Article pubs.acs.org/IECR
Measuring Mixture Adsorption by Temperature-Programmed Desorption Hans H. Funke,* Yiwei Luo, Michael Z. Chen, Grace C. Anderson, John L. Falconer, and Richard D. Noble Department of Chemical and Biological Engineering, University of Colorado, Boulder, Colorado 80309-0596, United States ABSTRACT: Temperature-programmed desorption (TPD) was used to measure single-component and mixture adsorption of CO2 and C3H8 in zeolite 13X and SAPO-34 crystals. The TPD was carried out with the adsorbing gases in the feed stream so that adsorption could be determined for weakly adsorbing molecules. The single-component loadings measured by TPD and by volumetric adsorption were within 5% of each other. Carbon dioxide preferentially adsorbs on zeolite 13X from CO2/C3H8 mixtures, whereas C3H8 preferentially adsorbs on SAPO-34. Adsorption equilibrium was quickly reached on zeolite 13X. However, more than 12 h were required to obtain equilibrium for CO2/C3H8 mixtures in SAPO-34 because C3H8 diffuses much slower than CO2 in SAPO-34 pores, but adsorbs more strongly and thus displaces half of the CO2 that initially adsorbs. This behavior helps explain the slow approach to equilibrium observed for CO2/CH4 separations with SAPO-34 membranes when C3H8 was added to the feed.
■
INTRODUCTION Adsorption isotherms of single components are routinely measured in static adsorption systems to characterize porous materials. However, mixture isotherms are important for adsorption-based processes, catalytic reactions, and membrane applications since several components are competing for adsorption sites. The amounts adsorbed in mixtures can be sensitive to the relative heats of adsorption and thus may be difficult to estimate from single-component measurements. Heats of adsorption calculated from the temperature dependence of single-component isotherms can be used to predict preferential absorption. However, reliable predictions and design of separation processes require mixture isotherms to verify the accuracy of adsorption models based on singlecomponent isotherms.1 Mixture adsorption isotherms have been determined by gravimetric,2 chromatographic,3,4 zero-length column,5,6 step change,7 and volumetric techniques.8−10 A comparison by Keller et al.11 and a recent review by Sircar12 discuss the advantages and disadvantages of the different approaches in detail. The final concentrations in the gas phase after equilibration are not controlled in the volumetric and gravimetric methods, and the void volumes, which are typically determined with helium, have to be known precisely. However, these void volumes may not be valid for other weakly adsorbing components because helium also weakly interacts with adsorbents. Measuring breakthrough curves may be less accurate when adsorption equilibrates slowly so that small concentration differences have to be determined over long time periods. Mixture adsorption on zeolites has only been reported in a few studies due to the difficulties in carrying out the measurements. For example, Mulgundmath et al.13 investigated adsorption of CO2/CH4 and CO2/N2 mixtures in zeolite 13X using a volumetric method and pulsed chromatography. Costa et al.10 and Calleja et al.8 measured coadsorption of CO2 with a © XXXX American Chemical Society
range of small hydrocarbons on zeolite 13X using a volumetric technique. Similarly, Brandini et al.5 reported the adsorption of CO2 mixed with C2H6 or C3H8 for several X-type zeolites at pressures below 12 kPa with the ZLC method. Rother and Fieback14 used a gravimetric method for adsorption of CO2 mixed with N2 or H2 on 13X, and Lamia et al.15 studied the adsorption of hydrocarbon mixtures on 13X. Temperature-programmed desorption (TPD) was used in the current study to measure mixture adsorption of weakly adsorbed gases in SAPO-34 and 13X zeolites. Fletcher et al.2 previously combined gravimetry with TPD in a flow system to obtain both adsorption kinetics and the adsorbed composition of a mixture. They used a mass spectrometer to determine the gas-phase concentration in close proximity to the sample in a thermogravimetric analyzer during TPD. The mass spectrometer was calibrated using single-component desorption data, and mixture isotherms for water and n-butane on activated carbon were reported. Foeth et al.16 used a precisely controlled temperature bath with low heating rates (∼1 K/min) to determine isotherms for activated carbon from a single TPD experiment. However, their method assumes that the adsorbent is quasi-equilibrated during TPD and is probably not feasible for molecules that slowly diffuse in zeolites. The current study was carried out with higher heating rates, and the same gas-phase composition that was used during adsorption was maintained during TPD to prevent weakly adsorbing molecules from desorbing before heating started. This is in contrast to most TPD measurements where the molecules are relatively strongly adsorbed so they do not desorb when an inert gas replaces the adsorbing gas.17 Unlike volumetric adsorption techniques, where the adsorbate Received: February 15, 2015 Revised: April 7, 2015 Accepted: April 21, 2015
A
DOI: 10.1021/acs.iecr.5b00667 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX
Article
Industrial & Engineering Chemistry Research
Autosorb AS-1C). The samples were outgassed at 473 K for 12 h between measurements. Temperature-Programmed Desorption. A schematic of the TPD system is shown in Figure 1. Three electronic mass
concentrations in the gas phase changes if one component preferentially adsorbs, the gas-phase concentrations can be controlled at the start of the TPD. The adsorbed loadings are then calculated from the amounts that desorb during TPD. This approach, however, requires that small concentration changes be measured for gases at high concentrations. This decreases the sensitivity compared to TPD in an inert gas. In addition, the volumetric flow rate of the effluent changes as molecules desorb and as gas in the sample cell expands due to the temperature increase, and these changes must be accounted for because they affect concentrations.18 The dilution due to desorbed molecules can be estimated from the decrease in concentration of the inert gas. However, at high inert concentrations, the signal noise is high. Therefore, the total flow rate was determined during TPD by directing the effluent into an evacuated buffer volume and measuring the pressure increase. Single-component CO2 and C3H8 loadings measured by TPD in SAPO-34 and 13X zeolite crystals were similar to those obtained by volumetric adsorption. The kinetic diameter of C3H8 (0.43 nm) is slightly larger than the SAPO-34 pores (0.38 nm), and thus C3H8 is expected to diffuse slowly.19 Indeed, TPD after different mixture adsorption times showed that C3H8 adsorbed much slower than CO2 in SAPO-34, and long times were needed for equilibrium. Carbon dioxide and C3H8 adsorption in SAPO-34 is of interest for natural gas processing because SAPO-34 membranes have been shown to have high fluxes and selectivities for removing CO2 from CH4 mixtures, even at high feed pressures.20 However, higher hydrocarbons that are often present at low concentrations in natural gas can affect the membrane performance. In a recent study, CO2 permeances decreased slowly, and several days were required to reach steady state after 5% C3H8 was added to a CO2/CH4 feed.21 The CO2 permeances decreased as C3H8 slowly displaced adsorbed CO2 and also inhibited CO2 permeation. The 13X zeolite was chosen because C3H8 and CO2 are much smaller than its pores (0.8 nm), and thus adsorption is not diffusion-limited. In addition, both single-component and mixture adsorption of different gases have been reported for this zeolite, but pelletized 13X samples were used in those studies, and thus loadings were lower than those found in this study.
Figure 1. Schematic of TPD system for measuring amounts adsorbed from binary mixtures.
flow controllers maintained the flow rates of the adsorbates and carrier gas through the sample, which was placed on a quartz wool plug inside a quartz tube (20 cm long, 12 mm O.D. at the inlet, 6 mm at the outlet). A thermocouple (0.79 mm O.D.) was placed in the center of the sample, and the sample was heated with a custom tube furnace that was controlled with a PID controller. The furnace consisted of a resistance wire wrapped around a quartz tube and was held in place with ceramic cement. This heating element was placed inside a second quartz tube, and the annulus between the tubes was filled with ceramic fibers for insulation. Fans near the top and bottom fittings of the sample holder prevented the elastomer O-rings in the fittings from overheating. Syringe injection ports near the inlet and outlet of the sample holder were used to expose the samples to known amounts of adsorbate and for manual calibration. The mass spectrometer (Pfeiffer Prisma QMA200/QME200) sampled a small fraction of the effluent stream (
