Search

Abstract: A technique to measure spatially resolved permeances through tubular membranes with a moveable mass spectrometer probe at the membrane surface is described. This technique was used to determine spatial permeance distributions for SAPO-34 zeolite membranes, alumina membrane supports, and a Vycor glass membrane. The distribution of local permeances provides more detailed information about membrane properties than global permeances. The CO2 and CH4 permeances of SAPO-34 membranes that had high permeances and selectivities at 4.6MPa were not spatially uniform across the membrane area. Membranes prepared using Al(OH)3 as the aluminum precursor had large regions that were nearly impermeable, whereas membranes prepared with Al(i-C3H7O)3 were more uniform, and thus had higher overall permeances. Scanning electron micrographs indicated that these differences in permeance were not due to differences in membrane thickness, and spatially uniform permeance distributions for alumina supports suggest that they were not due to non-uniformities in the supports. Defect locations in the SAPO-34 layers were detected using i-butane permeation, because i-butane is too large to enter the SAPO-34 pores, and the defect concentrations were low. These spatially resolved measurements indicate the potential to increase both CO2/CH4 selectivity and CO2 permeance, and help understand the factors that affect membranes properties. [Copyright &y& Elsevier]

Abstract: Separation of CO2/CH4 mixtures through tubular SAPO-34 membranes with high permeances was strongly affected by concentration polarization at feed pressures above about 1MPa. The faster-permeating CO2 was depleted near the membrane surface because it could not diffuse fast enough from the bulk phase. Concentration polarization increased as pressure increased at constant feed rates because fluxes through the membranes increased, gas diffusivities decreased, and the velocity across the membrane decreased because gas densities increased. The CO2/CH4 selectivity, at 298K and a feed pressure of 5.8MPa, increased by about 160% and the CO2 flux increased by about 80% when Teflon spacers were inserted inside the membrane tube and the feed flow rate was increased by a factor of three. The spacers and the increased feed rate increased the gas velocity and decreased the diffusion distance from the bulk phase to the membrane surface, and thus they decreased the polarization boundary layer. The spacers did not affect separations at low pressures, where gas velocities and diffusivities were higher and fluxes were lower. [Copyright &y& Elsevier]

Abstract: Two polycrystalline MFI membranes with significant flow through defects (non-zeolitic pores that are gaps between the crystals) are shown to have high ideal and mixture selectivities. The membranes were characterized at room temperature by permporosimetry, pervaporation, separations, and single-gas permeation. These measurements indicate that one membrane (B-ZSM-5) had a relatively large number of smaller defects, whereas the second membrane (silicalite-1) had a smaller number of somewhat larger defects. The relative contributions of these defects to the overall flux changed dramatically in the presence of n-alkanes and SF6. These molecules caused adsorption-induced expansion of the crystals, and this expansion shrank the defect sizes and thus changed the membrane permeation characteristics. The B-ZSM-5 membrane had 90% of its helium flux through defects at room temperature, but it had a H 2 /SF 6 ideal selectivity as high as 260 because of SF 6 -induced swelling that stopped 99% of the flux through the defects. In contrast, the silicalite-1 membrane had only 9% of its helium flux through defects, but the defects were large enough that crystal swelling only decreased the flux through them by 30%. Thus its selectivities were lower. These studies show that n-hexane, n-pentane, n-butane, n-propane, and SF6 swell MFI crystals when they adsorb, but benzene and CO2 do not. The changes in membrane microstructure due to crystal expansion not only significantly affect membrane separation ability, but also have implications on how to select appropriate characterization techniques for evaluating MFI membrane quality. [Copyright &y& Elsevier]

Abstract: SAPO-34 zeolite membrane synthesis was scaled up by preparing membranes on seven-channel monolith alumina supports. These monoliths have CO2 permeances and CO2/CH4 separations selectivities at 4.6MPa feed pressure that were similar to SAPO-34 membranes on single channel supports. They also exhibited the same pressure dependence. SAPO-34 membrane preparation was improved by adding SAPO-34 seed crystals to the synthesis gel instead of placing seeds on the support surface. Membranes prepared by seeded-gel synthesis had higher CO2 permeances, higher CO2/CH4 separation selectivities, and smaller standard deviations for these values. Using seeded-gel synthesis also decreased the number of synthesis steps, but increased the amount of seeds needed by two orders of magnitude. [Copyright &y& Elsevier]

Two types of CHA zeolite membranes (SAPO-34, SSZ-13) were used for CO 2 /CH 4 , N 2 /CH 4 , and CO 2 /i-butane separations at both low (270 and 350 kPa) and high (1.73 MPa) pressures. The SSZ-13 membranes were more selective, with CO 2 /CH 4 separation selectivities as high as 280 and N 2 /CH 4 separation selectivities of 12 at 270 kPa feed pressure. For both types of membranes, selectivities and permeances decreased as the feed pressure increased. The CO 2 /i-butane separation selectivities were greater than 500,000 for SAPO-34 membranes, indicating extremely low densities of defects because i-butane is too large to enter the CHA pores. The CO 2 /i-butane selectivities were smaller for SSZ-13 membranes (2,800–20,000), in part because the SSZ-13 layer was on the outside of the porous mullite tubes and sealing the membrane on the zeolite surface was more difficult than for the SAPO-34 membranes that were grown on the inside of glazed alumina tubes. Propane, in feed concentrations from 1 to 9%, significantly influenced separations by decreasing permeances in most cases. The effect was larger for N 2 /CH 4 than for CO 2 /CH 4 mixtures, apparently because the more strongly-adsorbing CO 2 competes better than N 2 with propane for adsorption sites. Although propane caused permeances to decrease significantly over time, selectivities decreased much less. Propane decreased permeances more for SAPO-34 membranes than for SSZ-13 membranes at 350 kPa, and at high pressure propane even increased CO 2 permeances and decreased CH 4 permeances in SSZ-13 membranes, thus significantly increasing CO 2 /CH 4 selectivities. Propane permeances reached steady state relatively quickly because its permeation was mostly through defects, but CO 2 , N 2 , and CH 4 permeances did not stabilize in the presence of propane, even after seven days. The effects of propane were reversible when it was removed from the feed and the membranes were heated. [ABSTRACT FROM AUTHOR]

Abstract: The CO2/CH4 separation selectivity of SAPO-34 membranes increased by 20–150% after soaking them in 1–2% aqueous β-cyclodextrin (CD) solution and drying them at 473K. The percentage increase was higher at higher pressures, and the selectivities increased more for membranes with more defects. The CD decreased the CO2 fluxes at high pressures by about 10–20%, whereas CH4 fluxes decreased by 30–60%. The flux of CF4, which is too large to enter the SAPO-34 pores and thus only permeates through defects, decreased 80% for one membrane (5.7MPa feed pressure) after CD deposition; this confirmed that defects were effectively blocked. Even before CD exposure, the CF4 permeance at 2.8MPa was only 0.25% of the CO2 mixture permeance for that membrane. Permporometry also showed that CD blocked defects; the maximum defect size decreased from greater than 30nm to less than 20nm. About 91% of the overall defect area in the original membrane was from defects smaller than 8nm, whereas 99% was for defects smaller than 2nm after CD deposition. The membrane separation performance did not change during 3 days of continuous measurements, nor did it change after long term storage at 473K in air. [Copyright &y& Elsevier]