Selective Transport of Sulfur Dioxide through Polymer Membranes. 2

Oct 22, 1979 - A novel technique is described for preparing cellulose triacetate/polyacrylate composite membranes for SOz ~s separations. The membrane...
1 downloads 0 Views 947KB Size
616

Ind. Eng. Chem. Process Des. Dev. 1980, 19, 616-623

Nomenclature

Literature Cited

a = constant, eq 3 A = membrane area, cm2 b = constant, eq 3 Bd = constant in free volume theory, eq 9 c = concentration of gas in membrane, cm3(STP)/cm3 c = concentration of polymer solution, g/g Lf = diffusion coefficient, cm2/s Do = preexponential constant for diffusion coefficient, cm2/s DT = thermodynamic diffusion coefficient, cm2/s f = average fractional free volume 1 = membrane thickness, cm P = pressure, cmHg or mmHg P(0) = pressure on upstream side of membrane, mmHg P(1) = pressure on downstream side of membrane, mmHg Q = permeability, cm3(STP)-cm/(cm2.s.cmHg) QO = preexponential constant for permeability, cm3(STP). cm/(cm2.s.cmH ) S = solubility, cm%(STP)/(cm3.cmHg) t = time, s T = temperature, K u = withdrawal rate of glass plate, cm/s ud = penetrant volume fraction V = volume of downstream chamber of manometric system, cm3 Greek Letters @(r= ) parameter in free volume theory, eq 9 y = plasticizing parameter, eq 4 t9 = time lag, s X = Constant in three-parameter exponential model, eq 7 p = viscosity of pure liquid or solution, P

Barrer, R. M., Fergusson, R. R., Trans. Faraday SOC., 54, 989 (1958). Carneil, P. H., Cassidy, H. G., J. Polym. Sci., 55, 233 (1961). Felder, R. M., Spence, R. D., Fenel, J. K., J. chem. Eng. Data, 20, 235 (1975). Frisch, H. L., J. Phys. Chem., 61, 93 (1957). Fujlta, H., Chapter 3 in "Diffusion In Polymers", p 75, J. Crank, G. S. Park, Ed., Academic Press, New York, 1968. General Electric, "General Electric Permselective Membranes", Medical Development Operation, Chemical and Medical Division, Schenectady, N.Y., 1970. Kuehne, D. L., Ph.D. Theds, Califomla InstiMe of Techndogy, Pasadena, CaiK, 1979. Kuehne, D. L., Friedlander, S.K., Ind. Eng. Chem. Process Des. Dev., foC lowing article in this issue, 1980. Levich, V. G., "Physicochemical Hydrodynamics", D 681, Prentlce-Hall. Engle. . wood Cliffs. N.J., 1962. Lonsdale, H. K., Riley, R. L., Lyons, C. R., Caroseila, D. P., Jr., in "Membrane Processes in Industry and Bbmedicine", p 101, M. Bier, Ed., Plenum Press, New York. 1971. Michaels, A. S., Pure Appi. Chem., 46, 193 (1978). Ponder, W. H., Stern, R. D., McGhmery, G. G., "SO2 Control TechnologiesCommerclai Availabilities and Economics", presented at the Third Annual International Conference on Coal Gasification and Liquefaction, Pittsburgh, Pa., Aug 3-5, 1976. Riley, R. L., Hightower, G. R., Lyons, C. R., Appi. Polym. Symp., No. 22, 255 (1973). Seibei, D. R., McCandless, F. P., Ind. Eng. Chem. RocessDes. Dev., 13. 76 (1974). Ward, W. J., 111, in "Recent Developments in Separation Science", Vol. 1. p 153, N. N. Ll, Ed., Chemical Rubber Co., Cleveland, Ohio, 1972. Ward, W. J., 111, Neulander, C. K., "Immoblllzed Liquid Membranes for Sulfur Dioxide Separation", PB-191-769, U.S.Department of Commerce, 1970. Weiions, J. D., Stannett, V., J. Polym. Sci., Part A- I , 4, 593 (1968). Zavaleta, R., McCandless, F. P., J. Membr. Sci., 1, 333 (1976).

Received f o r review October 22, 1979 Accepted June 12, 1980

Presented at the 87th AIChE National Meeting, Boston, Mass., Aug 1979.

Selective Transport of Sulfur Dioxide through Polymer Membranes. 2. Cellulose Triacetate/Polyacrylate Composite Membranes Donald L. Kuehne and Sheldon K. Friedlander Department of Chemical Engineering, California Institute of Technology, Pasadena, California 9 1 725

A novel technique is described for preparing cellulose triacetate/polyacrylate composite membranes for SOz ~s separations. The membranes were initially tested by measuring the permeabilities of pure Nz, COP,and SOz with a manometric system. SOz permeation data were observed to be pressure dependent and were correlated with a three-parameter exponential model. A continuous flow system was built to model the gas flow in a membrane separator, and selected membranes were tested with binary mixtures of N2 and SO2. Permeation rates for the mixtures were difficult to predict from data for the pure gases. Membrane performance was evaluated in terms of design goals for combustion and smelter gas applications.

Introduction In part 1 (Kuehne and Friedlander, 1980) we described the preparation of thin membranes of the polyacrylate and cellulose triacetate polymers by two conventional methods and the measurement of their permeability to N,, CO,, and SO,. The SO, permeability was pressure dependent for membranes of both polymers, so the SO, flux and selectivity were lower than expected at low SO, pressures. Consequently, the membranes were unable to meet SO2 flux and selectivity design goals for industrial applications. Polyacrylate was the better polymer for SO, separations because it was over ten times as permeable as cellulose *Chevron Oil Field Research Co., La Habra, CA 90631.

triacetate and its SO2 permeability was less dependent on pressure. The main problem with polyacrylate was the difficulty in casting thin, pinhole-free films. The method of casting films on mercury was tedious and for good results was limited to film thicknesses greater than 1 pm. In this paper it is shown how thin films of polyacrylate can be made by the glass-plate technique, which previously was applied only to nonsticking polymers. Membrane Preparation The preparation of polyacrylate films by the glass-plate technique (Carnell and Cassidy, 1961) required that the polymer be rendered nonsticking. This was accomplished by first applying an ultrathin layer of cellulose triacetate (CTA) to the glass plate. The CTA layer was made as thin

0196-4305/80/1119-0616$01.00/00 1980 American Chemical Society

Ind. Eng. Chern. Process Des. Dev., Vol. 19, No. 4, 1980

617

Table I. Preparation of Composite Membranes

membrane

CTA total thickness, thickness, w-n m

porous support

+

~

CTA/PA-43

0.00413

0.190

CTA/PA-52 CTA/PA-53 CTA/PA-58

0.001'7 0.001'7 0.001'7

0.179 0.461 0.485

CTA/PA-59

0.001'7

0.460

Millipore filterglossy side wet Celgud 3501-wet Celgard 3501-wet Millipore filterglossy side dry Celgard 2500-dry

as possible to minimize its effect on gas permeation. A thicker film of polyacrylate was formed on top of the CTA layer. The two layers were floated off the glass plate and placed on a microporous support. Details of this method were reported previously (Kuehne, 1979). The quality of thetie cellulose triacetate/polyacrylate (CTA/PA) composite!membranes depended on the type of glass and the porous support. Polished glass plates were used because the membrane would not come off unless the surface of the plate was very smooth. The membranes were supported on MI?-Millipore filters with 0.050 pm pore size, Celgard 2500, a porous hydrophobic polypropylene film, and Celgard 3501, a hydrophilic version of Celgard 2500. The Celgard products were obtained from Celanese. These microporous films gave satisfactory results but were not ideal. For example, when the Millipore filter was soaked in water, it swelled and needle-like projections appeared on its surface. Surface smoothness, pore size, and pore density of the porous support are critical in determining the minimum film thickness which can be supported. The thicknesses of the layers of the composite membranes were estimated from the theory of Levich (1962), as discussed in part 1. Film thicknesses were also determined by weighing known areas. Data for the composite membranes are given in Table I. The thickness of the CTA layer was more (criticalin determining the strength of the composite membrane than the thickness of the polyacrylate layer. Membrane CTA/PA-43 has a thicker CTA layer than the other membranes and was stronger and more stable. However, there is a trade-off between mechanical strength and permeability because the thicker the CTA layer is, the greater its resistance to permeation. The last column of Table I described the porous support. The designations "wet" and "dry" refer to the way in which the membrane and porous support were brought together. The porous support was either raised from the bottom of the water bath until it contacted the membrane floating on the surface (wet) or laid gently on top of the membrane (dry). Experimental Section The permeability of the composite membranes to pure N2, C02, and SO2 Was determined by the manometric method (ASTM D 1434-75) with the manometric system described in part 1. N2 and COz were tested a t 1 atm upstream pressure. SOZ was tested at upstream pressures of 100, 200, 400, and 760 mmHg to more accurately characterize its permeability which exhibited a strong pressure dependence. A steady-state flow system was designed and built to measure the permeability of mixtures of N2 and SO2. Design Equations; for Flow System. Equations are derived below for predicting the gas composition and permeation rate for a binary mixture of N2 and SOz. A schematic diagram of the flow system permeation cell is shown in Figure 1. A binary mixture of SO2mole fraction xu.,, enters the upstream chamber of the permeation cell

"d Xd

Figure 1. Schematic diagram of the flowsystem permeation cell.

a t a rate of nu,inmol/s and a pressure Puand leaves a t a rate of nu,outwith a mole fraction The downstream chamber of the cell is maintained at a lower pressure P d . N2and SO2permeate across the membrane as a result of their respective partial pressure driving forces. If the upstream chamber is well mixed, then the composition is xu throughout and is equal to xUput. From the definitions of gas flux and permeability, the Nz and SOz fluxes are

where 1 is the membrane thickness. By the conservation of mass (3) Substitution of (1) and (2) in (3) and manipulation of the resulting equation leads to a quadratic equation in x d axd2

+ bxd + c = 0

(4)

where (5)

"(

b=-

pd

xu-xu--

Q(so2) Q(N2)

1

)

+

--Q(s02) 1

Q(N2)

(6)

(7) Equation 4 is written in dimensionless form. The three dimensionless groups which determine the downstream mole fraction, x d , are the upstream mole fraction, xu, the pressure ratio, p u l p & and the permeability ratio, 8(SOz)/Q(N2).This is a useful result because it reduces the number of parameters which must be varied to characterize a membrane system. Once x d is found from the solution of (4), the permeation rate, n d , is easily derived from the above expressions

where A is the membrane area, R is the gas constant, and T,and P,are evaluated at STP (273 K and 76 cmHg). The N2 and SOz fluxes are estimated from permeability data for the pure gases; however, these estimates may be quite different from the observed fluxes because of interactions between the gases and the membrane. Flow System Equipment. Scale drawings of the stainless steel permeation cell are shown in Figure 2. A gas mixture enters the cell through a manifold in the top which distributes the gas to four inlet ports and exits through a larger port above the center of the membrane. This design produces good mixing in the upstream cham-

818

Ind. Eng. Chem. Process Des. Dev., Vol.

19, No. 4, 1980

TOP

Bottom

2%

NPT

Figure 2. Permeation cell.

ber. The membrane rests on a porous stainless steel disk which is supported by three posts in the bottom of the cell. A schematic diagram of the entire flow system is shown in Figure 3. Gas mixtures were prepared by diluting pure SO2 and 1 and 10% mixtures of SOz with pure Nz.The gas mixture flowed through the upstream chamber of the permeation cell at an elevated pressure controlled by a back-pressure regulator. After passing through a rotameter, the mixture was sent to an SO2 scrubber and then exhausted to a fume hood. The downstream chamber of the permeation cell was maintained at 0.2 atm by a diaphragm vacuum pump. Flow rates on the downstream side were measured with a rotameter and a bubble flow meter. Gas analysis was done by taking 1.0-mL samples from the upstream and downstream lines with a gas-tight syringe and injecting them into a thermal conductivity gas chromatograph (Beckman GC-2A). The column for the gas chromatograph was packed with Chromosil310 (Supelco, Inc.) and operated a t a helium flow rate of 50 mL/min.

Results and Discussion Data Analysis. Permeation data for pure gases were analyzed by the method described in part 1. N2 and COP permeabilities were independent of pressure, but the SO2 permeability exhibited a strong pressure dependence. SO2 permeability data were correlated with a three-parameter exponential model

”)+

1 = P(0) O‘ - P(1)lpl“)exp( P(I)

ASP

dP

(9)

The analysis of data for mixtures was somewhat simpler.

N2 and SOz permeabilities were calculated from

where u is the permeation rate at STP. In these equations the driving force for permeation is the partial pressure

“1 Vacuum pump

Permeation cei I

Samde

Scrubber

I

U

t

: : + ‘

t

Diaphragm vacuum pump Vent

Figure 3. Schematic diagram of the flow system.

difference across the membrane. Results for Pure Gases. The permeability results for pure gases are given in Table 11. The membranes are relatively free of pinhole defects except for CTA/PA-52, which has an abnormally large N2 permeability. The column headed “re1 error” tells how well the three-parameter exponential model correlated the SO2 permeability results. The last two columns in Table I1 give Permeability ratios which indicate how selective the membrane would be for SOz at low concentrations. The effect of the thickness of the CTA layer on permeability is seen from the SO2results. CTA/PA-43 has the thickest CTA layer and, correspondingly, the lowest value of Q,(SO,) and the highest value of 7s. The CTA layer offers a significant resistance to SO2permeation even though it is only 2.5% of the composite film thickness.

Ind. Eng. Chem. Process Des. Dev., Vol. 19, No. 4, 1980 619

Table 11. Permeability (ofCTA/PA Composite Membranes" membrane

l , m

q(N2)

@CO,)

Q0(S02)

7s

AS

CTA/PA-43 CTA/PA-52 CTA/P A- 5 3 CTA/PA-5 8 CTA/PA-59

0.190 0.179 0.46 1 0.48 5 0.460

2.05 11.6 2.96 1.13 1.98

23.6 35.2 42.2 22.6 31.0

50.6 165 311 125 178

0.0121 0.00810 0.00633 0.00767 0.00783

0.00219 0.00199 0.00132 0.00150 0.00150

x 1O'O; rS, AS, (mmHg)-'. a Units: Q, ~cm3(STI')~cm)/(cm2~s~cmHg) and hS are optimum values of the parameters in eq 9.

re1 QdSO, )I Q,(SO, )I error, % Q(N,) Q(C0,) 4.8 24.7 2.14 4.69 4.5 14.2 3.9 105 7.37 5.53 4.1 111 4.9 89.9 5.74

Q(N,) and WCO,) are mean values. Q,(SOJ, rS, 20,

I

1

I

I

1

/; Symbol

Membrane

o

CTA/PA-43 CTA/PA-58

e(pm)

Symbol

Upstream Pressure

+

I00 m m Hg 200 400 600 800

0

A

0 . q'

0

0.190 0.485

A

-

0 X

1

-

-

1

IO0

200 300 400 500 Mean Pressure irnm H g )

600

700

Figure 4. Pressure dependence of the SOz permeability of composite membranes supported on Millipore filters. The line segments represent the three-parameter exponential model (eq 9). The parameters are given in Table 11.

,

I

Symbol

Membrane

(Ipml

c

CTA/PA-52 CTA/PA-53

0.179 0.461

00 - - 01

i

Mean Pressure i m m H g )

Figure 5. Pressure dependence of the SOz permeability of composite membranes supported on Celgard. The line segments represent the three-parameter iaxponential model (eq 9). The parameters are given in Table 11.

The other four memlbranes have CTA layers which are one-third as thick as the one for CTA/PA-43, and their SO2 permeabilities are more characteristic of polyacrylate alone. Reducing the thickness of the CTA layer, however, made these membranes weaker and more subject to pinhole defects. The pressure dependence of the SO2 permeability is plotted in Figures 4 and 5 for membranes supported on Millipore filters and Celgard, respectively. The abscissa is the mean pressure, the average of the upstream and downstream pressures. The use of the mean pressure reduces the number of independent variables so that the pressure dependence is easier to illustrate graphically. Experimental data were taken a t 100, 200, 400, and 760 mmHg upstream pressure, and the permeability results were correlated with the three-parameter exponential model which is represented by the line segments in these figures. In both figures the thinner membrane has the lower SO2permeability because the CTA layer is a larger percentage of the total thickness. If these figures are

O d

Id0

2b0 360 400 500 Mean Pressure ( m m Hg)

600

760

Figure 6. Pressure dependence of the SOz permeability of CTA/ PA-59 from constant- and variable- A P experiments. The variableAP data were correlated with the three-parameter exponential model (eq 9), which is represented by the curve. The values of the parameters are given in Table 11. The letters next to the plotted points indicate the order in which the constant-AP permeation runs were performed. The order shows that the SOz permeability depended on time and previous conditioning of the membrane.

superimposed, it is observed that the membranes supported on Celgard are more permeable than those supported on Millipore filters. This results from differences in the surface properties of the two supports. Near the end of some of the permeation runs, there is a noticeable decline in the SO2 permeability which may be caused by the immobilization of SO2 molecules in clusters within the membrane. All the permeation data thus far were taken by the manometric method, in which the pressure difference (AP) across the membrane changes with time. Such experiments may yield different results from those in which AP is held constant, as in the flow system experiments. To compare the permeability results for the pure gases with those for the mixtures, the methods of measurement should give comparable results. Variable- AP and constant-AP experiments were run on CTA/PA-59 with pure SOz to see whether or not the method of measurement makes a significant difference. The variable-AP SO2runs were done immediately after testing the membrane with N2 and C02. At the conclusion of these experiments, the membrane was left in the permeation cell for a period of three weeks under vacuum. Then, constant-AP SOz permeation runs were made by holding the upstream pressure constant and maintaining the downstream pressure close to zero with the vacuum pump. The permeation rate was measured by closing the valve to the vacuum pump and noting the initial pressure rise on the downstream side of the membrane. After recording data for one run, the downstream side was reevacuated and the above procedure was repeated until the initial pressure rise was the same for consecutive runs, indicating steady-state permeation. The results for the variable- AP experiments were correlated with the three-parameter exponential model and are represented by the curve in Figure 6. The plotted

620

Ind. Eng. Chem. Process Des. Dev., Vol. 19, No. 4, 1980

Table 111. Gas Separation Results for CTA/PA-43 SO2 concentrations, % run 4J 4K 4L 4M 4N 40 4P 5A 5B 5c 5D 5E 5F a

flow ratea 2.0 2.0 1.0 2.0 4.0 4.0 2.0 2.0 2.0

L/min at 25 'C, 1 atm.

upstream inlet 0 2.02 5.06 9.94 9.94 9.94 100

upstream outlet 0 1.86 4.63 7.71 8.55 9.05 100

0

0

0.110 0.210 0.5 23 0.921 0

0.108 0.203 0.500 0.881 0

cm3(STP)/(cm2.min).

downstream 0 35.2 70.1 87.9 91.0 92.3 100 0 1.95 3.76 9.50 17.5 0

permeation rateb 0.547 1.09 2.32 4.31 5.15 6.11 12.78 0.541 0.606 0.574 0.608 0.632 0.531

Q(N2)' 2.32 3.05 3.06 2.36 2.12 2.16

Q(S02 I/ Q(N2)

Q(S02)' 136 203 261 288 3 26 596

2.30 2.53 2.35 2.34 2.23 2.26

44.6 66.3 111.0 136.0 151.0 28.1 29.4 32.6 38.1

71.1 69.0 76.4 85.0

(cm3(STP).cm)/(cm2.s.cmHg) x 10''.

points are data from the constant-AP experiments. The letters next to the points indicate the order of the permeation runs. Where two letters are next to the same point, the results for consecutive runs were identical. At 100 and 200 mmHg upstream pressure the SO2permeability varied over a wide range. The results at 200 mmHg were obtained over several hours and suggest that the polymer chains were undergoing a slow change in orientation to accommodate more SO2. At higher upstream pressures the rate of change was much faster so that no transient behavior was observed. After testing a t 800 mmHg upstream pressure, experiments were immediately rerun at 100 and 200 mmHg, and higher values of the SOz permeability were obtained than previously. This confirms that the polymer structure changes according to the amount of SO2 present. Figure 6 shows that there is good agreement between the variable-@ and constant-AP experiments at high SO2 pressures. At low pressures the SO2 permeability is very sensitive to previous exposure or conditioning of the membrane, so good agreement is expected only if the polymer chains are in the same configuration. Higher permeation rates can be achieved by operating a membrane at elevated pressure. Only CTA/PA-43 and CTA/PA-58 were able to withstand upstream pressures up to 10 atm. The pressure dependence of the N2 permeability for these membranes is shown in Figure 7. For CTA/PA-43 the N2 permeability increased linearly with pressure and returned to its initial value when the pressure was reduced. The performance of CTA/PA-58 was similar, except that the N2permeability did not return to its initial value a t 1 atm after testing at 10 atm. The pressure dependence of the N2 permeability was unexpected because Nzdoes not interact with the polymer. The pressure dependence may be caused by physical deformation of the membrane. A comparison of Figures 4 and 7 indicates that the SOz permeability has a much steeper pressure dependence than the N2 permeability. This means that increasing the upstream pressure will favor SOz permeation over N2 permeation and increase the selectivity for SOz. Results for Binary Mixtures. CTA/PA-43 and CTA/PA-58 were tested on the flow system with binary mixtures of SO2and N2. Typical operating conditions were room temperature (25-27 "C), 10 atm upstream pressure, 0.2 atm downstream pressure, and 2 L/min (25 OC, 1 atm) upstream flow rate. The elevated upstream pressure was selected to give higher permeation rates and greater selectivity for SOz. The downstream pressure was set at the minimum operating pressure of the diaphragm vacuum

3 C

I

CTA/PA-43 CTA/PA-58

0.51 0

0

2

8

PRESSURE

IO

(atm)

Figure 7. Pressure dependence of the N2 permeability.

pump. The flow rate was chosen to obtain reasonably good mixing in the cell without excessive consumption of the calibrated gas mixtures. The results for CTA/PA-43 and CTA/PA-58 are given in Tables I11 and IV. All runs with the same number were performed on the same day. The flow rate refers to the gas mixture entering the upstream chamber of the permeation cell. SO2 concentrations are given for the gas streams entering and leaving the upstream chamber and for the permeant stream. N2and SO2permeabilities were calculated from the measured SOzconcentrations and the permeation rate. The SO2 concentration of the gas mixtures entering the permeation cell ranged from 0.1 to 10%. Pure N2at 10 atm and pure SOz at 1.1atm were also tested for comparison with the results for the gas mixtures. The results in the concentration columns indicate that the permeant stream is greatly enriched in SOz. For example, CTA/PA-43 produced an enriched stream of 2% SO2 from 0.1% SO2 upstream, 35% from 2%, and over 90% from 10%. The performance of CTA/PA-58 was similar, The permeability results for N2 and SO2 suggest that there is some interaction between the two gases. The N2permeabilities of both membranes were enhanced when they were first tested with concentrated SO2mixtures and later returned to the values for pure N2. The N2 results for CTA/PA-58 are difficult to understand because the permeability of pure N2 changed several times. The SO2

Ind. Eng. Chern. Process Des. Dev., Vol. 19, No. 4, 1980

621

Table IV. Gas Separati'on Results for CTA/PA-58

SO, concentrations, % upstream upstream outlet inlet downstream ~

~

run NZ 1A 1B

1c

1D 41 5G 5H 51 55 5K 5L a

flow rate a

0 1.84 4.70 9.48 0 100 0 0.105 0.204 0.5 22 0.921 0

2.0 2.0 2.0

4.0 2.0 2.0 2.0

L/min a t 25 'C, 1 at,m.

~~

permeation rate b

0 34.1 65.9 86.2 0 100 0 2.44 4.45 11.1 21.3 0

0 1.79 4.19 7.70 0 100 0 0.102 0.194 0.495 0.872 0

Q(N2 1' 3.21 4.65 4.45 4.16 3.38

0.295 0.645 1.17 2.61 0.311 6.46 0.218 0.220 0.226 0.250 0.284 0.223

Q(S0z 1'

&(SO,)/

Q(Nz 1 44 63 95

205 279 395 758

2.36 2.32 2.34 2.4 2 2.44 2.42

103 97.0 104 138

44.4 41.5 43.0 56.6

cm3(STP)/(cm2.min). (cm3(STP)~cm)/cm2~s~cmHg) x 10'".

Table V. Effect of Uplstream Pressure and Flow Rate o n the SO, Permeability of CTA/PA-58

SO, concentrations, % run 4A 4B 4c 4D

4E 4F 4G 4H 41 a

flow rate"

upstream pressure

upstream inlet

upstream outlet

downstream

permeation rateb

1.0 2.0 4.0 1.2 0.8 2.0 2.0

10 10 10 10 3 2 2 3 1.1

0 9.94 9.94 9.94 34.0 54.9 50.4 32.3 100

0 7.26 8.09 8.90 30.8 48.8 48.0 30.2 100

0 86.1 89.0 91.0 99' 99' 99 99' 100

0.246 2.06 2.60 3.05 3.85 4.05 4.16 3.90 6.46

L/min a t 25 'C, 1 atm.

+

Q(N,)' 2.67 3.30 3.31 3.21

333 383 413 557 547 577 581 758

101 116 129

cm3(STP)/(cmZ~min). (cm3(STP)~cm)/(cm2~s~cmHg) X 10". 10

' " ' I

I

8 60

Pure SO, at I I otm upstream pressure

w

0

Downstream pressure = 0 2 a t m for all cases

.E

4-

x

-

21

.-

0

2.

2 a -

06

/

0

04

Downstream Pressure = 0 2 a t m for a11 cases

I

U p s t r e a m P a r t l a l P r e s s u r e o f SO,

( m m Hg)

Figure 8. Comparison of the permeability of pure SOz and the permeability of SOz with Nz present for CTA/PA-43. The permeation dqta for pure SOz were obtained from constant-hP experiments on the manometric systeim and were correlated with the three-parS = 9.50 rameter exponential model (eq 9 with Qo = 4.71 X X and AS = 1.59 X The curve connects the predicted values of the permeability of pure SOz for the upstream and downstream SOz partial presslures of NZ/SOzmixtures run on the flow system. The plotted points are data from Table I11 for the N2/SO2 mixtures and pure SOz.

permeabilities of the mixtures had the anticipated pressure dependence, though tlhey were not as large as for pure SO2 a t the higher concen1:rations. This can be seen by comparing the SOz permeabilities for runs 40 and 4P in Table 111. The upstream partial pressures of SO2are nearly the same, but the permeability of pure SOz is almost twice as large as the permeability of SOz in the mixture. To better understand how the SO2 permeability was affected by the presence of Nz, experiments were run on CTA/PA-58 at different flow rates and upstream pressures while holding the SO2 partial pressure nearly constant. The results are presented in Table V. Increasing the flow

6

8 IO

7 1.1 (pure SO, 1

20

60 80100 200 400 600 1000 Upstream Partial Pressure of SO, ( m m H g i

40

Figure 9. Comparison of the permeability of pure SOz and the permeability of SOz with Nz present for CTA/PA-58. The permeation data for pure SOz were obtained from constant-@ experiments on the manometric system and were correlated with the three-parameter exponential model (eq 9 with Qo = 6.64 X lo+, -yS = 1.06 X and AS = 2.17 X The curve connects the predicted values of the permeability of pure SOz for the upstream and downstream SOz partial pressures of NZ/SOzmixtures run on the flow system. The plotted points are data from Tables IV and V for the NZ/SOzmixtures and pure SOz. The two clusters of points in the upper right corner of the figure show the effects of increasing the flow rate and reducing the upstream pressure.

rate increased the SOzpermeability slightly, which means that there may have been some gas phase resistance or incomplete mixing in the upstream chamber. A much greater increase in the SO2 permeability was achieved by reducing the partial pressure of N2. This may be attributed to competition between the N2 and SO2molecules for the diffusion paths through the polymer. The above phenomena are illustrated in Figures 8 and 9 in which the permeability of pure SO2is compared with

622

Ind. Eng. Chem. Process Des. Dev., Vol. 19, No. 4, 1980

the permeability of SO2in the mixtures. The permeability of pure SOz was determined from constant-AP experiments on the manometric system, similar to those described previously for CTA/PA-59. The data were correlated with the three-parameter exponential model. The upstream and downstream partial pressures of SOz for each flow system run were then substituted into the model to obtain values for the permeability of pure SO2, which are represented by the curves in Figures 8 and 9. Pure SO2 at 1.1atm was also tested on the flow system; the permeability data agreed well with those obtained on the manometric system. At low SOz concentrations the results for the mixtures are higher than those for pure SOz. The simultaneous permeation of Nz may prevent the SOz molecules from becoming immobilized in clusters and thus enhance the SOz permeability. There is also greater uncertainty in the permeability of pure SO2 at low pressures because of its dependence on the history of the membrane. Above 200 mmHg upstream partial pressure the mixture data fall below the curve for pure SO2. The logarithmic scale in these figures makes the effect look smaller than it actually is. At 600 mmHg the SO2 permeability of the mixture is only 60% of the value for pure SOz. The effects of increasing the flow rate and reducing the upstream pressure are depicted by the two clusters of points in the upper right corner of Figure 9. Increasing the flow rate has almost negligible effect, but reducing the upstream pressure narrows the difference between the mixture data and the curve for pure SO2. Because the SOz permeability is affected by the presence of Nz, SO2 permeabilities for the mixtures cannot be predicted accurately from the data for pure SOz. Evaluation of Composite Membrane Performance. Ward and Neulander (1970) proposed five design goals for membranes for recovering SOz from combustion or smelter gases. These pertain to the SOz flux, selectivity, operating temperature, chemical stability, and life. The goal for the SOz flux was 4 X cm3 (STP)/(cmz.s.cmHg), based on economic analysis. For the results in Table 11, the SOz fluxes in the limit of zero pressure (QO(SO2)/l) fall below the goal by factors of 4-15. Higher SO2fluxes and selectivity can be achieved by operating the membranes at elevated pressures, as demonstrated by the flow system experiments. The second goal was an SOZ/CO2permeability ratio of several hundred for applications involving combustion gases and at least 10 for smelter gas applications. The SOz/COzpermeability ratios in Table I1 are less than 10, which means that these membranes would be limited to smelter gas and other applications involving high SOz concentrations. In practice, higher SOZ/CO2permeability ratios would be observed because of the pressure dependence of the SOz permeability. The third goal was a minimum operating temperature of 100 "C. The polyacrylate elastomers have a service temperature range of -40 to 200 "C; however, their transport properties would change drastically at high temperatures. Typically, the permeability of the noncondensable species would increase much faster than that of the condensable species, resulting in a decline in selectivity. The chemical stability and life of the membranes have not been studied specifically; however, CTA/PA-43 and CTA/PA-58 performed well over a period of 10 months of intermittent testing at 1 and 10 atm pressure. As with the single-layer membranes described in part 1,the composite membranes fall short in meeting Ward and Neulander's design goals. However, it is probably more appropriate to compare the composite membranes

with other membranes developed for SOzseparations. The immobilized liquid membranes of Ward and Neulander (1970) had high SO2fluxes but were unable to withstand even 1 atm pressure difference across them. The sulfolene-plasticized polyvinylidene fluoride membranes of Zavaleta and McCandless (1976) were effective in separating binary mixtures of Nz and SOz, but gas fluxes were very low. With the composite membranes, high SO2fluxes and good selectivity were achieved without losing mechanical stability.

Summary Polyacrylate has properties which make it attractive for separating SOz from a mixture of gases. A technique was described for casting 0.2-0.5-pm films of polyacrylate on glass plates on which a 0.002-0.005-~mlayer of cellulose triacetate, a nonadhesive polymer, had been previously applied. The resulting two-layer films were easily removed from the glass and placed on microporous supports. This technique is applicable to any polymer which is completely soluble in a common solvent. Cellulose triacetate/polyacrylate composite membranes were made by this method and initially tested with pure N2, COz, and SOz. The permeation of N2 and COz was nearly ideal. The SO2permeability increased rapidly with increasing pressure and was well correlated with a threeparameter exponential model. The composite membranes were then tested with binary mixtures of Nz and SOz at 10 atm upstream pressure with a steady-state flow system. A typical membrane produced an enriched permeant stream of 2% SOz from 0.1% SOz upstream, 35% from 2%, and over 90% from 10%. The N2 permeability was enhanced by the presence of SOz, particularly at high SOz concentrations. At low SOz partial pressures the SO2permeability of the mixtures was higher than the permeability of pure SOz a t the same pressure. The opposite was true at high SOz partial pressures where the presence of Nz inhibited SOz permeation. The interactions between Nz and SO2made it difficult to accurately predict the permeation rates of the mixtures from data for the pure gases. Membrane performance was superior to that of other membranes developed for SO2separations; however, it still fell short of design goals for combustion and smelter gas cleaning. The unique properties of the composite membranes make them attractive for instrumentation and other specialty applications. Nomenclature A = membrane area, cmz ,r = gas flux through membrane, cm3(STP)/(cm*.scmHg) I = membrane thickness, cm n d = flow rate leaving downstream chamber of permeation cell, mol/s nu,in= flow rate entering upstream chamber of permeation cell, mol/s nqmt = flow rate leaving upstream chamber of permeation cell, mol/s P = pressure, cmHg or mmHg ilp = pressure difference across membrane, cmHg or mmHg P(0) = pressure on upstream side of membrane, mmHg P(1) = pressure on downstream side of membrane, mmHg Pd= pressure on downstream side of permeation cell, cmHg P, = pressure on upstream side of permeation cell, cmHg P, = standard pressure of 76 cmHg Q = permeability, cm3(STP).cm/ (cm2.s.cmHg) Qo = preexponential constant for permeability, cm3(STP). cm/(cm2.s.cmHg) R = gas constant = 6236 cm3-cmHg/(mol.K) S = solubility, cm3(STP)/(cm3-cmHg)

Ind. Eng. Chem. Process Des. Dev. 1980, 19, 623-629

T,= standard temperature of 273 K u = volumetric flow rate leaving downstream chamber of

permeation cell, cm3(,STP)/s = so2mole fraction of gas in downstream chamber of permeation cell xu = SOz mole fraction of gas in upstream chamber of permeation cell xU,in= SOz mole fraction of gas entering upstream chamber of permeation cell xu,,ut = SO2 mole fraction of gas leaving upstream chamber of permeation cell Greek Letters y = constant in three-parameter exponential model, eq 9 h = constant in three-parameter exponential model, eq 9 xd

623

Literature Cited Carnell, P. H., Cassidy, H. G., J . Polym. Sci., 5 5 , 233 (1961). Kuehne, D. L., Ph.D. Thesis, California Institute of Technology, Pasadena, Cali., 1979. Kuehne, D. L., Friedlander, S. K., Ind. Eng. Chem. Process Des. Dev., preceding article in this Issue, 1980. Levich, V. G.. "Physicochemical Hydrodynamics", p 681, Prentice-Hall, Engla wood Cliffs, N.J., 1962. Ward, W.J., 111, Neulander, C. K., "Immobilized Liquid Membranes for Sulfur Dioxide Separation", PB-191-769, U. S. Department of Commerce, 1970. Zavaleta, R . , McCandless, F. P., J. Membr. Sci., 1, 333 (1976).

Received for review October 22, 1979 Accepted June 12, 1980

Presented at the 87th AIChE National Meeting, Boston, Mass., Aug 1979.

Gasification of Feedlot Manure in a Fluidized Bed Reactor. The Effect of Temperature K. Pattabhl Raman, Walter P. Walawender,' and L. T. Fan Department of Chemical Engineering, Kansas State University, Manhattan, Kansas 66506

Dried feedlot manure was gasified in a 22.9 cm i.d. fluidized bed reactor. The effect of reactor temperature on the yield, composition, and heating value of the produced gas and also on the energy recovery of the process was studied. The results were compared with those predicted by thermochemical calculations as well as with the data obtained for the gasification of municipal waste.

Introduction The conversion of agricultural residues into fuel has received increasing attention in recent years. The upward spiral in the price of fossil fuels coupled with impending shortages has made algricultural residues an attractive candidate as a supplemental source of energy. Agricultural residues represent a renewable, low sulfur energy resource, and they are available in significant amounts in select areas of the country. Feedlot manure is one such residue that is being considered for possible utilization. Several process routes have been pro;posed for converting manure into useful products, namely, anaerobic digestion, gasification at atmospheric pressure, hydrogasification, and liquefaction. Of these, atmospheric gasification would appear to be an attractive proposition on an economic basis (Engler e t al., 1975). When complex organic compounds such as feedlot manure are heated to higih temperatures, a series of physical and chemical changes occur resulting in the evolution of volatile products and a carbonaceous solid residue. The extent of volatile yield and composition will depend on the type of material used, the manner of heating (Anthony and Howard, 1976), and the environment under which the gasification reaction is performed. Many contacting devices such as the fixed bed (MRI, 1971), the moving bed (Smith et al., 1974), the entrained bed (Mikesell et al., 1978), and the fluidized bed (Burton, 1972; Huffman and Liberick, 1978), have been used for studying the gasification of manure. From the standpoint of gas production, fluidized beds are highly desirable because of their high heat transfer characteristics and their capabilities for maintaining a uniform high-temperature environment.

A survey of the literature on the gasification of manure indicated that the available experimental data are somewhat limited. Burton (1972) carried out two experimental runs with dried cow manure in a fluidized bed reactor. The reactor used was 38 cm in diameter and composed of sand as an inert matrix. The fluidizing gas for the reactor was generated by combusting methane or propane. The reactor operating temperatures used for the two runs reported were 1042 and 1022 K. Smith et al. (1974) published partial oxidation data obtained in a moving bed reactor using manure as the feed material. The experiments were conducted in a 5 cm diameter reactor and used recycled product gas and air as the gas medium. The data were obtained for a temperature range of 894 to 950 K. Bench scale operating data were obtained by Halligan et al. (1975) in a 5 cm diameter reactor, which was operated in a partial combustion mode with steam and air. The reactor was externally heated with electrical heaters, and the data were obtained between 966 and 1069 K. Mikesell et al. (1978) reported limited data on the flash pyrolysis of steer manure in a multiple hearth reactor. The operating temperatures for these experiments were between 873 and 1023 K. Recently, Beck et al. (1979) presented partial oxidation data on manure obtained in a pilot plant reactor. Steam and air were used as the fluidizing medium in the 450 kg/day pilot plant. The reactor used was 15 cm in diameter and had an axial temperature variation of about 500 K. The data were collected at an average temperature of 870 K in the reactor. To properly design a system for the gasification of manure or other biomass, it is necessary to develop systematic data based on the effect of operating temperature, particle

0196-4305/80/1119-0623$01.00/00 1980 American Chemical Society