Ind. Eng. Chem. Res. 1994,33,2879-2880
2879
CORRESPONDENCE Comments on “Concentrationand Recovery of COz from Flue Gas by Pressure Swing Adsorption” Arthur I. Shirley Technical Center, The BOC Group, Inc., Murray Hill,New Jersey 07974
Sir: The recent work of Kikkinides et al. (1993) presented a simulation comparison of the recovery of C02 from flue gas via pressure swing adsorption (PSA) using carbon molecular sieve (CMS) and activated carbon (AC). Their calculations showed that AC is superior t o CMS for CO2 recovery by producing product a t greater purity for the same level of C 0 2 recovery, leading those authors to conclude that “the kinetic selectivity (for COS over N2 in CMS) is of secondary importance as compared to the equilibrium selectivity (of C02 over N2 in activated carbon)”. They go on the state that better performance of the process using AC is “due to the lack of any diffusional resistances in the case of activated carbon”. While the results of their simulation are undoubtedly correct, it is debatable that the reasons for AC superiority have anything t o do with the kinetic selectivity for CMS for CO2 over N2, or the diffusional resistances of CMS. It is argued herein that the poorer performance of CMS relative t o AC is actually due to the lack of significant kinetic selectivity for C02 over 0 2 on CMS. In order to make this argument, both theoretical and experimental evidence can be cited. On the theoretical side, the time dependence of C02,02, and NZadsorption on CMS can be examined. The non-Fickian behavior of oxygednitrogen uptake on CMS (Koss et al., 1986) has previously been cited as evidence of the time dependence of the sieving action of CMS. One consequence of this is that, in the limit of long adsorption times (relative to the adsorption kinetics), the kinetic separation of gases on CMS will be more like the equilibrium separation of the same gases on AC (Shirley and LaCava, 1993). In the case of two gases with similar, rapid uptake kinetics, such as CO and 0 2 , their separation on CMS will also have a more equilibrium character, while the separation between the fast components and any slow components in the feed mixture will have the customary appearance of kinetic control. Since in the simulations of Kikkinides et al. the C02 and 0 2 are diluted in N2, it would be likely that the purity of the desorbed gas would be determined by the competition between these fast components and not the competition between fast C02 and slow N2. Evidence that this CO2-02 competition is the controlling factor can be seen from the following PSA experiment. A cocurrent purge process similar to that used by Kikkinides et al. was run using a mixture of 30% N20, 7% C 0 2 , and 63% air (approximately 13%0 2 and 50% N2 and Ar). The adsorbent beds contained commercial carbon molecular sieve (CMS) made by The BOC Group, Inc. The production pressure was about 3 psig, while the regeneration “pressure”was done under vacuum at about 200 mbar. 0888-5885l94l2633-2879$04.50/0
A full cycle of the PSA process was as shown in Table 1. The cycle time was 80 s (7 s equalization, 23 s feed and produce, and 10 s cocurrent purge). Except for the presence of N20, the components in this PSA experiment are the same as those studied by Kikkinides et al. The experiment can still be used for comparison since C02 and N2O have similar adsorption properties on AC (Robinson, 1942) and CMS (Koresh and Sofer, 1980a,b). Koresh and Sofer (1980a,b) have previously shown that C02 and N2O have rapid (and very nearly equal) adsorption kinetics on CMS. They also found 0 2 to be slightly slower than C02, in agreement with Kikkinides et al. Consequently, we will consider C02 and N2O to be a single component. The results of the PSA experiment above are given in Table 2. These results are comparable to one of the simulation data for CMS in Figure 4 of Kikkinides et al.: specifically, the simulation where the product C02 purity is approximately 86% (vs 86.2% here) and the C02 recovery is approxiately 82% (vs 88.5% in the experiment). Kikkinides et al. do not specify the composition of the non-CO2 portion of their product gas, but it can be seen from the experimental results above that the non-CO2 portion is about 80% 02, even though the non-CO2 portion of the feed was only 21% 0 2 . This behavior would explain why the simulation results for product C02 purity on CMS were so much poorer than those for AC at the same C 0 2 recovery: Poor selectivity between C02 and 0 2 causes a dilution of the C 0 2 product. Unfortunately, this comparison does not answer the question of the effect of difusional resistances on the separation of CO2 and N2 on CMS and AC. Although Kikkinides et al. did not compare the two adsorbents using a feed gas with only C02 and N2, an enlightening result from the literature can be cited of just such a comparison. Richter et al. (1984) compared the separation of a 24% C02 and 76% NZ mixture by PSA on narrow- and wide-pore CMS samples. The wide-pore CMS was more akin t o an uncoked AC, having low diffusional resistances, while the narrow-pore CMS exhibited the typical kinetic selectivity. The PSA cycle employed in these experiments was a simple adsorption a t 5.0-5.5 bar with subsequent desorption at vacuum. The narrow-pore CMS produced essentially pure NZ as an adsorption product and recovered nearly 100% of the C02. However, the desorption product from the narrow-pore CMS was of low (7040%) purity. The wide-pore CMS, on the other hand, produced an adsorption product of only about 90-95% N2, but its vacuum desorption product reached purities of essentially 100% C02. Richter et al. accounted for this by proposing that the slow desorption of N2 on the narrow-pore CMS limited the purity of the desorption product, whereas 0 1994 American Chemical Society
2880 Ind. Eng. Chem. Res., Vol. 33, No. 11, 1994 Table 1
step 1
2 3 4 5 6
adsorbent bed A adsorbent bed B feed pressurization with regeneration under vacuum waste release cocurrent purge with regeneration under vacuum waste release equalization regeneration under vacuum feed pressurization with waste release regeneration under vacuum cocurrent purge and waste release equalization
Table 2
fractional flow composition, % Nz 0 2
COz (+NzO) COZ(+NzO) recovery, % Nz + 0 2 recovery, % Nz recovery, % 0 2 recovery, %
waste
product
0.382
0.618
79.3 14.0 6.7 11.5 91.7 97.9 67.3
2.8 11.0 86.2 88.5 8.3 2.1 32.7
pure desorption product for the narrow-pore CMS by displacing N2 from the pores prior to the adsorption step. This cocurrent purge would need to be done at a slow enough rate to allow time for the Nz to diffuse out of the pores. Of course, for the wide-pore CMS the cocurrent purge can be done a t the same rate as the adsorption step since no diffusional limitations exist.
Literature Cited Kikkinides, E. S.; Yang, R. T.; Cho, S. H. Concentration and Recovery of COz from Flue Gas by Pressure Swing Adsorption. Ind. Eng. Chem. Res. 1993,32, 2714. Koresh, J.; Sofer, A. Study of Molecular Sieve Carbons: Part 1.-Pore Structure, Gradual Pore Opening and Mechanism of Molecular Sieving. J . Chem. SOC.,Faraday Trans. 1 1980a, 76, 2457.
in the wide-pore CMS the Na desorbed rapidly and was displaced into the initial desorption product, yielding a later, high-purity COS desorption product. One could envision that the addition of a cocurrent purge step between the adsorption and desorption steps of the Richter et al. experiments would yield a more
Koresh, J.; Sofer, A. Study of Molecular Sieve Carbons: Part 2.-Estimation of Cross-Sectional Diameters of Non-sphereical Moleculaes. J . Chem. Soc., Faraday Trans. 1 1980b, 76,2472. Koss, V. A.; Wickens, D. A.; Cucka, P.; LaCava, A. I. A Model of the Adsorption of Gases on Carbon Molecular Sieves with Langmuir's Kinetics and Simultaneous Diffusion. Presentation at Carbon '86, Baden-Baden, June 30-July 4, 1986. Richter, E.; Harder, K.-B.; Knoblauch, K.; Juntgen, H. New Developments in Pressure Swing Adsorption. Chem. Ing. Tech. 1984, 56 (91, 684. Robinson, C. S. The Recovery of Vapors;Reinhold: New York, 1942. Shirley, A. I.; LaCava, A. I. Novel Pressurization Methods in Pressure Swing Adsorption Systems for the Generation of HighPurity Gas. Ind. Eng. Chem. Res. 1993, 32, 906.