Table II.
Typical Results of Fluidized-Bed Contact of Sulfur Trioxide with Phosphate Rock
Temp., O
c.
250-500 310 286-320 297-324 27 5-3 31 294-302 327-332 320 330
Total Time, P z O ~ , A.P.A.a, F, Min. W t . 70 W t . 70 W t . % A. Contact with SO3 Alone .. 30.42 3.10 8.66 5.58 3.84 30 32.68 30 27.89 8.99 3.10 30 30.76 8.66 3.80 3.60 30 30.43 8.53 30 31.44 9.24 3.60 15 31.24 9.04 3.44 ... 30 30.3 7.72 B. Contact with SO3 and Air 15 28.6 6.9
s,
Wt.
y!
4.01 1.65 1.76 3.49 4.14 3.13 3.73
...
31.9 8.4 ... ... 28.8 9.2 ... ... 320 30 30.6 6.7 5 29.1 8.5 ... ... 325 320 10 29.6 8.6 ... ... C. Contact with SO3,Steam, and Air 320 15 22.1 10.46 ... ... 325 15 24.4 12.74 ... ... 335 3 23.1 12.30 ... ... 330 3 26.0 11.47 ... ... 330 1 31.4 9.40 ... ... 330 2 30.5 8.94 ... ... 320-340 5 28.8 10.09 ... 325 10 21.2 9.56 ... ... 330 10 29.3 10.60 2.98 10.0 325 15 23.7 13.12 1.70 10.3 15 24.7 13.73 1.70 5.6 325 ... ... 325 15 23.4 12.33 a Available phosphoric acid, sum of water-soluble and citrate-soluble PsOs by A O A C dejnitions. .
I
.
minutes, and all occurred within 15 minutes. The upper limit of conversion is probably due to a film of reacted material o n the particles of phosphate rock, but this phenomenon was not studied. The reaction temperature was usually about 325’ C. O b servation indicated that conversions dropped sharply above 340” C. This corresponds closely to the maximum tempera-
ture of the vapor-liquid region of the Hz0-S03system a t 1 atm. (6). Therefore the conversion limit a t 340’ C. is probably due to absence of liquid, since the fluidizing phase cannot wet the rock above this temperature. Stable fluidization was possible between about 315’ and 340’ C., and this dictated the choice of 325” C. as normal operating temperature. Sulfur dioxide produced only minor conversion of PZOS to available form a t contact times up to 1 hour. Detailed results are reported by Ross ( 8 ) . The product of the runs reported in Table I I C was brought in contact with a series of extractant solutions, in a n effort to discover whether or not the product \Yould yield more PzOs than the AOAC “availability.” The results may be summarized as follows: Basic and neutral solutions reduce the PpOb availability, and acid solutions improve the PpOb availability slightly (to 16.1% with KZS04 saturated solution a t 65’ C.). Both results are to be expected of highly polymerized PZOb. Ac knowledgrnent
The authors acknowledge the support of the Tennessee Corp. for this research. W. 0. Land, Jr., gave valuable assistance with experimental equipment. Literature Cited (1) Association of Official Agricultural Chemists, Washington, D. C., “Official Methods of Analysis,” 9th ed., 1966. (2) Baumgarten, P., Brandenburg, C., Chem. Ber. 72, 555-63
(1939). ( 3 ) Furman, N. H., ed., “Standard Methods of Chemical Analysis,” Vol. I, 6th ed., pp. 442-4, Van Nostrand, Princeton, N. J., 1962. (4) Giana, E., German Patent 219,680 (1907). ( 5 ) Hughes, A. E., Cameron, F. K., Ind. Eng. Chcm. 23, 1262-71 (1931). (6) Luchinskii, G. P., Zh. Fiz. Khim. 30,1207-22 (1956). (7) Pompowski, T., Zesrty Nauk. Politechn. Gdansk., Chem. 4, NO. 26, 3-28 (1962). (8) Ross, L. W., Ph.D. thesis, Georgia Institute of Technology, 1966. ( 9 ) Scheel, K., German Patent 966,264 (1957). (10) Snell, F. D., Biffen, F. M., “Commercial Methods of Analysis,” rev. ed., pp. 216-18, Commercial Publishing Co., New York, 1964. RECEIVED for review September 23, 1966 ACCEPTED May 1, 1967
FLUIDIZED BED DISPOSAL OF FLUORINE JOHN T. HOLMES, LOWELL B. KOPPEL,’ AND ALBERT A. JONKE Argonne National Laboratory, Argonne, Ill. 60439
s
volatility processes for the recovery of unspent fissionable and fertile materials from nuclear reactor fuels grow toward future commercial application, environmental contamination and waste disposal considerations will require evaluation of existing methods or development of new methods for the disposal of the toxic gaseous reagents used in the process. This paper describes the development of a new method for the disposal of fluorine. The work is part of a continuing effort a t the Argonne National Laboratory to develop methods for the disposal of gaseous fluoride volatility reagents and of volatile fission product compounds. The first requirement is that a fluorine disposal system have a high efficiency for the removal of fluorine from a gas stream.
A
FLUORIDE
Present address, Purdue University, West Lafayette, Ind. 408
l & E C PROCESS D E S I G N AND DEVELOPMENT
I t should also be economic, involve simple equipment and procedures, and c,onsume a minimum amount of chemical reactant. If the process is to be used in a nuclear fuel reprocessing plant, it should have a product which is suitable for packaging and storage as radioactive waste (preferably a free-flowing solid), and be able to remove fission product compounds associated with the fluorine-containing gas stream. Existing methods for fluorine disposal include the reaction of fluorine with liquids, gases, or solids. Gas scrubbers employing caustic solution are in common use for the disposal of fluorine (Liimatainen and Levenson, 1953; Slesser and Schram, 1951 ; Stainker, 1956). These scrubbers are efficient, but produce large volumes of liquid wastes, which is considered undesirable for radioactive applications. The reaction of fluorine with gases such as hydrogen, hydrocarbons (Long,
A fluidized bed process developed for the disposal of fluorine using activated alumina (AA) as the reactive solid, is over 99.9% effective in the removal of fluorine from a gas stream and utilizes the activated alumina to near the theoretical conversion. It has capability for high fluorine disposal rates and produces a free-flowing solid product for waste disposal. A factorial experiment was used to determine that increasing temperature (300" to 400" C,),increasing ratio of bed depth to diameter (3 to 61,and decreasing particle size (399 to 183 microns) significantly increased the capacity (grams of Fz per gram of AA) of activated alumina for fluorine removal. Changing the fluorine concentration from 5 to 75 volume % (v./o.) or the velocity from 1.25 to 1.65 times the minimum fluidizing velocity (V,,) had no significant effect. A higher velocity, 3.0 V,,, appeared to decrease the capacity slightly. Other solids reactants, which are less expensive than activated alumina were given preliminary evaluation. Soda ash appears especially promising.
1955; Turnbull, 1947), SO2 (Horton, 1965a), NH3 (Holmes, 1961), and steam (Smiley and Schmidt, 1954) has been used for fluorine disposal, but these methods produce other gaseous products which require further treatment. Many solids have been used as reactants in fluorine disposal schemes. Charcoal (Schmidt, 1957, 1959) has been used to dispose of large quantities of fluorine. T h e reaction products are mainly nontoxic carbon-fluorine gases and some condensables which may plug a n absolute filter system (Horton, 1965b). Packed beds of limestone, soda lime, and activated alumina have been used for fluorine disposal (Liimatainen and Levenson, 1953). Packed beds of activated alumina (AA) have been used routinely a t Argonne National Laboratory for disposal of fluorine from bench-scale and pilot plant-scale apparatus. U p to 85% of the activated alumina was consumed with high efficiencies for the removal of fluorine. Temperatures of over 1000" C . were generated in the packedbed reactors and caused sintering of the bed into a rigid mass which is difficult to rzmove from the reaction vessel. Other experiments a t Argonne have shown that packed beds of activated alumina have limited capabilities for trapping some of the volatile compounds of fission product tellurium and ruthenium. Since the activated alumina-fluorine reaction exhibits a number of desirable characteristics for a radiochemical process application, this systrm was studied in detail. A fluid-bed reactor was chosen rather than a packed-bed reactor in order to provide good heat transfer and thus prevent high temperatures which cause beti sintering. A free-flowing product can thus be achieved. A fluidized bed process can easily be automated with regard to addition of fresh solid reactants and withdrawal of the reaction product for waste storage. Determination of the efficiency of this fluidized bed process for fission product trapping is beyond the scope of the current work. Experimenta I
T h e experimental facility consists of four major parts: reagent supply, fluisd-bed reactor, potassium iodide (KI) scrubber, and back-up trap, as shown schematically in Figure 1. T h e reagent flows are measured with orifice meters and automatically controlled with pneumatically operated control valves. Fluorine is fed from cylinders, through traps of NaF and glass wool to remove any H F present. Nitrogen is used as the diluent gas. T h e fluid bed reactor is constructed from 2-inch diameter, 25-inch long, nickel pipe and is topped by a 3-inch diameter disengaging section. Two sintered nickel filters in the disengaging section remove particulate matter from the off-gas. Automatically controlled heaters and cooling coils are attached t o the outside walls of the reactor section. A 3/s-inch diameter tabular alumina sphere acts as a check valve for solids in the
4 VENT
REAGENT SUPPLY
FLUID' BED
K i SCRUBBER
REACTOR
BACK-UP TRAP
Figure 1. Experimental facility for studying the disposal of gaseous fluoride volatility reagents
cone-shaped gas distributor a t the inlet of the column. T h e reaction temperature is measured by three thermocouples, spaced a t I-inch vertical intervals starting 1-inch above the fluidizing gas inlet. The potassium iodide (KI) scrubber is a 2-foot long section of 3-inch i.d. glass pipe, packed with 5-mm. diameter glass beads t o a depth of 12 inches. One liter of 0.1M KI solution is used as the scrub solution. A vacuum pump and rotameter are used to bubble a IO-cu. foot per hour sample of the reactor off-gas continuously through the scrubber. A packed bed of activated alumina (back-up trap) is used to remove fluorine from the reactor off-gas during the breakthrough portion of a n experiment. T h e procedure for making a run consists of charging a given quantity of activated alumina (Alcoa, grade F-I) to the reactor, fluidizing the bed with nitrogen while the reactor is heated to the desired temperature, and then introducing the fluorine. T h e activated alumina is previously dried a t the reaction temperature. T h e x-ray diffraction pattern of the dried AA indicates that the compound is aA1203.H~O. T h e progress of the reaction is followed by taking periodic samples of the scrub solution and determining the iodine present (produced I- + 2F*/J2) using a standardized sodium by FP thiosulfate solution and a starch indicator. The concentration of fluorine in the off-gas from the reactor can thus be calculated. Corrections must be made for the change in volume of the scrub solution due to the periodic removal of samples. Tests of the scrubber, over the range 5 to 2600 p.p.m. of F2 in N P , showed the scrubbing efficiency to be essentially constant. T h e fluorine disposal experiments are terminated when the FS concentration in the off-gas exceeds 1000 to 2000 p.p.m. T h e sintered nickel filters are cleaned of any fines accumulation, using reverse pulses of high pressure nitrogen prior to removal of the reactor bed for weighing and sampling. T h e capacity is determined by chemical determination of fluorine in the bed. Capacity is defined as the grams of F2 reacting per gram of activated alumina charged to the reactor. Since nearly all of the fluorine is reacting when the experiment is
+
+
VOL. 6
NO. 4 O C T O B E R 1 9 6 7
409
terminated, the capacity a t the breakthrough point can be calculated by linear interpolation from the time of shutdown to the time a t breakthrough. The theoretical maximum capacity of 0.950 gram of F P per gram of AA, assuming the following reaction:
+
+
a constant temperature in the fluid bed near the gas inlet where most of the reaction takes place. I n the runs with high concentrations of Fz, axial temperature gradients of up to 200' C. were observed, but these large gradients did not appear to affect the operation. The data obtained from the K I scrubber were plotted against normalized time ( T N = actual time per time a t shutdown). Examples of the data from two typical runs are presented in Figure 2. All of the runs were characterized by a long initial period where the concentration of fluorine was less than 50 p.p.m. and then a short period where there was a rapid increase in concentration (breakthrough). Since the data from the scrubber give the average fluorine concentration from the time of the previous sample to the time of the present sample (dark line), a linear interpolation (fine line) was used to approximate the continuous concentration curve. The predried activated alumina is the monohydrate of aluminum oxide ( A 1 2 0 3 - H 2 0 )and, therefore, some H F is produced by the reaction of the H20 and Fa. The H F is not completely sorbed by unreacted activated alumina and small concentrations of H F were observed in the off-gas from the fluid-bed reactor. The H F concentration was highest during early portions of the run but never exceeded the concentration of fluorine in the off-gas and was usually considerably less. Results of Factorial Experiment, The half replicate factorial experiment is shown in Table I1 along with the values of capacity at breakthrough as defined a t the point a t which the off-gas concentration reached 200 p.p.m. and at the point a t which less than 99.9% of the fluorine was removed by the activated alumina. The experiments were made in a random sequence, as shown by the run numbers. The sequence was not randomized with respect to D,,since recalibration of the flowmeters was required whenever D , changed because of the difference in minimum fluidizing velocity. The capacity data in Table I1 were subjected to a n analysis of variance test with the aid of a digital computer. T h e results of this analysis using the capacity data a t either breakthrough point (200 p.p.m. or 99.9y0 removal) showed that for the range of the variables studied, temperature, particle size, and ratio of bed depth to diameter were the only variables which had significant effects on the capacity. Furthermore, these variables were significant a t the 99% confidence level,
+
A1203"zO 3Fz + 2AlFs H2O 1.5 0 2 After a series of shakedown runs to check out the equipment and procedure, a factorial experiment was used to determine the effect of the five most likely important independent variables on the capactiy of activated alumina for fluorine. T h e variables were: temperature ( T ) , particle size (D,), ratio of bed depth to diameter ( L / D ) , fluorine concentration (volume per cent, v./o.) and gas velocity ( V ) . For the ratio of bed depth to diameter, only the bed depth was varied, since the diameter was fixed a t 2.0 inches. For velocity, a multiple of the minimum fluidizing velocity (Vmf)was used. Each variable was studied a t the two levels shown in Table I. A complete study of the effects and interactions of all five variables at two levels would require a factorial experiment (Davies, 1954; Hicks, 1964) consistingof (2)b = 32 runs. A fractional factorial experiment was actually used in which only 16 runs were made (half replicate). This technique determines the effects of all the variables, the first-order interactions-e.g., T, D, interaction-and experimental error, but does not give any information on the effects of higher order interactions-e.g., T , D,, L I D interaction. Higher than first-order interactions are rather uncommon in physical situations. Results and Discussion
All of the runs were operationally very smooth. T h e product beds were always free-flowing and no significant pressure buildups were noted due to the deposition of fines on the sintered metal filters. T h e automatic filter blowback system was not used during any of the runs. All of the experiments a t 5 and 10 volume % F Pin N2 were made without the addition of coolant to the reactor walls. Runs a t 30 to 75 volume % FPrequired coolant to maintain Table 1.
Independent Variables
Low Level 300
High Level 400 3 6 48 to 100 (183) 28 to 48 (399) 1.25 1.65 5 10
Temperature, ' C. Bed size, L I D Particle size, mesh (microns) Velocity, multiple of V,, Conceritration,-volume %
F2AA-I 3 0
m O I
F2AA-5
:I 0
0
8IdN
LL!
Ba 8 -
8a * 0
s0 N
jiI
-
r!
0
c-
0
0
0.1 0.2 0.3 0.4
OB 0.6 0.7 0.8 0.9 1.0
TN
Figure 2. 410
rp -
0
0 0.1 0.2 0.3 0 . 4 ' 0 5 (16 07 0.8 0.9 1.0
TN . ..
Fluorine concentration in fluid-bed off-gas for typical runs
l&EC PROCESS DESIGN AND DEVELOPMENT
~~~
Table II.
v,x vln,
LID 6 6 6 6 3 3 3 3 6 6 6 6 3 3 3 3
1.65 1.65 1.25 1.25 1.65 1.65 1.25 1.25 1.65 1.65 1.25 1.25 1.65 1.65 1.25 1.25
which means there is less than one chance in 100 that the observed effects were due to experimental error. T h e effects of changing velocity, chamging concentration, and all first-order interactions of the variables were not significant. T h e magnitude of the effects (grams of Fz per gram of AA) of each variable can be calculated by taking the average capacity for those runs where the variable of interest was a t its high level and subtracting the average capacity for those runs where the variable was a t its low level. Inspection of Table I1 shows that the effects of the other four variables are cancelled and only the effect of the variable of interest is obtained in this manner. The magnitudes of the effects of the variables are given in Table 111. Temperature has the largest effect and ratio of bed depth to diameter and particle size have smaller effects. Changes in velocity and concentration over the range studied do not produce significant effects. T h e standard deviation of the capacity for breakthrough a t 200 p.p.m. was uo = 0.061 gram of FZper gram of AA and a t 77.7% removal Wac; uo = 0.058 gram of Fz per gram of AA. This means that the observed effects on Table I11 have 75% confidence limits of (-2) uJ2 = It0.06 gram of Fz per gram of AA and that the capacity data of Table I1 have 75% confidence limits of (.u2) uo = zk0.12 gram of F Pper gram of
AA. T h e results are presented in a more useful manner in Equations l and 2. Capacity a t 200 p.p.m. breakthrough
I
+
0.443 0.00307(T-623) f 0.0573 (L/D-4.5) 0.000750(0,-291) i n (1) grams of Fz per gram of AA
=
+
Capacity a t =-
breakthrough
~
~
Fractional Factorial Experiment for Fluid-Bed Disposal of Fluorine with Activated Alumina
lb, Microns 399 399 399 399 399 399 399 399 183 183 183 183 183 183 183 183
Run 5 2 1 8 4 7 3 6 12 11 16 13 10 15 14 9
~~~~~
-
+
0.410 0.00280(T-G23) 0.0653 (LID-4.5) - 0.00088(0,-271) in grams of Fz per gram of AA (2)
where Tis in O K., and D, is in microns. These equations represent the best (least square) linear interpolation over the range of variables studied (Table I). T h e equations should not be used to extrapolate outside of the ranges, and the user should not assume that the effects of velocity and concentration are not significant outside of the ranges studied. Within the range of variables studied, the maximum capacity can be obtained using T = 400' C., D, = 183 microns, L/D = 6, any concentration from 5 to 10 v./o., a n d any velocity from 1.25 to 1.65 Vm, (compare run 16 to other runs of Table I[).
T , C. 400 300 400 300 400 300 400 300 400
Volume % F2 in N a
Capacity, Gram of Fz per Gram of A A A t 200 A t 99.9Yo
0 627 0.294 0.660 0.344 0.206 0.153 0,429 0.187 0.740 0.388 0.797 0.412
5 10 10
5 10
5 5 10 10
5 5
300
400 300 400 300 400 300
10
5
0.642
0.25i 0.676 0.293
10 10 5
Table 111.
0,553 0.283 0.630 0.294 0.183 0.141 0,258 0.174 0.737 0.357 0.797 0.410 0.583 0.238 0.658 0.257
Effect of Independent Variables
Increasing Variable Temperature, 300" to 400' C. Bed size, L I D , 3 to 6 Particle size, 48-100 to 28-48-mesh Velocity, 1.25 to 1.65 V,,,, Concentration, 5 to 10 vol. % Mean capacity
1
Efect on Cajacity, Gram of FPper Gram of A A A t 200 p.p.m. At 99.9% 0.307 0.280 0.178 0.196 -0.162 -0.190 Not significant
0.443
0.410
Mechanism. I t is likely that the reaction of activated alumina with fluorine can be characterized by a continuous reaction model as described by Levenspiel (1962), since AA has a porous structure and high surface area (260 to 290 meters per gram). T h e model is shown schematically in Figure 3. I t is suspected that the reaction rate becomes pore diffusionlimited because of partial plugging of the pores by the buildup of reaction product. T h e product, AlFB, is stoichiometrically less dense than the reactant, AA, and would, therefore tend to fill the pores. There is an indication that diffusion through the gas film surrounding the particle does not limit the reaction rate, since gas velocity does not appear to have a significant effect o n capacity. Figure 4 shows the rate of a solid-gas reaction us. time with temperature as a parameter and L I D , D,, V , and volume per cent constant. These curves represent the maximum possible rate of reaction us. time for a reaction which becomes pore diffusion-controlled, The curve a t the higher temperature,
TI M E
L
RADIAL POSITION F i g u r e 3. C o n t i n u o u s reaction m o d e l VOL. 6
NO. 4
OCTOBER 1 9 6 7
411
-
FEED RATE
TIME
-
Figure 4. Maximum reaction rate for a solidgas reaction
T H I gives , a higher reaction rate due to the normal effects of temperature on chemical reaction rates and diffusion rates. For the F:!-AA reaction, the rate of consumption of F2 prior to breakthrough (BT) is essentially the feed rate of fluorine, since over 99.9% of the fluorine reacts. This being the case, it is not surprising that velocity and concentration had little effect on the capacity. The net consumption rate continues to be the feed rate of fluorine until that time when the maximum reaction rate profile is reached for the temperature of the reaction. T h e rate then rapidly decreases along the maximum reaction rate curve and a breakthrough of fluorine is observed. At the higher temperature, the breakthrough is postponed and a higher capacity is reached. The capacity is proportional to the feed rate multiplied by the time a t breakthrough or the area under the curve. If partial plugging of the pores is caused by stoichiometrically less dense reaction product such that pore diffusion resistance increases, it is not surprising that the smaller particles attain a greater capacity for fluorine. I t is suspected that the larger particles have only partially reacted cores (see Figure 3) a t the time of breakthrough. High values of L / D gave greater capacities. This observed effect is possibly due to poor solids mixing (Nicholson and Smith, 1966) in the deep beds. I n deep beds, some of the particles may remain in the bottom of the reactor for longer times and therefore will react to near the maximum theoretical capacity, since the concentration of fluorine may be high enough to overcome mass transfer resistance in the pores. T h e breakthrough of fluorine is postponed, since, in deeper beds, the particles a t the top have not yet fully reacted and are able to remove small amounts of fluorine coming from the lower portions of the bed. T h e net effect is that the capacity is higher in deep beds. There may be practical limits to the observed positive effect of increasing L I D when gas-solids contacting becomes poor because of slugging. Extreme Conditions. A series of five additional runs was made to determine the effect on capacity of operating with conditions outside the range of those used in the factorial experiment. Since there are practical limitations on using smaller particles of activated alumina and deeper beds (because of the size of the reactor), these variables were not investigated further. One of the five runs was made to determine the effect of predrying the activated alumina. One run of this series was made under the same conditions as in run 16 (which gave the highest capacity in the factorial 412
l&EC PROCESS DESIGN A N D DEVELOPMENT
experiment), except that the temperature was 450' C. compared to 400' C. in run 16. The resulting capacity was 0.803 gram of F2 per gram of AA, which is only slightly higher than 0.797 gram of F:! per gram of AA for run 16. This difference was too small to be significant. It is not surprising that there was only a small increase in capacity, since the capacity values are near the theoretical limiting capacity of 0.950 gram of F:! per gram of AA. A run was made using a higher velocity in a n attempt to increase the fluorine throughput capabilities of the system. The run was similar to run 12. except that the velocity was 3.0 Vm,. The capacity obtained was 0.637 gram of F2 per gram of .4A, which can be compared to 0.740 gram of Fz per gram of AA in run 12. The capacity a t the high velocity is lower than that of run 12 by an amount which is about 1.7 times the standard deviation. This means there is less than one chance out of 10 that the observed effect is due to experimental error. The decrease in capacity a t higher velocities is probably due to poorer gas-solids contact (Chakravarty et al., 1963). Two runs were made to determine the effect of high fluorine concentrations on capacity. Both runs were similar to run 12 except that 30 and 75 volume % fluorine were used. These experiments gave capacities of 0.757 and 0.739 gram of Fz per gram of AA compared to 0.740 for run 12. It is therefore concluded that F2 concentration has no significant effect of capacity over the range 5 to 75 volume yo. The runs a t the higher concentrations required external cooling and produced axial temperature gradients as high as 200' C. in the fluidized bed but did not adversely affect the operation. A single run was made which demonstrated that predrying the activated alumina had no significant effect on the capacity. R u n 14 was repeated, except that the bed was charged Lias received" rather than predried as in all of the other runs. The normal moisture loss on drying was about 3.2% a t 300' C. and 4.370 a t 400' C. The capacity obtained was 0.635 gram of Fz per gram of i\A compared with 0.676 gram of F2 per gram of AA for run 14. These results differ by only 0.3 times the standard deviation, so the effect of predrying was considered too small to be significant. Comparison with Packed Beds. I t is interesting to compare the throughput capabilities for the fluid-bed reaction system with that for a packed-bed reactor. Throughput is the feed rate of fluorine in pounds of Fa per hour and square feet of reactor cross section. Using 183-micron (48- to 100mesh) activated alumina and a concentration of 75 volume %, the fluorine throughput rate is about 40 lb./hr. sq. ft. Rates u p to 140 lb./hr. sq. ft., can be obtained with 28- to 48-mesh alumina but result in lower capacities. Data obtained earlier for packed beds of activated alumina gave maximum throughputs of about 3 lb./hr. sq. ft. Above this throughput rate sintering of the packed bed started to occur. I t is obvious that if the user requires a free-flowing reaction product, the fluid-bed technique will allow much higher fluorine throughput rates. The experiments with packed beds of AA gave maximum capacities a t breakthrough of about 0.85 gram of Fz per gram of AA for runs where bed temperatures were over 1000° C. and sintering occurred. The capacities achieved in fluidized beds (up to 0.8 gram of FP per gram of AA) were only slightly lower than the packed bed capacities. Nature of Solid Reaction Products. T h e solid reaction products appear to be suitable for waste disposal. T h e beds were free flowing a t the conclusion of all of the runs. No significant change in particle size is caused by the stoichiometrically less dense reaction product or by attrition due to the
Table IV.
Experimenlral Conditions for the Disposal of Fluorine Using Na2C03 and CaC03
Temperature, Ci. Bed size, L/D Particle size, mesh Concentration, vol. 70Fz Velocity, multiple of V,,,, O
400 6 -60, $100 10
1.65
turbulence of the fluid bed. T h e bulk density of the product from run 16 was 1.22 grams per cc. untapped and 1.36 grams per cc. tapped. Other Solid Reagents. Two other solid reagents, limestone (CaC03) and soda ash (Na&03), both of which are less expensive than activated alumina, have been tested for fluorine disposal using the conditions shown in Table IV. Qualitatively, the results of single experiments on each of the solids were similar to the results for activated alumina. The curves of fluorine concentration us. time were like those in Figure 2. T h e runs were characterized by a long period of high removal efficiency, followed by an abrupt breakthrough period. T h e capacity of soda ash \vas 0.32 gram of F Z per gram and the capacity of the limestone was 0.045 gram of Fz per gram. These capacities correspond to about 90 and 12% of ;he theoretical maximum c,apacities for the soda ash and limestone, respectively. I t appears that the reaction rate with limestone becomes diffusion-coni rolled and breakthrough occurs. T h e product of the soda ash reaction, sodium fluoride (NaF), apparently does not hinder the reaction, since the reaction product is stoichiometrically denser than the reactant and thus would not tend to produce a n increased resistance to mass transfer. Since NaF is known to be a n effective sorber for certain volatile fluoride compounds, the NaF product from the soda ash reaction m,iy effectively remove volatile fission producis associated with the fluorine stream. Conclusions
A fluidized bed process can be used effectively to dispose oi fluorine using activated alumina as the reactive solid. The process is capable of high fluorine disposal rates and efficiencies (over 99.9%) over a wide range of the independent process variables. A factorial experiment determined that increasing temperature (300” to 400’ C.), increasing ratio of bed depth to diameter (3 to G), and decreasing particle size (399 to 183 microns) significantly increase the capacity of the activated alumina for
fluorine removal to near the theoretical maximum value. There were no significant effects of changing the fluorine concentration from 5 to 75 volume % or of changing the velocity from 1.25 to 1.65 Vm,. Higher values of velocity (3.0 Vml) may slightly decrease the capacity of activated alumina for fluorine. Other solid reactants, which are less expensive than activated alumina, are also being evaluated. Soda ash is especially promising. Acknowledgment
The authors thank C. B. Schoffstoll for his help in construction and operation of the equipment. literature Cited
Chakravarty, R. K., Banerjee, S., Basak, N. G., Lahiri, A , , Indian J . Technol. 1, 423 (1963). Davies, 0. L., “Design and Analysis of Industrial Experiments,” Hafner, New York, 1954. Hicks, C. R.. “Fundamental ConceDts in the Design of ExDcriments,” Holt, Reinhart, and Winston, New York,?964. Holmes, J. T., Chem. Eng. 68,No. 26, 94 (1961). Horton, R. W., Oak Ridge National Laboratory, Oak Ridge, Tenn., private communication, September 1965a. Horton, R. W., Oak Ridge National Laboratory, Oak Ridge, Tenn., private communication, October 1965b. Levenspiel, O., “Chemical Reaction Engineering,” Wiley, New York. 1962. Liimatainen, R. C., Levenson, M., U. S. Atomic Energy Commission, Rept. ANL-50115 (1953). Long, G., U. K.Atomic Energy Authority, Rept. AERE C/M 260
-
- - ,.
1, 1 , 055)
Nicholson, W.J., Smith, J. C., Chem. EnE. Progr. Symp. Ser. 62, 83 11966). Schmidt, H. W.,National Advisory Committee for Aeronautics, Rept. NACA RM E 57E02 (1957). Schmidt, H. LV., National Aeronautics and Space Administration, Rept. NASA-MEMO :1-27-59E (1959). Slesser. C.. Schram. S. R.. “PreDaration. Prouerties and Technology of Fluorine and Organic’Fluoro Compbunds,” McGrawHill, New York, 1951. Smiley, S. H., Schmidt, C . R., Ind. Eng. Chem. 46,244 (1954). Stainker, S. H., U. S. 4tomic Energy Commission, Rept. CF56-12-128(1956). Turnbull, S. G., Benning, A. F., Feldmann, G. IV., Linch, A. L., McHarness, R. C., Richards, M. K., Ind. Eng. Chem. 39, 286 (1947). RECEIVED for review November 16, 1966 ACCEPTED May 15, 1967 Division of Nuclear Chemistry and Technology, 152nd Meeting, ACS, New York, N. Y., September 1966. ‘IVork performed under the auspices of the U. S. Atomic Energy Commission, Contract No. W-31-109-eng-38.
VOL. 6
NO. 4
OCTOBER 1967
413