isaev and Margolis, 1960; Keulks et al., 1971; Crozart and Germain, 1973; Aso et al., 1980). Many selective oxidation reactions proceed according to a redox mechanism (Cullis and Hucknd, 1982) although other mechanisms where surface oxygen is involved have been proposed for the partial oxidation of olefins (Cartlidge et al., 1975; Cort6s Corber6n et al., 1985). Praseodymium oxide has unstable lattice oxygen (Takasu et al., 1981) (up to 7.5% total lattice oxygen is desorbed between 298 and 873 K). On the other hand, this oxide exhibited a maximum in the rate of exchange of molecular oxygen in a series of lanthanide oxides (Minachev et al., 1977; Klissurski, 1979). Other studies showed that these oxides have a promoting effect for partial oxidation (Khiteeva and Rzakulieva, 1981). From these considerations it seems of interest to study the preparation of new selective catalysts by modification of known systems using praseodymium oxide. As part of a more comprehensive work on this type of material, in this paper the adsorption of propene and oxygen and the kinetics of acrolein formation over a Mo-Pr-Bi catalyst were studied. The rate equations that best describe the kinetic data are 0888-5885187 12626- 1419$01.5010
given and the methodology of “best fit” is examined.
Experimental Section Catalyst Preparation and Gases. The catalyst sample was prepared by double impregnation of silica (BASF D-11-11) previously heated for 4 h at 1073 K. After this treatment, the support had a BET specific surface area (measured by N2 adsorption at 77 K) of 135 m2 g-l and a pore volume of 0.8-1.1 cm3 g-l. First, the support was impregnated at pH 7 with 1.4 cm3 (per g of SO2)of an aqueous solution of (NH4)6M07024a4H20 (Merck, p.a.). The resulting precursor was first dried in a rotary evaporator at 323 K and 200 mmHg (1 mmHg = 133.3 N and then in an oven at 383 K. Finally it was calcined for 4 h at 723 K in flowing air. Second, the impregnation was effected with an acid solution of Pr(N0J2-5H20(Fluka AG) and Bi(N03)3(Koch Light). This precursor was dried as above and calcined for 16 h at 823 K in flowing air. The final catalyst has an atomic ratio Mo:Pr:Bi of 4:0.50.5 and a concentration in MOO, of 20 w t %. Its BET specific surface area was 55.6 m2 g-l; its pore volume and mean pore radius, both determined by mercury porosimetry, were 1.31 cm3 g-l and 40.3 nm, respectively. By means of X-ray diffraction, the phases Moo3, a-Bi203,and yBi2Mo06were found. No praseodymium compound was detected in the diffraction pattern. The gases used were propene (199%), oxygen (199.98%), and helium (99.998%) from Sociedad Espaiiola del Oxigeno. Adsorption Experiments. The adsorption experiments were carried out in a conventional high-vacuum volumetric apparatus (dynamic vacuum, lo4 mmHg) 0 1987 American Chemical Society
1420 Ind. Eng. Chem. Res., Vol. 26, No. 7, 1987
*I-
A
3 t
1
0
0
100
200
P, mmHg
Figure 1. Total (empty symbols) and reversible (solid symbols) adsorption of O2at 373 K on a clean surface (circles) and on a surface with preadsorbed propene (triangles).
equipped with an MKS capacitance pressure transducer with an accuracy of mmHg. Double isotherms at 373 K (total and reversible adsorption) of O2and propene on a clean surface (after pumping for 15 h a t 673 K) and on a surface with a gas preadsorbed (after completion of an isotherm of this gas a t 373 K and pumping for 1 h at 373 K) were determined. Between isotherms of total and reversible adsorption, the sample was pumped for 1 h at 373 K. Isotherms of total adsorption of 0, a t 515, 572, and 621 K on a clean surface were also performed. After a series of isotherms (0,-propene or propene-Oz), the sample was reoxidized (50 mmHg air, 1 h, 673 K) and outgassed as above. Kinetic Measurements. The kinetic measurements were carried out in a flow system with a Pyrex-glass tubular reactor (0.9-cm inner diameter) and a coaxial thermocouple for temperature measurements in the catalytic bed. Analyses of reactants and products were made by means of a 5830 A Hewlett-Packard chromatograph using columns of zeolite 13X for O2 and CO, and Porapak Q for other gases and liquids. Catalyst samples of 0.2-0.8 g and particle sizes between 0.42 and 0.59 mm were mixed with variable amounts of S i c (-0.59-mm particle size) to keep a constant volume in the catalytic bed (5.6 cm3). The propene flow was kept higher than 1.1X mol h-'. In these conditions, no diffusional effects were observed and the total conversion (X,) did not exceed 5%. Blank runs (no catalyst) for a reactant mixture of molar composition C3H6:02:HzO:He= 20:30:20:30 a t 653 K yielded a total conversion lower than 0.03%. The influence of the contact time, temperature, and partial pressure of the reactants on acrolein formation was examined. This last variable was studied keeping constant the partial pressure of oxygen (equal to 0.28 atm; 1atm = 1.01 X lo5 N or propene (equal to 0.19 atm) and varying the partial pressure of the other reactant (0.04-0.40 atm for propene; 0.06-0.50 atm for oxygen). The partial pressure of water was constant (0.20 atm) throughout all the experiments and that of helium balance to atmospheric pressure. The temperature interval studied was 593-653 K. Initial rates for acrolein formation were calculated by linear regression of y = ax (y, percent conversion; x , contact time).
Results and Discussion Adsorption of Oxygen and Propene. Adsorption and coadsorption experiments were performed at 373 K (where the overall chemical reaction does not take place) to obtain
I
I
I
100
*O0 P, mmHg Figure 2. Total (empty symbols) and reversible (solid symbols) adsorption of propene at 373 K on a clean surface (circles) and on a surface with preadsorbed O2 (triangles).
information on the adsorption centers. The results of these experiments are depicted in Figures 1 and 2. The isotherms for the total adsorption of 02,shown in Figure 1, are of Type I (Langmuirian); those for the reversible adsorption are Type I11 (concave upward). Preadsorption of propene does not appreciably affect the total adsorption of oxygen: the isotherms for a clean surface and for a surface with preadsorbed propene are coincident. Most of the oxygen adsorption is irreversible; the reversibly adsorbed fraction is only a small fraction of the total. Again the preadsorption of propene has very little effect on the amount of reversible adsorption. The isotherms for the adsorption of propene are shown in Figure 2. These are of the Henry type; the quantity adsorbed increases linearly with propene pressure. Comparison of Figures 1 and 2 shows that total propene adsorption is much less than the total oxygen adsorption over the pressure range studied. Preadsorption of oxygen does influence propene adsorption (Figure 21, in contrast to the reverse case (Figure 1). The total adsorption of propene on a surface with preadsorbed oxygen is about 25% lower than that on a clean surface. This decrease appears to be largely in the reversibly adsorbed part, since the irreversibly adsorbed part is almost the same for a clean surface and for one with preadsorbed oxygen. The decrease is presumed to result from a steric hindrance by the oxygen molecule in the subsequent adsorption of propene. The irreversibly adsorbed fraction of propene on a clean surface is only about 25% of the total adsorption. This situation contrasts with that for oxygen, where most of the adsorption is irreversible. Adsorption experiments with oxygen at higher temperatures, near those used in the catalytic reaction, are shown in Figures 3 and 4. Figure 3 contains the isotherms for total oxygen adsorption at temperatures of 515, 572, and 621 K. Figure 4 shows the oxygen adsorption isobars at pressures of 50 and 200 mmHg. The isobars exhibit an ascending branch above 550 K; this indicates a change in the nature of the adsorbed oxygen, probably associated with the appearance of dissociatively adsorbed oxygen above this temperature. The shape of the low-temperature isotherms and the ratio of reversible to total adsorption (which is higher for the hydrocarbon) indicate that the interaction of oxygen with the catalyst surface is significantly stronger than that of propene. This conclusion is consistent with earlier views that selective oxidation catalysts yield a bond of only
Ind. Eng. Chem. Res., Vol. 26, No. 7, 1987 1421
c 9 d
-5-
-7:/
-9
/
:/ -3
-2
-1
Ln Ph
Figure 6. Rate for acrolein formation as a function of propene partial pressure: (0) 593, (A) 623, (0)653 K. 2
L
I
p, "Ha
Figure 3. Total adsorption of O2at 515 (O),572 (A), and 621 K (0) on a clean surface.
-9
I/ -3
I 2
600
T, K
650
Figure 4. Isobar of O2adsorption at 50 (A) and 200 mmHg (0). Table I. Kinetic Constants for Acrolein Formation and Apparent Order of Reaction for the Power Rate Law T,K k, mol h-' ~ , - l a B 593 623 653
2.27 x 10-3 6.59 x 10-3 20.3 x 10-3
0.90 0.97 0.98
-1
-1t\
/ 550
Ln Po
Figure 6. Rate of acrolein formation as a function of O2partial 593, (A) 623, (0) 653 K. pressure: (0)
c
'500
-2
0.39 0.21 0.10
moderate strength with the hydrocarbon (Matsuura, 1977; Tascdn et al., 1985). Reaction Rate Equation. The experimental results for the rate of acrolein formation as a function of the partial pressures of reactants have been fitted to a power rate law, eq 1. Figures 5 and 6 are rectifying In-In plots: Figure ra = kPhaP,B (1) 5 shows the variation of rate with propene pressure, and Figure 6 shows the dependence on oxygen pressure. The calculated values of the kinetic parameters k , a,and 0 are listed in Table I. The apparent order of reaction with respect to propene is approximately equal to 1 and shows little variation with
i-5B -'I. 5
1.6 io3/ T,K-'
1.7
Figure 7. Kinetic constant as a function of the temperature: (0) k in power rate law; (A)k,; (0) k,, = hada
temperature over a 60 K range. This first-order behavior is expected on the basis of the relatively weak bonding of propene to the surface (see adsorption results above). The apparent order of reaction with respect to oxygen, however, is relatively low and decreases, from 0.4 to 0.1, as the temperature increases over the range from 593 to 653 K. Keulks and co-workers (Krenzke and Keulks, 1980; Monnier and Keulks, 1981; Keulks and Krenzke, 1983) found the apparent orders of reaction for propene and oxygen to be a function of temperature, both being lower than 1below 673 K and equal to 1 and 0, respectively, above this temperature. Our findings are reasonably consistent with those of Keulks and co-workers. In Figure 7 an Arrhenius plot (In k vs. 1/T) of the variation of rate constant with temperature is given; the fit to a straight line is quite good. The apparent activation energy is calculated to be 117 kJ mol-'. This value is similar to that found for propene oxidation catalyzed by USb3OI0(Keulks and Krenzke, 1983) but considerably lower than that reported over bismuth molybdate (Krenzke and Keulks, 1980). Kinetic Models. The experimental data have also been fitted to some kinetic models often used for catalytic ox-
1422 Ind. Eng. Chem. Res., Vol. 26, No. 7, 1987 Table 11. Fits of the Experimental Results for Variable Propene Partial Pressure type of linearization In rn = cy In p h + B h (Ph/r,)'/'
l/rn =
= AhPh
model 1
+ Bh
2a, 2b
+ Bh
Ah(l/Ph)
" Units depend
3a, 3b, 4 5a, 5b 6a, 6b
T,K
Aha
593 623 653 593 623 653 593 623 653
0.90 0.97 0.98 10 2.5 0.8 558 184 53.4
22
13.40 7.24 55.40 36.80 5.67
correlation coeff 0.990 0.997 0.999 0.799 0.592 0.637 0.993 0.999 >0.999
F 200 772 1884 7.04 2.16 2.73 280 1853 6756
IL 0.17 0.09 0.06 0.74 0.99 0.94 0.15 0.06 0.03
on the chosen kinetic equation.
Table 111. Fits of the Experimental Results for Variable Oxygen Partial type of linearization model T,K A," In rs = fl In Po + Bo 1 593 0.39 623 0.21 0.10 653 1266 l / r , = A , , ( l / P ~ / 2 )+ Bo 3b, 5b 593 6b 623 181 653 22 l / r , = Ao(l/P0) + E, 3a, 5a 593 236 6a 623 33.60 653 4.08 ra = '42'0 4 593 0.001 623 0.003 653 0.010
" Units depend
Bh' -6.58 -5.32 -4.04
Pressure Bo" -7.55 -6.59 -5.50 705 593 233 2198 806 259 0 0 0
correlation coeff 0.991 0.982 0.982 0.995 0.993 0.996 >0.999 0.997 0.997 0.675 0.682 0.691
F 211
108 110 378 302 484 4983 607 774 1.56 4.30 4.90
11. 0.17 0.23 0.23 0.13 0.14 0.11 0.03 0.10 0.09 1.32 3.10 7.60
on the chosen kinetic equation.
idation. The first of these is the Langmuir-Hinshelwood model (Hinshelwood, 19401, which assumes that the controlling step is the surface reaction between the two adsorbed reactants. Two cases are considered. If propene and oxygen compete for the same adsorption sites, the model leads to eq 2 for the rate law. In this equation, n
equals 1 if oxygen does not dissociate (model 2a) or 'Iz if oxygen adsorption is dissociative (model 2b). If, on the other hand, adsorption of oxygen and propene takes place on different sites (noncompetitiveadsorption), the Langmuir-Hinshelwood model leads to eq 3. Again, n = 1 for nondissociative adsorption of oxygen (3a), and n = ' I 2for dissociative adsorption (3b).
ra =
k K h P h( K z o )
[1 + K h p h l [ 1
+ (KPo)"]
(3)
In the Eley-Rideal mechanism (Bond, 1962), the controlling step is assumed to be the reaction between adsorbed propene and gaseous oxygen. The rate equation then becomes eq 4. ra =
KhPhPo
1+ K h p h
+ (KZo)"
(4)
In the redox mechanism (Mars and van Krevelen, 1954), the oxidation proceeds stepwise: the hydrocarbon is oxidized by the (oxidized form of the) catalyst surface, and the oxidized form of the catalyst surface is regenerated in subsequent reoxidation of the reduced surface by gaseous oxygen. The steps reach equal rates in the stationary state, and the resulting rate equation for acrolein formation is eq 5. The catalyst reduction step is generally assumed r, =
kThXkoxPoY k?h"
-k koxPoy
(5)
to be first order with respect to propene (i.e., x = 1) (Uda
et al., 1980; Baussart et al., 1982); orders of 1 (Uda et al., 1980; Baussart et al., 1982) or 1/2 (Viswanathand Chanda, 1978; Ono and Kubokawa, 1982) with respect to oxygen have been reported for the reoxidation step. For this reason, two cases are considered here: x = 1,y = 1 (model 5a) and x = 1,y = 'Iz (model 5b). The "stationary state of adsorption" model (Jaswal et al., 1969) is similar to the redox model; the difference is that in the former case, the adsorbed oxygen, rather than lattice oxygen, participates in the reaction. The rate equation for this model is
re =
kPhkadsPon akPh
+ kadsPon
where n = 1 (6a) or (6b) for nondissociatively or dissociatively adsorbed oxygen, respectively, and a is the number of oxygen atoms required by the reaction stoichiometry. The experimental data for acrolein formation have been fitted to linearized forms of the various rate equations given above. Our experimental measurements are for initial rates, and the partial pressure of water vapor was kept constant in all runs. In order to compare the different models, the correlation coefficients, Fisher's F (Fisher, 1955) and Exner's J/ (Exner, 1966)) were chosen as parameters for discrimination (better fits are indicated by lower J/ and higher F values, respectively). The influence of propene (or oxygen) partial pressure was examined by fitting the rate data vs. p h (or Po),keeping Po (or ph) constant. Tables I1 and I11 summarize the results of the data fitting for the several models. Table I1 lists the discrimination parameters for experiments in which propene pressure was varied; Table I11 does the same for those runs in which oxygen pressure was varied. The models corresponding to the linearized rate equations are listed in column 2 of each table. The best fits in Table I1 (variable P h ) are for the last group, which includes models 3a, 3b, 4,5a, 5b, 6a, and 6b. The best fits in Table I11 (variable Po) are for the third
Ind. Eng. Chem. Res., Vol. 26, No. 7, 1987 1423 Table IV. Acrolein Formation T,K ka 593 25 (4) x 10-3 623 33 (9) x 10-3 653 192 (20) x 10-3
Kh, atm-l 9.93 (0.3) X 20.0 (5.5) x 10-2 10.6 (3.2) X lo-'
KO,atm"
k,b
9.31 (0.04) 24.0 (1.6) 63.5 (4.7)
1.79 (0.26) X 5.43 (0.30) x 10-3 18.7 (0.60) X
kox = kadsb 4.24 (0.14) X 29.7 (2.9) X 245 (21) x 10-3
In mol of acrolein h-I gcacl. In mol of acrolein a t d h-' gcac'
e/
af
I2 10
,b,,10-4m~(e~ ti' g;b, Figure 8. Calculated vs. experimental rate for acrolein formation for models 1 ( O ) , 3a (A), and 5a and 6a (A).
group, models 3a, 5a, and 6a. The fit to model 1, the power rate law, is not as good; however, this model would rank second in both tables. Mechanistic Considerations. Although data fitting of kinetic parameters is not a reliable method of determining the mechanism for catalyzed reactions, it is instructive to consider possible implications of the analysis just given. Of those models listed as giving the "best fits" in Tables I1 and 111,only models 3a, 5a, and 6a are in the top group for both tables. Table IV lists the values of the kinetic parameters for these three models, along with their confidence limits (in parentheses). The rates of acrolein formation calculated from these parameters for the three models are plotted in Figure 8 against the measured (experimental) rates. The rates for model 1calculated from Table I are, also, represented. All four models show a reasonable fit to the expected diagonal, over almost 2 orders of magnitude. A further measure of consistency is the behavior of an Arrhenius plot for the rate constants. Figure 7 contains such plots for models 1, 5a, and 6a; these show the expected linear variation of In k , In k, and In k , (kox= kab/a) with 1/T. The corresponding plots for model 3a, In k and In kh vs. 1/T, are not shown because they deviate substantially from linearity; this model is therefore discarded. The apparent activation energies deduced from the slopes of the lines in Figure 7 are 125 f 3 kJ for reduction of the catalyst and 216 f 3 kJ for reoxidation of the catalyst. Krenzke and Keulks (Krenzke and Keulks, 1980), in comparing their kinetic data over Bi2Mo06with those of Uda et al. (Uda et al., 1980), concluded that propene oxidation is controlled by reduction of the catalyst at high temperatures and by reoxidation of the catalyst at low temperatures. In our case, k,, (=kads) > k,, and the difference becomes larger as the temperature increases. If one assumes that ko,Poy > k,PhXin eq 5 or kadZon> ak,Ph in eq 6, eq 5 and 6 can be transformed into the simple expression ra = k g h x
(7)
where x = 1in eq 6. Equation 7 , which is a power rate law in form, applies when the apparent order of reaction for oxygen tends to zero.
Our results illustrate a major caution that has to be observed when the methodology of "best fits" to models is used to rule in or out particular models. As the discussion above makes clear, the most satisfying fit to the experimental kinetic data for propene conversion was given by models 5a and 6a. Both of these are based on nondissociative adsorption of oxygen. By contrast, the experimental adsorption data for oxygen shown in Figures 3 and 4 clearly imply that at temperatures above about 570 K (i.e., in the range where reaction kinetics were studied), dissociative adsorption occurs. Furthermore, the strong bonding of oxygen to the catalyst at elevated temperatures is also consistent with chemisorption of oxygen atoms, not physical adsorption of oxygen molecules. We conclude that the data- and model-fitting procedures described are useful in arriving at rate equations that best describe the global data for kinetics but that it is hazardous to push this analysis to conclusions about detailed mechanism. Of interest are the facts that (a) a power rate law, eq 7 , can be deduced if the question of oxygen dissociation is obviated by zero-order dependence on oxygen pressure and (b) the power rate law, eq 1, does fit the kinetic data quite well, as shown in Tables I1 and 111. Acknowledgment We are indebted to CAICYT and CSIC for sponsorship of this work. Nomenclature A,, B, = slope and origin ordinate of the linearized rate expression for variable partial pressure of reactant k = kinetic constant for acrolein formation, mol h-' gcaL1 k e d s = kinetic constant for oxygen adsorption k,, k,, = kinetic constant for reduction or reoxidation of the catalyst, mol of C3H,0 atm-' h-l gCac1 K, = equilibrium constant for adsorption of reactant i P, P, = equilibrium pressure (mmHg) or partial pressure of reactant i (atm) q = gas adsorbed, molecule g-' T = temperature, K r, = initial rate for product i formation, mol produced h-' gaC1 W / F = contact time, g,,, h (mol of propene-') XT = total conversion of propene, ?& Greek Symbols a , 6 = apparent orders of reaction for propene and oxygen Subscripts
a = acrolein h = propene i = reactant or product o = oxygen w = water Registry No. 02, 7782-44-7; Mo, 7439-98-7; Bi, 7440-69-9; Pr, 7440-10-0; HZC=CHCH,,
115-07-1; HZC=CHCHO, 107-02-8.
Literature Cited Aso, I.; Furukawa, S.; Yamazoe, N.; Seiyama, T. J. Catal. 1980, 64, 29. Baussart, H.; Delobel, R.; Le Bras, M.; Le Maguer, D.; Leroy, J. M. J. Chem. SOC.,Faraday Trans. 1 1982, 78, 485. Bond, G. C. Catalysis by Metals; Academic: New York,1962; p 128. Cartlidge, J.; McGrath, L.; Wilson, S. H. Trans. Inst. Chem. Eng. 1975, 53, 117.
1424
Ind. Eng. Chem. Res. 1987, 26, 1424-1434
Corti% Corberin, V.; Corma, A.; KremeniE, G. Ind. Eng. Chem. Prod. Res. Deu. 1985, 24, 62. Crozart, M.; Germain, J. E. Bull. SOC.Chim. Fr. 1973, 1973, 2498. Cullis, C. F.; Hucknall, D. J. Catalysis; Bond, G. C., Webb, G., Eds.; The Royal Society of Chemistry: London, 1982; Vol. 5, Chapter 7. Exner, 0. Collect. Czech. Chem. Commun. 1966, 31, 3222. Fisher, T. Statistical Methods for Chemists; Wiley: New York, 1955. Hinshelwood, C. N. The Kinetics of Chemical Change; Oxford University Press: New York, 1940; p 207. Hucknall, D. L. Selective Oxidation of Hydrocarbons; Academic: London, 1974. Isaev, 0. V.; Margolis, L. Y. Kinet. Katal. 1960, 1, 237. Jaswal, I. S.; Mann, R. F.; Junsola, J. A.; Downie, Y. Can. J. Chem. Eng. 1969, 47, 284. Keulks, G. W.; Yu, Z.; Krenzke, L. D. J . Catal. 1983, 84, 38. Keulks, G. W.; Rosynek, M. P.; Daniel, Ch. Ind. Eng. Chem. Prod. Res. Deu. 1971, 10, 138. Khiteeva, V. M.; Rzakulieva, Sh. M. Russ J . Phys. Chem. 1981,55, 1202. Klissurski, D. G. Proceedings of the Climax Third International Conference on the Chemistry and Uses of Molybdenum; Barry,
H. F., Mitchell, P. C. H., Eds.; Climax Molybdenum Company: Ann Arbor, MI, 1979; p 123. Krenzke, L. D.; Keulks, G . W. J . Catal. 1980, 64, 295. Mars, P.; van Krevelen, D. W. Chem. Eng. Sci. (Spec. Suppl.) 1954, 3, 41. Matsuura, I. Proceedings of the 6th International Congress on Catalysis; Bond, G. C., Wells, P. B., Tompkins, F. C., Eds.; The Chemical Society: London, 1977; p 819; Vol. 2. Minachev, Kh. M.; Antoshin, G. V.; Guin, N. K.; Klissurski, D. G.; Abadzhijeva, N. Ts. Chem. Phys. Lett. 1977,48, 515. Monnier, J. R.; Keulks, G. W. J . Catal. 1981, 68, 51. Ono, T.; Kubokawa, Y. Bull. Chem. Soc. Jpn. 1982, 55, 1748. Takasu, Y.; Matsui, M.; Tamura, H.; Kawamura, S.; Matsuda, Y.; Toyoshima, I. J . Catal. 1981, 69, 51. Tasc6n, J. M. D.; Cort6s CorberBn, V.; KremeniC, G.; Gonzglez Tejuca, L. J . Colloid Interf. Sei. 1985, 106, 269. Uda, T.; Lin, T. T.; Keulks, G. W. J . Catal. 1980, 62, 26. Viswanath, V. J.; Chanda, M. J . Appl. Chem. Biotechnol. 1978,28, 1. Received for review September 19, 1986 Accepted March 6, 1987
Modeling of Exothermic Solid-Solid Noncatalytic Reactions Jan Puszynski, Jan Degreve, and Vladimir Hlavacek* Department of Chemical Engineering, State University of New York at Buffalo, Buffalo, New York 14260
A certain class of refractory materials can be synthesized in the sintering or in the self-propagating regime by a direct reaction between the metal powder and carbon, boron, or silicon. A detailed analysis and modeling of such processes are given. Criteria are derived for stable, unstable, and degenerated combustion as well as approximate relations for the velocity of a constant-pattern profile, as a function of the heat-transfer coefficient. The approximate formulas are verified by a detailed numerical study. Different types of combustion (constant-pattern profiles, standing waves, and complex oscillatory behavior) have been observed. There exists an important class of heterogeneous noncatalytic reactions of the type solid-gas and solid-solid which are accompanied by a substantial generation of heat. The most important are oxidation of metals and nonmetallic compounds, direct reaction of nitrogen or hydrogen with certain metals, synthesis of carbides, borides, sulfides, or silicides of metals from pure components, reduction of metallic oxides by aluminum, synthesis of intermetallic compounds (e.g., Ni-AI), combustion of solid propellants, and many others. In several of the examples mentioned above, the process takes place as follows: heat of reaction or externally transferred heat evaporates the material, the reaction occurs in the gas phase, and the product is obtained as a solid or a liquid. In other cases, the exothermic chemical reaction results in an excessive gas-phase formation (oxidation of propellants). In this paper, we will address reactions occurring without gasphase formation. Such reactions are sometimes referred to as gasless combustion reactions. Many of these reactions are of major technological importance. Reduction or roasting of ores, thermite reactions in ordnance applications, and synthesis of new refractory materials from transition or rare earth metals and boron or carbon are a few examples. The exothermic noncatalytic reaction can proceed through the reacting mixture in a uniform way, and the concentration of the components at any instant is almost the same everywhere (the kinetic or “sintering” regime). However, if the system was locally ignited by external 0S8S-5885/87/2626-1424$01.50/0
means, a steep reaction front can propagate through the mixture. Evidently, the latter phenomenon is closely related to flame propagation in gaseous mixtures and is frequently referred to as the self-propagating regime. Synthesis of borides and carbides of transition and rare earth metals can be performed both in sintering and self-propagating regimes. The synthesis in the sintering regime requires a substantial energy input, a high reaction temperature, and a long reaction time. On the other hand, the self-propagating synthesis calls only for a short-range initial energy input. The reaction is self-sustaining owing to the high values of the activation energy and heat of reaction (Merzhanov, 1974). Recent experimental observations by Novikov et al. (1974) and Maksimov et al. (1981) and asymptotic analysis by Margolis (1983) reveal that there may exist several types of propagating waves (constant-pattern profiles, oscillatory waves, and spinning and fingering fronts). However, in this paper we will analyze only the one-dimensionalsituation. Breaking of symmetry in the transverse direction will be the topic of another paper. There is extensive combustion-oriented literature on propagation of fronts in reacting systems. Many of the results can be applied to the theory of noncatalytic exothermic reactions in solid-solid and solid-gas systems. In addition, there is a recent extensive treatment of combustion systems based on activation energy asymptotics. In the present paper, we make an attempt to amalgamate the dispersed information on propagation phenomena in 0 1987 American Chemical Society
