Ind. Eng. Chein. Process Des. Dev., Vol. 18, No. 2, 1979
Table 11. Comparison of Calculated to Computer Predicted Values. Composition in Mol % space time
C3H, C3H6 CJL C,H, CH, H? HZO
calcd
comp
80.10 6.00 0.01 3.00 3.10 2.50 3.50
81.33 6.24 0.014 3.07 3.08 263 3.62
--0.5 s calcd
coinp
66.00 9.31 0.10 7.00 7.05 6.00 3.10
66.00 9.40 0.08 7.08 7.20 6.10 3.20
Temperature 650 ' C
_--
space time 0.1 s C,H, C3H6
C2H6
C2H4
CH, H?
of Research Equipment and computer time. Literature Cited
Temperature 600 C 0.1 s
245
0.5 s
calcd
prep
calcd
prep
69.48 8.90 0.10 6.23 6.40 5.30 3.26
69.48 8.83 0.073 6.38 6.44 5.39 3.39
51.20 12.82 0.34 11.70 12.35 9.54 3.00
50.90 13.00 0.33 11.70 12.48 9.60 3.00
HZO Acknowledgment The author gratefully acknowledges the assistance of the Department of Chemical Engineering and Chemical Technology, Imperial College, London, for the provision
Benwn, S.W., O'Neal, E., in "Kinetic Data on G% phase Unimolecubr Reactiom", p 385, 1970. Blakernore, J. E., Barker, J. R., Corcoran, W. H..Ind. Eng. Chem. Fundam., 12, 147 (1973). Bradley, J. N., "Flame and Combustion Phenotwna", Chapter 6, Methuen, London. 1969. Bradley, J. N., Proc. n. SOC.London, Ser. A, 337. 199 (1974). Calvert, J. G.,J . Am. Chem. SOC.,61,1206 (1957). Cullis, C. F., Hardy. F. R. E., Turver, D. W., Proc, R , SOC. London, Ser. A 244, 573 (1958). Fish, A,, Adv. Chem. Ser., No. 76, 71 (1968). 55, 468 (1959). Gray, P., Trans. faraday SOC., Herriot, G. E., et al., AIChE J., 18(1), 84 (1972). 1611 (1970). Kerr, J. A., Trotrnan-Dickenson, A. F., J . Chem. SOC., Kershenbaurn, L. S.,Sena, M. P., "Kinetic Modelling of Gas Phase Reactions Involving Free Radicals", presented at 65th Annual Meeting of the AIChE, 1972. Knox, J. H., Trans. Faraday SOC.,55, 1362 (1959). Knox, J. H., Trans. Faraday Soc., 56, 1225 (1960). Knox, J. H.,TrotmanDickenson, A. F., Trans. Faraday SOC.,54, 1509 (1958). Layokun, S. K., Ph.D Thesis, University of London, 1975. Layokun. S . K., Slater, D. H., Ind. Eng. Chem. Process Des. Dev.. precedlng paper in this issue, 1979. Lin, M. C., Laidler, K. J., Can. J . Chem., 44, 2927 (1986). Martin, R., Niclause, M., Scacchi, G., ACS Symp. Ser. 32, 37 (1976). McMillan, G. R., Caivert, J. G., Oxid. Combust. Rev., 1. 83 (1965). Niclause, M., et al., Can. J . Chem., 43, 1120 (1965). Pratt, G. L., "Gas Kinetics", p 159, Wiley, New York, N.Y., 1969. Purnell, J. H., Quinn, C. P.,Proc. R . SOC.London, Ser. A, 270, 246 (1962). Sarnpson, R . J., J . Chem. SOC.,5, 5104 (1963). Taylor, J. E , Kulich, D. M., ACS Symp. Ser., 32, 72 (1976). Voievodsky, V. V., Kondratiev. V. N., Prog. Reacf. Kinef., 1, 41 (1961). I
Received for review September 15, 1977 Accepted October 18, 1978
Gaseous Diffusion in Carbon with Particular Reference to Graphite Ralph 1. Yang" and Rea-Tling Llu Brookhaven National Laboratory, Upton, New York 11973
Binary gaseous diffusion coefficients in a nuclear graphite (H-451) along with the pore volume distribution were measured. The diffusion models of Johnson and Stewart and of Wakao and Smith were applied to this material. Diffusion coefficients predicted by the latter model were higher than the measured values at a ratio ranging from 1.2 to 1.6, whereas this ratio was about 4-5 with the former model. The models were also applied to other graphites whose diffusion and pore data were available in the literature. The predicted diffusion coefficients were all higher than the experimental data and the model by Wakao and Smith was consistently better. These results were compared with the literature data on commercial catalysts and inferences were drawn for the differencesin the characteristics of diffusion in these two types of materials.
Introd uction Problems involving gaseous diffusion in carbon occur in many technically important areas. Knowledge of gaseous diffusion is also essential in understanding the reactions between gases and carbonaceous materials. Among the considerable number of studies on this subject, most have been dealing with coal and nuclear graphite as they relate, respectively, to problems in coal mining and the gas-cooled nuclear-power reactors. Also, authors of these studies have been primarily concerned with gaseous permeability which is caused by pressure as well as concentration gradients. Nevertheless, one of the important common goals of these *Address correspondence to this author at the Department of Chemical Engineering,State University of New York at Buffalo, Amherst, N.Y. 14260. 0019-7882/79/1118-0245$01.00/0
studies has been to develop an understanding of the relationship between the rate of diffusion or flow and the pore structure. The subject of diffusion and flow of gases in graphite has been reviewed by Hewitt (1965). In the field of heterogeneous catalysis, gaseous diffusion in the porous catalysts has long been an area of significant importance. Several models have been developed for the prediction of gaseous diffusion rate from the information on pore structure and other physical properties. Among the more popular ones are the models of Johnson and Stewart (19659, Wakao and Smith (19621, Foster and Butt (1966), and others. This subject has been recently reviewed (Satterfield, 1970; Smith, 1970; Youngquist, 1970; Aris, 1975; Peterson, 1974; Carberry, 1976; Jackson, 1977). Among these models, the most tested ones are the Johnson and Stewart (J--S)model and the Wakao and Smith (W-S) model. In the J - S model, the pores with a 0 1979 American Chemical Society
246
Ind. Eng.
Chem. Process Des. Dev., Vol. 18, No.
2, 1979
known pore size distribution f(r) are assumed to be straight, nonintersecting, random in direction, and there are no dead-ends. The W-S model was developed for porous solids with basically bi-modal pore size distribution; each mode possesses an average pore size and they are termed macro- and micropores. These two models have been tested and compared by using a good number of commercial catalysts and catalyst supports all of which had basically bi-modal pore size distributions (Satterfield and Cadle, 1968a,b; Brown et al., 1969). Although both models displayed excellent predictive abilities (mostly within 50%), the J-S model seemed to be slightly better, especially a t higher pressures (to 65 atm total pressure). In this study, it was desired to learn how well these models could be applied to the graphitic materials and in the process to shed light on the characteristics of gaseous diffusion in the carbonaceous materials. The nuclear core-grade graphite H-451 was chosen for this study. This graphite has been developed recently and it is anticipated for use in the high temperature gas-cooled reactors in this country. The two aforementioned models have also been applied to two other graphites whose pore and diffusion information is available in the literature. The pore structure of H-451 has been carefully characterized and the results will also be presented. Diffusion coefficients of two gas pairs were measured a t various temperatures ranging from 23 to 75 "C at a pressure of 765 mmHg. The transient technique developed previously by the author (Yang et al., 1977) was used for the diffusion measurements. Experimental Section Measurement of the Diffusion Coefficient. The technique and the apparatus used have been described in detail (Yang e t al., 1977). Briefly, the technique was a transient type wherein the amount of the residual gas in the porous solid after a certain period of transient diffusion was measured and the effective diffusion coefficient was calculated. It has been shown in the same publication that this technique is best suited for measuring diffusion coefficients in the range of to cmz/s. The period of diffusion time in this work was 3 min. The residual gases were analyzed with a gas chromatograph. Pore Structure Characterization. The total porosity was measured by helium displacement. The surface area was obtained with the B E T method (N, a t 77 K). The micropore size distribution was measured by analyzing the nitrogen desorption isotherms. Descriptions of the method of computation are given by several investigators (Barrett e t al., 1951; Pierce, 1953; Anderson, 1964). The macropore size distribution was analyzed with a mercury porosimeter (manufactured by Micromeritics Co.). The two measurements overlapped a t the pore radius of 200 A. The size of the graphite was 60-80 Tyler mesh for both pore size distribution and the surface area measurements. Materials. The graphite (€3-451) was manufactured by Great Lakes Carbon Corp. It was machined to spheres of a diameter of 1.90 cm. The total volume of the diffusion chamber was 11.99 cm3 (Yang et al., 1977). The graphite samples were cleaned ultrasonically to eliminate possible blockage of the pore openings by the dust. The gases used were of the following grades: CHI, CP grade (99%); COz, Coleman Instrument (99 % ); and N2, prepurified (99.996% ' ). Application of Models The Johnson-Stewart Model. The original model was derived for one-dimensional diffusion through a cylinder as normally being used in the Wicke-Kallenbach (1941) type of diffusion apparatus. The model states that
NAzL = K{cDAB/[(MA/MB)'/' - 111 J-1.
0
([l+
XA,[(MA/MB)"~ - 1 1 + ( D A B / D A K ) I /+ [~ f(r)dr (1) XAL[(MA/MB)'"- 11 (DAB/DAK)]) To use the above equation to calculate the effective diffusion coefficient, Deff,we may substitute the following equation (2) Deff = NAZL / (CAO - CALI into eq 1. Now, Deffis dependent on the.boundary concentrations although the dependency may be quite weak. In a typical Wicke-Kallenbach type of experiment, pure gases A and B are flushed over the two boundaries, i.e., xA0= 1 and x a = 0. In our experiments (Yang et al., 1977) which are not a t steady state, X A ( t = 0) = 1 and X A (boundary) = 0. The two sets of boundary conditions are quite similar, especially considering the fact that our measurements were for the initial period of 3 min. Using these conditions, we obtain
The Wakao-Smith Model. The original model (eq 19 of their paper) is not repeated here due to its lengthiness. After substitutions similar to the foregoing, we obtain In [1/(1 - 41 + DAB/DKAa = t,2 In Deff DAB - + DAB/DKAa €:a 4€a(l - ea)
+
+
+
X
1 - a / 2 + DAB/DKAi 1 [(I - €a)2/€i2] ./{I - a / 2 -t DAB/[DAKi (1 e?)/(1 - Ea)'])
(4)
Here, a = 1+ NB/NA. For equimolar counterdiffusion, we may use CY = 1- (MA/MB)'-*. To calculate the average Knudsen diffusivities, DKAaand DW, we may first use the following equations to evaluate the average pore radii
The cumulative pore volume is obtained with mercury porosimetry and the Nz desorption method. Results and Discussion Macropore Size Distribution by Mercury Porosimetry. A contact angle of 130" was used for volume as a function of the pore radius. The pore volume distribution was calculated by taking the slopes of the tangential lines to the curve of the cumulative pore volume vs. radius. Results of these calculations are shown in Figure 1. The radii in this figure range from 200 to 10000 A. A clear peak is shown a t about 1000 A. Micropore Size Distribution by N2Desorption. The derivative of the cumulative pore volume with respect to radius is plotted against the radius as shown in Figure 2. Again, a clear and sharp peak was obtained. The peak appeared a t a radius of about 20 A. The complete pore distribution was obtained by splicing the micropore and the macropore distributions a t 200 A. At this pore radius, both techniques yielded a value of cm3/g.A for AV/Ar. From the completed about 1 X picture, it is clear that the pores have a bi-modal size distribution. The bi-modal pore size distribution is quite
4
----r7
t 0L
Ind. Eng. Chem. Process Des. Dev., Vol. 18, No. 2, 1979
Table 11. Gaseous Diffusion in Graphite H-451 Deff
1
1
J
E k
1000
'200 500
4000
6000
8000
IO000
i Figure 1. Macropore volume distribution of graphite H-451 from mercury porosimetry. PORE RADIUS ( r ) ,
247
temp, ("C)
De, X
x
103 (calcd),= cm2/s
103 (calcd J-S model), cm2/s
DeffaX
10
(calcd W-S model), cm 2/s
26 38 52 75
For CHJN, 1.69 8.28 1.83 8.50 1.95 8.87 2.00 9.29
2.65 2.79 2.93 3.19
23 36 51 61
F o r CHJCO, 1.61 6.41 1.72 6.71 1.84 7.20 1.90 1.62
2.07 2.16 2.33 2.43
Measured with a transient technique a t a total pressure of 1 a t m (Yang e t al. 1 9 7 7 ) .
the Chapman-Enskog equation (Bird et al., 1960) and DAK was calculated DAK = 9.7
40
80
I20
I60
ZOC
DORE RADIUS ( r ) ,
Figure 2. Micropore volume distribution of graphite H-451 from nitrogen desorption measurements. Table I. Physical and Pore Size Characteristics of Graphite H-451 total porosity E (helium penetration) = 17.7% microporosity e i ( r < 200 A , N, desorption) = 4 . 0 % macroporosity e a ( r > 200 A , mercury penetration) = 13.7% pore volume distribution: 23% ( r < 200 A ) ; 54% ( r = 200-2000 A ) ; 24% ( r = 2000-10000 A ) helium density = 2 . 1 1 g/cm3 BET surface area (N, at 77 K) = 4.98 m'/g
common to graphite and also t o some coals (Gan et al., 1972). Other physical properties of the H-451 graphite were measured with standard techniques and are shown in Table I. The only noteworthy point is on the macroporosity, ea. As shown in Figure 1, the upper limit of the pore radius was taken as 10000 A. Apparently, there existed pores of radii greater than this value. In fact, we did observe the penetration by mercury in the range of 1-50 pm. However, since the sample material was 177-250 wm in size, and the amount of mercury penetrated above 1pm pore radius greatly exceeded what could be accounted for, it was clear that the penetration of mercury was primarily in the interparticle space and the irregular open cracks on the surfaces of the particles. The value of ea was, therefore, by subtraction of the microporosity from the total porosity. The Measured Diffusion Coefficients. Diffusion coefficients were measured with the transient technique (Yang et al., 1977) as aforementioned. Two binary systems were measured: CH4/N2and CH4/C02. The results are shown in Table 11. Comparison with the J-S Model. Equation 3 was the working equation for computing Defffrom the pore size distribution. In the computation, D, was calculated with
X
lo-' P
(6)
The integral in eq 3 was computed grahically. The integration was carried out for r ranging from 15 to l 0 W A with 12 equal increments in r. The value of the geometric constant, K , was taken as assuming that the graphite was isotropic in pore orientation. The values of Defffor four temperatures were thus computed and shown in Table 11. The values predicted by the J-S model are about four to five times higher than the experimental values. Further discussion will be presented jointly with the comparison with the W-S model. Comparison with the W-S Model. Equations 4 and 5 were used to compute Defffor the W-S model. To use eq 5, we first had to determine or choose the integration limits. The radius of 200 A was used as the demarcation line for the micro- and macropores because it corresponded approximately to the minimum value of AV/& in the pore distribution curve. The upper limit was again taken as 10 000 A. The average micro- and macropore radii were then calculated, by graphical integration, to be 45 and 1424 A, respectively. DABand DAK were calculated in the same manner as described above. With these values, Deffwas then calculated with eq 4 and the results are shown in Table 11. Application of the Models to Other Graphites. Pore size distribution and diffusion data are both available in the literature for two nuclear graphites. AGOT graphite, a product of National Carbon Co., is a conventional material in which control over particle sizing for optimum packing and pitch impregnation has produced a graphite of basic density 1.70 g/cm3. This graphite has been a standard reactor grade in the United States and does not differ in structure from the more exotic gas-purified reactor grades. The mercury porosimetry data for AGOT were reported by Eatherly et al. (19581, and the binary diffusion data of helium and argon in AGOT were reported by Evans et al. (1961). The pore size and diffusion data for a British graphite were reported by Jenkins et al. (1964). This graphite, designated number 7 in their work, was developed for the British high-temperature gas-cooled (DRAGON) reactor project. The pore size data for these two graphites were reproduced in Figure 3 along with our data on H-451. T o use the W-S model for AGOT, the line of demarcation for the micro- and macropores was chosen to be 1.25 pm because there is a clear minimum in the pore volume distribution curve (Evans et al., 1961). The void for the
248
Ind. Eng. Chem. Process Des. Dev., Vol. 18, No. 2 , 1979
--
&--_-,-
7
7
v
-
3
1O-6E
,i
L
104
IO5
,0-7LLuLLLL~IL_LLLLLli , ' 1 1 ,
IO'
IO2
103
PORE SADIE ( r \
E
Figure 3. Pore volume distribution for three graphites: H-451 (A), AGOT (0) and No. 7 (0). Arrows indicate the lines of demarcation for micro- and macropores. Table 111. Gaseous Diffusion in Graphites AGOT and No. 7
diffusion system
Detf3X (calcd 10 J-S (exptl),' mod.), cm'/s cmz/s
(calcd W-S mod.), cm2/s ~~~~
He-Ar in AGOT, 20 'C, 1.249 a t m He-N, in No. 7, 22.3 "C, 94.98 cmHg
6.17
44.0
1.82
28.1
-
13.6
6.94
' Measured with a transient technique (Yang e t a]., 1977). macropores e,, was calculated to be 0.083 (by graphical integration of their curve). The void for the micropores, cl, was obtained by subtraction oft, from the total voids that they reported, and t i = 0.137. The pore volume distribution curve reported in their work was, however, not completed in the micropore range in that the curve terminated a t a radius of 2500 A. The void for the pores in the size range 25oct12500 A was 0.018 and, therefore, voids of 0.119 belonged to the pores with radii less than 2500 A. Because the two graphites, AGOT and H-451, have similar base stock materials, we assumed that their microstructures are also similar. Consequently, we completed the pore volume distribution by drawing a parallel curve to H-451 such that the voids for r < 2500 A equaled 0.119. This curve is shown as the dashed line in Figure 3. With this completed distribution, eq 5 yielded pa = 40486 A and Fi = 2046 A. The W-S model (eq 4) gave the Deffvalue of 1.36 X lo-* cm-2/s for the diffusion system referred in Table 111. With the J-S model (eq 3), the value of Defffor AGOT was calculated as 4.4 X lo-* cmz/s for the same system. Here, the completed curve in Figure 3 was again used and K was assumed to be 1/3. In the above calculations for AGOT, with both models, the boundary conditions were xA0 = 0.993 and x A L = 0.023. Now we proceed to calculate the values of Defffor the no. 7 graphite. The pore volume distribution data of Jenkins et al. (1964) showed that no pores existed at radius greater than about 10 000 A, and these measurements ended a t a radius of about 70 A. We consider only this range to calculate DefP For the W-S model, the distribution curve (Figure 3) is clearly of bi-modal type and the demarcation for the micro- and macropores was taken as 550 A. Using eq 5 the following values were obtained: ii, = 2044 A, pi = 206 A, t, = 0.134, and ei = 0.036. The W-S cm2/s model (eq 4) yielded the value for Deffof 6.94 x for the diffusion system referred in Table 111. For the same
system, the value of De, for the J-S model (with K = 1/3) was 2.81 x 10-2/s. In the calculations for the no. 7 graphite, the boundary conditions used were x A 0 = 0.954 and xAL = 0.168. The experimental data for Deffare compared with the calculated results for the two diffusion systems in Table 111. Diffusion Characteristics in Graphite and Application of Models. From Tables I1 and 111, we can draw two conclusions. First, the diffusion rates predicted by the J-S and the W-S models are consistently higher than the experimental results. Second, the W-S model is consistently closer to the experimental data than the J-S model. These two facts are especially interesting when compared with the results with commercial catalysts obtained by Satterfield and Cadle (1968a, b) and by Brown et al. (1969). These authors have shown that for the catalysts which were studied, values of Deffcalculated with the models were either higher or lower than the experimental data in a random manner. They also showed the two models predicted quite well the diffusion coefficients and perhaps the J-S mode1 was slightly better than the W-S model. It should be noted that the "J-S model" used by Satterfield and Cadle (196813) was actually of a modified form. A brief review of the nature of the pores in graphite would be helpful for further discussion. As a result of the packing of crystallites, particles, and disorganized carbon, graphite contains pores of various sizes most of which are interconnected by finer channels (Nightingale, 1962; Lang and Magnier, 1967). The shapes of the pores and the interconnecting channels are obviously highly irregular. Now we consider the possible false information that mercury porosimetry may lead to for solids with interconnected pores. Here by interconnected pores it is meant that the large pores are connected by fine pores. The fine connecting pores may be very short. In mercury porosimetry measurements, the volume of the connected large pores, Le., the large pores which are fed by the fine pores or necks, is all counted as the volume of the fine pores with radii of the connecting pores. This results in an overestimate of the fine pores and, therefore, peaks in the true AV/Ar vs. r distribution curve should be shifted toward greater r values. Predicted diffusion rates based on the mercury porosimetry data would be lower than the experimental results. However, as shown in Tables I1 and 111, the predicted values for Deffare consistently higher than the measured. There appear to be a t least three factors which may contribute to the high calculated values of Deff;they are dead-end pores, tortuosity, and irregularity in shape. The dead-end pores contribute to the pore volume measured by porosimetry and desorption methods but do not contribute to gaseous diffusion. Tortuosity may be more intensive in graphite than in the commercial catalysts, especially for the graphite which contains the onion-skin shaped pores. Finally, irregularity of the pore shape may also be considered. As reviewed and discussed by Eldridge and Brown (1976), irregularities in both the axial and the radial directions of a pore can result in a lower diffusion coefficient than a uniform, circular pore. The noncircularity of the pore only affects Knudsen diffusion, not the molecular diffusion which is considered as the major contributing mechanism for graphite. The fact that the W-S model gives better results for graphite than the J-S model suggests that the physical structure of graphite bears more resemblance to that used in the W-S model. Here, one recalls that in the J-S model the pores do not intersect, while in the W-S model the
Ind. Eng. Chem. Process Des. Dev., Vol. 18, No. 2. 1979
macropores are always connected, either through a neck (the macro-macro term) or through micropores (the macro-micro-macro, etc. terms.) Diffusion in graphite may be better described by the S.l'--S model, and its predominant mechanism seems to be the diffusion in the large pores which are interconnected by fine channels, with both having a highly tortuous path. Nomenclature A, B = gaseous species CAO, c a = molal concentrations of A at z = 0 and L. g-mol/cm3 c = total molar concentration, g-mol/cm3 D A B = binary molecular diffusion coefficient, cm'/s DAK = Knudsen diffusion coefficient of A, cm'js Dee = effective diffusion coefficient = iV/[L.(c, - cAo)]. cm'/s f ( r )= fraction of voids occupied by pores with radii between r and r + dr L = total length of diffusion path, cm M = molecular weight N = molar flux, r = pore radius, r?, r?, = average pore radii of the micro- and macropores, 8, t = time, s t = total (as subscript) T = absolute temperature, K u = cumulative pore volume, cm3 V = total cumulative pore volume, cm3 x = mole fraction z = direction of diffusion in the WickeeKallenbach cell C Y = ratio of diffusion rates = 1 + lvB/.vA 5. = infinitesimal increment c, = void fraction of the macropores ti = void fraction of the micropores K = geometric constant Literature Cited
249
Catalysts", Vol. 1, Clarendon Press, Oxford. Chapter 1. 1975. Barrett. E. P.. Joyner, L. G., Halenda, P. P., J . Am Cbem. SOC.,73, 373 (1951). Bird, R. B., Stewart, W. E.. Lightfoot, E. N., "Transport Phenomenon", p 511, Wiley, New York, N.Y., 1960. Brown. L. F., Haynes, H. W., Manogue, W. H., J . Catal., 14, 220 (1969). Carberry. J. J . , "Chemical and Catalytic Reaction", Chapter 9, McGraw-Hill, New York, N.Y.. 1976. Eatherly, W. P., Janes, M., Mansfield, R. L. Bourdeau, R A , . Meyer, R. A , Proc. Inl. Conf. Peaceful Uses At. Energy. 2nd. 15, 708 (1958). Eldridge, B. D., Brown, L. F., AICbE J , , 22, 942 (1976). Evans, R. B., 111, Truitt. J., Watson, G. M., J . Cbem. Eng. Data. 6 . 522 (1961). Foster, R. N.. Butt, J . B., AICbE J . , 12, 182 (1966). Gan, H., Nardi, S. P., Walker, P. L., Jr., f u e l , 51, 272 j1972). Hegedus, L., Petersen, E. E.. Catal. Rev., 9, 245 (1974). Hewitt. G. F., Cbem. Pbys. Carbon, 1, 74-121 (1965). Jackson, R.. "Transport in Porous Catalysts", Elsevier, New York. N.Y., 1976. Jenkins, T. R. Morris, J, B., Roberts, F., in "Advances in Materials". publ. by Pergamon for Inst. Chem. Eiig. (England), 1964 Johnson, M. F. L., Stewart, W. E., J . Catal., 4, 248 (1965). Lang, F. M., Magnier. P., "Studies of the Macroporosity of Carbons", in "Porous Carbon Solids". R. L Bond. Ed., Academic Press, New York and London, 1967. Nightingale, R. E.. "Nuciear Graphite", Chapter 5 , Academic Press, New York and London, 1962. Pierce, C. J.. J . Pbys. Cbem., 57, 149 (1953). Satterfield. C. N.. "Mass Transfer in Heterogeneous Catalysis", Chapter 1, MIT Press, Cambridge, Mass.. 1970. Satterfield. C. N., Cadle. P. J., Ind. Eng. Cbem. Process Des. Dev., 7 . 256 (1968a). Satterfield, C. N.. Cadle. P. J.. Ind. Eng. Cbem. Fundam , 7 , 202 (1968b). Smith, J. M., "Chemical Engineering Kinetics", 2nd ed. Chapter 11. McGraw-Hill, New York, N.Y.. 1970. Wakao, N., Smith, J. M., Cbem. Eng. Sci., 17. 825 (1962). Wicke, E., Kallenbach, R.. Kolloid-Z., 97(2), 135 (1941). Yang, R. T., Liu, R. T., Steinberg, M., Ind. Eng. Cbem. Fundam., 16, 486 (1977). Youngquist, G. R., in "Flow Through Porous Media", pp 57-70, America1 Chemical Society, Washington, D.C.. 1970.
R m e i i d f o r r a w u October 3, 1977 .4cc~ptedNovember 16, 1978
This work has been performed under the auspices of the Office of Chemical Sciences. Division of Basic. Energy Sciences, U.S. Energy Research and Development Administration. Washington,
Anderson, R. B., J . Catal., 3, 50 (1964). Aris, R., "The Mathematical 'Theory of Diffusion and Reaction in Permeable
D.C.
Kinetics of Formation of HCI(g) by the Reaction between NaCl(s) and SO,, 02,and H,O(g) Marianne Henriksson and Bjorn Warnqvist' Swedish Forest Products Research
Laboratory, PO Box
5604, S-774 86 Stockholm, Sweden
+
+
+
The kinetics of the reaction NaCl(s) -k '/,S02(g) 1/402(g) 'i2HpO(g) HCI(g) '/,Na,SO,(s) has been elucidated by laboratory studies at 600 OC (300-900 "C). The rate of reaction, expressed as the amount (moles) of HCI formed per unit of time (and surface), dnHC,/dt,obeys the following rate law: dn,,,/df = k~pso,"2-pop"2, where k is a specific rate (temperature dependent), and pso2and po, are partial pressures. The rate is practical1 independent of pH This rate law may be explained in terms of a rate-limiting reaction step: SO,(ads, 2") -t /,O,(ads, 2') &ads), between SO, and 0,molecules adsorbed on two sites ( * ) each, on the surface of the NaCl phase. The SO,(ads) then may be assumed to react with adsorbed H,O and with NaCI, probably via the intermediates The magnitude of the specific rate, k , is generally H,SO,(ads) or NaHSO,, to form the final products HCI and Na,SO,. low (reaction rate slow) which may explain the limited rate of HCI formation under recovery furnace conditions. +
-
Y
Background Hydrogen chloride, HCl(g), may be formed by reaction of sulfur dioxide, oxygen, and water vapor with sodium chloride '/$02 + 1/402 + l/*H20(g)+ NaCl(s, 1) HC1(g) + 1/2Na2S04(st l)
-
0019-7882/79/1118-0249$01 .OO/O
There are patented processes for production of HCl(g) at elevated temperatures by this reaction (Gmelin's Handbook, 1968). It has also been found that sodium chloride in flue gas dust in kraft recovery furnaces is converted into HCl by the same reaction (Warnqvist and Norrstrom, 1976; Warnqvist and Bernhard, 1975). This fact provides a way of removing chloride from the recovery 0 1979
American Chemical Society
