- Email: [email protected]
Solid—solid interaction between particles of a fluid bed catalyst
Shorter Communications [14] Blander M., Molten salts chemistry, Interscience, New York, 1964. [IS] Delimarskii Yu K. et Markov B. F., The Electrochemistry of fused salts Sigma Press Washington l%l. [16] Braunstein J., Mamantov G. et Smith G. P., Advances ‘in molten salts chemistry Plenum Press New York 1973. [17] Sundheim B. R., Fused salts McGraw Hill, New York (1964). [18] Delahay P., New Instrumental methods in electrochemistry Interscience, New York 1953. [19] Rius A., Polo S., Llopis J., Anal. Real. Sac. Fis. Quim. Madrid 1949,45 1029. [20] Peters D. G. et Lingane J. J., J. Electroanal. hem. 1%12 1. [21] Brown 0. R., J. Electroanal. them. 197234 419. [22] Can00 C. et Claes P., Electrochim. Acta 1974 19 37.
Chwicd
Engineering
Science,1975, Vol. 30, pp. 15334535.
P.~gamon Press.
1533
[23] Barale G., Comtat hf. et Mahenc J., Bull. Sot. chim. Fr. 1%9 5 1585. [24] Reinmuth W. H., Anal. hem. l%l 33 485. [25] Delahay P. et Mamantov G., Anal. Chem. 195527 478. [26] Laity R. W. et McIntyre J. D. E., J. Am. hem. Sot. 196587 [27] t!Etat M., These Doctorat Toulouse 1974. [28] Jaeger J. C. et Clarke M., Proc. Roy. Sot. Edinburgh 1942 A61 229. [29] Von Stackelberg M., Pilgram M., Toome V., 2. Eleckfro&em, 195357 342. [30] Handbook ofphysical chemistry, 48th Edition, The Chemical Rubber Co., 1%7.
Printed in Great Britain
Solid-solid interaction between particles of a fluid bed catalyst (Received 17 February 1975; accepted 7 July 1975)
The quality of gas fluidizations is sometimes inlluenced by cohesive forces between particles. Apart from electrostatic or magnetic forces[l, 21, the effect of van der Waals and capillary forces might become relevant in certain circumstances. Irregularities in fluidixationcurves of fine powders and hollow glass spheres have been referred to van der Waals interactions[3]. Also, the nucleation of microcavities observed in expanded beds of catalysts rmd, plastics within the bubble-free range of fluidization of these materials has been related to interparticle forces. These presumably were of van der Waals and capillary type because, due to the almost complete lack of solids motion in bubble-free fluid&d beds, electrostatic charging was expected to be negligible[4,5]. According to the theory[6-8], capillary and van der Waals forces betweed nndeformable particles are respectively given by: F, =, k,$f$
I
2
which applies when wetting involves only local bridge-type bonds between particles and
Constants k, and k, depend on the system being studied, and R,R,/(R, t R,)is a reduced curvature radius related to curvature radii of the two particle surfaces at the contact point. Evaluation of R, and R, requires detailed information on particle surface geometry. Observations made to this end on a number of fluid bed powders[S], indicated that, due to surface asperities, the reduced curvature radius R,R,/(R, t RJ was certainly much smaller than particle radius Dp /2 for all powders tested, but that only for some of them, the type of surface roughness was such as to permit quantitative evaluation of local curvature radius. A particularly simple surface geometry was shown by a silicacatalyst (1.51 g/cm’ particle density), produced by atomization and drying of Ludox colloidal silica 130A. The scope of the present note is that of showing how, for this catalyst, forces predicted by eqns (1) and (2) on the basis of observed radii of curvature R, and RP compare with experimental values of cohesive forces between particles. The study concerns particles sampled from a 40-45pm batch of the material. Cohesion measurements between Ltrdox catalyst particles and substrates made of the same material have been taken by means of the centrifoge me&lb, lo]. Characteristics of this technique are its measuring sensitivity and, because of the simultaneous determination of the adhesive forces of many particles, statistical accuracy. The latter is an advantage specially if contact geometry
varies because of surface roughness. Obviously, the method is applicable only if cohesion forces are greater than particle weight, which was so with the material investigated. A Martin Christ centrifuge was used. The rotor was fitted with eight 18x 8 mm plates, on which catalyst was glued to form the substrates. These were preliminarly centrifuged at high speed to detach loose particles. Substrates were subsequently examined under a microscope to check whether any glue was left on the surface. Substrates selected for use were seeded with Ludox particles up to about 1% of their area. A cover plate, smeared with Vaseline, was then fitted in front of each substrate, to collect particles coming off. The rotor was then’centrifugcd up to a certain speed, which was kept constant for a run’$me of at least 15mm. Afterwards, the cover plates were removed and the collected particles counted under a microscope. New cover plates were fitted to the rotor, and a higher speed run carried out. This procedure was repeated at progressively increased speeds until no more particles detached from the substrates. The relative humidity was recorded for each run and varied between 48 and 65%. Results are presented in Fig. 1 as the histogram of incremental percentages of particles coming off at different centrifuge forces F. Marked lines indicate the average in detachment percentages from all the eight substrates. Confidence limits at 95% are given by light lines. Particles of Ludox catalyst were microspheres (Fig. 2). Under a Jeol electron scanning microscope particle surface appeared to be covered with spheroidal sub-particles (Figs. 3 and 4), sometimes grouped in clusters (Fig. 5). Photographs taken at an enlargement ratio of 30,000/l have been used to determine sub-particles size and distribution density on particle surface. Several particles were examined and, for each of them, different surface spots were considered. In reading the photographs, sub-particles of diameter D,down to Oalpm were recorded. For this size, measurement accuracy was 220%. Hystogram of sizes of sub-particles is reported in Fii. 6. Their average ‘density was about 10” sub-particles per square millimeter. With the silica material tested and the low relative humidity at which cohesion measurements have been carried out, capillary forces are given by eqn (I), and k, value of 60 dynlmm can be assumed[‘l, 81.Vahre of k, in eqn (2) is 8 dyn/mm[6]. Concerning RIR2/(RI+Rz), the mostprobable contact arrangement is that between a sub-particle on one particle and the main surface of the partner particle. From sub-particles distribution it can be shown that the probabilities of direct contact between the main spherical surfaces of facing particks and of contact between sub-particles are of the order of magnitude of 10-3-10-‘. Then, R, = Dp /2 and R, = DJ2, and considering that 4/D, is above 100,R>R,./(R, t R,) is about D./2. On the basis of these considerations,
1534
14
12
IO
%
6
6
4
2
a 2
F, m
dyn
Fig. 1. Histogram of particles leaving substrates for a given detaching force.
Fig. 2. A Ludox catalyst particle.
Fig. 3. Detail of sub-particles.
Fig. 4. Sub-particles in view on the particle surface.
Fig. 5. Sub-particles grouped in clusters.
1535
Shorter Communications
shown in Fig. 7, particles weight was 6.10-’ dyn. A spread in the distribution of cohesive forces will also be noted both from direct centrifuge measurements and from calculations based on subparticles size. This is in agreement with the nucleative character of bed expansion in the bubble-free range of lhtidization of the material tested[5].
40
30
%
Acknowledgment-Authors like to thank Prof. Dr. Ing. D. Technhc. H. Rumpf for discussion and Mr. A. Hargreaves for the careful execution of the cohesion measurements; they are also indebted to Miss Zucchii for accurate micrograph readings.
20
tlstituto di Chimica Industriaie e Impianti Chimici, Universitri; Laboratorio di Ricerchesulla Combustione,C.N.R.-Napoli,Italy.
IO
G. DONSI’t S. MOSERS L. MASSIMILLAt
tlnstitut fiir mechanische Verfahrenstechnik, UniversitiitKarlsruhe, West Germany. Fig. 6. Hystogram of sub-particle sizes REFERENCFS
sub-particle size distribution of Fig. 6 can be immediately converted into cohesive forces at contact point between catalyst particles. In Fig. 7, values of cohesive forces for both capillary and van der Waals type, calculated according to eqns (1) and (2) on the basis of surface geometry evaluation, are compared with cohesive forces directly measured with the centrifuge method. Diagrams are expressed in terms of cumulative frequencies. It appears that the values calculated on the basis of photographic information are of the correct order of magnitude in respect to forces directly measured with the centrifuge method. Measured forces are larger than those predicted assuming the occurrence of van der Waals interactions, but are smaller than those expected in the case where these were of capillary type. In comparison with cohesive forces
I.0 -
E 0.6
0 0
0
A
0
0
0
O0
DO00
A
O-6 -
I
00
0
A
j
g
A
AA
[1] Ciborowski J. and Wlodarski A., Chem. EngngSci. 19621723. [2] Agbim J. A., Nienow A. W. and Rowe P. N., Chem. Engng Sci. 197126 1293. [3] Baems M., I&EC Fund 19665 508. [4] Massimilla L., Donsi’ G. and Zucchini C., Chem. Engng Sci. 197227 2005. [5] Donsi’ G. and Massimilla L., A.LCh.E.J. 19736 1104. 161Kmpp H., Ado. Coil. Int. Sci. 1%7 1 111. [7] Schubert H., Chem. Ing. Techn. 197345 3%. 181Schubert H. Herrmann W. and Rumof _ H.. Powder Techn. 197511 121. [9] Bijhme G., Krupp H., Rabenhorst H. and Sandstede G., Trans. Inst. Chem. Engrs. 1%2 40 252. [lo] Polke R., Chem. Engng Techn. 196848 1057.
q
ooooo
-
Fjo4_ A
q
0
0.2 t
0
A
0
0
0 0 I
I
I
2
4
6
mdyn’
I
I
10
I
12
I
14
F, Fig. 7. Distribution of cohesive forces. 0, measured with centrifuge method; 0, calculated according to eqn (1); A, calculated according to eqn (2).
Chwicd
Engineering
Science,1975, Vol. 30, pp. 15334535.
P.~gamon Press.
1533
[23] Barale G., Comtat hf. et Mahenc J., Bull. Sot. chim. Fr. 1%9 5 1585. [24] Reinmuth W. H., Anal. hem. l%l 33 485. [25] Delahay P. et Mamantov G., Anal. Chem. 195527 478. [26] Laity R. W. et McIntyre J. D. E., J. Am. hem. Sot. 196587 [27] t!Etat M., These Doctorat Toulouse 1974. [28] Jaeger J. C. et Clarke M., Proc. Roy. Sot. Edinburgh 1942 A61 229. [29] Von Stackelberg M., Pilgram M., Toome V., 2. Eleckfro&em, 195357 342. [30] Handbook ofphysical chemistry, 48th Edition, The Chemical Rubber Co., 1%7.
Printed in Great Britain
Solid-solid interaction between particles of a fluid bed catalyst (Received 17 February 1975; accepted 7 July 1975)
The quality of gas fluidizations is sometimes inlluenced by cohesive forces between particles. Apart from electrostatic or magnetic forces[l, 21, the effect of van der Waals and capillary forces might become relevant in certain circumstances. Irregularities in fluidixationcurves of fine powders and hollow glass spheres have been referred to van der Waals interactions[3]. Also, the nucleation of microcavities observed in expanded beds of catalysts rmd, plastics within the bubble-free range of fluidization of these materials has been related to interparticle forces. These presumably were of van der Waals and capillary type because, due to the almost complete lack of solids motion in bubble-free fluid&d beds, electrostatic charging was expected to be negligible[4,5]. According to the theory[6-8], capillary and van der Waals forces betweed nndeformable particles are respectively given by: F, =, k,$f$
I
2
which applies when wetting involves only local bridge-type bonds between particles and
Constants k, and k, depend on the system being studied, and R,R,/(R, t R,)is a reduced curvature radius related to curvature radii of the two particle surfaces at the contact point. Evaluation of R, and R, requires detailed information on particle surface geometry. Observations made to this end on a number of fluid bed powders[S], indicated that, due to surface asperities, the reduced curvature radius R,R,/(R, t RJ was certainly much smaller than particle radius Dp /2 for all powders tested, but that only for some of them, the type of surface roughness was such as to permit quantitative evaluation of local curvature radius. A particularly simple surface geometry was shown by a silicacatalyst (1.51 g/cm’ particle density), produced by atomization and drying of Ludox colloidal silica 130A. The scope of the present note is that of showing how, for this catalyst, forces predicted by eqns (1) and (2) on the basis of observed radii of curvature R, and RP compare with experimental values of cohesive forces between particles. The study concerns particles sampled from a 40-45pm batch of the material. Cohesion measurements between Ltrdox catalyst particles and substrates made of the same material have been taken by means of the centrifoge me&lb, lo]. Characteristics of this technique are its measuring sensitivity and, because of the simultaneous determination of the adhesive forces of many particles, statistical accuracy. The latter is an advantage specially if contact geometry
varies because of surface roughness. Obviously, the method is applicable only if cohesion forces are greater than particle weight, which was so with the material investigated. A Martin Christ centrifuge was used. The rotor was fitted with eight 18x 8 mm plates, on which catalyst was glued to form the substrates. These were preliminarly centrifuged at high speed to detach loose particles. Substrates were subsequently examined under a microscope to check whether any glue was left on the surface. Substrates selected for use were seeded with Ludox particles up to about 1% of their area. A cover plate, smeared with Vaseline, was then fitted in front of each substrate, to collect particles coming off. The rotor was then’centrifugcd up to a certain speed, which was kept constant for a run’$me of at least 15mm. Afterwards, the cover plates were removed and the collected particles counted under a microscope. New cover plates were fitted to the rotor, and a higher speed run carried out. This procedure was repeated at progressively increased speeds until no more particles detached from the substrates. The relative humidity was recorded for each run and varied between 48 and 65%. Results are presented in Fig. 1 as the histogram of incremental percentages of particles coming off at different centrifuge forces F. Marked lines indicate the average in detachment percentages from all the eight substrates. Confidence limits at 95% are given by light lines. Particles of Ludox catalyst were microspheres (Fig. 2). Under a Jeol electron scanning microscope particle surface appeared to be covered with spheroidal sub-particles (Figs. 3 and 4), sometimes grouped in clusters (Fig. 5). Photographs taken at an enlargement ratio of 30,000/l have been used to determine sub-particles size and distribution density on particle surface. Several particles were examined and, for each of them, different surface spots were considered. In reading the photographs, sub-particles of diameter D,down to Oalpm were recorded. For this size, measurement accuracy was 220%. Hystogram of sizes of sub-particles is reported in Fii. 6. Their average ‘density was about 10” sub-particles per square millimeter. With the silica material tested and the low relative humidity at which cohesion measurements have been carried out, capillary forces are given by eqn (I), and k, value of 60 dynlmm can be assumed[‘l, 81.Vahre of k, in eqn (2) is 8 dyn/mm[6]. Concerning RIR2/(RI+Rz), the mostprobable contact arrangement is that between a sub-particle on one particle and the main surface of the partner particle. From sub-particles distribution it can be shown that the probabilities of direct contact between the main spherical surfaces of facing particks and of contact between sub-particles are of the order of magnitude of 10-3-10-‘. Then, R, = Dp /2 and R, = DJ2, and considering that 4/D, is above 100,R>R,./(R, t R,) is about D./2. On the basis of these considerations,
1534
14
12
IO
%
6
6
4
2
a 2
F, m
dyn
Fig. 1. Histogram of particles leaving substrates for a given detaching force.
Fig. 2. A Ludox catalyst particle.
Fig. 3. Detail of sub-particles.
Fig. 4. Sub-particles in view on the particle surface.
Fig. 5. Sub-particles grouped in clusters.
1535
Shorter Communications
shown in Fig. 7, particles weight was 6.10-’ dyn. A spread in the distribution of cohesive forces will also be noted both from direct centrifuge measurements and from calculations based on subparticles size. This is in agreement with the nucleative character of bed expansion in the bubble-free range of lhtidization of the material tested[5].
40
30
%
Acknowledgment-Authors like to thank Prof. Dr. Ing. D. Technhc. H. Rumpf for discussion and Mr. A. Hargreaves for the careful execution of the cohesion measurements; they are also indebted to Miss Zucchii for accurate micrograph readings.
20
tlstituto di Chimica Industriaie e Impianti Chimici, Universitri; Laboratorio di Ricerchesulla Combustione,C.N.R.-Napoli,Italy.
IO
G. DONSI’t S. MOSERS L. MASSIMILLAt
tlnstitut fiir mechanische Verfahrenstechnik, UniversitiitKarlsruhe, West Germany. Fig. 6. Hystogram of sub-particle sizes REFERENCFS
sub-particle size distribution of Fig. 6 can be immediately converted into cohesive forces at contact point between catalyst particles. In Fig. 7, values of cohesive forces for both capillary and van der Waals type, calculated according to eqns (1) and (2) on the basis of surface geometry evaluation, are compared with cohesive forces directly measured with the centrifuge method. Diagrams are expressed in terms of cumulative frequencies. It appears that the values calculated on the basis of photographic information are of the correct order of magnitude in respect to forces directly measured with the centrifuge method. Measured forces are larger than those predicted assuming the occurrence of van der Waals interactions, but are smaller than those expected in the case where these were of capillary type. In comparison with cohesive forces
I.0 -
E 0.6
0 0
0
A
0
0
0
O0
DO00
A
O-6 -
I
00
0
A
j
g
A
AA
[1] Ciborowski J. and Wlodarski A., Chem. EngngSci. 19621723. [2] Agbim J. A., Nienow A. W. and Rowe P. N., Chem. Engng Sci. 197126 1293. [3] Baems M., I&EC Fund 19665 508. [4] Massimilla L., Donsi’ G. and Zucchini C., Chem. Engng Sci. 197227 2005. [5] Donsi’ G. and Massimilla L., A.LCh.E.J. 19736 1104. 161Kmpp H., Ado. Coil. Int. Sci. 1%7 1 111. [7] Schubert H., Chem. Ing. Techn. 197345 3%. 181Schubert H. Herrmann W. and Rumof _ H.. Powder Techn. 197511 121. [9] Bijhme G., Krupp H., Rabenhorst H. and Sandstede G., Trans. Inst. Chem. Engrs. 1%2 40 252. [lo] Polke R., Chem. Engng Techn. 196848 1057.
q
ooooo
-
Fjo4_ A
q
0
0.2 t
0
A
0
0
0 0 I
I
I
2
4
6
mdyn’
I
I
10
I
12
I
14
F, Fig. 7. Distribution of cohesive forces. 0, measured with centrifuge method; 0, calculated according to eqn (1); A, calculated according to eqn (2).