CH4 separation of a facilitated transport membrane for natural gas sweetening

Journal of Membrane Science 423–424 (2012) 150–158

Contents lists available at SciVerse ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Effect of monoethylene glycol and triethylene glycol contamination on CO2/CH4 separation of a facilitated transport membrane for natural gas sweetening ¨ n Mohammad Washim Uddin, May-Britt Hagg Department of Chemical Engineering, Norwegian University of Science and Technology (NTNU), NO-7491, Trondheim, Norway

a r t i c l e i n f o

a b s t r a c t

Article history: Received 5 March 2012 Received in revised form 29 July 2012 Accepted 1 August 2012 Available online 16 August 2012

A CO2-facilitated transport composite membrane made of PVAm/PVA blend was exposed to a humid synthetic natural gas mixture with monoethylene glycol (MEG) and triethylene glycol (TEG). The effects of different parameters such as relative humidity, types of impurities, exposure temperature were analyzed to understand the real mechanism of interaction and their effects on the CO2/CH4 separation performance. Both the CO2 and CH4 permeances were increased after the exposure of hygroscopic MEG and TEG, except in one case. The CO2/CH4 selectivity was slightly reduced by the exposure to MEG since MEG plasticized the membrane a little bit, whereas the selectivity was slightly increased by the exposure to TEG. Water plays a significant role in the overall performance. For this facilitated transport PVAm/PVA blend membrane, the high relative humidity helps the facilitated transport of CO2 through the PVAm/PVA blend composite membrane to maintain its permeation properties to a value very close to that of a fresh membrane. This study reports the effect of MEG and TEG on the CO2/CH4 separation of a PVAm/PVA blend composite membrane, and documents a positive step forward for using this membrane in a rigorous environment of natural gas sweetening where entrained glycol is considered as a potential threat to the membrane. & 2012 Elsevier B.V. All rights reserved.

Keywords: Monoethylene glycol Triethylene glycol Blend membrane Facilitated transport Humidity

1. Introduction The importance of removing acid gases such as carbon dioxide (CO2) and hydrogen sulfide (H2S) from natural gas is well known. Membrane technology for removal of CO2 from natural gas was introduced already around 30 years ago, and is expected to take more shares from other conventional capture processes in upcoming years. Even for the unconventional source like shale gas where the reserve is thought to be more than five times the proven amount of conventional natural gas, membranes are now going to be used to remove CO2 from shale gas in British Columbia, Canada [1]. However, the performances of a CO2 removal membrane may be reduced by different components which are present in the gas. Most of these components are originally present in the geological reservoirs (water, heavy hydrocarbons, etc) but others may also come from the units/ chemicals used in the natural gas pre-treatment chains either accidentally (glycols, compressor oils) or intentionally (well additives, corrosion inhibitors).

n

Corresponding author. Tel.: þ47 73594033; fax: þ 47 73594080. ¨ E-mail address: [email protected] (M.-B. Hagg).

0376-7388/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.memsci.2012.08.011

Natural gas generally coexists with water in the geological reservoir. Water in the liquid phase causes corrosion or erosion problems in pipelines and equipment, particularly when carbon dioxide and hydrogen sulfide are present in the gas—this may cause the formation of hydrates with hydrocarbons [2–3]. Dehydration with glycol is a widely used absorption technology in natural gas processing to meet pipeline specifications [4]. Glycol dehydrators and chillers are typically used upstream of membrane modules to prevent hydrate formation or freeze-up [5]. However, glycol dehydrators may also pose a potential hazard to membrane systems: for instance, an upset in dehydrator operation may flood the membrane modules with triethylene glycol (TEG), commonly used as the working fluid [6]. Glycol dehydrators are not well suited for use on small gas stream or on offshore platforms, increasingly common source of natural gas [7]. Instead, glycols can be added to the well stream as antifreeze to inhibit the formation of hydrates before the raw natural gas is being transported through long distance pipelines, quite common for offshore fields. For instance, monoethylene glycol (MEG) is used as the antifreeze in the Statoil operated Snøhvit field in Barents Sea with 145 km multiphase-flow pipeline running to the LNG plant in Hammerfest, Norway [8,9]. Though the glycol–water mixture is separated, a trace amount of glycol may be carried by the natural gas. A recent simulation study on the MEG and TEG

M. Washim Uddin, M.-B. H¨ agg / Journal of Membrane Science 423–424 (2012) 150–158

injection to the natural gas shows that most of the glycol lost is associated with the vaporization of the glycols into the processed natural gas [10]. However, almost no study is found in the literature which describes the effects of these carried-over glycols on the CO2/CH4 separation performance of a membrane. Most of the CO2/CH4 separation studies with membranes deal with either pure gas permeation experiments or binary mixed-gas experiments [11]. For application in industry, it is important to examine multiple mixed-gas environments and interactions with the components not only present geologically in natural gas wells but also the components added artificially or accidentally in natural gas streams such as glycols. Among the three common varieties of glycols, diethylene glycol (DEG) is the most toxic. TEG is the most commonly used glycol to dehydrate natural gas by glycol dehydration unit, whereas MEG is popular for being injected into raw natural gas as antifreeze (inhibitor to gas hydrate and ice formation). In this study, MEG and TEG are chosen as the model impurities to study the effects of these two glycols on the performance of a polymeric blend composite membrane in CO2/CH4 separation. The PVAm/PVA blend composite membrane acts as a CO2 facilitated transport membrane in the presence of water and shows impressive permeation performance in CO2/CH4 separation as reported earlier in [12,13]. The study of the effects of hydrogen sulfide and hydrocarbons on this membrane is being reported separately in parallel. In the current article, the effects of MEG and TEG on membrane permeation properties in presence of water are reported. Static durability experiments of the PVAm/PVA blend membrane under the exposure of MEG and TEG were performed, and the CO2 permeation properties of the membrane were compared before and after different exposure conditions. The term static exposure is used since the membrane is kept in a durability chamber filled with the test gases during the entire exposure. The impurity gases are hence not forced to pass through the membrane as they do in a standard permeation module. The performance of the membrane is checked before and after exposure. The support of this blend composite membrane is polysulfone (PSf), a well-known glassy polymer, on a non-woven polypropylene (PP). The rigidity and highly ordered structure of the PSf glassy polymer might make them more susceptible to the harmful effects of condensable impurities [11]. MEG and TEG are highly condensable, and have very low vapor pressure. Hence, the effects of these condensable additives in natural gas were studied on the performance of the composite PVAm/PVA–PSf membrane. The need to study the impacts of entrained glycol impurities of natural gas streams on the performance of a membrane was mentioned almost a decade ago by Vu et al. [11], however, very few studies were found in the literature. TEG is reported by Ito et al. [14–16] to be used in liquid membranes to separate out volatile organic compound (VOC), while the dehydration behaviour of ethylene glycol (MEG) water mix in different pervaporation membranes is documented by several researchers [17–22]. Polyethylene glycol (PEG) is introduced in numerous blends to make different membranes. The early reverse osmosis cellulose diacetate or triacetate membranes [23,24] were treated with glycerol as a conditioning agent, which allows easy rewetting with water and physically holds open the pores of the asymmetric microporous membrane structure; however, for gas separation membranes glycerol is not an option since it may physically inhibit the passage of gases [6]. The effects of CO2 on glassy polymer are documented by several researchers in [25–30]. A number of studies describing the effects of H2S, hydrocarbons, water, and both condensable and non-condensable impurities on different membranes are likewise found in the literature [31–44]. However, no such studies are found on the effects of MEG or TEG in CO2/CH4 separation membrane. Glycols are well known as hygroscopic compounds and are also used as plasticizers [45–46].

151

Since the transport behaviour of gases like CO2, CH4 in PVAm/PVA blend membrane are strongly correlated with relative humidity, it was expected that MEG and TEG should have some effects on the permeation behaviors of this facilitated transport membrane in presence of H2O. One of the most important findings of this research was that the permeation behaviors of the membrane were found to be dependent on the type of glycols, temperature and more profoundly on relative humidity. Moreover, the effects of MEG and TEG on the CO2, CH4 transports were studied with respect to relative humidity and hydrophilicity of the membrane. The hydrophilicity of the membrane was increased by the exposure in both cases, but the separation performances of the membrane were affected differently by MEG and TEG. The study also gave an indication about the potential operational relative humidity range where the CO2 selective PVAm/PVA blend membrane may perform best in presence of these glycols in natural gas sweetening.

2. Experimental 2.1. Membrane preparation, characterization and permeation experiment The composite membrane is composed of a thin selective layer of PVAm/PVA blend on a polysulfone (PSf) on non-woven polypropylene (PP) support membrane (Alfa Laval, MWCO 50,000). The membrane was prepared the same way as reported earlier in [12,47,48], with the exception that a high molecular weight PVAm (BASF, MW 340,000) was used after successive purification steps as described in [49]. The membrane was characterized by a field emission scanning electron microscope (FESEM, Zeiss Ultra 55 Limited Edition) and water contact angle on the membrane was measured by a CAM 200 (KSV Instruments Ltd. Finland) equipped with a high speed camera. All permeation experiments reported here were carried out at temperature of 25 1C, feed pressure at 2 bars, with 10.0% CO2 and 90.0% CH4 (from YaraPraxair) as feed and with nitrogen as sweep in a permeation set-up as described in [47]. The sweep was saturated with water, while the relative humidity of the feed was controlled in a range of 20–92%. 2.2. Static exposure experiments with impurities The pre-mix gas used in this study was supplied by YaraPraxair at 100 bars and the compositions is 10.0 mol% of CO2 and 90.0 mol% of CH4. The static exposure of the blend membrane is done with this gas mixture saturated with water and MEG/TEG (one at a time) and kept at 2 bars and 10 bars each for 1 week in durability chambers made of stainless steel of a volume of 0.60 L. Both the MEG (Purity grade, Z99.5%) and TEG (Bioultra grade, anhydrous, Z99.0%) used are from Sigma Aldrich. The schematic structures of MEG [50] and TEG [51] are shown in Fig. 1. Fig. 2 shows a schematic diagram of the set-up used for the static exposure experiments. The system was first evacuated and then the exposure gas was passed through the water humidifier to become saturated before being fed to the durability chamber. Since MEG and TEG are very hygroscopic in nature a special procedure is followed to ensure that the feed gas is saturated with

Fig. 1. Schematic structure of (a) MEG and (b) TEG.

152

M. Washim Uddin, M.-B. H¨ agg / Journal of Membrane Science 423–424 (2012) 150–158

Fig. 2. Schematic diagram of the experimental set-up used for static durability exposure experiment.

both water and MEG/TEG and is described below. The reason to choose 2 bars is that both MEG and TEG have very low saturated vapor pressure at room temperature, which means difficult to vaporize, and it would be even more difficult at higher pressure. Since the aim of this study to analyze the effects of MEG and TEG on the membrane permeation properties, 2 bars was chosen as the pressure of static durability experiment at temperatures of 25 1C and 50 1C. All experiments were performed within a confined environment securing no leakage to surroundings. The experimental protocol is described as follows, and wherever MEG/TEG is mentioned, the procedure was the same for both, but only one was tested at a time: 1. The performance of a fresh membrane sample was measured as described in Section 2.1 with pre-mix gas with a composition of 10.0 mol% of CO2 and 90.0 mol% of CH4, the measured performance is defined as the ‘‘before exposure value’’. Then the membrane was taken out from the permeation set-up and placed in the durability chamber. 2. The durability chamber containing the membrane was evacuated for 10 min. and then slowly filled with the exposure gas saturated with water to reach a pressure just higher than the atmospheric. The three-way valve indicated by no. 3 in Fig. 2, was then opened in the direction to the valve no. 4. The gas was kept flowing through the humidifier, durability chambers connected in series at a flow rate of 1.5 N ml/s for 1 h and vented to the atmosphere through valve no. 9—this step is sufficient enough to saturate the water in the humidifier, with the pre-mix feed gas. Then three-way valve no. 3 was opened in the direction to the valve no. 5. The gas was kept flowing through the water humidifier, MEG/TEG column and durability chambers connected in series at a flow rate of 1.5 N ml/s for 1 h—this step was sufficient enough to saturate the MEG/TEG column, with the pre-mix feed gas. Then the durability chamber was evacuated with a vacuum pump for 10 min. by connecting the outlet valve no. 9 with the vacuum pump. During this evacuation, both valve no. 4 and 5 were kept closed. After 10 min of evacuation, valve no. 5 was slightly opened very slowly and carefully and then closed again. The opening and closing of valve no. 5 was repeated several times. The aim of this step is to fill the durability chambers with MEG/TEG vapor. Since both MEG and TEG have very low vapor pressure, this special step was taken to ensure the evaporation of MEG/TEG under vacuum. Then both the outlet valve no. 9 and valve no. 6 were closed and vacuum pump was stopped. The (3-way) valve no. 3 was opened to the direction of valve no. 4 and valve no. 4 was slightly opened. The durability chambers were allowed to be filled with the humid feed gas to the desired pressure of 2 bars, which already contained MEG/TEG vapor. Then all the inlet valves were closed and the

two chambers were disconnected from the gas feed line and left for one week. 3. The sample membranes were kept in the sealed durability chambers one at a temperature of 25 1C and another at 50 1C, at the desired pressures of 2 bars for one week. 4. After the exposure time set, the durability chamber was depressurized by releasing the gas gently at a flow rate of 1.5 N ml/s. The durability chamber was flushed with nitrogen for 1 h and then kept at vacuum for 10 min. The nitrogen flushing and subsequent evacuation steps were repeated twice in order to ensure the desorption of contaminant from the membrane. 5. The exposed membrane was taken out of the chamber and mounted in the permeation set-up. The membrane was flushed by the permeate gas mixture overnight and the performance was measured as stated in step 1, and is defined as the ‘‘after exposure value’’.

3. Results and discussion The main purpose of the current investigations is to detect and understand any changes in the PVAm/PVA membrane material upon exposure to impurities in natural gas – the membrane material itself used here is not optimized with respect to separation performance – separate studies are currenly being carried out with an optimized membrane—results are not yet published. The structure of the blend composite membrane will be discussed before analysing the effects of the glycols on the permeation properties of the membrane as the structure will be associated with any structural changes which may take place. The scanning electron microscopic (SEM) images of the cross-sectional view of the composite membrane (before exposure) are shown in Fig. 3, where (a) shows approximately 0.77 mm dense layer of PVAm/PVA blend on the relatively porous polysulfone support. The cross-sectional view of the whole composite membrane as shown in Fig. 3b indicates the total thickness of approximately 284 mm which includes the polypropylene (PP) non-woven part, the PSf substrate layer and with the PVAm/PVA top layer. In CO2/CH4 separation, CO2 transport is facilitated through the membrane as HCO3 ions at water swollen conditions, the amino groups of PVAm help to initiate the interactions of CO2 with water. However, at high pressures the carriers will become saturated, and the facilitated transport will be of less importance—this is not discussed in the current paper. Methane is transported only by the solution– diffusion mechanism. The permeance performance of the membrane reported here is not as high as reported earlier [12,13], but it should be noted that the aim of this research is to analyze the effect of glycols on the performance. The membranes used in this study were not prepared by the optimized techniques as reported in the mentioned references.

M. Washim Uddin, M.-B. H¨ agg / Journal of Membrane Science 423–424 (2012) 150–158

153

CO2 Permeance, [m3 (STP) / (m2 h bar)]

0.77 µm

CO2 (before exposure) CO2 (after MEG exposure) CH4 (before exposure) CH4 (after MEG exposure)

0.12 0.10

0.004

0.08

0.003

0.06

0.002

0.04 0.001

0.02 0.00

1 µm

20

40 60 80 Relative Humidity, [%]

CH4 Permenace, [m3 (STP) / (m2 h bar)]

0.005

0.14

0.000 100

25

Before exposure After MEG exposure at 25 οC

20 15 10 5 0 20

284 µm

40 60 80 Relative Humidity, [%]

100

20 µm

Fig. 3. SEM images of the cross-sectional PVAm/PVA blend composite membrane (before exposure) showing, (a) approx. 0.77 mm of dense skin layer on the polysulfone (PSf) support, (b) approx. 284 mm of total membrane thickness which includes PVAm/PVA, PSf and PP.

3.1. Comparing permeation performances before and after static exposure to MEG at 25 1C and 50 1C The membrane was placed in the static durability chamber and exposed by a humid mix-gas of a dry percentage 10.0% CO2, 90.0% CH4 almost saturated with monoethylene glycol (MEG) and kept for 1 week at a pressure of 2 bars and temperatures of (a) 25 1C and (b) 50 1C. The membrane was taken out of the chamber, and the permeation performance of the membrane was measured with a mixed-gas (10.0% CO2, 90.0% CH4) in the permeation set-up at 2 bars, 25 1C and a relative humidity (RH) in a range of 24% to 92%. The permeation performance of the membrane after the MEG exposure was compared with the permeation performance measured before the exposure. The results are shown in Fig. 4 (at 25 1C) and Fig. 5 (at 50 1C). Fig. 4 (25 1C), (a) shows the permeance of CO2 and CH4 with relative humidity both before and after exposure (note the different ranges on the two y-axes). As expected, the permeances of CO2 and CH4 increase with relative humidity as the water swollen membrane helps to facilitate the transport; however CO2 permeance increases at a much higher rate than that of CH4, resulting in higher CO2/CH4 selectivity as shown in Fig. 4b. The CO2/CH4 selectivity is found to be somewhat reduced after MEG exposure at 25 1C due to the increase in CH4 permeance as seen in

Normalized Permeance, after/before [-]

2.0

1.2

1.8

1.0

1.6 0.8

1.4 1.2

0.6

1.0

0.4

0.8

CO2 permeance CH4 permeance CO2/CH4 selectivity

0.6 0.4 20

40

60

80

0.2

0.0 100

Normalized CO2/CH/4 Selectivity, after/before [-]

CO2/CH4 Selectivity, [-]

30

Relative Humidity, [%] Fig. 4. Performance of the PVAm/PVA blend membrane before and after exposed to MEG at 2 bars, 25 1C for 1 week. (a) permeance, (b) selectivity, (c) normalized performance.

Fig. 4a. Hardly any permeance change is detected for CO2 upon exposure. In order to investigate the relative changes in permeance of both gases a normalized plot is drawn as shown in Fig. 4c—this plot is based on permeances at only a few experimental points of plots 4a and 4b. The normalized performance is defined by the ratio of the performance after the exposure to the performance before the exposure to MEG at 25 1C. The interesting effect of the MEG in the permeance behaviour is evident in Fig. 4c. Both the CO2 and CH4 permeance are increased by almost 40–45% at a low relative humidity of 25%. This may be attributed by the presence of MEG in the composite membrane. MEG is hygroscopic by nature, and the easily condensable MEG may enter the membrane matrix both into the hydrophobic polysulfone support as well the top layer made of the blend PVAm/PVA where PVA is a hydrogel. The PVAm/PVA blend is a dense layer whereas PSf is porous. The presence of hygroscopic MEG in the membrane matrix can hold water in the support and selective layer even at this

154

M. Washim Uddin, M.-B. H¨ agg / Journal of Membrane Science 423–424 (2012) 150–158

0.005 CO2 (before exposure) CO2 (after MEG exposure) CH4 (before exposure) CH4 (after MEG exposure)

0.15

0.004 0.003

0.10 0.002 0.05 0.00

0.001

20

40 60 80 Relative Humidity, [%]

CH4 Permeance, [m3 (STP) / (m2 h bar)]

CO2 Permeance, [m3 (STP) / (m2 h bar)]

0.20

0.000 100

Before exposure After MEG exposure at 50 οC

25 20 15 10 5

Normalized Permeance, after/before [-]

0 20

40 60 80 Relative Humidity, [%]

100

2.0

1.2

1.8

1.0

1.6 0.8

1.4 1.2

0.6

1.0

0.4

0.8

CO2 permeance CH4 permeance CO2/CH4 selectivity

0.6 0.4

20

40 60 80 Relative Humidity, [%]

0.2

0.0 100

Normalized CO2/CH4 Selectivity, after/before [-]

CO2/CH4 Selectivity, [-]

30

Fig. 5. Performance of the PVAm/PVA blend membrane before and after exposed to MEG at 2 bars, 50 1C for 1 week. (a) permeance, (b) selectivity, (c) normalized performance.

low relative humidity and thus influence the permeation of especially CH4, hence reducing the selectivity in the low humidity range. The diffusional transport will be the main transport in this range. However, with the increase of relative humidity from 25 to 40%, the facilitated transport of CO2 is starting to become important, and, the normalized permeance of CO2 being reduced to unity at a much higher rate than CH4. At the normalized value of 1, the facilitated transport of CO2 is the main transport both in the fresh membrane (before exposure) and in the one exposed to MEG—the relative humidity is now very high, and the diffusional transport of CO2 becomes less important. Hence, the effect of MEG on the CO2 permeance of the membrane is almost negligible or less evident in the relative humidity range higher than 40% (Fig. 4c). The relative change in permeance for CH4 is less pronounced at high relative humidity, but still somewhat increased, resulting in a reduced selectivity for CO2/CH4 of around 12% when exposed to MEG at 25 1C (Fig. 4c). The membrane is highly swollen by water at high

relative humidity which is confirmed by noticing an increase in the overall thickness of the membrane this will be discussed in Section 3.3. In addition, the presence of hydrophilic MEG holds the water in the membrane, i.e., in the relatively hydrophobic PSf support, on non-woven polypropylene and the hydrophilic PVAm/PVA top layer. This extra water may help the permeation of both the gases at the expense of comparatively loose structure of membrane matrix and will reduce the CO2/CH4 selectivity as pointed out. It may be concluded that MEG seems to plasticize the membrane at 25 1C. In Fig. 5a and b the permeation behaviour after the exposure of MEG at 50 1C is presented. The behaviour is found to be similar to that at 25 1C but in less pronounced at low relative humidity as shown. There are two possible reasons of this behaviour: one may be that the condensation of MEG at 50 1C is less at 25 1C, and the other is related to the amount of MEG in the chamber. The chamber was saturated with MEG at 25 1C and then the closed chamber was placed in an oven at 50 1C, which means the absolute amount of MEG inside the chamber was about the same for both temperatures. As the temperature of the chamber was increased, the degree of saturation of MEG in vapor phase was reduced. And at 50 1C, the gas inside the chamber is way below saturation which eventually reduced the probability of condensation of MEG during this experiment and hence less effect of MEG on the membrane permeation behaviour. Less condensation of MEG means less amount of MEG absorbed inside the membrane and hence also less bound water associated with the MEG. This effect will be especially clear when testing permeance with gas in the low relative humidity region. Hence the normalized plots for CH4 and CO2 after the MEG exposure experiment at 50 1C (Fig. 5c) show lower deviations after exposure than those at experiment at 25 1C (Fig. 4c). The normalized CH4 permeance was found to remain almost the same with change of relative humidity. The CO2/CH4 selectivity is thus influenced almost solely by the change of CO2 permeance and the normalized CO2/CH4 selectivity follows the same shape as the normalized CO2 permeance changes with the relative humidity. The mechanism of transport is the same as described at 25 1C. At the high humidity of 90%, the CO2/CH4 selectivity was restored to almost 98% of the pre-exposure value (Fig. 5c). In conclusion, it seems to be clear that the negative effect of MEG exposure at 50 1C is less pronounced due to less sorption of hygroscopic MEG in the membrane and hence less swelling associated with MEG. This also helps the facilitated transport of CO2 over diffusional transport of CH4. The CO2 permeance is noted to increase by almost 9% after the exposure in MEG at 50 1C, at a relative humidity of around 90% (Fig. 5c).

3.2. Comparing permeation performances before and after static exposure to TEG at 25 1C and 50 1C The effects of triethylene glycol (TEG) on the performance of membrane are also investigated at both 25 1C and 50 1C. The permeation performances of the membranes were measured by the same pre-mix gas already stated both before and after the static exposure. The exposure of the membrane was done with the mix-gas (10.0% CO2, 90.0% CH4) almost saturated with TEG at humid conditions, at 2 bars and 25 1C and 50 1C for 1 week in the durability chamber. The performances of the membranes in both conditions (before and after exposure) are shown in Figs. 6 and 7. The CO2 and CH4 permeances and CO2/CH4 selectivity of the samples are documented to be strongly dependent on relative humidity of the permeance gases both before and after exposure, shown in Fig. 6a and b. The normalized performance is shown in Figs. 6c and 7. The normalized performance is defined as the ratio of the performance after static exposure of TEG to the performance before exposure. For both temperatures it can be seen

M. Washim Uddin, M.-B. H¨ agg / Journal of Membrane Science 423–424 (2012) 150–158

CO2 (before exposure) CO2 (after TEG exposure) CH4 (before exposure) CH4 (after TEG exposure)

CO2 Permeance, [m3 (STP) / (m2 h bar)]

0.14 0.12 0.10

0.0035 0.0030

0.08

0.0025

0.06

0.0020

0.04 0.0015

0.02 0.00

20

40 60 80 Relative Humidity, [%]

CH4 Permeance, [m3 (STP) / (m2 h bar)]

0.0040

0.16

0.0010 100

25 20 15 10 5 0 20

Normalized Permeance, after/before [-]

Before exposure After TEG exposure at 25 οC

40 60 80 Relative Humidity, [%]

100

1.8

1.2

1.6

1.0

1.4

0.8

1.2

0.6

1.0

0.4

0.8

CO2 permeance CH4 permeance CO2/CH4 selectivity

0.6 0.4

20

40

60

80

0.2

0.0 100

Normalized CO2/CH4 Selectivity, after/before [-]

CO2/CH4 Selectivity, [-]

30

Relative Humidity, [%]

1.8

1.2

1.6

1.0

1.4

0.8

1.2 0.6 1.0 0.4

0.8

CO2 permeance CH4 permeance

0.6

0.2

CO2/CH4 selectivity

0.4 20

40

60 80 Relative Humidity, [%]

Normalized CO2/CH4 Selectivity, after/before [-]

Normalized Permeance, after/before [-]

Fig. 6. Performance of the PVAm/PVA blend membrane before and after exposed to TEG at 2 bars, 25 1C for 1 week. (a) permeance, (b) selectivity, (c) normalized performance.

0.0 100

Fig. 7. Normalized performance of the PVAm/PVA blend membrane before and after exposed to TEG at 2 bars, 50 1C for 1 week.

155

that the effect of exposure to TEG is much less pronounced than exposure to MEG. Focusing on TEG exposure at 25 1C (Fig. 6), the CO2 permeance and CO2/CH4 selectivity is only slightly increased (unlike MEG), whereas the CH4 permeance after the exposure of TEG hardly shows any effect at all (Fig. 6c). The exceptional difference in permeation behaviour of the membrane being exposed to MEG and TEG may be explained as follows: Both MEG and TEG have low vapor pressures, although the saturated vapor pressure of TEG is several folds lower than that of MEG. For instance, the saturated vapor pressure of MEG is 11 Pa [50] whereas the saturated vapor pressure of TEG is less than 1 Pa [45], at 20 1C. This implies the maximum amount of MEG in vapor phase would be 55 ppm whereas the maximum amount of TEG would be less than 5 ppm at a total pressure of 2 bars and 20 1C. At a same operational temperature and pressure, for instance at 2 bars and 25 1C, the saturated concentration of TEG in vapor phase is almost 10 times less than the saturated amount of MEG in the vapor phase in the durability chamber. Moreover, TEG is larger in size due to the additional ether and methylene groups (–CH2–) as shown in Fig. 1, and it may be more difficult to diffuse into the PSf matrix. Both of these two factors may contribute less amount of TEG to be adsorbed inside the membrane matrix than the amount of MEG. TEG is more likely to be adsorbed on the top selective layer. And the less amount of adsorbed TEG inside the PSf matrix means less amount of associated bounded water. Hence at a low relative humidity of around 27%, the diffusional permeance increase of CO2 or CH4 induced by the bound water of TEG (Fig. 6c) will be less pronounced than it happened for MEG (Fig. 4c). Another interesting observation is that both in Figs. 4 and 5c, at a relative humidity of approximately 40%, the mechanism of transport changes abruptly as the facilitated transport is taking over for CO2. The adsorbed MEG in the PSf matrix keeps the water bound below this relative humidity. When the relative humidity is increased above this value, the availability of free water will make the facilitated transport dominating over the diffusional transport. And the bound water associated with MEG looses the ground of importance in diffusional transport in presence of large amount of free water. Although the amount of adsorbed TEG and associated bound water inside the membrane matrix is less than the amount of adsorbed MEG and associated bound water at the same conditions, the attraction between TEG and bound water is much stronger than that between MEG and bound water. Because TEG is more hygroscopic in nature than MEG [10,45], TEG has extra hydrophilic ether groups in addition to the hydrophilic alcohol groups which are common in MEG (see Fig. 1). This strong affinity between TEG and surrounding bound water, makes the free water to start the facilitated transport at a higher relative humidity than in the case of MEG, and facilitated transport start to dominate over diffusional transport at this higher relative humidity. In the case of TEG, Figs. 6c and 7 show that it occurs around relative humidity of approx. 55%, a higher relative humidity than that for MEG (compare Figs. 4 and 5c, with Figs. 6c and 7). In natural gas processing industry, TEG is extensively used in glycol dehydration units as the working fluid, and because of TEG’s higher hygroscopic nature than MEG, lower amount of TEG is required to achieve the same performance and lower loss of TEG (lower vapor pressure) [10,51,52]. Summing up, with the increase of relative humidity, the facilitated transport of CO2 takes over and the permeation properties after the TEG exposure are close to the value before exposure. Moreover, the permeance of CO2 was found to increase by 7% by the TEG exposure at the relative humidity of 91% (Fig. 6c). This may be attributed with the higher permeation of the facilitated transport of CO2 as hydrated form (HCO3 ions) which is increased by the strong hydrophilic TEG in the membrane. Sata et al. [46] also reported that more hydrated ions transport is much influenced by

156

M. Washim Uddin, M.-B. H¨ agg / Journal of Membrane Science 423–424 (2012) 150–158

the hydrophilic glycols impregnated in a membrane. In this study, the hydrophilic TEG was not intentionally impregnated in the membrane, but the positive effect on permeation of this PVAm/PVA composite membrane was documented. The CO2/CH4 selectivity is also slightly increased by the exposure to TEG. For instance, CO2/CH4 selectivity is increased by 4% at a relative humidity of 91% (Fig. 6c). This may be associated with the increase of CO2 permeance by the TEG exposure. The bulky TEG adsorbed on the surface of the dense PVAm/PVA blend, and may increase the hydrophilicity of the membrane; hence CO2 in the humid gas mixture may permeate more easily than CH4. This behaviour seems to be confirmed by the reduction of water contact angel measured on the blend membrane after exposure—this is more closely discussed in Section 3.3. TEG has an extra hydrophilic ether groups in addition to the hydrophilic alcohol groups and these ether groups are not present in MEG (see Fig. 1). This extra hydrophilic group of TEG may help the CO2–CH4 separation performance to restore or improve a bit even at this over swollen state of the PVAm/PVA blend membrane in high relative humidity. Similar to the exposure of TEG at 25 1C as shown in Fig. 6b and c, the CO2/CH4 selectivity in high relative humidity is found to be increased by 7% after the exposure of TEG at 50 1C (Fig. 7). The possible reason is described above. However, both the permeance of CO2 and CH4 after the exposure of TEG at 50 1C (Fig. 7), was found to be decreased slightly in the high relative humidity range compared to the permeance of the membrane before exposure—this was quite surprising. For instance, a decrease of CO2 permeance by almost 3% at the relative humidity of 91% can be noted in Fig. 7. Although the behaviour is not very pronounced, it was found to be different from the other three cases discussed here for MEG and TEG and is difficult to explain. A combination of different phenomena may contribute this unusual behaviour. One possibility might simply be the experimental error, which might have been checked with repeated experiment at the same conditions. Another explanation may be related to the exceptional behaviour of TEG compared to MEG. TEG is a bulky molecule with very high hydrophilicity due to the extra ether groups in addition to hydrophilic alcholic groups. In the TEG experiment at a high temperature of 50 1C, the amount of the condensation/adsorption of the TEG is less likely and results in a very small amount of TEG adsorbed in the membrane matrix and the selective top layer of PVAm/PVA blend. This tiny amount of the bulky but very hydrophilic TEG in the membrane matrix may behave differently at high relative humidity condition than discussed above. However, further study is required to establish this unusual behaviour of TEG at high temperature. The unusual effect of highly sorptive impurities in glassy polymer is well known, higher concentration of these large penetrant molecules may lead to plasticization while lower concentration

sometimes may lead to antiplasticization [53–57]. A speculation may thus be that the fairly low concentration of absorbed TEG at 50 1C may have resulted in some antiplasticization. The permeation properties of the membrane after the exposure of MEG and TEG at 25 1C and 50 1C are summarized and the normalized values are given in Table 1 both in a low relative humidity range of 24–27% and at a high relative humidity of 91%. Although the normalized plots help to analyze the effects of MEG or TEG on the performance of the membrane in relation to the transport mechanisms, it only shows the relative magnitude. The real magnitude of the performance of the membrane are evident from the plots a and b, which also indicate the contribution of facilitated transport over diffusion transport for higher relative humidity. Normalized plots give an understanding of the transport of gases especially at low relative humidity where the permeances are not big enough to be evident in normal plots like a and b. However, all the plots presented here, a–c, are important to analyze the real nature of the effects and to know the exact value of the permeation performance of the membrane in CO2/CH4 separation after the exposure to impurities like the glycols MEG and TEG in natural gas at two different temperatures. Every experimental point was taken after the system came to equilibrium. Moreover, the performance was checked twice using two different samples with identical conditions. 3.3. Characterization of membrane The contact angle of water on the membrane was measured both before and after exposure conditions and the results are shown in Table 2. It was found that the contact angle of the membrane was reduced after exposure to MEG and TEG, which implies increase in hydrophilicity of the PVAm/PVA blend membrane by the glycol exposure. The decrease of contact angle was larger for TEG than MEG, supporting the fact that TEG has a stronger hygroscopic nature than MEG. The hydrophilic–lipophilic balance (HLB value) of MEG and TEG was reported in the literature to be 9.85 and 10.55, respectively [46]. A higher HLB value for TEG indicates the higher affinity of TEG to water. Moreover, the contact angle reduction by the exposure to MEG at 50 1C was less than the exposure at the lower temperature of 25 1C. An interesting observation is that the contact angle of TEG after exposure at 50 1C, has only slightly decreased. These contact angle measurements seem to confirm our hypothesis on how the MEG and TEG is also affecting the permeation performance of the membrane. Other researchers have reported introducing polyethylene glycol in PSf to increase the hydrophilicity of the blend [58]; and as already mentioned, the hydrophilicity of an ion-exchange membrane was reported to be increased by impregnation of the hydrophilic glycols into the membrane [46]. In the current research, it was found that MEG and TEG also behaves the same

Table 1 Normalized permeation performance values at low and high relative humidity, different static exposure gases at 2 bars, 25 1C and 50 1C for 1 week. Static exposure gas mixture

Normalized permeation performance At low RH (24–27%)

At high RH (91%)

P CO2

P CH4

P CO2 /P CH4

P CO2

P CH4

P CO2 /P CH4

Before exposure After MEG exposure at 25 1C

1.00 1.41 (25%)

1.00 1.45 (25%)

1.00 0.97 (25%)

1.00 1.00

1.00 1.13

1.00 0.88

After MEG exposure at 50 1C

1.15 (25%)

1.25 (25%)

0.92 (25%)

1.09

1.11

0.98

After TEG exposure at 25 1C

1.15 (27%)

0.97 (27%)

1.18 (27%)

1.07

1.02

1.04

After TEG exposure at 50 1C

1.15 (24%)

1.11 (24%)

1.04 (24%)

0.97

0.90

1.07

M. Washim Uddin, M.-B. H¨ agg / Journal of Membrane Science 423–424 (2012) 150–158

157

Table 2 Contact angles of water on the sample membranes before and after exposure to MEG and TEG. Sample

Contact angle with water (1)

Before exposure After MEG exposure at 25 1C After MEG exposure at 50 1C After TEG exposure at 25 1C After TEG exposure at 50 1C

80.1 70.1 74.9 70.8 77.8 70.1 67.3 70.2 67.1 70.9

way as other glycols though the effect of each is different in magnitude. SEM picture were taken for the membrane both before exposure and after exposure to MEG and TEG, in order to see detect any swelling of the composite membrane. The cross-sectional view of the membranes after the exposure to MEG is shown in Fig. 8, which shows an increased thickness from 284 mm for the unexposed membrane (Fig. 3b) to approximately 317 mm and 327 mm for the whole composite membrane after the exposure at 25 1C and 50 1C, respectively. This was checked by repeated measurements of several samples, all indicating the same: the membrane thickness was increased by the exposure to MEG and TEG. The increase in composite membrane thickness is associated with the adsorbed MEG or TEG in the membrane matrix which holds moisture and swells the whole membrane a little bit.

317 µm

20 µm

325 µm 4. Conclusion The permeation performance of the PVAm/PVA blend composite membrane was analyzed before and after the static exposure to MEG and TEG. The effects of different parameters such as relative humidity, type of impurities, exposure temperature were investigated to understand the mechanisms of transport and interaction between the membrane and hygroscopic glycols. The CO2 and CH4 permeance were slightly increased after the exposure of hygroscopic MEG and TEG; however an exception was the exposure to TEG at 50 1C where it was found a slight decrease. The increase of permeance may be attributed to the extra bounded water to the hygroscopic glycols which help the diffusional transport of both gases throughout the composite membrane. An increase in hydrophilicity was also confirmed by the decrease of water contact angle on the blend membrane after exposure to MEG and TEG. At a high relative humidity of 91%, the maximum increase of CO2 permeance was around 9%. Water plays a significant role in the overall performance and effects hygroscopic impurities like MEG and TEG. For this facilitated transport PVAm/PVA blend membrane, the high relative humidity helps the facilitated transport of CO2 through the PVAm/PVA blend composite membrane to maintain its permeation properties at a value very close to that of a fresh membrane. The overall CO2/CH4 selectivity is almost maintained by the exposure to both MEG and TEG. The selectivity is slightly reduced by the exposure of MEG since MEG plasticized the membrane slightly. However, the selectivity was slightly increased by the exposure to TEG. This may be attributed to the relatively highly hygroscopic nature of TEG and a comparatively lower vapor pressure than small size MEG molecule. The comparatively less amount of adsorbed TEG (bulky in size and more highly hygroscopic) in the membrane matrix induces a slight increase of CO2/CH4 selectivity. It may concluded that the study of MEG and TEG on the PVAm/PVA blend composite membrane is a positive step toward the applicability of this membrane in a rigorous environment of natural gas sweetening where entrained glycols are considered as

100 µm

Fig. 8. SEM images of the cross-sectional blend composite membrane after exposed to MEG showing increased thickness of the composite membrane, (a) approx. 317 mm (MEG exposure at 25 1C), (b) approx. 325 mm (MEG exposure at 50 1C).

a potential threat to the membrane. However, further study is needed to investigate the permanent effects over a longer period of time.

Acknowledgements The authors would like to thank colleagues in Memfo R&D group, Dr. Taek-Joong Kim and Dr. Liyuan Deng for making the ˚ membrane, and project partners Knut Ingvar Asen, Eivind Johannessen and Øystein Engen (all Statoil), Magne Lysberg (SINTEF MC) for good discussions. The authors greatly acknowledge the financial support of the GASSMAKS program in the Research Council of Norway and of Statoil ASA.

References [1] UOP Separex membranes remove contaminants from natural gas, Membr. Technol., 2010 (2010) 2. [2] N.P. Cheremisinoff, P. Rosenfeld, Chapter 1—the petroleum industry, in:, Handbook of Pollution Prevention and Cleaner Production—Best Practices in The Petroleum Industry, William Andrew Publishing, Oxford, 2009, pp. 1-97.

158

M. Washim Uddin, M.-B. H¨ agg / Journal of Membrane Science 423–424 (2012) 150–158

[3] S. James G, Chapter 4—production of hydrocarbons from natural gas, in: Handbook of Industrial Hydrocarbon Processes, Gulf Professional Publishing, Boston, 2011, pp. 127-162. [4] R.W. Baker, K. Lokhandwala, Natural gas processing with membranes: an overview, Ind. Eng. Chem. Res. 47 (2008) 2109–2121. [5] D. Dortmundt, K. Doshi, Recent developments in CO2 removal membrane technology, in, UOP LLC, Des Plaines, Illinois, 1999. [6] L.S. White, Evolution of natural gas treatment with membrane systems, in: Y. Yampolskii, B. Freeman (Eds.), Membrane Gas Separation, John Wiley & Sons, Ltd, Chichester, West Sussex, 2010, pp. 313–332. [7] R.W. Baker, Membrane Technology and Applications, second ed., John Wiley & Sons, Ltd, Chichester, West Sussex, 2004. [8] Statoil, Underground storage of carbon dioxide—emission mitigation which counts, in: Statoil’s official pamphlet, Statoil ASA, Stavanger, 2009. [9] Statoil, The long road to LNG—follow the Snøhvit gas through the process plant, in, Statoil, in /http://statoil.com/en/OurOperations/pipelines/Docu ments/veien_til_lng_engelsk_enkel.swfS, 2005. ˚ [10] A. Brathen, Development of processes for natural gas drying—further exploring the TEG injection concept, Master Thesis, June 2008, in: Department of Energy and Process Engineering, Norwegian University of Science and Technology (NTNU), Trondheim, 2008, pp. 1–108. [11] D.Q. Vu, W.J. Koros, S.J. Miller, Effect of condensable impurity in CO2/CH4 gas feeds on performance of mixed matrix membranes using carbon molecular sieves, J. Membr. Sci. 221 (2003) 233–239. ¨ [12] L. Deng, T.-J. Kim, M. Sandru, M.-B. Hagg, PVA/PVAm blend FSC membrane for natural gas sweetening, in: E.A. Hassan, G.V.R. Reklaitis, G.V.R.R. Mahmoud, M. El-HalwagiA2 - Hassan, E. Alfadala, M.E.-H. Mahmoud (Eds.), Proceedings of the 1st Annual Gas Processing Symposium, Elsevier, Amsterdam, 2009, pp. 247–255. ¨ [13] L. Deng, M.-B. Hagg, Swelling behavior and gas permeation performance of PVAm/PVA blend FSC membrane, J. Membr. Sci. 363 (2010) 295–301. [14] A. Ito, G. Sui, N. Yamanouchi, VOC vapor permeation through a liquid membrane using triethylene glycols, Desalination 234 (2008) 270–277. [15] G. Sui, J. Li, A. Ito, Separation of VOC vapor from air by a surface-soaked liquid membrane module using triethylene glycol, Sep. Purif. Technol. 68 (2009) 283–287. [16] Y. Shimoyama, A. Ito, Calculation of permeabilities and numerical simulation of separations for volatile organic compound vapor through triethylene glycol derivative liquid membranes, J. Membr. Sci. 369 (2011) 448–454. [17] X. Feng, R.Y.M. Huang, Pervaporation with chitosan membranes. I. Separation of water from ethylene glycol by a chitosan/polysulfone composite membrane, J. Membr. Sci. 116 (1996) 67–76. [18] F.R. Chen, H.F. Chen, Pervaporation separation of ethylene glycol–water mixtures using crosslinked PVA–PES composite membranes. Part I. Effects of membrane preparation conditions on pervaporation performances, J. Membr. Sci. 109 (1996) 247–256. [19] F.R. Chen, H.F. Chen, A diffusion model of the pervaporation separation of ethylene glycol–water mixtures through crosslinked poly(vinyl alcohol) membrane, J. Membr. Sci. 139 (1998) 201–209. [20] S. Yong Nam, Y. Moo Lee, Pervaporation of ethylene glycol–water mixtures: I. Pervaporation performance of surface crosslinked chitosan membranes, J. Membr. Sci. 153 (1999) 155–162. [21] R. Guo, X. Ma, C. Hu, Z. Jiang, Novel P.V.A.–silica nanocomposite membrane for pervaporative dehydration of ethylene glycol aqueous solution, Polymer 48 (2007) 2939–2945. [22] Y. Wang, M. Gruender, T.S. Chung, Pervaporation dehydration of ethylene glycol through polybenzimidazole (PBI)-based membranes. 1. Membrane fabrication, J. Membr. Sci. 363 (2010) 149–159. [23] D. Nguyen Thanh, K. Chan, T. Matsuura, S. Sourirajan, Effect of shrinkage on pore size and pore size distribution of different cellulosic reverse osmosis membranes, Ind. Eng. Chem. Prod. Res. Dev. 23 (1984) 501–508. [24] W.M. King, D.L. Hoernschemeyer, C.W. Saltonstall, Cellulose acetate blend membranes, in: H.K. Lonsdale, H.E. Podall (Eds.), Reverse Osmosis Membrane Research, Plenum Press, New York, 1972, pp. 131–162. [25] A.G. Wonders, D.R. Paul, Effect of CO2 exposure history on sorption and transport in polycarbonate, J. Membr. Sci. 5 (1979) 63–75. [26] E.W. Funk, N.N. Li, Purification of natural gas by membranes, in: E.F. Vansant, R. Dewolfs (Eds.), Process Technology Proceedings, Vol. 8, Gas Separation Technology, International Symposium on Gas Separation Technology, Elsevier Science Publishers B. V. Amsterdam, Antewerp, Belgium, 1989, pp. 355–372. [27] S.M. Jordan, M.A. Henson, W.J. Koros, The effects of carbon dioxide conditioning on the permeation behavior of hollow fiber asymmetric membranes, J. Membr. Sci. 54 (1990) 103–118. [28] S.M. Jordan, W.J. Koros, Characterization of CO2-induced conditioning of substituted polycarbonates using various exchange penetrants, J. Membr. Sci. 51 (1990) 233–247. [29] S.M. Jordan, W.J. Koros, J.K. Beasley, Characterization of CO2-induced conditioning of polycarbonate films using penetrants with different solubilities, J. Membr. Sci. 43 (1989) 103–120. [30] G.K. Fleming, W.J. Koros, Carbon dioxide conditioning effects on sorption and volume dilation behavior for bisphenol A-polycarbonate, Macromolecules 23 (1990) 1353–1360.

[31] L.S. White, T.A. Blinka, H.A. Kloczewski, I.f. Wang, Properties of a polyimide gas separation membrane in natural gas streams, J. Membr. Sci. 103 (1995) 73–82. [32] R. Quinn, D.V. Laciak, Polyelectrolyte membranes for acid gas separations, J. Membr. Sci. 131 (1997) 49–60. [33] J.D. Wind, D.R. Paul, W.J. Koros, Natural gas permeation in polyimide membranes, J. Membr. Sci. 228 (2004) 227–236. [34] S. Li, G. Alvarado, R.D. Noble, J.L. Falconer, Effects of impurities on CO2/CH4 separations through SAPO-34 membranes, J. Membr. Sci. 251 (2005) 59–66. [35] M. Al-Juaied, W.J. Koros, Performance of natural gas membranes in the presence of heavy hydrocarbons, J. Membr. Sci. 274 (2006) 227–243. [36] J. Hao, P.A. Rice, S.A. Stern, Upgrading low-quality natural gas with H2S- and CO2-selective polymer membranes: Part II. Process design, economics, and sensitivity study of membrane stages with recycle streams, J. Membr. Sci. 320 (2008) 108–122. [37] J.S. Lee, R.T. Adams, W. Madden, W.J. Koros, Toluene and n-heptane sorption in Matrimids asymmetric hollow fiber membranes, Polymer 50 (2009) 6049–6056. [38] J.S. Lee, W. Madden, W.J. Koros, Antiplasticization and plasticization of Matrimids asymmetric hollow fiber membranes—Part A. Experimental, J. Membr. Sci. 350 (2010) 232–241. [39] J.K. Ward, W.J. Koros, Crosslinkable mixed matrix membranes with surface modified molecular sieves for natural gas purification: II. Performance characterization under contaminated feed conditions, J. Membr. Sci. 377 (2011) 82–88. [40] C.A. Scholes, S.E. Kentish, G.W. Stevens, The effects of minor components on the gas separation performance of polymeric membranes for carbon capture, in: Y. Yampolskii, B. Freeman (Eds.), Membrane Gas Separation, John Wiley & Sons, Ltd, Chichester, West Sussex, 2010, pp. 201–226. [41] C.J. Anderson, W. Tao, C.A. Scholes, G.W. Stevens, S.E. Kentish, The performance of carbon membranes in the presence of condensable and noncondensable impurities, J. Membr. Sci. 378 (2011) 117–127. [42] C.A. Scholes, G.Q. Chen, W.X. Tao, J. Bacus, C. Anderson, G.W. Stevens, S.E. Kentish, The effects of minor components on the gas separation performance of membranes for carbon capture, Energy Procedia 4 (2011) 681–687. [43] C.A. Scholes, S.E. Kentish, G.W. Stevens, The effect of condensable minor components on the gas separation performance of polymeric membranes for carbon dioxide capture, Energy Procedia 1 (2009) 311–317. [44] C.A. Scholes, G.W. Stevens, S.E. Kentish, The effect of hydrogen sulfide, carbon monoxide and water on the performance of a PDMS membrane in carbon dioxide/nitrogen separation, J. Membr. Sci. 350 (2010) 189–199. [45] Dow, in: Dow (Ed.), Triethylene Glycol, Dow Chemical Company, 2007, pp. 1–30. [46] T. Sata, K. Mine, K. Matsusaki, Change in transport properties of anionexchange membranes in the presence of ethylene glycols in electrodialysis, J. Colloid Interface Sci. 202 (1998) 348–358. ¨ [47] L. Deng, T.-J. Kim, M.-B. Hagg, Facilitated transport of CO2 in novel PVAm/PVA blend membrane, J. Membr. Sci. 340 (2009) 154–163. ¨ [48] T.-J. Kim, M.W. Uddin, M. Sandru, M.-B. Hagg, The effect of contaminants on the composite membranes for CO2 separation and challenges in up-scaling of the membranes, Energy Procedia 4 (2011) 737–744. ¨ [49] M. Sandru, S.H. Haukebø, M.-B. Hagg, Composite hollow fiber membranes for CO2 capture, J. Membr. Sci. 346 (2010) 172–186. [50] Safety Data Sheet—Ethylene Glycol—33068, in, Sigma-Aldrich, 2012, pp. 1–7. [51] A. Rojey, C. Jaffret, S. Cornot-Gandolphe, B. Durand, S. Jullian, M. Valais, Natural Gas—Production, Processing, Transport, Editions Technip-Paris, Paris, 1997. [52] D.L. Christensen, Gas dehydration—thermodynamic simulation of the water/ glycol mixture, Master Thesis, February 2009, K10, in: Section of Chemical Engineering in Department of Biotechnology, Chemistry and Environmental Engineering, Aalborg University Esbjerg, Esbjerg, 2009, pp. 1–77. [53] Y. Maeda, D.R. Paul, Effect of antiplasticization on selectivity and productivity of gas separation membranes, J. Membr. Sci. 30 (1987) 1–9. [54] Y. Maeda, D.R. Paul, Effect of antiplasticization on gas sorption and transport. II. Poly(phenylene oxide), J. Polym. Sci., Part B: Polym. Phys. 25 (1987) 981–1003. ˜ o, D.R. Paul, Gas permselectivity properties of high free [55] F.A. Ruiz-Trevin volume polymers modified by a low molecular weight additive, J. Appl. Polym. Sci. 68 (1998) 403–415. [56] N.M. Larocca, L.A. Pessan, Effect of antiplasticisation on the volumetric, gas sorption and transport properties of polyetherimide, J. Membr. Sci. 218 (2003) 69–92. [57] I.C. Omole, D.A. Bhandari, S.J. Miller, W.J. Koros, Toluene impurity effects on CO2 separation using a hollow fiber membrane for natural gas, J. Membr. Sci. 369 (2011) 490–498. [58] S.P. Nunes, K.-V. Peinemann, Part I, Membrane materials and membrane preparation, in: S.P. Nunes, K.-V. Peinemann (Eds.), Membrane Technology in the Chemical Industry, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 2006, pp. 3–75.