Enhanced Pressure Swing Membrane Absorption Process for CO2

Article pubs.acs.org/IECR

Enhanced Pressure Swing Membrane Absorption Process for CO2 Removal from Shifted Syngas with Dendrimer−Ionic Liquid Mixtures as Absorbent Xingming Jie, John Chau, Gordana Obuskovic, and Kamalesh K. Sirkar* Otto H. York Department of Chemical, Biological, Pharmaceutical Engineering, Center for Membrane Technologies, New Jersey Institute of Technology, Newark, New Jersey 07102, United States ABSTRACT: Using pure ionic liquid [bmim][DCA] or its dendrimer-containing solutions as absorbents, studies of a novel fivevalve pressure swing membrane absorption (PSMAB) process for CO2 removal from lower temperature shifted syngas were carried out with porous hydrophobized polyether ether ketone (PEEK) hollow fiber based membrane modules. At first how the preferred absorption step duration in each cycle was determined was presented. Then we investigated the effects of the following aspects of the membrane module: reducing the tube-side dead volume in the hollow fiber module; decreasing the fiber packing density leading to increased absorbent liquid layer thickness around any fiber; increasing the tube-side gas volume. All three design features in the membrane module greatly enhanced the gas absorption behavior and improved the product qualities. Increasing temperature reduced the product qualities when pure [bmim][DCA] was used. Adding polyamidoamine (PAMAM) dendrimer Gen 0 to the ionic liquid greatly improved its CO2 absorption capability especially at higher temperatures. Among all absorbents tested, 20.0 wt % dendrimer in [bmim][DCA] showed the best absorption enhancing effect at 100 °C. For a feed gas pressure of 250 psig, He-rich product having a CO2 concentration around 24.7% (actual value should be around 16.7% based on dead volume influence elimination) and CO2-rich product with CO2 concentration around 85.7% (actual value should be higher) have been achieved. We have also carried out a second stage PSMAB test with a 14.0% CO2 in helium as feed gas, and a He-rich product with CO2 concentration around 8.55% (actual value should be around 5.50%) could be achieved at 100 °C. A new module having a larger tube-side gas volume yielded a CO2-rich stream having 90.7% CO2. The newly developed five-valve PSMAB process shows great application prospects for CO2 removal from precombustion shifted syngas; significant membrane module redesign is, however, needed.

1. INTRODUCTION Demands for removal of greenhouse gases as much as possible from all headstreams are acquiring great urgency because of severe related environmental problems. There have been quite a few scientific studies and theoretical analyses showing that CO2 is playing a dominant role within all greenhouse gases.1 Carbon dioxide capture and storage (CCS) is therefore an area of considerable deliberation in terms of research and development around the world with special concern about the longterm effects of geological storage.2,3 The largest source of CO2 emission is coal-based electrical power stations; these accounted for ∼67% of 2011 reported emissions.4 Three main methods have been suggested and applied for the capture of CO2 from power systems: 1. precombustion capture: capturing CO2 from shifted synthesis gas after CO is converted to CO2 2. postcombustion capture: capturing CO2 from the flue gas produced after the fuel has been fully burnt with air 3. O2-enrichment combustion capture: capturing CO2 after the fuel has been burnt with a high O2 concentration air5−7 Here we focus on CO2 removal from precombustion shifted syngas since CO2 is present at a much higher partial pressure than that in postcombustion capture processes. Further, this gas is presumed to be provided at a lower temperature. Processes applied for separation of the CO2 in a gas mixture are mainly based on chemical, physical, or hybrid absorption; adsorption; membrane separation; or cryogenic separation. Often some of © 2014 American Chemical Society

these processes are used together so that the advantages of different methods could be combined to achieve the best separation effect.8−12 We describe here further enhancements of a cyclic separation process of CO2 absorption and recovery, the pressure swing membrane absorption (PSMAB) process. This process combines the specific advantages of several basic separation techniques: highly selective CO2 absorption in a nonvolatile absorbent at higher temperatures; pressure swing absorption (PSAB) process simulating a pressure swing adsorption (PSA) process (which however uses adsorbent particles); hollow microporous fibers providing per unit device volume a large surface area of nondispersive contact between the gas stream in the hollow fiber bores and the liquid absorbent present as a thin stagnant absorbent layer in between microporous hollow fibers on the shell side. For the proposed process, the liquid absorbent and the membrane module will be two factors of major importance dictating the performance of the PSMAB process. Ionic liquids (ILs) have high thermal stability and essentially no vapor pressure; most important, compared with hydrogen, it has been widely reported that CO2 has a much higher solubility in ILs Received: Revised: Accepted: Published: 3305

October 24, 2013 January 7, 2014 January 22, 2014 January 22, 2014 dx.doi.org/10.1021/ie403596b | Ind. Eng. Chem. Res. 2014, 53, 3305−3320

Industrial & Engineering Chemistry Research

Article

providing a basis for its applicability to PSMAB.13−19 In our first paper,20 we provided a preliminary report on the PSMAB process at higher temperatures and pressures with an ionic liquid based on earlier studies using room temperature and atmospheric pressure operation with aqueous absorbent solutions.21,22 We have already defined20 what an ideal membrane contactor should have for such a process: a high breakthrough pressure; appropriate surface hydrophobicity; low pore transport resistance; significant gap between neighboring fibers so that there is sufficient liquid for absorption without sacrificing a high surface area per unit volume. In addition, we had identified the need for a reduction in the tube-side dead volumes in the membrane module. Earlier studies19,23−28 indicated also that a polyamidoamine (PAMAM) dendrimer of generation zero (Gen 0) can reversibly react with CO2 and improve membrane performance and can improve PSMAB process performance20 as well. Modeling of the PSMAB process performance for a pure IL has also been implemented.29 Ionic liquids have emerged as important absorbents for CO2 capture and natural gas sweetening. Karadas et al.30 published a review containing a large collection of experimental and theoretical studies to infer the viability of ILs as an alternative to current amine-based absorption. Albo et al.31 reported their results with the IL 1-ethyl-3-methylimidazolium ethylsulfate as solvent in a cross-flow membrane contactor for CO2 capture from flue gases. A focused review of recent advances and future challenges for gas−liquid membrane contactors in acid gas removal is available from Zhang and Wang.32 The conditions of our high pressure PSMAB studies especially with regard to temperature are radically different from what is being currently studied in the area of gas−liquid contactors for CO2 removal. In this paper we focus first on various aspects of the membrane module design as they influence the separation: the effect of the tube-side dead volume into which the feed gas is introduced and from which the two product streamsa Herich stream and a CO2-rich streamare withdrawn from the two ends of the tube side; hollow fiber packing density controlling the gap between contiguous fibers wherein lies the absorbent; the tube-side gas volume vis-à-vis the dead gas volume. The effects of the running temperature on the product qualities of the PSMAB process when [bmim][DCA] was used as absorbent was also systematically investigated. Continued attempts to add dendrimer to IL and form a mixed absorbent to enhance PSMAB performance especially at higher temperatures have also been made. The performances of different mixed absorbents with a variable dendrimer concentration were studied to determine the preferred mixed absorbent. The effect of feed gas pressure up to 250 psig (1724 kPag) on product quality at a high temperature was studied next to show the stability of this system and the membrane module itself. Then an analysis of the influence of the dead volume was proposed and a calculation was made to show what the actual product quality should be in the absence of the dead volumes. Near the end, a two-stage PSMAB process was proposed to achieve satisfactory product qualities at both ends since the single modules used did not have sufficient length and surface area for the gas processing rates used. At the end the achievement of a much better CO2-rich product quality at higher temperatures and pressures was experimentally explored using a new module having a larger tube-side gas volume which countered the reduction in the quality of the two product streams by the gas in the tube-side dead volumes.

2. EXPERIMENTAL PROCEDURES 2.1. Materials. Ionic liquid (IL) [bmim][DCA] was purchased from EMD Chemicals (Philadelphia, PA) and used as received for membrane module breakthrough pressure tests. It was selected as the absorbent because of its excellent reported CO2 absorption behavior.19 For example, it has a CO2/He solubility selectivity of 10 and a CO2 solubility of 0.21 mole fraction at 4 atm CO2 pressure at 50 °C. Further, it is stable in the presence of moisture at a high temperature. It is also highly miscible with PAMAM dendrimer to enhance CO2 absorption and improve the final separation results. A 20 wt % solution of PAMAM in [bmim][DCA] has a high CO2/He solubility selectivity of 40 at 50 °C in the presence of moisture.19 However, it has a very high viscosity of 41 cP at 50 °C. Polyethylene glycol 400 (PEG-400) was purchased from Chemicals Direct (Roswell, GA) and used as received for membrane module breakthrough pressure tests. PAMAM dendrimer (Gen 0) (structure shown in Figure 1) purchased from Dendritech (Midland, MI) was received as a

Figure 1. Structure of PAMAM dendrimer of generation 0.

dendrimer−methanol solution in which the dendrimer concentration was 64.05 wt %. To get pure dendrimer, the solution was vacuumed for several days under a relatively high temperature around 60 °C to remove methanol. Simulated precombustion syngas gas containing 40.67% CO2, He balance as surrogate for H2, was purchased from Air Gas (Salem, NH). This dry gas was used as feed gas quite often. This gas was humidified sometimes (when mentioned) and used. At the end a gas mixture of 14.28% CO2, He balance as a surrogate for H2 (Air Gas, Salem, NH), was used as feed gas for a second-stage absorption process. Three types of polyether ether ketone (PEEK) membrane modules were purchased from Porogen (Woburn, MA). The membrane module details are listed in Table 1. The modules were identified as PEEK-L-I, PEEK-L-II, and PEEK-L-III. Each module used exactly the same hollow fiber. All the fibers are helically wound so that the length of the whole module is shorter than the actual fiber length. Further, the fibers are tightly bunched together as in a strand (Figure 2a). The absorbent liquid is on the shell side of these hollow fibers, while the gas is on the tube side (Figure 2b). Two kinds of modules, PEEK-L-I and PEEK-L-II, have almost the same effective area. (The addition of “L” distinguishes these modules from the much smaller PEEK-S modules used often in the earlier study.20) The main difference between them is the fiber packing density with the PEEK-L-II module having a higher spacing between the contiguous fibers. On the other hand, the 3306

dx.doi.org/10.1021/ie403596b | Ind. Eng. Chem. Res. 2014, 53, 3305−3320


Article

Table 1. Dimensional Characteristics of the Membrane Modules for the Pressure Swing Membrane Absorption Processa module

o.d./i.d. (cm)

L (cm)

pore sizeb (Å)

VVF

fiber no.

surf. areac (cm2)

A/V (cm−1)

packing density (%)

PEEK-L-I PEEK-L-IId,e PEEK-L-III

0.0452/0.029 0.0452/0.029 0.047/0.0272

117 41.0 41.0

∼20 ∼20 ∼20

∼0.4 ∼0.4 ∼0.4

208 568 908

3452 3420 5500

27.2 54.7 57.4

67.0 21.8 27.2

o.d., outer diameter of fiber; i.d., inner diameter of fiber; L, effective fiber length; VVF, void volume fraction. bPore size of the outside surface. Based on outer diameter of fibers. dPEEK-L II module has a packing density around 21.8% that was defined as the ratio between total fiber volume and the real volume they occupied (total fiber volume plus space between the fibers in the fiber strands wound helically in the module). ePEEK-L module with PTFE bead-filled tube-side headers in the module. a c

Figure 2. (a) Hollow fiber winding configuration in a PEEK-L module. (b) Gas liquid membrane contacting in a membrane contactor. (c) Membrane module of hollow fibers showing locations of dead volume in the tube side. (d) Schematic diagram of pressure swing membrane absorption apparatus having PTFE balls in the tube-side headers of the membrane module.

PEEK-L-III module has a considerably larger fiber surface area and therefore a considerably larger feed gas volume vis-à-vis the dead volumes in the tube side of the module and the connections on the tube side (Figure 2c). Teflon balls of 1/8 in. diameter were bought from Engineering Laboratories (Oakland, NJ). During PSMAB tests, these balls were sometimes added to both sides of a PEEK-L module to reduce any existing dead volume in the tube-side headers of the module as much as possible; this is illustrated in the experimental setup shown in Figure 2d (described in more detail later). 2.2. Breakthrough Pressure Test for Membrane Modules. Before a new membrane module is used, it is important to determine its breakthrough pressure for different liquids to limit how high the feed gas pressure for PSMAB studies could go. Breakthrough pressure is determined mainly by two factors: the pore size of hollow fiber membranes at the liquid−membrane interface on the shell side; the surface tension of the IL-based absorbent. During the test, one port of the membrane module shell side was connected to a small cylinder containing [bmim][DCA] (the other port was kept closed) and the module shell side was filled with IL. The IL cylinder was also connected to a N2 cylinder to develop the

desired pressure. The membrane module tube side had a low N2 flow to bring out any possible breakthrough of IL when the pressure was gradually increased. The leaked ionic liquid will fall on a piece of paper next to the outlet. When leaked IL could be detected from the tube side, the test pressure was defined as the breakthrough pressure. Liquids other than ILs were also tested to explore the capabilities of the membranes as well as compare ILs against other potential liquids. 2.3. Pressure Swing Membrane Absorption (PSMAB) Process. Figure 2b shows a microporous hydrophobized PEEK hollow fiber; in the separator device, there will be many such fibers as in Figure 2c. Figure 2c illustrates also the various sources of the dead volume in the tube side of the hollow fiber module: the volumes of the two tube-side headers and the two potted sections. In addition, the connections on the tube-side outside the module contribute also to the dead volume. Surrounding the hollow fibers is the nonvolatile absorbent filling the shell side of the separator. The pores in the wall of the hollow fiber are gas-filled. In the test apparatus (Figure 2d), the membrane contactor module was put inside the PV-222 oven from Espec (Hudsonville, MI) so that the exact temperature could be set and controlled. 3307



Article

Figure 3. Schematic diagram of the five-valve pressure swing membrane absorption process.

There are six steps in each cycle in the case of the five-valve system as listed below: Step 1. Valve 1 is opened and all others are closed; fresh feed gas is introduced into the tube side until the desired pressure is established. Step 2. All valves are closed; absorption happens between the gas and the absorbent. Step 3. Valve 3 is opened and all others are closed; He-rich product is withdrawn. Step 4. Valve 4 is opened and all others are closed; middle part gas is withdrawn. Step 5. Valve 2 is opened and all others are closed; CO2-rich product is withdrawn by applying a vacuum. Step 6. Valve 5 is opened and all others are closed; middle part gas is recycled into the membrane tube side as initial feed gas. This step completes one cycle. The main advantage of the five-valve system is that after the gas absorption step we can divide the gas mixture in the tube side of the module into three parts: He-rich product, middle part gas, and CO2-rich product. The middle part gas will be collected first and then recycled into the tube side at the beginning of next cycle. In this case a better product quality in the CO2-rich stream should be expected. The data presented in this study were obtained after any run was conducted for at least 30 cycles; usually after 5−10 cycles the separation performance was found to be stable. Further, the tests were repeated three times to confirm reproducibility.20 One may wonder why this cyclic process is being implemented instead of having one conventional absorption unit and another conventional desorption (stripping) unit with the absorbent liquid circulating in between and achieving a continuous process. First, it is well-known that pressure swing adsorption (PSA) processes achieve continuous operation by having two to two plus units operating in tandem; the same would be valid for this PSMAB process. Therefore achievement of a continuous process by having several units operating with appropriate time lags in a given cycle is doable and realistic. Second, ionic liquid viscosity is high. We have already shown20 that as the temperature increases to 100 °C the performance of ionic liquid deteriorates significantly; the addition of PAMAM dendrimer substantially improves the separation performance. The dendrimer addition increases the viscosity considerably. A

The shell side of the module was filled with the absorbent supplied from the absorbent container connected to a N2 cylinder to maintain the desired pressure. Feed gas was introduced into the membrane tube side where it contacted the absorbent through the micropores to be absorbed. Usually the absorbent pressure in the shell side was kept about 20 psig higher than the feed gas pressure in the tube side to avoid any possible gas breakthrough. Before entering the tube side, feed gas may enter the water bath and be humidified to investigate the water vapor influence on the absorption behavior; an HP 1100 model water pump (Waldbronn, Germany) was used to inject water into the system. The CO2 product side was connected to a vacuum pump to supply the driving force for product withdrawal; a certain number of pneumatic valves were used to control exactly the time period for different steps in one absorption cycle. This valve control system was realized via a programmable logic controller (PLC) scheme installed by PneuMagnetic (Quakertown, PA). Both He-rich and CO2-rich product streams were sometimes analyzed by an HP 5890A gas chromatograph (GC) (Santa Clara, CA). One Hayesep D. 100/120 packed column from Alltech Associates (Deerfield, IL) was used to analyze the gas products. Helium was used as the carrier gas with the oven temperature at 100 °C and the thermal conductivity detector (TCD) temperature at 125 °C. A Quantek Model 906 CO2 analyzer (Grafton, MA) was used much more often so that the data collected allowed estimation of real time CO2 concentration fluctuations in different product gas streams. Usually there is a very small concentration range for each product stream; an average value is reported in terms of the CO2 concentration. A pressure transducer installed inside the oven and directly connected to the tube side of the membrane module revealed detailed pressure changes during gas absorption. This allowed a better understanding of the PSMAB process as we have seen earlier.20 For example, from the pressure drop in the absorption step, we could estimate how much gas was absorbed and find out which absorbent was better. 2.4. Five-Valve System for PSMAB Process. The newly designed five-valve system introduced earlier20 for the pressure swing membrane absorption process is shown in Figure 3 for the sake of illustration. 3308



Article

dendrimer Gen 0). As shown in Table 2, when ionic liquid [bmim][DCA] was used as the solvent, for PEEK-L-I, the feed pressure could go up to 250 psig. PEEK-L-II and PEEK-L-III modules could also easily withhold liquids at high pressure up to 250 psig. The symbol “>300 psig” in Table 2 means it was tested up to 300 psig and there was no leakage. 3.2. Long Duration Test to Determine the Preferred time duration for the Absorption Step. As mentioned in our previous work,20 the absorption step is important since most gas−liquid contact and absorption will happen during this period. To find out the preferred duration for the absorption step with various PEEK modules, a 900 s long absorption process was carried out with PEEK-L-I and PEEK-L-II modules at room temperature with a feed gas pressure of 150 psig and pure [bmim][DCA] as absorbent. Detailed results are presented in Figure 4. The dead volumes in the tube-side headers at both sides of PEEK-L-I and PEEK-L-II modules were greatly reduced by adding small Teflon balls before testing (detailed explanations provided in section 3.3). From Figure 4, clearly similar pressure drop tendencies were observed in both PEEK-L-I and PEEK-L-II: at the beginning sharp pressure drops were encountered and then a relatively gradual and a slow pressure drop could be seen that should be mostly attributed to the progress toward absorbent liquid saturation. Under the same testing conditions during 900 s absorption, module PEEK-L-II generated a pressure drop of ∼17.71 psig, which was much larger than 13.08 psig in PEEKL-I. Table 1 shows that the main difference between PEEK-L-I and PEEK-L-II is the latter’s much lower fiber packing density, indicating that much more absorbent liquid exists between contiguous fibers in PEEK-L-II. More liquid will lead to more gas absorption, which yields a larger gas pressure drop. This is the reason for different pressure drops observed in modules PEEK-L-I and PEEK-L-II for the same absorption time. Figure 4 shows also the percentage of the total gas pressure drop as a function of the absorption time. After 60 s, which was only 6.67% of the total 900 s absorption duration, 62.75% of the total pressure drop for PEEK-L-I and 57.68% of that for PEEK-L-II have been achieved. In other words, around 60% absorption has been completed within the first 60 s. Therefore,

number of serious problems will be created if we have to pump this highly viscous liquid through the tiny interfiber channels on the shell side.29 Third, the volume of the costly IL used is reduced considerably in the PSMAB mode of operation. To start with, due to flow pressure drop it will increase the breakthrough pressure considerably, which will lead to further pore size reduction to achieve a stable gas−liquid interface and considerable additional resistance to mass transfer. Nonuniform liquid flow distribution on the shell side is a major problem in membrane contactor processes; it will be a major problem here, too. In addition, the pumping energy cost for the highly viscous solution will become significant. However, the mass transfer resistance will be reduced due to flow as opposed to a stagnant liquid here. On the other hand, since gas absorption takes place here over a very short period of time around 30−60 s, the rate of gas absorption to the stagnant liquid is not going to be low since it is inversely proportional to the time period t as shown below:33 rate of physical absorption ∝

DCO2 πt

In fact, the rate of physical absorption will be high.

3. RESULTS AND DISCUSSION 3.1. Breakthrough Pressure of Membrane Modules. All breakthrough pressure test results for three types of PEEK-L membrane modules with different liquids are listed in Table 2. Table 2. Breakthrough Pressure Test Results module type

water

[bmim][DCA]

PEG 400

PEEK-L-I PEEK-L-II PEEK-L-III

>260 psiga >300 psig >250 psig

>260 psig >250 psig >250 psig

>260 psig >250 psig N/A

a The symbol “>260 psig” here and other data means that it was tested up to 260 psig.

Three kinds of liquids were used for testing; only [bmim][DCA] was applied as the absorbent (with or without PAMAM

Figure 4. Pressure drop magnitude and the percentage of the total tube-side pressure drop during 900 s absorption for PEEK-L-I and PEEK-L-II modules at 25 °C for a feed pressure of 150 psig. 3309



Article

liquid. Thus it seems reasonable that a better CO2-rich product will be generated at a higher feed gas pressure. On the other hand, increasing feed gas pressure will have a negative influence on the He-rich product quality since more gas will be introduced into the module tube side for a fixed gas−liquid contacting area. A much higher gas absorption burden from the higher feed gas pressure will inevitably cause degradation in the quality of the He-rich product. The same effect of increasing feed gas pressure could be seen from Figure 6 for the PEEK-LII module with full and reduced dead volumes. However, compared with the PEEK-L-II module with full dead volume, after dead volume reduction by adding small Teflon balls on both sides, the PSMAB process generates much better product qualities at both CO2 and helium product ends. For example, when feed gas pressure was kept at 250 psig, CO2-rich product from the reduced dead volume module had a CO 2 concentration about 88.2%, which was much higher than 79.9% for the module having full dead volume. For He-rich product, its CO2 concentration after dead volume reduction was around 21.5%, which was much lower than 30.25% for the case of full dead volume. As mentioned at the beginning of this section, feed gas existing in the dead volume has no chance to contact the shellside absorbent: thus it will not undergo any gas absorption. Dead volume reduction means a higher fraction of the tube-side gas will have an opportunity to contact the shell-side absorbent; thus better product qualities are expected. Since we have demonstrated product quality improvement by reducing dead volume, for all tests listed below, dead volume on both sides of membrane module will be reduced as much as possible by adding small Teflon balls (unless otherwise mentioned, for example, for the module PEEK-L-III). Another aspect about the PSMAB results in Figure 6 is that the He-rich product qualities are not very satisfactory. This is because we deliberately took out most of the gas in the tube side after the absorption step as He-rich product as shown in Table 3. In our previous work20 we have reported that He-rich product quality could be monitored by changing its withdrawal quantity in each cycle. In this section our strategy was to take out most gas as He-rich product so that we can achieve the best CO2-rich product quality. Continued efforts of how to improve He-rich product quality will be discussed later. 3.4. Comparison between PEEK-L-I and PEEK-L-II Modules for Runs at Different Temperatures. The main difference between PEEK-L-I and PEEK-L-II membrane modules is that the latter has a much lower fiber packing density, implying that more absorbent liquid will exist in the space between contiguous fibers in PEEK-L-II. Figure 4 shows already that under the same testing conditions a larger pressure drop caused by more gas absorption took place in PEEK-L-II. To investigate how this fiber packing density will influence the PSMAB process and its product qualities, tests were carried out at different temperatures for a feed gas pressure of 150 psig. Pure [bmim][DCA] was used as the absorbent. The same cycle step durations were applied as in section 3.3. Pressure changes in the tube side during one complete cycle were close to the values for 150 psig in Table 3 (the values are: 18 psig for the third step, 5.0 psig for the fourth step, and −3.5 psig for the sixth step). Pressure drops in the absorption step have been presented in detail in Figure 7. If we compare pressure drops in Figure 7a and Figure 7b, the first conclusion is that compared with PEEK-L-I, under the same running conditions, PEEK-L-II showed a larger pressure

for further tests, the preferred duration of the absorption step for PEEK-L-I and PEEK-L-II was set as 60 s unless otherwise mentioned. 3.3. Performance Improvement by Reducing Dead Volume of PEEK-L-II Module. Here “dead volume” is defined as the space in a membrane module where the existing gas has no chance of contacting the absorbent liquid on the shell side. In a membrane module, two locations contribute to the dead volume: the tube-side header section of the membrane module and related connections, and the potted section of the module at both ends of the module. Nothing can be done about the potted section dead volume; for the tube-side header section and related connections, the dead volume can be considerably reduced by adding small diameter Teflon balls. To find out how much improvement could be achieved after dead volume reduction by this method, tests were carried out at room temperature at different feed gas pressures with the PEEK-L-II membrane module. The dead volume formed by header section connections for PEEK-L-II has been calculated to be about 35.75 cm3; the total volume of Teflon balls added to both ends is around 28.5 cm3. Pure [bmim][DCA] was used as absorbent. The cycle time was set as 5 s for feed gas introduction, 60 s for absorption, 2 s for He-rich product withdrawal, 2 s for middle part gas withdrawal, 60 s for CO2-rich product withdrawal, and another 2 s for middle part gas recycle. Detailed pressure changes in the module tube side during one complete cycle under different feed gas pressures are summarized in Table 3. Pressure drops within the absorption step are presented in Figure 5; product quality comparison is shown in Figure 6. Table 3. Pressure Changes in Membrane Module Tube Side during One Complete Cycle under Different Feed Gas Pressures feed gas press. (psig)

first step (psig)

second stepa (psig)

third step (psig)

fourth step (psig)

fifth step (psig)

sixth step (psig)

100 150 200 250

102 153 207 261

97.1/96.2 145.3/144.6 196.8/195.6 237.5/236.3

14.0 25.0 33.0 43.0

4.9 10.0 16.0 21.0

−14.5 −14.5 −14.5 −14.5

−5.0 −2.0 1.0 4.0

The first pressure value is for PEEK-L-II with dead volume and the second value is for reduced dead volume. a

From Figure 5 two brief conclusions could be made: First, higher feed gas pressure will lead to more gas absorption and thus a larger pressure drop in the module tube side can be generated; this is clearly seen from both Figure 5a and Figure 5b. Second, after the module header dead volume was reduced, a relatively larger pressure drop in the tube side under the same running conditions could be seen since for this case the ratio of the gas in the tube side which has a chance to directly contact with the absorbent on the shell side and get absorbed to the total tube-side volume will increase. Of course, this arrangement of Teflon ball based dead volume reduction does not reduce the dead volume in the potted section of the module. That would require an enhanced tube-side gas volume vis-à-vis the potted section of the module (considered later). Figure 6 presents the effect of feed pressure on product qualities and the effect of reducing the dead volume in the tube side of the membrane module. From Figure 5 we already know that increasing the feed gas pressure will generate a larger pressure drop due to higher gas absorption by the shell-side 3310



Article

Figure 5. Pressure drop in PEEK-L-II module tube side during the absorption step at 25 °C for different feed gas pressures: (a) with full dead volume; (b) with reduced dead volume.

drop. This is in accord with results shown in Figure 4. With regard to the effect of temperature, PEEK-L-I performs differently from PEEK-L-II. Pressure drop in the tube side of PEEK-L-I did not vary very much under different test temperatures. However, the test temperature had great influence on the performance of PEEK-L-II: higher temperature brought a larger pressure drop in the tube side. This could be explained by the fact that there is a very limited amount of absorbent surrounding the hollow fibers in PEEK-L-I; thus it will get saturated very quickly and we cannot see obvious differences from different temperatures. Unlike PEEK-L-I, PEEK-L-II has a much lower packing density; thus there is much more absorbent liquid existing between contiguous fibers. As the temperature increases, the rate of absorption increases due to an increased diffusivity and lowered viscosity. Although the equilibrium CO2 solubility is decreased significantly as the temperature is increased, the 60 s duration of absorption is far from equilibrium. Therefore, higher temperature has facilitated

gas absorption in PEEK-L-II during the short gas absorption period leading to higher pressure drop as shown in Figure 7b. Influence of the operating temperature on product qualities from both PEEK-L-I and PEEK-L-II membrane modules is shown in Figure 8. First we will discuss the influence of temperature on product qualities. Unlike the effect of increased feed gas pressure, which improves the quality of the CO2-rich product but reduces the quality of the He-rich product, increased test temperatures showed negative effects for both product streams; this means at higher temperatures both Herich product and CO 2 -rich product will face quality degradation. In our previous studies on gas solubility in [bmim][DCA], we have demonstrated that at higher temperature the solubility selectivity of CO2 over helium in IL will decrease.19 This should be the reason for product quality degradation at higher temperatures. If we compare the product qualities from PEEK-L-I and PEEK-L-II under the same testing conditions, it is clear that both CO2-rich product and He-rich product qualities of PEEK3311



Article

Figure 6. Comparison of the product qualities of PEEK-L-II module between full and reduced dead volumes at 25 °C for different feed gas pressures.

on the shell side of PEEK-L-II module and then PEEK-L-III module. The goal is to find out whether the dendrimer addition improves the separation performance at higher temperatures. First, a set of tests were carried out with a 10.0 wt % dendrimer in [bmim][DCA] solution as the absorbent for 900 s long absorption tests at different temperatures for a feed gas pressure of 150 psig. The corresponding pressure drops in the module tube side are shown in Figure 9a. The total pressure drop within 900 s did not appear to have been influenced much by the test temperature: it varied between 22 and 24 psig. Note, however, that within the first 60 s higher temperatures yielded much higher pressure drops. For example, at the end of 60 s absorption, the following results are obtained: at room temperature, there was a pressure drop of ∼8.5 psig; with the temperature increased to 50 °C, there was a larger pressure drop of ∼9.8 psig; tests at 75 °C showed a much larger pressure drop of about 15.2 psig; the 100 °C test had a slightly increased pressure drop of ∼15.8 psig. This is due to two reasons: first, high temperature increases initially the gas absorption rate; second, adding dendrimer to pure IL will greatly increase its viscosity while a temperature increase will lead to lower liquid viscosity and increase the diffusion coefficient, which is favorable for gas absorption. Furthermore, the reaction kinetics of CO2 absorption vis-à-vis the dendrimer will be enhanced. Therefore, for a dendrimer−IL mixed absorbent, a higher running temperature will be more favorable for the PSMAB process. Since the duration of the absorption step is 60 s or lower, Figure 9b provides a view of these differences between the different temperatures for a 10 wt % dendrimer−IL system within this shorter time range relevant for the PSMAB cycle. Continued PSMAB studies were carried out at different temperatures and 150 psig with the same cycle step durations as described in section 3.3. Pressure changes in the tube side were close to the second line data in Table 3. The detailed pressure drops during the absorption step have been presented in Figure 9b. It has been revealed that the test temperature showed great influence on gas solubility when the dendrimer− IL mixture was used as absorbent. At room temperature because of the high viscosity of mixture absorbent, gas solubility was restricted; thus a pressure drop of only 6.45 psig was

L-II were better than those of PEEK-L-I. The differences for the two product qualities were different: He-rich product qualities of PEEK-L-II were only a little bit better than those of PEEK-LI, while, for CO2-rich product quality PEEK-L-II performed much better than PEEK-L-I. This could be explained as follows: because of its much lower packing density, PEEK-L-II has much more absorbent surrounding fibers and thus much more gas will be absorbed during each cycle. More absorption will of course bring a better CO2-rich product quality. The reason we did not see an obvious enhancement in the He-rich product quality between PEEK-L-I and PEEK-L-II may be attributed to the following two points: first, compared with the absorbed gas, there is still a much larger amount of gas in the module tube side; second, we almost took most of this gas out as the He-rich product. This super large withdrawal amount masked the influence of the test temperature on the He-rich product quality. In addition, one needs a longer fiber length to produce a much higher quality He-rich stream; our membrane modules did not have the length for the volume of high pressure feed gas introduced unlike that in PEEK-S membrane modules studied earlier.20 It has been proven that PEEK-L-II shows much better separation capability over PEEK-L-I. Thus all our continued tests will only focus on PEEK-L-II except when we discuss the performance of the module PEEK-L-III. 3.5. Effect of Adding PAMAM Dendrimer Gen 0 to [bmim][DCA] for High Temperature Performance Improvement. Shifted syngas is generated at a high temperature and high pressure; its CO2 removal will be preferably required to be carried out at a high temperature. We already know that, with pure [bmim][DCA] as absorbent, the PSMAB process will face product quality degradation problems at higher temperatures. There are a number of reports indicating that amine groups of polyamidoamine (PAMAM) dendrimer can react with CO2;20,23−28 therefore, addition of dendrimer to [bmim][DCA] and forming a mixed absorbent is expected to enhance its CO2 absorption performance. Jie et al.20 made a preliminary investigation of the effect of adding dendrimer to the ionic liquid in a ceramic module. Here we have investigated the effect of adding PAMAM dendrimer Gen 0 to [bmim][DCA] present 3312



Article

Figure 7. Pressure drop in the membrane module tube side during absorption step at different temperatures for a feed gas pressure of 150 psig: (a) PEEK-L-I; (b) PEEK-L-II.

temperature was increased to 75 °C, CO2 concentration in the He-rich product was reduced to 22.8% and the CO 2 concentration in CO2-rich product was increased to 80.5%. We can see a great improvement at both product ends. This is primarily due to a reduction in viscosity resulting in faster diffusion and therefore more CO2 absorption. More reaction with the dendrimer is also taking place as a result, leading to better product quality at both ends. When the temperature was increased to 100 °C, a slight degradation in product quality could be seen. The solubility of CO2 in the ionic liquid is significantly reduced at the higher temperature,19 affecting product quality. Even though from Figure 9b we can see very close pressure drops in the tube side during the absorption steps for 75 and 100 °C tests, we believe that compositions of the absorbed gas have been changed a lot because of the temperature change. On the basis of these test results, the continued question will be: Will an increase in dendrimer

observed. When the temperature was increased to 50 °C, the pressure drop in the absorption step increased to 8.73 psig correspondingly since high temperature will decrease the mixture viscosity and enhance the gas diffusion rate. Continued high temperature tests at 75 and 100 °C did not show great differences since both had a higher pressure drop of around 12.5 psig. In terms of product qualities, as shown in Figure 10, a temperature increase showed an obviously positive effect when the dendrimer−IL mixture was used as the absorbent. At room temperature He-rich product with a CO2 concentration as high as 29.3% and CO2-rich product with a CO2 concentration around 59.8% were generated when a 10 wt % dendrimer in [bmim][DCA] mixture was used as absorbent. The very high viscosity of the solution due to the low temperature of 25 °C and dendrimer addition reduced CO2 absorption; therefore, there is considerable CO2 in the He-rich product stream and not enough CO2 in the CO2-rich product stream. When the 3313



Article

30.0 wt % did not bring a pressure drop increase since it showed a value of around 14.3 psig. All these results are in accordance with the long duration absorption test results in Figure 11. Although it is not strictly correct, we may state that product qualities from the PSMAB process may be partially predicted by the pressure drop behavior during its absorption step. Influence of the dendrimer concentration on PSMAB performance in terms of product quality is shown in Figure 13. It is clear that adding dendrimer to IL greatly improved the product qualities at higher temperatures. Among all tested absorbents, pure [bmim][DCA] generated products with the worst qualities: CO2 concentration in He-rich product was as high as 26.2% and at the same time in CO2-rich product it was only 76.5%. When a 10.0 wt % dendrimer−IL mixture was used, the former value decreased to 24.8% and the latter value increased to 78.5%; both have been improved. Continued dendrimer concentration increase to 20.0 wt % did show further improvement since the CO2 concentration in He-rich product decreased to 22.3% and the CO2 concentration in CO2-rich product increased to 84.9%. When the dendrimer concentration was increased to 30.0 wt %, we did not see an obvious further improvement since both product qualities showed a little degradation when compared with the 20.0 wt % mixture. This could be explained by the fact that the 30.0 wt % dendrimer−IL mixture has a much larger viscosity as pointed out earlier. On the basis of the test results of this section, it is clear that adding PAMAM dendrimer Gen 0 to pure IL to form a mixed absorbent will greatly improve its absorption capability at higher running temperatures. Among all mixtures tested, 20.0 wt % dendrimer in IL showed the best performance; therefore, we are going to use it as the main absorbent for further investigations. 3.7. High Feed Pressure Performance of PSMAB Process with 20.0 wt % Dendrimer in [bmim][DCA] Mixture as Absorbent. Most of our tests have been run at a relatively low feed gas pressure of 150 psig. Since precombustion shifted syngas will be generated at a high temperature and high pressure, it is important to carry out the PSMAB process at higher temperatures with a much higher feed gas pressure than 150 psig. In this section, with 20.0 wt % dendrimer in [bmim][DCA] mixture as absorbent, a set of tests were carried out with feed pressure up to 250 psig and temperature as high as 100 °C. At first a set of 900 s duration absorption tests were carried out under different feed gas pressures; the pressure drop results are presented in Figure 14. Feed gas pressure showed great influence on the tube-side gas pressure drops since it was the driving force for gas absorption. Actually this increase was proportional. If we divided the total pressure drop during 900 s absorption by its beginning absolute pressure in the tube side, we got almost the same value of around 17.2−17.5% for different feed pressures. Pressure changes in the module tube side during one complete PSMAB cycle at different feed gas pressures are presented in Figure 15. As stated before, our strategy is trying to take out most of the gas in the tube side after the absorption step as He-rich product; thus we can have a better CO2-rich product. Detailed pressure drops in the tube side during the absorption step at different feed pressures are also listed in Figure 15. Similar to the results shown in Figure 14, we can

Figure 8. Comparison of product qualities between PEEK-L-I and PEEK-L-II modules for different running temperatures when the feed gas pressure is 150 psig.

concentration in IL increase help further improve PSMAB performance at higher temperatures? 3.6. Influence of Dendrimer Concentration in Mixed Absorbents on PSMAB Process Performance at Higher Temperatures. Adding dendrimer to the ionic liquid has been proven to be a possible choice to improve PSMAB process performance at higher temperatures. To determine an optimal dendrimer concentration in IL for the best enhancement, a set of tests using dendrimer−IL mixtures having different dendrimer concentrations were carried out at 100 °C and 150 psig. Similar to what was done earlier, at first 900 s duration absorption process tests were studied and detailed pressure drops in the module tube side with time were recorded (Figure 11). It can be clearly seen that, at higher temperatures up to 100 °C, adding dendrimer could greatly improve the gas solubility in absorbent. That is, in Figure 11, a larger pressure drop was generated. When pure [bmim][DCA] was used as absorbent, a pressure drop of around 21.3 psig could be seen at the end of 900 s absorption duration. Under the same test conditions, a 10.0 wt % dendrimer−IL mixture could generate a larger pressure drop of about 24.6 psig. This value increased to 27.3 psig when a 20.0 wt % dendrimer−IL mixture was used as absorbent, and continued dendrimer concentration increase did not bring a larger pressure drop. A 30.0 wt % dendrimer−IL mixture led to a lower pressure drop of around 26.0 psig. We would like to attribute this abnormal phenomenon to the very high viscosity of 30.0 wt % dendrimer−IL mixture even at a higher temperature. It is likely that a much higher temperature test may show its enhancing effects better. Based on the long duration absorption test results, PSMAB process tests were carried out using dendrimer−ionic liquid mixtures with different dendrimer concentrations at 100 °C with a feed gas pressure of 150 psig. Detailed pressure drops during the absorption step are presented in Figure 12 for the first 65 s. As one can see from Figure 12, when pure [bmim][DCA] was used as absorbent, the lowest pressure drop of around 12.1 psig could be seen. A 10.0 wt % dendrimer−IL mixture showed a slightly larger pressure drop about 12.5 psig, and when dendrimer concentration in [bmim][DCA] was increased to 20.0 wt %, we found a much larger pressure drop of 14.8 psig. Continued dendrimer concentration increase to 3314



Article

Figure 9. (a) Influence of test temperature on pressure drop in module tube side during 900 s absorption for PEEK-L-II for a feed pressure of 150 psig when a 10 wt % dendrimer in [bmim][DCA] mixture was used as absorbent. (b) Pressure drop in PEEK-L-II membrane module tube side during absorption step at different temperatures for a feed gas pressure of 150 psig when a 10 wt % dendrimer in [bmim][DCA] mixture is the absorbent

degradation to 23.6%, and 250 psig generated He-rich product with the highest CO2 concentration around 24.7%. For the CO2-rich product side, we can see an obvious quality improvement when feed gas pressure was increased from 100 to 150 psig since the corresponding CO2 concentration increased from 76.9 to 84.9%. (This could be attributed to a large increase of absorbed gas quantity.) Continued feed gas pressure increase did not show any obvious improvement in product quality since all CO2 concentrations were around 84.9−85.7% even when we did see a pressure drop increase during the absorption step. We do need to recognize that solubility selectivity of CO2 over helium deteriorates at a high temperature such as 100 °C, compared to that in IL at room temperature.19

clearly see a proportional increase in pressure drop when the feed gas pressure was increased. In terms of product qualities shown in Figure 16, as was observed earlier with room temperature tests in section 3.3, feed gas pressure increase showed a positive effect for the CO2rich product quality and a negative influence on the He-rich product quality. Feed gas pressure increase in the tube side means more feed gas needed to be treated for a fixed gas−liquid contacting area and absorbent amount; therefore, the higher burden of CO2 absorption will degrade He-rich product quality. For a feed gas pressure of 100 psig, He-rich product with a CO2 concentration around 18.7% was generated; when feed gas pressure was increased to 150 psig, we got an increased He-rich product with a CO2 concentration around 22.3%. Continued feed gas pressure increase to 200 psig led to He-rich product 3315



Article

Figure 10. Influence of running temperature on product qualities for a feed gas pressure of 150 psig for a 10 wt % dendrimer in [bmim][DCA] mixture used as absorbent (module PEEK-L-II).

Figure 11. Pressure drops in PEEK-L-II module tube side during 900 s absorption from different dendrimer−[bmim][DCA] mixtures at 150 psig and 100 °C.

3.8. Additional Analysis and Second-Stage Tests for Improvements in PSMAB Product Qualities. Based on our test results, by using a 20.0 wt % dendrimer in IL solution as absorbent, when a one-stage PSMAB process is run with a PEEK-L-II module (with dead volume reduced) at 100 °C and 250 psig feed gas pressure, we can have a He-rich product with a CO2 concentration around 24.7% and a CO2-rich product with CO2 concentration around 85.7%. These product qualities are still far from the desired goals. In section 3.3, we demonstrated how much the presence of dead volume in the tube side will affect the final product qualities at both product ends. Even though we have greatly reduced dead volume in the tube side header by adding small Teflon balls and achieved improved product qualities, there is still quite a large dead volume. Therefore, a detailed calculation and analysis will be carried out now based on this information.

As stated earlier the dead volume in the module tube side is composed of two parts: the tube-side header section and related connections of the membrane module; the potted section of the module at both ends. The volume of the first part has been reduced from 35.75 to 7.25 cm3 by adding 28.5 cm3 small Teflon balls on both sides; for the potted sections, the tube-side volume of PEEK-L-II membrane module has been calculated to be 58.29 cm3 with an actual length of 20.0 cm. The total length of the potted section on both sides is 8.0 cm; thus we can easily calculate the dead volume for this section to be 23.32 cm3. The total dead volume on the tube side will be 30.57 cm3, that is, about 52.4% of the effective tube-side volume of 58.29 cm3 (feed gas in this space has an opportunity to contact with the absorbent on the shell side and be absorbed). Here we need to assume that the feed gas in the dead volume will keep its composition unchanged at the original level during 3316



Article

Figure 12. Pressure drop in the PEEK-L-II membrane module tube side during absorption step from different absorbents at 100 °C for a feed gas pressure of 150 psig.

Figure 13. Influence of dendrimer concentration on product qualities when tested at 100 °C for a feed gas pressure of 150 psig.

PEEK-S modules in series.20 In fact, it just follows from the basic behavior of any pressure swing adsorption based purification of the least adsorbed species, namely, helium here and hydrogen in the actual process. Provide length in the feed gas flow direction and achieve high purification of the less adsorbed species. We propose an additional scenario here. One reason for obtaining a lower level of purification in the He-rich stream in this work is that our strategy to generate a best CO2-rich product degrades the He-rich product quality; on the other hand, precombustion shifted syngas treatment needs to be carried out at extremely high temperature and pressure increasing the absorption load on the membrane module. One possible way to improve the He-rich product by introducing a second absorption stage has been reported in our previous work.21 With 20.0 wt % dendrimer in [bmim][DCA] as absorbent and 14.0% CO2 with helium balance as feed gas for

the PSMAB process since it has no chance of contacting the absorbent (axial diffusion will change it with time). On the basis of this assumption, and according to the ratio between the dead volume and the effective volume for absorption, it can be easily calculated that the actual CO2 concentration in the He-rich product will be around 16.7%; that is much better than 24.7% that we got by mixing the purified highly He-rich gas with a CO2-rich gas in the dead volume in the He-rich product end of the system. Correspondingly the actual CO2 concentration in the CO2-rich product also should be higher; as we can see from Figure 6, that means a better quality. The actual CO2 concentration in the He-rich product of around 16.7% is higher than the desired value. We believe that when we put two modules in series we will achieve the desired value as we have pointed out at the end of section 3.9 based on the values we have experimentally obtained earlier with two 3317



Article

pressure. Both products could be recycled as feed gas to avoid helium loss. 3.9. Tests for PSMAB Product Qualities Using the Module PEEK-L-III. There are two important features in the PEEK-L-III module. First, it has a considerably larger tube-side volume resulting from a considerably larger membrane surface area of 0.55 m2 compared to 0.34 m2 in the other two module types studied here. Second, it has a packing density similar to that of the module PEEK-L-II. Further, we did not fill up the tube-side headers with Teflon balls. Table 5 provides the process performance results for this module vis-à-vis the two product streams for 20 wt % dendrimer-containing IL absorbent. One can clearly see an interesting result: for a feed gas at 100 °C and 250 psig pressure, the CO2-rich stream has achieved a CO2 concentration of 90.7%, higher than any value reported so far in this study. Additional studies29 using this PEEK-L-III membrane module and pure ionic liquid [bmim][DCA] indicated that, at 25 °C and 250 psig feed gas pressure, the CO2-rich stream achieved a 92.9% CO2 concentration. Further, the He-rich stream had a lower CO2 concentration than those from other modules. (Further, this study29 provides information about the production rates of the two product streams.) These results suggest that a redesigned membrane absorption module having a larger feed gas volume considerably reduced tube-side dead volumes and a lower interfiber spacing-based packing density will yield much better product qualities. It is of interest to note here also that, using a simple ionic liquid as absorbent, two much smaller PEEK-S modules (each having a membrane surface area of 0.116 m2) in series achieved He-rich product streams20 with a low CO2 concentration of 5.5% at a lower feed pressure of 140 psig and room temperature. These PEEK-S modules had limited dead volumes and fibers which were spaced apart. However, their potting was not suitable for higher pressure and higher temperature studies. On the other hand, the standard PEEK modules which can withstand higher pressures and temperatures have features which are not

Figure 14. Pressure drops in the module tube side during 900 s absorption for a 20.0 wt % dendrimer−[bmim][DCA] mixture at 100 °C for different feed gas pressures.

the second absorption stage, similar PSMAB process tests were carried out at 100 °C. The results are listed in Table 4. As Table 4 shows, PSMAB process tests with 14.0% CO2 in helium as feed gas were carried out at 100 and 150 psig. This feed pressure range is mostly like what we are going to get from the first-stage PSMAB process with a feed gas pressure of up to 250 psig. From the pressure change between the second and third steps, it is clear that most gas in the tube side has been taken out as the He-rich product. The CO2 concentrations in the He-rich products from two test feed pressures are very close around 8.30−8.55%. If we deduct the dead volume influence, the actual CO2 concentration in the helium-rich product will be 5.46%. This means that, after two-stage PSMAB processing, we can achieve a He-rich product with a helium concentration higher than 94.0%. For the CO2-rich product side, 150 psig feed gas pressure generated a CO2 concentration around 26.2%, which was higher than 20.9% with the 100 psig feed gas

Figure 15. Pressure drop in membrane module tube side during one complete PSMAB cycle and absorption step from 20.0 wt % dendrimer− [bmim][DCA] mixture at 100 °C with different feed gas pressures. 3318



Article

Figure 16. Influence of feed gas pressure on product qualities with 20.0 wt % dendrimer−IL mixture as absorbent and tested at 100 °C for PEEK-LII module.

Table 4. Pressure Changes in Membrane Module Tube Side during One Complete Cycle and Product Quality Variation under Different Feed Gas Pressures

a

feed gas press. (psig)

first step (psig)

second step (psig)

third step (psig)

fourth step (psig)

fifth step (psig)

sixth step (psig)

He-rich producta (%)

CO2-rich productb (%)

100 150

96.8 149.8

92.6 143.6

9.75 19.7

1.42 6.84

−14.5 −14.5

−5.30 −1.07

8.30 8.55

20.9 26.2

All product qualities are in terms of CO2 concentration. bAll product qualities are in terms of CO2 concentration.

temperature of the shifted syngas. A prerequisite for a useful preliminary economic analysis is achievement of the required product stream compositions in any given cycle; further, it is desirable to achieve it without a second stage design. Experiments with a redesigned module are needed as a first step to demonstrate achievement of such product compositions from both ends.

Table 5. Product Qualities at Different Temperatures and Feed Pressures for PEEK-L-IIIa with 20 wt % Dendrimer in [bmim][DCA] as the Liquid Absorbent for a Five-Valve PSMAB System

a

press. (psig)

temp (°C)

CO2-rich product (% CO2)

He-rich product (% CO2)

100 200 250 100 200 250 200 250

50 50 50 75 75 75 100 100

85.9 88.3 89.5 81.9 88.4 89.3 89.8 90.7

20.9 24.7 25.4 23.3 24.4 26.5 25.2 25.9

4. CONCLUDING REMARKS Based on our previous PSMAB process studies, we have demonstrated here that decreasing the dead volume in the tube side by adding small Teflon balls, increasing space between contiguous fibers by decreasing fiber packing density, and increasing the tube-side gas volume show considerable positive effects for PSMAB product qualities. At room temperature, [bmim][DCA] shows satisfactory absorption capability for CO2 removal from precombustion syngas, while when running at high temperature such as 100 °C, its suitability as absorbent has been obviously compromised because of a substantial reduction in gas solubility. The primary amines of PAMAM dendrimer Gen 0 can react with CO2 even in the absence of moisture; thus its addition to IL can substantially improve the PSMAB process performance at higher temperatures. This assumption has been verified by continued tests with dendrimer−[bmim][DCA] mixtures at 100 °C; 20.0 wt % dendrimer in IL seems to be an appropriate concentration since at this point the enhancing effect of adding dendrimer will not be compromised by the negative effect of inevitable viscosity increase. High feed gas pressure tests up to 250 psig have been carried out and stable PSMAB products could be generated, showing the suitability of PEEK hollow fiber membrane modules for this designed

Without any Teflon balls to reduce the tube-side dead volume.

beneficial to the PSMAB process. Therefore, variation in the PEEK module design was employed here. What should be mentioned also is that for practical applications membrane modules with much larger effective areas (longer fibers) will be required, which means the dead volume ratio over the effective absorption volume is going to be much lower. Thus its influence on product quality will be greatly reduced and a better product quality close to its actual state will be achieved. This could satisfactorily explain the meaning of our efforts for dead volume analysis here. This study and earlier publications, namely, Jie et al.20 and Chau et al.,29 have essentially explored the feasibility of the proposed process for separation of shifted syngas at higher temperatures which are still significantly lower than the original 3319



Article

(15) Yuan, X.; Zhang, S.; Chen, Y.; Lu, X.; Dai, W.; Mori, R. Solubilities of Gases in 1,1,3,3-Tetramethylguanidium Lactate at Elevated Pressures. J. Chem. Eng. Data 2006, 51, 645. (16) Yuan, X.; Zhang, S.; Liu, J.; Lu, X. Solubilities of CO2 in hydroxyl ammonium ionic liquids at elevated pressures. Fluid Phase Equilib. 2007, 257, 195. (17) Raeissi, S.; Peters, C. J. A potential ionic liquid for CO2separating gas membranes: selection and gas solubility studies. Green Chem. 2009, 11, 185. (18) Sudhir, N. V. K. A.; Mellein, B. R.; Saurer, E. M.; Brennecke, J. F. High-pressure phase behavior of carbon dioxide with imidazoliumbased ionic liquids. J. Phys. Chem. B 2004, 108, 20355. (19) Chau, J.; Obuskovic, G.; Jie, X.; Tripura, M.; Sirkar, K. K. Solubilities of CO2 and He in ionic liquid containing Poly(amido amine) dendrimer Gen 0. Ind. Eng. Chem. Res. 2013, 52, 10484. (20) Jie, X.; Chau, J.; Obuskovic, G.; Sirkar, K. K. Preliminary studies of CO2 removal from pre-combustion syngas through pressure swing membrane absorption process with ionic liquid as absorbent. Ind. Eng. Chem. Res. 2013, 52, 8783. (21) Bhaumik, S.; Majumdar, S.; Sirkar, K. K. Hollow-fiber membrane-based rapid pressure swing absorption. AIChE J. 1996, 42, 409. (22) Obuskovic, G.; Poddar, T. K.; Sirkar, K. K. Flow swing membrane absorption-permeation. Ind. Eng. Chem. Res. 1998, 37, 212. (23) Kovvali, A. S.; Chen, H.; Sirkar, K. K. Dendrimer Membranes: A CO2-selective molecular gate. J. Am. Chem. Soc. 2000, 122, 7594. (24) Kovvali, A. S.; Sirkar, K. K. Dendrimer liquid membranes: CO2 separation from gas mixtures. Ind. Eng. Chem. Res. 2001, 40, 2502. (25) Kosaraju, P.; Kovvali, A. S.; Korikov, A.; Sirkar, K. K. Hollow fiber membrane contactor based CO2 absorption-stripping using novel solvents and membranes. Ind. Eng. Chem. Res. 2005, 44, 1250. (26) Duan, S.; Kouketsu, T.; Kazama, S.; Yamada, K. Development of PMAM dendrimer composite membranes for CO2 separation. J. Membr. Sci. 2006, 283, 2. (27) Taniguchi, I.; Duan, S.; Kazama, S.; Fujioka, Y. Facile fabrication of a novel high performance CO2 separation membrane: Immobilization of poly (amidoamine) dendrimers in poly(ethylene glycol) networks. J. Membr. Sci. 2008, 322, 277. (28) Duan, S.; Kouketsu, T.; Kai, T.; Kazama, S.; Yamada, K. PAMAM dendrimer composite membrane for CO2 separation: formation of a chitosan gutter layer. J. Membr. Sci. 2007, 287, 51. (29) Chau, J.; Obuskovic, G.; Jie, X.; Sirkar, K. K. Pressure swing membrane absorption process for shifted syngas separation: modeling vs. experiments for pure ionic liquid. J. Membr. Sci. 2014, 453, 61. (30) Karadas, F.; Atilhan, M.; Aparicio, S. Review on the use of ionic liquids (ILs) as alternative fluids for CO2 capture and natural gas sweetening. Energy Fuels 2010, 24, 5817. (31) Albo, J.; Luis, P.; Irabien, A. Carbon dioxide capture from flue gases using a cross-flow membrane contactor and the ionic liquid 1ethyl-3-methylimidazolium ethylsulfate. Ind. Eng. Chem. Res. 2010, 49, 11045. (32) Zhang, Y.; Wang, R. Gas-liquid membrane contactors for acid gas removal: recent advances and future challenges. Curr. Opin. Chem. Eng. 2013, 2, 255. (33) Danckwerts, P. V. The absorption of gases in liquids. Pure Appl. Chem. 1965, 10, 625.

PSMAB process. Additional analysis shows that, in the absence of the effect of dead volume, using a two-stage PSMAB process, with 20.0 wt % dendrimer in IL as absorbent, keeping a feed gas pressure up to 250 psig, and running at 100 °C, a He-rich product with CO2 concentration as low as 5.46% and a CO2rich product with CO2 concentration higher than 85.7% will be achieved. Using a membrane module having a larger gas volume allowed the achievement of a CO2-rich product having a CO2 concentration of 90.7%.

■

AUTHOR INFORMATION

Corresponding Author

*Tel.: 973-596-8447. Fax: 973-642-7854. E-mail: sirkar@njit. edu. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This material is based upon work supported by the U.S. Department of Energy, National Energy Technology Laboratory, under Award No. DE-FE0001323.

■

REFERENCES

(1) Lashof, D. A.; Ahuja, D. R. Relative contributions of greenhouse gas emissions to global warming. Nature 1990, 334, 529. (2) Azar, C.; Lindgren, K.; Larson, E.; Mollersten, K. Carbon capture and storage from fossil fuels and biomasscosts and potential role in stabilizing the atmosphere. Clim. Change 2006, 74, 47. (3) Rubin, E. S.; Chen, C.; Rao, A. B. Cost and performance of fossil fuel power plants with CO2 capture and storage. Energy Policy 2007, 35, 4444. (4) http://epa.gov/ghgreporting/ghgdata/reported/index.html. (5) Kanniche, M.; Gros-Bonnivard, R.; Jaud, P.; Valle-Marcos, J.; Amann, J.; Bouallou, C. Pre-combustion, post-combustion and oxycombustion in thermal power plant for CO2 capture. Appl. Therm. Eng. 2010, 30, 53. (6) Tan, Y.; Douglas, M. A.; Thambimuthu, K. V. CO2 capture using oxygen enhanced combustion strategies for natural gas power plants. Fuel 2002, 81, 1007. (7) Merkel, T. C.; Lin, H.; Wei, X.; Baker, R. Power plant postcombustion carbon dioxide capture: An opportunity for membranes. J. Membr. Sci. 2010, 359, 126. (8) Romeo, L. M.; Bolea, I.; Escosa, J. M. Integration of power plant and amine scrubbing to reduce CO2 capture costs. Appl. Therm. Eng. 2008, 28 (8−9), 1039. (9) Loo, S.; Elk, E. P.; Versteeg, G. F. The removal of carbon dioxide with activated solutions of methyl-diethanol-amine. J. Pet. Sci. Eng. 2007, 55, 135. (10) Figueroa, J. D.; Fout, T.; Plasynski, S.; McIlvried, H.; Srivastava, R. D. Advances in CO2 capture technologythe US Department of Energy’s carbon sequestration program. Int. J. Greenhouse Gas Control 2008, 2 (1), 9. (11) Hwang, H. T.; Harale, A.; Liu, P. K. T.; Sahimi, M.; Tsotsis, T. T. A membrane-based reactive separation system for CO2 removal in a life support system. J. Membr. Sci. 2008, 315 (1−2), 116. (12) Amelio, M.; Morrone, P.; Gallucci, F.; Basile, A. Integrated gasification gas combined cycle plant with membrane reactors: technological and economical analysis. Energy Convers. Manage. 2007, 48 (10), 2680. (13) Shiflett, M. B.; Yokozeki, A. Solubilities and diffusivities of carbon dioxide in ionic liquids: [bmim][PF6] and [bmim][BF4]. Ind. Eng. Chem. Res. 2005, 44, 4453. (14) Yokozeki, A.; Shiflett, M. B. Hydrogen purification using roomtemperature ionic liquids. Appl. Energy 2007, 84, 351. 3320


Enhanced Pressure Swing Membrane Absorption Process for CO2

Recommend Documents