Effect of the reactive amino and glycidyl ether terminated polyethylene oxide additives on the gas transport properties of Pebax® bulk and thin film composite membranes

This paper considers Pebax® MH 1657 as a material for the CO2/N2 separating layer in thin film composite (TFC) membranes. The CO2 permeability of Pebax® can be improved via blending with various poly(ethylene oxide) (PEO) based materials without loss of CO2/N2 selectivity. Analogous blends containing PEOs with reactive end groups have been investigated for the possibility of a network formation within the Pebax® matrix. The formation of network is possible through the reaction between two types of additives containing two reactive end groups. The thick film samples and TFC membranes were prepared from mixtures of Pebax® MH 1657, PEG DG526 and JEFFAMINE® with different molecular weights. The samples were characterized by single gas permeation measurements, DSC, and NMR. The samples with incorporated networks show improved and stable gas transport properties compared to the original polymer for both thick films and TFC membranes.


Introduction
Carbon dioxide (CO 2 ) is considered as a major anthropogenic greenhouse gas affecting global climate change. In 2014 the global carbon dioxide emission stood at 32.3 billion tons, the same amount as in 2013. 1 Coal-red power plants are responsible for 28% of this enormous value. 2 Among the activities to minimize the harmful effect of greenhouse gas emission on the climate the approach of CO 2 capture and storage (CCS) during energy production is one of the most crucial since such types of emission are very localized and constant.
For the implementation in existing fossil fueled power plants the post-combustion approach of CCS based on either absorption or separation with polymeric membranes is among the most analyzed. The use of separation systems based on polymeric membranes with high CO 2 /N 2 selectivity and high CO 2 permeance as well as implementation of developed systems into the post-combustion separation process can lead to signicant CO 2 emission reduction but, unfortunately will reduce the efficiency of the energy production process. 3,4 The current trend in polymeric membrane development is the use of materials able to transport CO 2 either by active transport or by simple solution-diffusion mechanisms. Active transport membranes are currently prepared in small pilot scale and cannot be considered as materials of immediate or near future availability. 5 Polymeric membranes based on poly(ethylene oxide) (PEO) have come to larger pilot scale production and are tested at several industrial sites for performance and stability in the environment of natural gas or coal red power plants ue gas streams. 6,7 The permeance and selectivity of developed PEO membranes lead to effective extraction of CO 2 from the ue gas while the modelling of separation process economy shows that membrane separation systems are competitive to other considered methods of CCS. 8 The selectivity of the developed membranes has reached the limit dened by the ideal selectivity of PEO based materials used for formation of the selective layer. At the same time the permeance can be varied by either reduction of the selective layer thickness or by development of new materials with higher CO 2 permeability coefficient. The selective layer thickness of currently available thin lm composite membranes has reached approximately 70 nm and hardly can be signicantly reduced, meaning that new materials with high CO 2 permeability coef-cients are needed.
It was shown that blending of PEO containing block copolymers with low molecular weight PEO can signicantly increase CO 2 permeability coefficient without loss of the CO 2 /N 2 selectivity. 9 For example experiments with physical blending of low molecular weight poly(ethylene glycol)s (PEG) with PEO based Pebax® MH 1657 showed that permeability of the blend material at room temperature can be increased by 4-5 times. 10,11 The increase in permeability at the same loading level depends on the molecular weight and end group nature of the additive to the polymer. The largest increase in permeability was observed for the materials containing PEG additive with the lowest molecular weight or shortest PEO chain length. The inuence of PEO chain length on the CO 2 solubility was studied in detail. 12 However, stability tests of these blend materials at 50 C, which is considered to be the operating conditions of CCS postcombustion process, showed signicant leaching out of the PEO additive from the polymer matrix resulting in decay of gas transport properties. It was shown 13 that the best results in terms of material's gas transport properties and stability can be achieved when additives having molecular weight more than 500 g mol À1 are used.
Alternatively compositions containing PEO blocks have been used in the studies of organic/inorganic hybrid membranes prepared from g-glycidyloxypropyltrimethoxysilane (GPTMS) and poly(ether) diamines (PAPE). In this case the materials with the longest (2000 g mol À1 ) PEO chain have shown the highest CO 2 permeability. The CO 2 /N 2 selectivity was increased for the membranes containing amine functionalized compounds. 14,15 Current work was targeted at development of materials for CO 2 separation membranes with increased permeability and optimized selectivity. JEFFAMINE® poly(ether) diamines, as well as glycidyl ether terminated polyethylene oxides were considered as suitable materials for Pebax® MH1657 modication. JEFFAMINE® poly(ether) diamines in combination with PEO containing reactive end groups such as diglycidyl ether could lead to the formation of a polymer within the Pebax® matrix which can be considered as a network similar to certain extent to the concept of the interpenetrating networks. 16 In this work the incorporation of the network into the matrix of Pebax® MH 1657 is presented. The membrane materials and thin lm composite membranes were characterized using single gas permeation measurements, thermal and spectroscopy methods.

Materials
The Pebax® MH 1657 (further Pebax®) was received in a pellet form from Arkema. PEG diglycidyl ether DG526 (M w : 526 g mol À1 ), JEFFAMINE® ED series (O,O 0 -bis(2-aminopropyl) polypropylene glycol-block-polyethylene glycol-block-polypropylene glycol) ED600 and ED2003 having molecular weights of 600 and 2000 g mol À1 respectively were obtained from Sigma-Aldrich. Ethanol (Th. Geyer, 99.9%) and high purity deionized water (0.06 mS cm À1 ) were used as solvents. A porous PAN membrane developed at HZG 17 with average pore size 20 nm and 15% surface porosity coated with a 150 nm layer of PDMS was used for formation of thin lm composite membranes.

Membrane preparation
Films having thickness of approximately 100 mm were prepared according to the procedure reported elsewhere 13 by casting from a 3 wt% Pebax® solution in a solvent mixture of ethanol and water (EtOH/H 2 O) with a ratio of 70/30 (wt%). For the preparation of binary polymer compositions, different amounts of JEFFAMINEs® and PEG DG526 were added to a 3 wt% Pebax® solution and stirred at room temperature for 20 min. The prepared solution was poured on a Teon® coated glass plate and the ethanol/water solvent was evaporated at 30 C within 36-48 h.
Samples with incorporated network structures formed by reaction of amino-and glycidyl-functionalized PEGs were prepared as follows: equimolar mixtures of JEFFAMINE® and PEG DG526 were stirred for 20 min at 40 C in EtOH/H 2 O mixtures used as solvent. A predetermined amount of the prepared JEFFAMINE®/PEG DG526 solution was added to the Pebax® solution and stirred at room temperature for 20 min. Thick lms were cast from the resulting Pebax®/ JEFFAMINE/PEG solution in the same way as described above. The mechanism of amine-glycidyl ether reaction at an initiation temperature 50 C is known since decades and thus no reaction has happened during the lm preparation. 18 In the current study the reaction in the prepared polymer lms was initialized by membrane sample conditioning at 80 C for 48 h in a vacuum oven equipped with a turbomolecular pumping unit. It ensured complete degassing of the sample from traces of solvents or other volatile compounds.
The thin lm composite membrane (TFCM) has been prepared from a 0.5 wt% polymer solution in EtOH/H 2 O. Binary blends of Pebax® with JEFFAMINE® ED600, ED2003 and PEG DG526 were used for selective layer formation along with threecomponent blends of Pebax® and an equimolar mixture of ED600/DG526. TFCMs were further treated for 48 h in the vacuum oven at 80 C in order to achieve the amine-glycidyl reaction. As supporting layer for the preparation of TFCMs polyacrylonitrile (PAN) microporous membrane prepared on non-woven polyester, coated with a poly(dimethylsiloxane) (PDMS) layer, 19 was used. PDMS served as a "gutter layer" distributing gases penetrating through the selective layer to the pores of the support and protecting the porous support from impregnation with components of the selective layer during the TFCM preparation. 20,21 Additionally, a thin layer of ethyl cellulose was deposited on the surface of the PDMS layer to provide better compatibility between PDMS and Pebax®. The permeance of the ethyl cellulose/PDMS/PAN support was signicantly higher than the permeance of the nal membrane and did not inuence signicantly the selectivity of the resulting CO 2 selective TFCM.

Membrane characterization
For the characterization of the produced thick lms, spectroscopy and thermal analysis experiments were accomplished. 1 H-NMR and 1 H-HR-MAS NMR experiments were performed on the thick lms. Thermal tests on the produced lm samples as well as on the initial materials were done by differential scanning calorimetry (DSC). The properties obtained in these experiments were correlated to the gas transport measurements at different temperatures, and results were compared to the weight change of lm samples studied for gas transport properties obtained from the weighting of lms on an analytical balance before and aer gas transport measurements at various temperatures. The lms thickness was measured with a digital micrometer DELTASCOPE® FMP10.
For the NMR experiments Bruker AVIIIHD 500 MHz spectrometer (Bruker, Karlsruhe, Germany) equipped with BBFO and HR-MAS probes was used. The solvent for the membrane swelling measurements was deuterated chloroform with TMS as internal standard (CDCl 3 , Sigma-Aldrich, Schnelldorf, Germany). The experiments were performed at room temperature. HR-MAS experiments were made using a 4 mm Bruker HR-MAS probe and -rotors at rotation frequencies of 10 kHz. The relaxation delays were chosen in that way that the sample was fully relaxed.
DSC experiments were done with the use of the calorimeter DSC 1, from the company Mettler Toledo (Mettler-Toledo, Greifensee, Switzerland), with temperature measurement range from À60 C to +280 C. The heating rate was 10 K min À1 . The calculations presented in this work are based on the rst cooling and the second heating intervals. The measurement was performed in argon atmosphere. Approximately 10 mg of the polymer were placed in an aluminum pan of 10 ml.
The instrumentation used for the thermogravimetric analysis experiments was the TG 209 F1 Iris (Netzsch, Selb, Germany). The experiment was done in a temperature range from 25 C to 500 C, and heating rate of 10 K min À1 . The measurement was performed in a nitrogen gas ow.

Gas transport properties
The permeability coefficients of pure carbon dioxide and nitrogen were measured by a constant volume variable pressure technique 22 (time-lag) in the temperature range from 30 C to 90 C and back from 90 C to 30 C. The feed pressure was 600, 450 and 350 Torr, in consecutive measurements for all gases. The permeability coefficient P [cm 3 (STP) cm cm À2 s À1 cmHg À1 ] of single gas was determined as: where V P is the constant permeate volume (cm 3 ), l is the lm thickness (cm), A is the effective area of membrane (cm 2 ), R is the gas constant (8.314 J mol À1 K À1 ), Dt is the time for permeate pressure increase from p P1 to p P2 (s), p f is the feed pressure (cmHg). 23 The use of pressures instead of fugacities is justied since the absolute pressures on the feed side were sufficiently low. The diffusion coefficient D (cm 2 s À1 ) was calculated from membrane thickness l (cm) and time-lag q (s) determined graphically as intersection of the line drawn through the linear region of the pressure increase curve to intersection with the time axis: The solubility coefficient S (cm 3 (STP) cm À3 cmHg À1 ) was calculated according to the following equation: The ideal permeability selectivity of the material can be expressed as the ratio of permeability coefficients of two penetrants and according to eqn (3) it is a function of diffusion and solubility selectivity, leading to the equation: The single gas permeances of prepared TFCMs were determined using the "pressure increase" facility designed and built at HZG. The pressure increase facility 19 utilizes the same principle as the aforementioned "time-lag" facility with difference in the data acquisition routine adjusted for investigation of membranes with thin selective layers. In case of the "pressure increase" the aim is the determination of the pressure increase rate in the calibrated permeate volume without considering any effects related to the thickness of the membrane. The basic principle of "pressure increase" measurements was described elsewhere. 24 Single gas permeation data were determined at 500 mbar feed pressure and in the temperature range from 30 C to 80 C. In order to control the property stability for each defectfree TFCM sample, the measurements were repeated in the temperature range from 80 C to 30 C.
The permeance L (m 3 (STP) m À2 h À1 bar À1 ) of the membrane can be calculated using the equation: where V P is the constant permeate volume (m 3 ), R is the gas constant (8.314 J mol À1 K À1 ), A is the effective area of membrane (m 2 ), t (h) is the time of measurement and p f , p 0 , and p P(t) (bar) are pressures at the feed, permeate side at the beginning, and permeate side at the end of the measurement, respectively.

Results and discussion
Preliminary results obtained for the system Pebax® MH1657/ LMW PEG have already indicated that the gas transport properties of lms containing PEGs with different molecular weights and concentration up to 40 wt% in the polymer matrix can be improved by a factor of 3 to 5. The best results were obtained for Pebax® blended with PEG having molecular weight less than 300 g mol À1 , but at temperatures interesting for application (50-80 C) such low molecular weight compound tends to leave the polymer matrix (leaching-out). This can be explained by the fact that heat excites the mobility of PEO chains and provokes the separation of ller and polymer. On the other hand, membranes containing PEOs with higher molecular weight up to 1000 g mol À1 exhibit increased crystallinity which leads to material brittleness and damage of the thick lm sample during the cooling part of gas permeation experiment. The Pebax® MH 1657 used in this study contains 60% of exible PEO blocks promoting the high CO 2 selectivities over other light gases. Commercial availability of this material makes it attractive for use as a selective material of TFCMs for CO 2 /N 2 separation. In the present study JEFFAMINEs® of ED series and their reaction product with epoxy groups containing PEGs were used to increase the permeability of the CO 2 selective material.

Binary systems of Pebax® and PEGs
Samples of blended Pebax®MH1657 having thicknesses of about 100 mm were prepared as described above and tested for the gas transport properties. The temperature conditions of measurements in the range 30 C to 90 C were chosen according to previous work. 13 Addition of 10 wt% PEG DG526 increases the permeability of the material by 15% whilst almost no change in the CO 2 /N 2 selectivity was observed. Increasing of the concentration of PEG DG526 up to 40 wt% resulted in 1.5 times permeability increase whilst the CO 2 /N 2 selectivity was slightly decreased (Fig. 1). The increase of permeability in polymers blended with PEG depends on their molecular weight, end groups and concentration in Pebax® MH 1657. For this work an optimal concentration of 20 wt% ller in the total membrane weight in order to maintain the integrity of the thick lms was selected. Thus, for each reagent concentration is only 10 wt%, which explains the increase of the CO 2 permeability only 1.5 times. The deviations in CO 2 permeability coefficients for pure Pebax® MH 1657 and blends containing 10 wt% and wt 40% of LMW PEG DG526 were 0.7% and 6%, respectively, for the results obtained at the same temperatures during the heating and cooling process.
The CO 2 permeability for samples containing 20 wt% polyether amines ED600 and ED2003 is depicted in Fig. 1. The permeability increment for samples with lower molecular weight JEFFAMINE® (ED600) is greater than that of the higher molecular weight samples (ED2003) at temperatures above 60 C. The selectivity of the sample with ED600 decreases signicantly compared to the pure Pebax®, which may indicate a surprising incompatibility of the low molecular weight PEG terminated with amino groups. An interesting effect is observed at temperatures above 60 C, where the ED600 blend shows the same selectivity as other studied materials. This can be considered as complete dissolution of the ED600 in the PEO blocks of Pebax®. Further attempts to increase the concentration of JEFFAMINE® ED600 up to 30 wt% lead to sample's mechanical instability due to strong phase segregation. ED2003 in the PEBAX® matrix shows completely different behaviour. The relatively long PEG chain masks the effect of the presence of amine groups and at concentration 20 wt% the blend shows good CO 2 permeability and CO 2 /N 2 selectivity. The increase of ED2003 concentration to 40 wt% still gives one the possibility to handle the prepared thick lm indicating good mechanical properties of this blend material. The gas transport properties show that ED2003 starts to behave differently from the surrounding Pebax® matrix: the CO 2 permeability temperature dependence has a signicant step which starts at 30 C and ends at 50 C. At a temperature below 30 C, the blend shows very low CO 2 /N 2 selectivity which can be explained by the presence of a crystalline ED2003 phase surrounded by highly disordered and loosely packed PEG chains. In the temperature range 30 C to 50 C, the ED2003 crystals start to melt and the molten phase becomes available for CO 2 transport. At temperatures above 50 C the blend of ED2003 and Pebax® shows gas CO 2 /N 2 selectivity usual for materials with high PEG content.

Three-component system
A three-component system was prepared by incorporating in Pebax® matrix two additional components that are able to react with each other even in solid (solvent-free) state, resulting in a stable membrane. This approach is different from formation of interpenetrating network within the polymer matrix. Bifunctional PEGs in equimolar mixture can give only linear product since the reactivity decreases signicantly from primary to secondary amines. At the same time it was found earlier that incorporation of linear PEGs having molecular weight higher than 1000 does not give the desired effect of gas permeability increase. Additionally, PEGs with the molecular weight above 1500 g mol À1 have strong tendency for crystallization at room temperature. Formation of a linear polymer as a product of a reaction between amino-and glycidyl-end groups of low molecular weight PEGs can lead to interesting results in terms of gas transport properties because when such a compound is embedded into the Pebax® matrix it will depict a lower probability for crystallization, since PEG blocks will be interrupted by blocks formed from the amine-glycidyl reaction. This effect can increase membrane's permeability and reduce the probability for leaching-out of the low molecular weight additive for the Pebax® matrix. Films of Pebax® blended with PEG DG526 and JEFFAMINE® ED600 having equal amount of PEG monomeric units in the chain (n ¼ 9) were prepared following the same procedure as described above. In Fig. 2 the CO 2 permeability coefficients for three-component system Pebax®/ED600/DG 526, where the total content of the latter two components was 20 wt% and 30 wt%, is presented. The effect in permeability increase was smaller than anticipated, the CO 2 /N 2 selectivity decreased sharply. In Fig. 2, curves for the heating interval of "time-lag" measurements are presented since material instability leading to sample damage was observed during the cooling down from 90 C to 30 C. The increase in permeability was observed for the samples Pebax®/ED600-DG 526 20 wt%, treated at 80 C for 48 hours; the values of the CO 2 /N 2 selectivity were decreased compared to the neat Pebax®. The deviation of the observed gas transport properties during the heating up and cooling down part of the "time-lag" experiment was less than 1.5% for this sample, indicating the reaction of the amine with the glycidyl ether and formation of network within PEO blends. When the concentration of additives in the Pebax® matrix of the thermally treated samples was raised to 30%, the CO 2 permeability has decreased compared to thermally untreated material; but the CO 2 /N 2 selectivity maintained at values of the neat Pebax® MH 1657. The degree of amine glycidyl ether consumption during the reaction in solid polymer matrix was controlled by a NMR study.
To verify the completion of the amine-glycidyl reaction during the thermal treatment of solid lms the samples of the three component system were investigated with High Resolution Magic Angle Spinning ( 1 H HR-MAS NMR). The 1 H HR-MAS NMR spectra are dominated by the presence of Pebax® 1567, so only slight variations can be seen. At 3.7 ppm all the (-CH 2 -CH 2 -O-) signals from JEFFAMINE® ED600, Pebax 1567 and PEG DG526 overlap (Fig. 3), but the signals of the -CH 3 group of the JEFFAMINE® are visible below 1.25 ppm. The signals of these -CH 3 groups experiencecompared to the 1 H NMR measurements in liquids of a binary system PEG DG526/ JEFFAMINE® ED600a small shi due to the Pebax® environment, but nevertheless they still can be identied. Fig. 3 clearly shows, that one of the signals of the -CH 3 groups (b, 1.1 ppm) loses intensity and a new signal arises (c, 1.05 ppm). Because of that loss it can be concluded that the JEFFAMINE® and the PEG glycidyl ether have reacted leading to formation of a stable blend system even when not all of the reactive compounds were consumed.
Processes observed by changes in CO 2 permeability can be conrmed by DSC analysis. The DSC experiments were carried out for all initial components as well as binary and ternary component systems. The results are summarized in Table 1. For the blend systems, the glass transition of the PEG block is observed at approximately À50 C, which is comparable with the values of the neat Pebax®. Slightly stronger effect is observed for the crystallization and melting temperatures of the PEG block. The temperatures T PEG c and T PEG m vary from À10 C to 10 C and from 15 C to 40 C, respectively. The crystallization temperature attributed to the polyamide block seems to be stable with the only exception in the case of the Pebax®/ED600 (20%) blend, where a shi of approximately 20 C is observed. This could be attributed to the fact that the low molecular weight PEG contains a relatively large amount of -NH 2 groups that could form hydrogen bonds with the respective polyamide block. The melting temperature of the polyamide block remains at similar values to that of neat Pebax®, at approximately 205 C for binary systems.
The effect of changes in gas transport properties caused by introduction of additional PEG with different molecular weight  : glass transition temperature of the polyethylene glycol block, C, T PEG c : crystallization temperature of the polyethylene glycol block, C, T PEG m : melting temperature of the polyethylene glycol block, C, T PA c : crystallization temperature of the polyamide block, C, T PA m : melting temperature of the polyamide block, C, n.a.: not applicable, -: no conclusive data, nw: thermally treated at 80 C. and end groups into the Pebax® matrix can be clearly seen in Fig. 4, where the solubility and permeability coefficients of CO 2 as well as the CO 2 /N 2 selectivity are plotted against the PEO content in the system. Amino terminated PEO ED600 exhibits the highest CO 2 solubility coefficient among the other studied materials but has not the leading permeability coefficient due to incompatibility of this Jeffamine® to the polymer matrix and thus presence of additional resistance for CO 2 diffusion between phases. From the low CO 2 /N 2 selectivity of the Pebax®/ ED600 system one can conclude that CO 2 can be even partially immobilized on amino groups while all N 2 molecules have the freedom to pass through the polymer system. As soon as amino groups are consumed in e.g. reaction with glycidyl groups the selectivity of the system becomes similar to the one of neat Pebax®.
The behaviour of the system containing diglycidyl ether DG526 is even more interesting. This system shows the highest permeability at 30 C amongst the other studied materials, the CO 2 /N 2 selectivity does not differ from the neat Pebax® at all DG526 loadings, but the solubility coefficient is the lowest of all other materials meaning that DG526 signicantly and nonselectively increases diffusion coefficients of both carbon dioxide and nitrogen.
Jeffamine® ED2003 when distributed in Pebax® gradually increases the CO 2 solubility coefficient but at temperature below melting temperature causes gas diffusion path distortion as well as changes in material packing which results in lower CO 2 permeability and CO 2 /N 2 selectivity at high loadings. As soon as the experimental temperature becomes higher than the melting point, the permeability coefficient dependence becomes more gradual and the selectivity reaches the level of neat Pebax®.
Gas transport characteristics of a ternary system Pebax®/ ED600/DG526 before and aer thermal treatment follow those of the Pebax®/ED2003 at low loadings. The solubility coefficient increases gradually with a rise in PEO content, the permeability of the thermally not treated system does not change, as well as the selectivity. The decrease of the CO 2 permeability and CO 2 /N 2 selectivity at low loading in case of the thermally treated system we attribute to too low concentrations of reactive groups dispersed in the polymer matrix. In this case the reaction in solid Pebax® is hindered, and amino containing compounds can form a separate phase as discussed above. When the concentration of both additives to the Pebax® matrix increases, the probability of reaction between amino-and glycidyl-groups increases and the resulting system exhibits properties usual for PEO containing materials. It is interesting to observe, that at 50 C all studied systems except amino-groups containing have very similar selectivities.
As was discussed above, the presence of the crystalline phase gives strong effects on both permeability and selectivity. Table 2 illustrates that PEO crystallinity of the studied systems obtained from the analysis of DSC results varies signicantly. For crystallinity calculations literature value of the PEO melting enthalpy of 8.67 kJ mol À1 have been used. 12,[25][26][27][28][29][30] The lowest level of crystallinity was observed for neat Pebax® and for the ternary system. All binary systems demonstrate 5-9% higher crystallinity compared to the neat Pebax®. Since crystallinity values are obtained from the DSC curves and one can consider (to certain extent) this crystallinity as crystallinity at melting temperature for systems with low molecular weight PEO, higher values do not inuence the gas transport at temperatures well above the melting point. In case of the system containing ED2003 the crystallinity at 30 C presents a signicant problem for both CO 2 permeability coefficient and CO 2 /N 2 selectivity but as soon as the membrane is heated above the melting temperature, the CO 2 permeability surpasses those of other studied systems while the selectivity reaches values characteristic for other studied materials.

Thin lm composite membranes
A practical application of rubbery polymeric materials for membrane gas separation is possible in form of thin lm composite (TFC) membranes. This type of membranes has a long history of successful use in numerous applications. 31,32 Currently at least two membranes with selective layers made of materials having high PEG content are undergoing rigorous pilot tests for the separation of CO 2 from the off gases of natural gas (Polaris, MTR) and coal red (PolyActive™, HZG) power plants. 7,33 Development of new highly permeable and, if possible, more selective materials as well as their implementation in TFC membranes can facilitate a big advance in membrane separation systems for CO 2 capture.
The most promising materials investigated in the current study in a form of thick isotropic lms have been used for the formation of the selective layer of the TFC membrane. The membrane used as a support consisted of non-woven providing mechanical strength, porous PAN support and PDMS layer serving as a smooth gutter layer. The PDMS layer provides smooth adhesive surface for application of thin selective layer and additionally serves for distributing the gas stream penetrating through the selective layer to the nears pore of the porous support (gutter layer approach). 34,35 In order to improve compatibility of the gutter layer and solution used for deposition of the selective layer one more layer of ethyl cellulose was deposited onto the PDMS membrane. The CO 2 permeance of the prepared ethyl cellulose TFC membrane was 4.1 Nm 3 m À2 h À1 bar À1 and CO 2 /N 2 selectivity 16.8 at 30 C. This permeance is three times higher than the CO 2 permeance of the pure Pebax® TFC membrane (1.37 Nm 3 m À2 h À1 bar À1 at 30 C) and thus layers supporting Pebax® based selective layer did not inuence the total membrane selectivity which could occur when inappropriate support is used for TFC membrane formation. 36,37 The permeance of the TFC membrane having pure Pebax® selective layer is not competitive compared to other available CO 2 selective membranes and should be improved for the use in industrial processes. 38 As seen on the  Fig. 5, the permeance divergence was observed for the threecomponent system (not thermally treated before the gas permeation experiment) Pebax®/ED600/DG 526 20% during the heating and cooling parts of measurements in the temperature range from 20 C to 90 C. Exposure to the elevated temperatures facilitates leaching out of the low molecular weight components from the selective layer and leads to reduced membrane permeance observed during the cooling from 90 to 20 C part of the gas transport experiment. The same membrane but thermally treated in vacuum for 10 hours at 80 C showed stable gas transport properties in the investigated temperature range in case of both heating and cooling parts of the experiment. It is important to mark that the membrane heated to 80 C relatively fast, with the heating rate provided by the vacuum oven shows higher permeance compared to the membrane which was heated up stepwise during the gas transport experiment. This observation is especially valuable since it is possible to provide intensive thermal treatment of the membrane aer the solvent drying procedure in the membrane production facility and by this intensive IR treatment to foster fast reaction between amino-and glycidyl-end groups of low molecular weight PEOs dispersed in the Pebax® matrix. The CO 2 /N 2 selectivity of both ternary system based TFC membranes was identical to the selectivity of the neat Pebax® and reects the quality of the membrane manufacturing process. The CO 2 permeance of the thermally treated membrane sample was above 2.2 Nm 3 m À2 h À1 bar À1 at 30 C and indicates a good potential of the developed material for production of competitive TFC membranes.

Conclusions
Two and three-component systems based on the commercially available polymer Pebax® MH 1657, JEFFAMINE® diamine and PEG diglycidyl ether were investigated for their gas transport properties. The JEFFAMINE® and the PEG diglycidyl ether reacted in the polymer matrix and formed a network stably distributed in the Pebax® matrix. Thermally treated samples present stable gas transport properties in comparison with binary blends of PEG or JEFFAMINE® with Pebax®. TFCM with incorporated networks have demonstrated improved CO 2 permeance and unchanged CO 2 /N 2 selectivity compared to the TFC membrane with pure Pebax® selective layer. The thick lm of a thermally treated ternary system has shown stable gas transport properties aer multiple heating/cooling cycles. The same properties were achieved for the thin lm composite membranes indicating that proposed ternary system can be used for large scale CO 2 selective membrane production. The developed membranes are to be tested at operating conditions temperatures for the removal of CO 2 from the ue gas of coal red power plants.