Ind. Eng. Chem. Res. 1993,32, 159-165 Le Van, M. D.; Vermeulen, T. Binary Langmuir and Freundlich Isotherms for Ideal Adsorbed Solutions. J. Phys. Chem. 1981,85, 3241-3250. Liapis, A. I. Modelling Affinity Chromatography. Sep. Purif. Methods 1990,19(2),133-210. Liapis, A. I.; McCoy, M. A. Theory of Perfusion Chromatography. J . Chromatogr. 1992,599,87-104. Liapis, A. I.; Anspach, B.; Findley, M. E.; Davies, J.; Hearn, M. T. W.; Unger, K. K. Biospecific Adsorption of Lysozyme onto Monoclonal Antibody Ligand Immobilized on Nonporous Silica Particles. Biotechnol. Bioeng. 1989,34,467-477. Mao, Q.M.; Johnston, A,; Prince, I. G.; Hearn, M. T. W. High performance liquid chromatography of amino acids, peptides, and proteins. CXIII. Predicting the performance of non-porous particles in affinity chromatography of proteins. J. Chromatogr. 1991,548,147-163. Muller, A. J.; Carr, P. W. Examination of Kinetic Effects in the High-Performance Liquid Affinity Chromatography of Glycoproteins by Stopped-Flow and Pulsed Elution Methods. J. Chromatogr. 1984,294,235-246. Onyegbado, C. 0.;Susu, A. A. Theoretical Study of Nonlinear Adsorption Isotherm by Chromatography. 1. Application to Methyl Linoeate and Methyl Oleate Adsorption on Supported Copper Catalyst. Chem. Eng. Commun. 1990,91,79-90. Place, H.; SBbille, B.; Vidal-Madjar, C. Split-Peak Phenomenon in Nonlinear Chromatography. 2. Characterization of the Adsorption Kinetics of Proteins on Reversed-Phase Supports. Anal. Chem. 1991,63,1222-1227. Ruthven, D. M. Principles of Adsorption and Adsorption Processes; Wiley: New York, 1984;pp 49-52. Skidmore, G. L.; Horstmann, B. J.; Chase, H. A. Modelling SingleComponent Protein Adsorption to the Cation Exchanger S Sepharose' FF. J. Chromatogr. 1990,498,113-128.
159
Van Cott, K. E.; Whitley, R. D.; Wang, N.-H. L. Effects of Temperature and Flow Rate on Frontal and Elution Chromatography of Aggregating Systems. Sep. Technol. 1991,1,142-152. Wade, J. L.; Carr, P. W. Split Peaks in Non-Linear Chromatography and Their Effect on Sample Throughput in Large Scale Separations. J. Chromatogr. 1988,449,53-61. Wade, J. L.; Bergold, A. F.; Carr, P. W. Theoretical Description of Nonlinear Chromatography, with Applications to Physiochemical Measurements in Affinity Chromatography and Implications for Preparative-Scale Separations. Anul. Chem. 1987,59,1286-1295. Whitley, R. D. Dynamics of Nonlinear Multicomponent Chromatography-Interplay of Mass Transfer, Intrinsic Sorption Kinetics, and Reaction. Ph.D. Dissertation, Purdue University, 1990. Whitley, R. D.; Wachter, R.; Liu, F.; Wang, N.-H. L. Ion-Exchange Equilibria of Lysozyme, Myoglobin, and Bovine Serum Albumin: Effective Valence and Exchanger Capacity. J.Chromatogr. 1989, 465,137-156. Whitley, R. D.; Van Cott, K. E.; Berninger, J. A.; Wang, N.-H. L. Effects of Protein Aggregation in Isocratic Nonlinear Chromatography. AZChE J. 1991a,37 (4),555-568. Whitley, R. D.; Berninger, J. A.; Rouhana, N.; Wang, N.-H. L. Nonlinear Gradient Isotherm Parameter Estimation for Proteins with Consideration of Salt Competition and Multiple Forms. BiotechnoL Prog. 1991b,7 (6),544-553. Wilson, E. J.; Geankoplis, C. J. Liquid Mass Transfer a t Very Low Reynolds Numbers in Packed Beds. Znd. Eng. Chem. Fundam. 1966,5 , 9-14.
Received for review March 30, 1992 Revised manuscript received October 6, 1992 Accepted October 26, 1992
Adsorption Equilibrium and Kinetics for Multiple Trace Impurities in Various Gas Streams on Activated Carbon Timothy C. Golden* and Ravi Kumar Air Products and Chemicals, Znc., Allentown, Pennsylvania 18195-1501
Equilibrium and kinetic adsorption data for seven trace impurities (propylene, Freon-12 (CF2C12), n-butane, methylene chloride, acetone, n-hexane, toluene, and Freon-22 (CHFCl,)) from various carrier gases (helium, nitrogen, methane, carbon dioxide, and a mixture of methane and carbon dioxide) are provided. Activated carbon a t several temperatures and pressures is used as the adsorbent. Two empirical characteristic curves, one relating equilibrium isotherms of trace impurities with their physical properties and the other relating mass-transfer coefficients with equilibrium properties, are generated. These can be used to predict equilibrium capacities and mass-transfer zone lengths for multiple trace impurities from a carrier gas and design a thermal swing adsorption clean-up system.
Introduction Equilibrium isotherms and mass-transfer zone lengths for trace impurity removal are needed to design thermal swing adsorption (TSA) systems for cleaning gas streams before bulk gas separation can be carried out (Golden et al., 1991a,b; Kumai and VanSloun, 1989). Some of the gas streams of interest, such as landfill gas, contain hundreds of trace impurities which have to be removed prior to further processing of the bulk gas (Kumar and Golden, 1991). This preaents a challenging experimental problem. In addition, the types of impurities and their concentrations vary from landfill to landfill. This emphasizes the need for a general approach to design TSA clean-up systems in the case of landfill gas purification. In this study, we demonstrate an approximate but practical approach to handle these types of problems with minimum experimental data. As an example, removal of several trace
impurities from landfill gas is discussed. The approach used was to identify seven key components from a long list of possible trace impuritiea in landfill gas. This list contained representative species from each of the following categories-chlorofluorocarbons, alkanes, olefins, oxygenated species, and aromatics. Carbon dioxide, the most strongly adsorbed gas species in the bulk of landfii gas which consists of a mixture of carbon dioxide and methane, was chosen as the carrier gas. Since carbon dioxide is more strongly adsorbed than methane, depression in trace component loading due to carbon dioxide co-adsorption is expected to be higher than the corresponding depression from a mixture of carbon dioxide and methane. Therefore, carbon dioxide carrier gas provided a conservative estimate for TSA design. A recirculating apparatus was constructed for measuring equilibrium isotherms and mass-transfer coefficients for
0888-588519312632-0159$04.00/00 1993 American Chemical Society
160 Ind. Eng. Chem. Res., Vol. 32, No. 1, 1993 ADSORBENT BED
PREHEATER
BYPASS LOOP INDICATOR
SEPTUM 'PORT
DRY TEST METER
SEPTUM PORT I 0
RECIRCULATING PUMP
U
Figure 1. Schematic of the recirculating apparatus used to measure equilibrium and kinetic data.
the seven trace impurities from carbon dioxide carrier gas. The recirculating apparatus was chosen because (1)it requires small quantities of gas and adsorbent, (2) the total gas inventory is low (therefore, safety hazards associated with some of the trace impurities are reduced), (3)it provides "mixturen equilibrium and mass-transfer data in a single experiment, and (4) it is much faster and more accurate than any other technique for measuring these properties. The equilibrium data for the seven trace impurities from the carrier gas were then correlated to a single characteristic curve with physically known properties of the trace impurities. The mass-transfer coefficients were then correlated with equilibrium capacity generating a second empirical characteristic curve. Therefore, two characteristic curves were obtained, one which predicts equilibrium capacities from known trace impurity physical properties and one which predicts mass-transfer coefficients from the estimated equilibrium capacities. These were then used to design a TSA gas clean-up system for removing any trace impurity with known physical properties from landfill gas. Details of this approach and its application are discussed below.
Experimental Technique Figure 1 provides a schematic of the recirculating apparatus used in this study. The apparatus is constructed of stainless steel including 1/4-in.tubing, mixing vessels, an admrbent cell (-10 cm3),various ball valves, and a high speed Teflon diaphragm pump. The gas flow rate in the experiments was 10 L/min. The adsorbent employed in all experiments was type BPL activated carbon from Calgon (6 X 16 mesh). Adsorption measurements were conducted by placing a known amount of the trace impurity in the recycle or bypass loop of the apparatus at a known temperature, pressure, and system volume. Generally, the trace impurity was added as a pure component and then diluted with carrier gas. A Perkin-Elmer 3420 gas chromatograph equipped with a flame ionization and a thermal conductivity detector was used to analyze the trace impurity concentration. The volume of sample was much less than withdrawn for measurement (- 1cm3) the total system volume (--4ooo cm3). The trace impurity in the recycle loop was then directed to the adsorption bed which contained regenerated adsorbent, and the concentration decrease was monitored over time yielding data on the impurity uptake rate. In all cases,the activated carbon was regenerated in flowing N2at 150 "C. Once sufficient time was allotted for a constant impurity concentration (- 1 h), the final equilibrium concentration was noted and the extent of impurity adsorption was calculated by difference. Other adsorption or desorption points were ob-
100 150 200 IMPURITY CONCENTRATION, ppm
50
250
Figure 2. Equilibrium isotherms for trace impurities from carbon dioxide carrier gas at 721 kPa, 311 K on activated carbon: (1) Freon-12, (2) n-butane, (3) propylene, (4) n-hexane, ( 5 ) acetone, (6) toluene, and (7) methylene chloride. Table I. Henry's Law Constants for Key Impurities (Carrier Gas, CO,: T = 311 K: Adsorbent. Activated Carbon) Henry's law constant (g-mol/g) compound P (kPa) = 721 P (kPa) = 110 propylene 0.10 0.135 Freon-12 0.33 0.504 n-butane 1.28 2.32 methylene chloride 1.36 acetone 3.17 n-hexane 17.8 toluene 55.7
tained by either increasing or decreasing, respectively, the impurity concentration in the bypass loop. Due to the large recirculating flow, the system behaved in an essentially isothermal manner with a maximum temperature rise of 2 "C.
Adsorption Equilibrium If the initial (t = 0) loading of the trace impurity is ni and initial gas-phase mole fraction is yo, then the final gas-phase mole fraction of the trace impurity at equilibrium (t = tf), yf, can be used to calculate equilibrium capacity of the trace impurity (nf) at pressure P and temperature T by mass balance on the trace impurity (Appendix, eq A2):
Equilibrium isotherms obtained in this manner are plotted in Figure 2, and the corresponding Henry's law constants are listed in Table I. Henry's law constants for trace impurities increase as the total system pressure decreases. This suggests that as the total system pressure decreases, the effect of carbon dioxide co-adsorption decreases resulting in a more favorable condition for the adsorption of trace impurities and therefore higher Henry's law constants.
It is apparent from Figure 2 that equilibrium isotherms for the trace impurities are significantly different from each other, and given the fact that landfill gas contains hundreds of trace impurities, one can only painstakingly measure all the neceesary equilibrium isotherms. An approach, based on the Dubinin-Radushkevich equation (Dubinin et al., 19471,was taken to correlate the equilibrium isotherms for all the trace impurities. This equation has been used previously to correlate equilibrium adsorption data for pure components and is being employed in the current study to correlate equilibria from a binary gas mixture treated as a pseudo pure component isotherm.
Ind. Eng. Chem. Res., Vol. 32,No. 1, 1993 161 Table 11. Physical Properties of the Trace Impurities molecular weight (g/mol) 42
compound name
chemical formula C3HB CClzFz CIHlO CHzClz CH3COCH3 CBH14 C6H5CH3 CHFClz
propylene Freon-12 (dichlorodifluoromethane) n-butane methylene chloride acetone n-hexane toluene Freon-22
1x10 1
121
58 50.5 58
ia
92
mo1ecu1ar volume, V (cm3/g-mol) 85.7 96.6 100 69.9 15.7 134.9 107.4
parachor, p (cm3/g-mol) 151.2 171.6(2) 190.3 150.4 161.7 210.8
245.7 156
p+44
1.0
saturation, P, (kPa) 1540
8, ( p / p m # / *
902
0.89 0.95
425
1
101
0.89 0.92
53.7 39.5 8.0 405
1.19
1.14 0.91
callgmols
%”
KH
grnole 9
q = 4 2 6 3 cal/grnole
0.10
0.01 ..
2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0
103/T, (OK)”
-4 i8.. v i n I”
0
1
2
3
4
5
6
1
8
(RT /p en PS 1 P Y ) x~10.’ Figure 3. Modified Dubinin-Radushkevich plot for the equilibrium isotherma of trace impurities from carbon dioxide carrier gas at 721 kPa, 311 K on activated carbon: (1) Freon-12, (2) n-butane, (3) propylene, (4) n-hexane, (5) acetone, (6) toluene, and (7) methylene chloride. W,= 0.5 cm3/g (-); PT = 110 kPa.
The governing Dubinin-Radushkevich equation is given below:
In this equation the affinity coefficient, 8, which compares the strength of the adsorptive interaction of the adsorbate in question to that of some reference substance, was defined as
B = (P/P,,31’2
(3)
where p is the molecular parachor of the adsorbate. The molecular parachor is a secondary derived function dependent on the primary properties of surface tension, density, and molecular weight (Quayle, 1953). It was developed in an attempt to correlate the structure of molecules with their physical and chemical properties. The parachor of a molecule is given by the equation p = vma1/4
(4)
where V, is the molar volume and a is the surface tension. Thus, a comparison of parachors of liquids is equivalent to a comparison of molar volumes under conditions that the liquids have the same surface tension. To the authors’ knowledge, this is the first time a value of /3 as the square root of the ratios of parachors has been used to correlate adsorption data Table I1 summarizes necessary physical parameters and affinity coefficients, 8. These values of /3 were arrived at by arbitrarily choosing n-butane as the reference compound for the trace impurities. Other ref-
Figure 4. Effect of the carrier gas on Henry’s law constant for nitrogen, Freon-12 on activated carbon at 721 P a : ( 0 )helium, (0) (X) 43% carbon dioxide + 57% methane mixture, and (e) carbon dioxide.
erence compounds could have been chosen without effecting the fit of the plot. Figure 3 plots the seven isotherms of Figure 2 according to eq 2. It is observed that equilibrium isotherms for such different species correlate well by eq 2. It should be noted that even though the correlation in Figure 2 appears quite good, the DubininRadushkevich plot is very insensitive. Nonetheless, Figure 3 combined with eqs 2-4 provides a general correlation to predict the equilibrium capacities of trace impurities at the test conditions from carbon dioxide carrier gas. This correlation can be used to predict equilibrium isotherms for trace impurities on activated carbon if the physical properties of the trace impurity are known. The value of W o(micropore volume) obtained from the correlation is 0.5 cm3/g. The total open pore volume of BPL carbon from porosimetry data is 0.7 cm3/g, and the micropore volume determined by C02 adsorption is 0.4 cm3/g. The effect of total pressure on the equilibrium capacity of the trace impurities was also measured in the recirculating apparatus. As discussed above, the amount of adsorbed trace impurity increased with decreasing total pressure of the carrier gas. The equilibrium isotherms for trace impurities are related by another characteristic curve described by eq 2 and plotted in Figure 3. The same characteristic curve failed to describe the equilibrium isotherms at two different total pressures. This pinta out the need to measure equilibrium data at the design pressure of the system. Insufficient data at low values of the adsorption potential parameter exist to accurately determine the micropore volume of the activated carbon, which should be equal to that obtained from the high pressure data. The effect of carrier gas on the equilibrium capacity of the trace impurities was also measured in the recirculating apparatus. In addition to carbon dioxide, helium, nitrogen, methane, and a mixture of 43% carbon dioxide and 57%
162 Ind. Eng. Chem. Res., Vol. 32, No. 1, 1993 Table IV. Mass-Transfer Coefficients for Key Impurities (Carrier Gas, COz; T = 311 K; Adsorbent, Activated Carbon; P = 721 kPa) compound k. (8-I) propylene 5.42 x 10-3 Freon-12 3.21 x 10-3 n-butane 1.00 x 10-3 methylene chloride 8.01 x 10-4 acetone 4.57 x 10-4 n-hexane 5.11 x 10-5
q = 4 1 7 9 caligrnole
/ KH
//
gmole 9
q = 3424
caligrnole
q = 2607
caligmole
effective mass-transfer coefficient (k,) in the linear driving force model: dn/dt = k,(y - 9 ) (5)
Af
0.10
u
3.0
3.2 3.4 IO~IT, ( O K ) ”
3.6
Figure 5. Effect of the carrier gas on Henry’s law constant for Freon-22 on activated carbon a t 721 kPa: ( 0 )methane, (X) 43% carbon dioxide + 57% methane mixture, and (a) carbon dioxide. Table 111. Effect of Carrier Gas on the Henry’s Law Constants and Mass-Transfer Coefficients (Total Pressure 721 kPa)
-
He
NZ
43% CO:, + 57% CH4
COZ
CHI 43% Cop + 57% CH, COZ
Freon-12 104 3.19 104 3.19 47 3.55 50 3.53 70 3.40 96 3.24 70 3.40 100 3.22 106 3.18 239 2.58
4.40 2.20 1.40 1.30 0.82 0.50 0.46 0.33 0.27 0.08
Freon-22 68 3.41 95 3.25 50 3.53 69 3.41 91 3.27 70 3.40 100 3.20
0.63 0.45 0.36 0.29 0.23 0.13 0.10
3.67 5.00
X X
3.21
X
3.89 X 4.44 x 10-3 4.14 4.78
X X
methane were chosen as the carriers. Figures 4 and 5 respectively plot Henry’s law constants for Freon-12 and Freon-22 from these carrier gases as functions of temperature. From adsorption isotherms at different temperatures, the isosteric heat of adsorption can be calculated using the Clausius-Clapeyron equation. This equation is thermodynamically correct for pure components only; hence, the heat of adsorption calculated from mixture data is an “apparent” heat of adsorption. It is interesting to note that Henry’s law constants and apparent heats of adsorption for the trace impurities increase as the expected strength of adsorption for the carrier gas on activated carbon decreases (He < N2< CH, < CH,/COp mixture < COJ. Also, temperature variation of trace Henry’s law constant from carrier gas demonstrate the same behavior as expected for pure trace components. The values plotted in Figures 4 and 5 are listed in Table 111. This study demonstrates the importance of carrier gas in designing TSA systems and provides an additional degree of freedom in developing an optimized process (Kumar and Golden, 1991). Adsorption Kinetics The mole fraction of a trace impurity as a function of time was measured, as described above, from the recirculating apparatus. This information was used to calculate
As shown in the Appendix, if the left hand side of eq A5 is plotted against time on a semilogarithmic scale, then a straight line is indicative of the validity of the linear driving force model given by eq 5 or its equivalent: dn/dt = k,(ii - n) (6) It should be emphasized that, due to the presence of a carrier gas, there is some degree of counterdiffusion in these experiments. Therefore, some deviation from linearity is expected. The plots in Figure 6 confirm the validity of the linear driving force model for these compounds. Effective mass-transfer coefficients, k,, for six of the trace impurities are listed in Table IV. As noted in Table IV, there is a large variation in effective mass-transfer coefficient for the trace impurities, and therefore a generalized correlation is needed. Figure 7 plots the effective mass-transfer coefficient for the trace impurities against the corresponding Henry’s law constant. It is observed that as the Henry’s law constant increases, the effective mass-transfer coefficient decreases. This suggests that more strongly adsorbed species diffuse more slowly on the carbon surface. Therefore, Figure 7 can be used to obtain an effective mass-transfer coefficient for the trace impurity from COPcarrier gas on activated carbon at the test conditions. The effect of carrier gas and temperature on the mass-transfer rates was also measured for the trace impurities. Table I11 shows that the carrier gas also has a significant effect on the rate of mass transfer for the tested trace compounds. This again demonstrates the importance of the carrier gas on TSA designs. Even though mass-transfer coefficients were determined at only two temperatures, it appears that the maas-transfer coefficients showed an Arrheneius type temperature dependence, as noted in Figure 8. This rules out bulk (T1.5) or Knudsen (P5) diffusion as the controlling mechanism for mass transfer. In addition, the activation energies are about half the corresponding heats of adsorption (Figures 4 and 5). It has been shown that the temperature coefficients of surface diffusivity for adsorbates (D,) me given by D,= Da0exp[-xq/RT] (7) where q is the isosteric heat of adsorption, D , O is a constant, and x can vary between zero and unity, but typically has a value of 0.45 for physical adsorption of pure gases (Sladek et al., 1974). This suggests that surface diffusion may be the controlling mechanism of mass transfer for these species. At the gas flow rate and adsorbent particle size used in this study, film diffusion resistance was neg ligible. Also, from the pore size distribution of BPL carbon, the macropore diffusivity was calculated for each impurity with a technique shown in the literature (Wakao and Smith, 1962). The experimentally measured masstransfer coefficients were greater than the calculated ma-
Ind. Eng. Chem. Res., Vol. 32, No. 1,1993 163 1
5 L. H. S. OF 0.2 EQUATION A-5 0.1
0.02 0.05
EQUATION
1
0.005 O.O1 0 ~
8 8
2
6
4
8
10
0
"1
7
0 3RD ADSORPTION
0
0.1
12
14
16
1
0.05
I 200
I 100
OF
0.2
0.05 0.02
EQUATION A.5
0
1
I B 2
4
6
8
400
e 0.1
1
I
0
0
0 O. 0. O015 0
I
I 300
TIME (SEC)
TIME (MIN)
L. H. S. OF
10
12
14
16
0.02 0 .' 0 15 0.005
e
e 2
0
4
6
8
1
0
1
2
1
1
7
L. H. S.
::::I A
EQUATION A-5
A
0.1
7
A 0.005 0.01 0.02
..
2
4
7
0.1
0.05
0.01
0
0
7
0.5
L. H. S.
i
4
TIME (MIN)
TIME (MIN)
EQUATION A-5
1
2ND ADSORPTION
0
EQUATION A-5 0.1
@
7 1ST ADSORPTION
L. H. S. 0.5
8
6
10
12
14
16
I 2
0
4
6
8
10
12
14
16
TIME (MIN)
TIME (MIN)
Figure 6. Trace impurity uptake from C02carrier at P = 721 kPa, T = 311 K (a) propylene, (b) Freon-12, (c) n-butane, (d) methylene chloride, (e) acetone, and (D n-hexane. 1x10
I
Freon-22 1 1
-
kn S'
' kn
lXd-
rx1ci51 0.01
(secl)
I
0.10
I
1 .o
I 10.0
1C
E = 3 8 4 8 callgmole
0
K" gmole / g
Figure 7. Variation of ma-transfer coefficient on activated carbon with Henry's law constant for various trace impurities: PT = 721 kPa; T = 311 K; carrier gas, carbon dioxide.
cropore mm-transfer coefficient indicating a pardel pore and surface diffusion mechanism (Golden et d., 1991b). thenfurther to explain the linear dependence of k, on KH,as demonstrated by Figure 7. Column Performance Successful design of a thermal swing adsorption system depend upon the ability to predict impurity breakthrough
1x10 3.2
3.4
3.6
1031T, ( O K ) "
Figure 8. 'ariation of mass-transfer coefficients on activated carbon with temperature: PT = 721 kPa; carrier gas, 43% carbon dioxide + 57% methane. SYmbOls: ( X ) Freon-12; ( 0 )Freon-22.
from a column, Therefore, the above outlined approach was tested bv measuring breakthroueh curves for two trace impurities: Freon-12 &d Freon-22 From 43% COz + 57% CH, carrier gas. Mass-transfer and equilibrium values obtained from the recirculating apparatus as outlined in the above discussion were used to predict the breakthrough
164 Ind. Eng. Chem. Res., Vol. 32, No. 1,1993 Table V. Summary of Breakthrough Exwriments P = 721 kPa T = 283 K L = 1.5 f t Carrier Gas, 43% COz + 57% CH, ~b 30 lb/ft3 Estimated Adsorption Parameters from Recirculating Apparatus KH (gmol/g) k, (8-9 Freon-12 1.32 2.85 x 10-3 Freon-22 0.36 3.65 x 10-3
Appendix prior to the start of the experiment (t C 0), the gas-phase mole fraction of the impurity in the constant volume (V) recirculating apparatus is yi. The corresponding adsorbent loading in equilibrium with the initial conditions (P, T, yi) is ni. For a 'clean" adsorbent, ni and yi = 0. At time t = 0, the adsorbent is exposed to a known amount of impurity in the gas phase, yo. Since, only a trace amount of impurity is introduced, it is assumed that the uptake measurement is isobaric and isothermal. Therefore, if the total weight of the adsorbent is W, mass balance at any time t > 0 gives
dn PV dY w-=---
RT dt
dt
P:M
20 10
I I J/ cI L / y// "
0
X
I
.
or dn = -a dY dt
I
I
I
10
12
14
where
I
0
2
6
4
8
1
1
16
18
20
a3 s
tx 1
Figure 9. Comparison of experimental data and theoretical predictions (-) for Freon-12 ( 0 )and Freon-22 (X) breakthrough curves on activated carbon: PT = 721 P a ; T = 283 K carrier gas, 43% C02 + 57% CH,; L = 1.5 ft. Parameter flow rate in (g-mol/cm2)/s.
curves using Rosen's equation (Rosen, 1954) for a linear isothermal system:
-Y- - -1 erfc Yfeed
2
(
dt
)
1 - tl/tm 2(l/k,tm)1/2
(8)
where, neglecting gas-phase accumulation, t, = Lp6(H/G. Table V lists the details of the experimental unit used to measure breakthrough curves and operating conditions. Table V also lists the Henry's law constants and masstransfer coefficients estimated from the recirculating apparatus. Figure 9 compares the predicted breakthrough curves (eq 7) against data for Freon-12 and Freon-22 from 43% carbon dioxide + 57% methane carrier gas. The match is reasonable. In addition, breakthrough time (tl) for Freon-12 under similar operating conditions at a flow rate of 1.40 X (g-mol/cm2)/s was observed to be 8190 s and predicted to be 8272 s from eq 8 (not shown in Figure 9). Again, this shows a good match between experimental data and predictions using the above approach. Therefore, it is concluded that independently measured mass-transfer and equilibrium data from the recirculating apparatus can be used to predict column performance.
Conclusion The two characteristic curves, Figure 2 for equilibrium capacity prediction with molecular parachor as the correlating parameter and Figure 7 for mass-transfer coefficients with Henry's law constant as the correlating parameter, provide a method to predict column performance for multiple trace impurities of known physical properties. This method can be generalized for any system (at a given temperature, pressure, and carrier gas) and provides an easy and accurate method to design TSA systems with a minimum amount of experimental data. Acknowledgment The authom are thankfd to Air Products and Chemicals, Inc., for their permission to publish this work.
a = PV/
WRT
Integration from t = 0 to any time t yields n=ni+a(yo-y) tLO (A21 It should be noted that n is in equilibrium with y only at the final time at which the loading corresponding to the gas-phase mole fraction of yf is nf. Also, ni is not in equilibrium with yo. If the rate of uptake is given by dn dt = ky(y - 9) and the equilibrium isotherm is linear where 9 represents the gas-phase mole fraction inside the adsorbent in equilibrium with ita loading. Equation A3 can therefore be rewritten as
dn dt = k y ( y - 2 ) Substituting the left hand side from eq A1 yields
(
- a dy z = k y y - - H:) Further substitution by eq A2 yields
Integrating this equation from initial conditions (t = 0, y = YO, = ni) to anytime during the experiment yields the following expression for the uptake of trace impurity:
where y is the mole fraction of the trace impurity at time t in the recirculating apparatus. Similarly, if the rate of uptake is given by dn/dt = k,(ii - n) (A61
Ind. Eng. Chem. Res. 1993,32,165-172
166
It can be shown that eq A5 is obtained by replacing k, with
Pb = bulk density of packed bed, lb/ft3
KHkW
Subscripts
Nomenclature C 5 flow rate, (g-mol/cm2)/s KH = Henry's law constant for trace impurity, g-mol/g k, = effective mass-transfer coefficient in linear driving force model based upon solid-phase loading, 8-l k, = effective maas-transfer coefficient in linear driving force model based upon gas-phase mole fraction, (g-mol/g)/s L = column length, ft n = solid-phase loading, g-mol/g P = pressure, kPa P, = saturation vapor pressure, kPa q = heat of adsorption, cal/g-mol R = gas constant T = temperature, K t = time tl = breakthrough time t, = midpoint time, Le., time when y / y f d = 0.5 V = volume V,,,= molecular volume (=molecular weight/liquid density), cm3/g-mol W = weight of the adsorbent, g W o= adsorbent micropore volume, cm3/g y = mole fraction of the impurity in the gas phase yo = mole fraction of the impurity in the gas phase prior to ita exposure to the adsorbent
f = final equilibrium value i = initial value
Greek Letters a = surface tension
Superscript - = corresponding equilibrium value
Literature Cited Dubinin, M. M.; Zaverina, E. P.; Radushkevich, L. V. Zh.Fir. Khim. 1947,21,1351. Golden, T. C.; Hsiung, T. H.; Snyder, K. E. Removal of Trace Iron and Nickel Carbonyls by Adsorption. Ind. Eng. Chem. Res. 1991a,30, 502-507. Golden, T. C.; Sircar, S.; Rao, M. B. Adsorption Equilibrium and Kinetics of Trace Hydrocarbons from C02 on Activated Carbon. 20th Biennial Conference on Carbon, Santa Barbara, CA, 24 June 1991b. Kumar, R.; Golden, T. C. Thermal Swing Adsorption Process for Removing Trace Impurities from a Multicomponent Gas Mixture: (Landfill Gas). Gas Sep. Purif. 1991,5,21-24. Kumar, R.; VanSloun, J. K. Purification by Adsorptive Separation. Chem. Eng. h o g . 1989 (Jan), 34. Quayle, 0. R. The Parachors of Organic Compounds. Chem. Rev. 1953,53,439-589. Rosen, J. B. General Numerical Solution for Solid Diffusion in Fixed Bed. Znd. Eng. Chem. 1954,46,1590-1594. Sladek, K. J.; Gilliland, E. R.; Baddour, R. F. Diffusion on Surfaces. 11. Correlation of Diffusivities of Physically and Chemically Adsorbed Species. Znd. Eng. Chem. Fundam. 1974,13,100. Wakao, N.; Smith, J. M. Diffusion in Catalyst Pellets. Chem. Eng. Sci. 1962,17,825-834.
j3 = ratio of parachors given by eq 3 = molecular parachor, molecular volume of a liquid with
Received for review March 3, 1992 Revised manuscript received July 6,1992 Accepted October 6,1992
p
surface tension equals unity, cm3/g-mol
Simultaneous Absorption of H2S and C 0 2 in NaOH Solutions: Experimental and Numerical Study of the Performance of a Short-Time Contactor Vasilis Bontozoglout a n d Anastasios J. Karabelas* Chemical Process Engineering Research Institute a n d Department of Chemical Engineering, Aristotle University of Thessaloniki, P.O.Box 1517, GR 54006 Thessaloniki, Greece
Experimental data are reported on the simultaneous absorption of H2S and C02from steam, using NaOH solutions of varying strength. The application motivating this study is the selective removal of H2Sfrom geothermal steam. A short small-diameter tube is used as a cocurrent contact device favoring selective H2S removal. The absorption rates achieved are compared with the results of similar experimental studies in different types of devices, and the hydrodynamic characteristics of the tubular contactor are discussed. The process is computer simulated, using a numerical scheme to calculate absorption rates at the tube cross section and integrating along the contactor. The average mass-transfer coefficients, resulting from a fit of the numerical results to the data, are among the highest appearing in the literature. Introduction Hydrogen sulfide and carbon dioxide coexist in a variety of gas streams in the process and power industries. Because of environmental, safety, and corrosion considerations, it normally is desirable to remove only HzS. However, the similarity of the two gases with respect to their Physical and chemical properties results in the unneceaeaTy removal Of large quantities Of ' O Z with H2S, with obvious consequences on equipment size and + Present address: Department of Mechanical Engineering, University of Thessaly, Pedion Areos, 38334 Volos, Greece.
operating expenses of the purification plant. Development of appro&& processes fir selective Ha removal is dearly in demand. High-enthalpy geothermal steam, used for power generation, is of main interest in the present work. Typically, the stSam contains H a at concentrations of a few hundred parts per million accompanied by much larger quantities of C02 (14%). The objective of inexpensively H2S without significant enthalpy losses is confounded in this application by the need for a relatively simple R~OCM in terms of operation and control. This need st& from the fact that the geothermal installations are frequently located in remote areas and are operated by personnel with
0888-5885/93/2632-0165$04.00/00 1993 American Chemical Society