1791
Znd. Eng. Chem. Res. 1991,30, 1791-1795 Rewatkar, V. B.; Raghava b o , K. S. M. S.; Joehi, J. B. Some Aspects of Solid Suspension in Mechanically Agitated Reactors. MChE
J. 1989,35, 1577. Rewatkar, V. B.; Joshi, J. B. Effect of Impeller Design on Liquid Phase Mixing in Mechanically Agitated Reactors. Chem. Eng. Commun., 1991, in press. Smith, D. N.; Ruether, J. A.; Shah, Y.T.; Badgujar, M. N. Modified Sedimentation-Dispersion Model for Solid in a Three-phase Slurry Column. AZChE J. 1986, 32, 426. Subbarao, D.; Taneja, V. K. Three Phase Suspension Agitated Vessels. h o c . 3rd Eur. Conf. Mixing 1979, 1, 229.
Voit, H.; Mersmann, A. General Statement for the Minimum Stirrer Speed During Suspension. Cer. Chem. Eng. 1986,9, 101. Weismann, J.; Efferding, L. E. Suspension of Slurries by Mechanical Mixtures. AIChE J. 1960,6,419. Wichterle, K. Conditions for Suspension of Solids in Agitated Vessels. Chem. Eng. Sci. 1988, 43, 467. Zwietering, T. N. Suspending of Solid Particles in Liquid by Agitators. Chem. Eng. Sei. 1958,8, 244.
Received for review August 14, 1990 Accepted February 25, 1991
Kinetics of Anthraquinone Reduction with Sodium Sulfide in Alkaline Medium Julio F.Tijero, Mercedes Oliet, Jose C. Burillo, and F r a n c i s c o Rodriguez* Departamento Ingenieria Quimica, Facultad de Quimica, Universidad Complutense, 28040 Madrid, Spain
The kinetics of anthraquinone reduction with sodium sulfide in alkaline medium has been studied in a reactor by performing kinetic runs at different temperatures, sodium sulfide concentrations, initial average radius of anthraquinone particles, and sodium hydroxide concentrations. The product was the disodium salt of 9,lO-dihydroxyanthracene. The kinetic data treatment was carried out by using the isothermal shrinking-core model for solid-fluid reaction with changing particle size. A reaction rate expression was proposed. The kinetic equation is first order in relation to sodium sulfide concentration. The studied anthraquinone reduction can be classified as having a sufficiently high reaction rate, and the surface chemical reaction is the controlling step of the overall process. The use of quinone compounds are redox catalyst additives in the pulp and paper manufacture is one of the most successful industrial improvements in this sector lately. The early studies about the efficiency of these compounds were realized by Holton and Chapman in 1977 (Holton, 1977; Holton and Chapman, 1977). The results showed that the quinone compounds, with the exception of the benzoquinones, improve the wood delignification rate during soda and kraft cookings, with anthraquinone (AQ) and its alkylate derivatives having the best performance in that sense. Later, results similar to Holton’s were attained by many other authors (Blain, 1979; Lmhenal, 1979; Lachenal et al., 1979; McDonough and Van Drunen, 1980; Fossum et al., 1980a,b;Bhandari et al., 1982; Wandelt and Surewicz, 1983; El-Saied et al., 1984; Irvine and Nelson, 1986; Haldar and Bhattacharya, 1987; Eagle and McDonough, 1988; Nelson and Irvine, 1989).. The analysis of the AQ mechanism has shown that its reduced form in alkaline medium, the dianion of 9,lOdihydroxyanthracene (AQ2-),soluble in these conditions, is the active agent of the catalytic process. Subsequently, the direct use of 9,10-dihydroxyanthracene,as disodium salt in the the studied case, instead of AQ presents several additional advantages. Fullerton (1978,1979)was the first to propose this substitution, which allows as the main advantage introduction of the additive in dissolution, improving its contact with the wood and consequently reducing the catalyst induction time. The aim of this work is to study the heterogeneous solid-liquid reaction kinetics between AQ and sodium sulfide in an alkaline medium. The AQ reduction viability has been previously verified with the use of sodium dithionite as reduction agent in an alkaline medium (Burillo et al., 1988). In this case the AQ reduction compound employed was sodium sulfide in dissolution of sodium hydroxide, and in concentrations similar to liquors of the kraft pulping process, with a view to its integration on an industrial scale.
Runs with different size particles of AQ and sodium sulfide/ hydroxide mixtures of several strengths were carried out in a pressure reactor in the 125-180 “C temperature range. Kinetic data were fitted according to the isothermal shrinking-core model (SCM) for spherical particles with no interference due to formation of a layer of material on the surface and with the surface chemical reaction as the controlling step of the overall process. Rate constants were experimentally determined for the above temperature range. The activation energy was obtained from the Arrhenius law and was found to be 63.4 kJ/mol. The application of the SCM to solid-liquid systems is an interesting fact since limited information is available from the literature. Theoretical Section Thermodynamic Equilibrium. The reaction between sodium sulfide and AQ in an alkaline medium can be represented by the equation 2Na2S + 4AQ + 6NaOH Na2S203+ 4Na2AQ + 3H20 (1)
-
where both sodium sulfide and sodium thiosulfate are in solution and AQ is a solid. This is a redox process where the following half-reactions are involved: 2S2- + 60HSz032-+ 3H20 + 8eEo=-0.006 Q
(2)
AQ + 2e- AQ2(3) Taking into account the standard reduction potentials and the Nernst equation, the equilibrium constant can be calculated as K = (nF/RT)(EoAQ/AQ” - EO S2032-/Sz-) (4) Q
There are no data available derived from the literature for the standard potential of the AQ/AQ2-system although
0888-5885/91/2630-1791$02.50/00 1991 American Chemical Society
1792 Ind. Eng. Chem. Res., Vol. 30, No. 8, 1991
the existence of dianion in an alkaline medium has been proved elsewhere (Landucci, 1980). For its estimation the following reactions have been considered (Ksenzhek et al., 1977; Bard, 1975):
+ eAQ- + eAQ
--
AQ-
Eo = -0.520 V
(5)
AQ2-
Eo = -0.585 V
(6)
@ vz
DL
i
assuming that the last system has a standard potential similar to the standard potential of l-chloroanthraquinone anion as (Ksenzhek et al., 1977) 1-ClAQ- + e-
-
1-C1AQ2-
(7)
This assumption is supported by the fact that the difference between the standard potential of AQ/AQ- and AQ-/AQ2- systems, 0.065 V, is next to values determined by Ksenzhek et al. (1977) for the same equilibrium in other solvents. The relationship between the water ionic product, pKw, and the temperature can be derived from the data of Gustafsson and Teder (1969) and Patai and Rappoport (1971). According to resultant correlation pKw = 14.629 exp(-1.82
X
10-3T/0C)
(8)
at 100 "C a pKw value of 12.2 is obtained. For a sodium hydroxide concentration of 1 M (pKw = pH), the values of equilibrium conditional constants (pH function) of the AQ/AQ- and AQ-/AQ2- systems at 100 "C are 4.86 X 10'' and 2.30 X lo9, respectively. Therefore, it can be considerated that under these conditions the reaction between AQ and sodium sulfide in alkaline medium is practically displaced toward the AQ2- formation and the reverse reaction is neglected. Exhaustive information on the thermodynamics of this reaction can be obtained from Burillo's work (1988). Description of Kinetic Model. Since the solubility of AQ in aqueous media is extremely low, even at high temperatures, the associated solid diffusion phenomena in the liquid film can be rejected. Obviously, chemical reaction either in the liquid film or in the bulk never takes place. The system studied in this paper deals with a solid-liquid heterogeneous noncatalytic procea. The solid phase was nonporous and the liquid phase was a sulfide/ sodium hydroxide solution. In an alkaline medium the reaction products are all in solution, so the formation of a layer of material on the surface does not take place during a run. The kinetic data treatment was carried out by using the isothermal SCM for solid-fluid reaction with changing particle size (Levenspiel, 1972). The following reaction scheme has been considered:
A (fluid) + bB (solid)
-
C + D (fluids)
(9)
where b is the stoichiometric coefficient of the solid dispersed phase of AQ. In the development of the model, overall process rate control by surface chemical reaction was assumed. Taking into account the pseudo-steady-state approximation and integrating the mass flow balance equations for spherical solid particles, the reaction time was obtained as a function of the solid fractional conversion, X g : t/TQ
= 1 - (1 - XB)lI3
(10)
where the complete conversion time of the particle, T ~ is, given by TS
= (P&
/ (&$A")
(11)
P
O1 SAMPLING DRAW
Figure 1. Scheme of the equipment employed for kinetic rune.
When diffusion is the controlling step, the reaction time, as a function of X g , is t / T D = 1 - (1 - XB)'l3 (12) where rDis the complete conversion time. A check of the proposed model was carried out by determining if the influence of the different kinetic variables on the required complete conversion time was in agreement with the model-predicted influence.
Experimental Section Equipment and Reaction Procedure. The equipment used to perform the experimentation, represented in Figure 1, has two parts: the reaction system and the feed line of the sodium sulfide solution. The reaction system consists of a 2-L stainless steel pressure reactor, provided with sampling draw, measurement and control pressure, heating system, agitation, and measurement and control temperature. Once the AQ suspension in alkaline solution reachea the fixed temperature, the sodium sulfide solution is fed from the auxiliary tank by overpressure. In order to avoid the oxidation of the reduction product, air was purged during the heating phase of both vessels by means of El and E2 solenoid valves. The temperature never varied more than 0.5 "C during any of the experiments performed. Analysis. The AQ fractional conversion was calculated from the disodium salt of 9,lO-dihydroxyanthracene (NagiQ) analysis. The direct measurement of this reaction product is difficult due to its easy oxidation in the presence of air (Armentrout, 1981). To deal with this difficulty, the equivalent amount of AQ was computed by quantitative oxidation of Na2AQwith dilute hydrogen peroxide, which gives rise to the precipitation of AQ. A filtered sample of known volume is withdrawn from the reactor and oxidized to AQ, which is separated by vacuum filtration, dried, and dissolved in cyclohexanone. Afterward, the samples were analyzed by high-performance liquid chromatography (Rodriguezet al., 1989). The analysis of the sodium sulfide solutions has been carried out by iodometry. Operating Conditions. The variables studied and their application ranges are given in Table I. In a preliminary run carried out with stoichiometric amounts of reactants for reaction 1, 9,lO-dihydroxyanthracene deposition was detected due to a medium pH drop below 9.7. In order to ensure the absence of a solid reaction product, a value of a 1.0 M sodium hydroxide concentration was chosen as minimum, which exceeds the stoichiometric amount. The fractional conversion function, f(XB) = 1 - (1 X,)1/3,was obtained and plotted vs reaction time. From this representation, the complete conversion time of the solid can be derived. According to eq 10, ita value corresponds to the slope of the straight line.
Ind. Eng. Chem. Res., Vol. 30, No. 8, 1991 1793 Table I. Owratina Conditions of the Exwriments Performed concn of agitation speed, rpm temp, O C expt AQ, M X 1 2.4 1200 165 4.8 2 1200 165 165 7.2 3 1200 2.4 4 lo00 165 2.4 5 1200 165 2.4 1400 6 165 2.4 7 1200 i25 2.4 8 1200 145 2.4 9 1200 165 2.4 10 1200 180 2.4 11 1200 165 2.4 12 1200 165 2.4 165 13 1200 2.4 165 14 1200 2.4 15 1200 165 2.4 165 16 1200 2.4 17 1200 165
le
Table 11. Values of Complete Conversion Time and Rate Constant at Different Owrating Conditione complete conversion rate const, time, min cms-1 x 10s expt 1 159.5 7.02 156.3 2 7.15 154.6 7.23 3 25.6 4 6.82 26.6 5 6.56 27.7 6 6.29 311.9 7 1.28 123.9 3.22 8 51.1 9 7.82 29.9 10 13.34 75.4 11 7.06 78.7 12 7.15 118.0 13 7.38 7.41 14 53.9 7.81 15 51.1 13.54 16 91.1 23.03 17 90.6
Results and Discussion When the SCM is applied for particles of changing size with chemical reaction as the rate-controlling step, film diffusion must be eliminated; in addition, the constancy of fluid reactant concentration has to be satisfied. In order to verify whether these conditions are attained in the operating ranges of the variables studied, two sets of runs described below were carried out. Agitation Rate Effect on Solid Fractional Conversion. When diffusion control predominates, a rise in the agitation rate leads to a solid-liquid mass-transfer coefficient increase and consequently a higher fractional conversion is attained. From Table I1 (experiments 4-6) it can be clearly seen that the complete conversion time is practically unaffected by varying the agitation speed. This fact shows that the surface chemical reaction controls the overall process in the agitation speed range studied. In accordance with these results, a stirrer speed of 1200 rpm was chosen for all the experiments. Dispersed Phase Fraction Effect on Solid Fractional Conversion. During a run, sulfide concentration constancy has to be supported. This condition is obtained when operating with sufficiently high Na2S/AQ initial concentration ratios. A set of runs (experiments 1,2, and 9) with several AQ amounts for the minimum sodium sulfide concentration chosen was carried out. Table I1 shows that AQ complete conversion time is also unaffected by the solid dispersed-phase fraction in the tested range. A minimum AQ concentration value of 2.4 X M was selected for every experiment. This con-
concn of Na2S, M X 10 1.0 1.0 1.0
2.8 2.8 2.8 2.8 2.8 2.8 2.8 2.1 2.8 2.8 2.8 2.8 2.8 2.8
solid particle radius, cm 2.20 2.20 2.20 0.96 0.96 0.96 2.20 2.20 2.20 2.20 2.20 3.10 4.80 2.20 2.20 6.80 11.50
concn of NaOH, M 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.0 2.0 1.5 1.5
centration assures an excess amount of sulfide with respect to the solid sufficient to consider CN, constant throughout a test. Effect of Variables. In order to verify the model validity, the influence of variables on the complete conversion time was studied. Afterward the kinetic constant as a temperature function was calculated. The reaction rate constant, ks,is an exponential function related to temperature. Therefore, a plot of In T~ vs 1/T should yield a straight line. Any variation in the slope of this line will obey either a kinetic equation change or a change in the controlling step of the overall process rate. From data of Table I1 (experiments 7-10) it can be established that the relationship between In T~ and 1/T is linear, which indicates that the overall process rate is controlled by the surface chemical reaction step, confirming the validity of the proposed model in the investigated temperature range. The SCM with chemical reaction as the control step predicts an inverse relation between sodium sulfide concentration and complete conversion time, as eq 11 indicates. It can be seen from data of Table I1 (experiments 1,9, and 11)that the representation of complete conversion time against the inverse of the sodium sulfide concentration gives rise to a straight line that passes throught the origin. This behavior is in accordance with the model predictions in relation to the fluid reactant concentration, proving its validity to the range studied. On the other hand, if the logarithm of complete conversion time is represented against the logarithm of inverse of sodium sulfide concentration, a straight line with a slope of 1.086 is obtained, which means according to eq 11 that the reaction is first order with respect to sodium sulfide concentration. According to eq 11, complete conversion time is also a linear function of the initial average solid particle radius. From Table I1 (experiments 6,9, 12, 16, and 17) T~ vs the average particle radius at 165 "C can be represented. I t can be seen that for an average particle radius smaller than approximately 50 pm the system behavior agrees well with the spherical geometric model chosen. However, for an initial particle radius greater than 50 l m , the proposed model is inappropriate. The possibility that larger particles are not spherical, but perhaps rod shaped with a 50-pm radius, has been rejected by microscopy and in application of the SCM for particles of cylindrical geometry. Therefore, this fact can be explained by presupposing that relative large crystalline particles are easily fragmented along cleavage planes. Consequently, a different and irregular particle size dis-
1794 Ind. Eng. Chem. Res., Vol. 30, No. 8, 1991 Table 111. Temwrature Effect on Reaction Rate Constanta temD. OC rate const. cms-l x 10a 125 1.28 145 3.22 165 7.14 180 13.34 a To calculate the rate conetanta, experiments with initial average solid particle radius greater than 50 pm were rejected.
tribution is obtained. It was therefore presumable that the value of 50 pm would be the minimum solid radius where the particle break takes place under the studied conditions. Reaction model selection was made taking into account that alkali strength affects only the equilibrium, displacing it to the AQ reduction. Therefore, when chemical reaction is the controlling step, the complete conversion time should be independent of sodium hydroxide Concentration. In Table I1 (experiments 9,14, and 15) it can be verified that in the selected concentration range there is not a remarkable variation in the complete conversion time. The difference between the maximum value obtained for 1.0 M and the rest is less than 6%. Thus, according to the selected model, the complete conversion time does not depend on the sodium hydroxide concentration. Kinetic Parameter Estimation. The rate constant ks can be worked out by substituting the complete conversion time, rS (s), and the corresponding experimental kinetic variables in eq 11. As described above, runs carried out with an initial average solid particle radius greater than 50 pm do not satisfy the conditions imposed by the model. In consequence, its practical application is restricted to the range where the data are in concurrence with the kinetic model. Hence these values were rejected when both the rate constants and the activation energy were calculated. The stoichiometric coefficient, b, takes a value of 2, and the solid molar density when the reaction starts is 6.907 M. Once the different kS values were calculated, the average values for several temperatures were determined. Table I11 shows the rate constants obtained in accordance with Arrhenius law. The activation energy in the investigated range was found to be 63.4 kJ/mol. Conclusions The analysis of the results was made according to an isothermal SCM for spherical particles with no interference due to formation of a layer of material on the surface. It was shown that the average solid particle size effect agreed well with the model only in the 22-50-pm particle radius interval. Rate constants were worked out for temperatures from 125 to 180 "C and their values varied from 1.28 X lo4 to 13.34 X lo* cm/s, the activation energy being 63.4 kJ/ mol. From the results it can be concluded that, in AQ reduction with sodium sulfide in an alkaline medium, high ratio conversions are reached at relative short times. Furthermore, the SCM is applicable to solid-liquid systems, with the operating conditions of the pseudosteady-state approximation. Bischoff (1963) and Wen (1968) have found restriction limits in applying the pseudo-steady-state assumption to this kind of system. For gases and very dilute solutions CA/pB < 0.001 is satisfied. In the present work this relation is in 0.01-0.06 range; nevertheless, a good agreement between complete conversion times obtained and those predicted by the model was attained. This agreement supports the pseudosteady-state approximation used and is in concurrence with the Lindman and Simonsson (1979) work in applying the SCM to solid-liquid systems.
Nomenclature A = sodium sulfide reactant in eq 8 B = anthraquinone reactant in eq 8 b = stoichiometric coefficient of AQ reactant = 2 C = product in eq 8 CA = liquid-phase concentration, M or mol/cm3 CAQ= dispersed-phase fraction, M or mol/cm3 dissolution CNeOH= sodium hydroxide concentration, M or mol cm3 ,C = sodium sulfide concentration, M or mol/cm D = product in eq 8 E, = activation energy, kJ/mol Eo = reduction standard potential, V F = Faraday's constant K = thermodynamic equilibrium constant k s = surface chemical reaction rate, cm/s n = number of electrons or reaction order referred to sodium sulfide concentration R = gas constant Ro = AQ particle initial radius, cm t = reaction time, s or min T = temperature, "C or K V, = stirrer speed, rpm XB = solid fractional conversion
I
Greek Letters
solid molar density, mol/cm3 complete conversion time with diffusion control, min complete conversion time with chemical reaction control, min
pB =
TD = TS =
Registry No. AQ, 84-65-1;Na2S, 1313-82-2.
Literature Cited Armentrout, D. N. Liquid Chromatography Determination of Anthraquinone in Soda and Kraft Pulping Liqour, Pulp, Air Filters, and Wastewater. Tappi 1981,64(9),165-169. Bard, A. J. Encyclopedia of Electrochemistry; Marcel Dekker: New York, 1975;Vol. 4,p 277. Bhandari, K. S.; Srivastava, A.; Singh, S. P.; Sharma, Y. K. A Preliminary Note on Alkaline-Anthraquinone Pulping of Eucalyptus Grandis. Indian For. 1982,108,455-459. Bischoff, K. B. Accuracy of the Pseudo Steady State Approximation for Moving Boundary Diffusion Problems. Chem. Eng. Sci. 1963, 18,711-713. Blaii, T. J. Low-SuKdity Pulping with Anthraquinone. Tappi 1979, 62 (61,53-55. Burillo, J. C. Purificacion Industrial de Antraquinona. Ph.D. Dissertation, The yniversidad Complutense of Madrid, 1988. Burillo, J. C.; Rodriguez, F.; Adrados, L. F.; Tijero, J. F. Kinetics of Anthraquinone Reduction with Sodium Dithionite in Alkaline Medium. AIChE J. 1988,34,865-869. Eagle, A,; McDonough, T. J. A Kinetic Study of High-Yield AqSulfite Pulping of Lobolly Pine. Appita 1988,41, 138-140. El-Saied, H.; Nada, A. M. A.; El-Ashmawy, A. E. Soda Anthraquinone Pulping of Whole Cotton Stalks. Holzforschung 1984,38, 167-170. Fossum, G.; Hagglung, S.; Lindquist, B. Alkaline Pulping of Pine and Birch with Anthraquinone as Additive: Kraft Pulping. Suenks. Papperstidn. 1980a,83,430-435. Fossum, G.; Hagglung, S.; Lindquist, B. Alkaline Pulping of Pine and Birch with Anthraquinone as Additive: Soda Pulping. Suenks. Papperstidn. 1980b,83,455-460. Fullerton, T. J. Soda Pulping Appita - - with Anthrahydroquinones. .1978,32,117-118. Fullerton. T. J. Soda-Anthraauinone PulDing. The Advantanes of Using Oxygen-Free Conditions. Tappi 1f79,62 (a),55-57: Gustafsson, L.;Teder, A. Alkalinity in Alkaline Pulping. Suenks. Papperstidn. 1969,72, 795-801. Haldar, R.; Bhattacharya, P. K. Studies on Kraft and Soda Anthraquinone Pulping of Indian Mixed Hardwoods. Tappi J. 1987, 70 (6), 129-132. Holton, H. H. Soda Additive Softwood Pulping: A Major New Process. Pulp Pap. Can. 1977,78,T218-T223. Holton, H. H.; Chapman, F. L. Kraft Pulping with Anthraquinone. Laboratory and Full-scale Mill Trials. Tappi 1977, 60 (ll), 121-125.
Ind. Eng. Chem. Res. 1991,30,1795-1801 Irvine, G.M.; Nelson, P. J. Studies on Soda-Anthraquinone Pulping. Appita 1986,39, 289-292. Ksanzhek, 0. S.;Petrova, S. A,; Oleinik, S. V.; Kolodyazhnyi, M. V.; Moskovskii, V. Z.Study of the Redox Properties of the Quinone Structure Compounds. Anthraquinone and ita Derivative Compounds. Elektrokhimyya 1977,13, 182-190. Lachenal, D. Cuiseons Alcalines en Presence d’Additifs: Etude de la Cuiseon Soude-Anthraquinone dee Bois RBsineux. ATIP 1979, 33, 260-268. Lachenal, D.; De Choudens, C.; Monzie, P. Cuisson Soude Anthraquinone: Cas des Bois Feuillus. ATZP 1979,33, 213-190. Landucci, L. L. Quinones in Alkaline Pulping Characterization of an Anthrahydroquinone-Quinone Methide Intermediate. Tappi 1980,63 (7),95-99. Levenspiel, 0.Chemical Reaction Engineering; Wiley: New York, 1972. Lindman, N.; Simonsson, D. The Application of the Shrinking Core Model to Liquid-Solid Reaction. Chem. Eng. Sci. 1979,34,31-35. Nelson, P. J.; Irvine, G.M. Two-Stage Soda-AQ Kraft-AQ Pulping of Radiata Pine. Appita J. 1989,42, 271-274.
1795
McDonough, T. J.; Van Drunen, V. J. Pulping to Low Residual Lignin Contenta in the Kraft-Anthraquinone and Kraft Processee. Tappi 1980,63 (ll), 83-87. Patai, S.; Rappopoh, Z.Handbook of Chemistry and Physics; Weast, R. C., Ed.; The Chemical Rubber Co.: Cleveland, OH, 1971; Section D. Rodriguez, F.; Adrados, L. F.; Burillo, J. C.; Tijero, J. F. Simultaneous Determination of Polycyclic Aromatic Hydrocarbons by High-Performance Chromatography in the Manufacture of Anthraquinone from Anthracene Cake. Analyst 1989, 114, 1241-1244. Wandelt, P.; Surewicz, W. Catalyzed Alkaline Sulfur-Free Pulping. Selection of the Catalyst and Its Dose. Cellul. Chem. Technol. 1983,17, 543-552. Wen, C. I. Non-Catalytic Heterogeneous Solid-Fluid Reaction Models. Ind. Eng. Chem. 1968, 60, 34-39.
Received for review July 23, 1990 Revised manuscript received March 1, 1991 Accepted March 18, 1991
Nonisothermal Analysis of the Kinetics of the Combustion of Coked Sand Liang C. Lin, Milind D. Deo,* Francis V. Hanson, and Alex G. Oblad Department of Fuels Engineering, University of Utah, Salt Lake City, Utah 84112
The kinetic parameters for the combustion of carbonaceous residue or coke deposited on the sand during the fluidized bed pyrolysis of tar sands have been determined. Nonisothermal thermogravimetric experiments at a fixed heating rate were performed at various oxygen partial pressures on coked sands produced at three different retorting temperatures. Direct Arrhenius plot analysis which invoked a first-order assumption with respect to the unreacted fraction of coke yielded an oxygen partial pressure dependence of 0.5 and activation energies in the range of 127-148 kJ/mol. A second-order assumption with respect to the residual carbon gave an oxygen partial pressure dependence of 0.75 and activation energies of 175-217 kJ/mol. The first-order model was found to be applicable for lower oxygen partial pressures (5 kPa). In the overall range of oxygen partial pressures, the second-order model seems to provide a better fit to the data. These observations were consistent with the chemisorption/surface reaction mechanism proposed for the combustion of small carbonaceous particles. The Freeman and Carrol analysis indicated a reaction order between 1 and 2 with respect to the unconverted coke fraction, with high oxygen partial pressures resulting in reaction orders close to 2 and lower oxygen partial pressures yielding orders nearer to 1.
Introduction Tar sands are bitumin-impregnated rocks. Tar sand reservoirs, the world over, contain an estimated 2100 billion barrels of bitumen in place, with about 50 billion barrels in the United States. The state of Utah has about 45% of the United States’s (lower 48 states) exploitable tar sand resource. Pyrolysis is one of several processing techniques used for the recovery of bitumen, the organic constituent of the tar sands. Numerous studies have been undertaken to understand the recovery of liquid fuels by fluidized bed pyrolysis of tar sands (Venkatesan, 1980; Wang, 1983; Dorius, 1984). Pyrolysis of tar sand results in the formation of a carbonaceous residue on the spent sand. This char or coke on the sand can contribute a substantial amount of energy and thus feature significantly in the energy balance calculations of the overall process. Prior to this work, there have been no studies on the kinetics of the combustion of coked sand formed by the pyrolysis of tar sand. The purpose of this work was to study the intrinsic kinetics of the combustion of this material.
* Author to whom correspondence should be directed.
Depending on the dimensions of the sand particles and the conditions of combusion, the overall rate of the process may be controlled either by the intrinsic kinetics of the process or by the mass-transfer rate of oxygen. Dockter (1976),Mallon and Braun (1976),and Dockter and Turner (1978)performed studies on the combustion of oil shale char in large block samples at high temperatures (>loo0 K) and concluded that the rates were limited by oxygen diffusion. In their studies on the combustion of carbonaceous residue from retorted oil shale, Kim and Sohn (1985)postulated that, at certain conditions, the controlling mechanism switched from chemical rate control to oxygen transport control. For particle sizes of about 100 pm the rate is usually controlled by a chemical reaction at lower temperatures (
