Improving CO₂/N₂ separation performance using nonionic surfactant Tween containing polymeric gel membranes

Abstract

Polymeric gel membranes containing 0 to 65 wt% of Tween20, Tween21 and Tween80 in PEBA2533 were prepared using a solvent casting method. The effect of Tween with different lengths of alkyl chain on the properties and performance of the gel membranes was discussed. The results showed a better compatibility of PEBA2533 with Tween21 than Tween20 and Tween80. Both the pure CO₂ permeability and CO₂/N₂ selectivity increased with the increase in the number of N-alkyl groups in Tween. On the other hand, with the increase of Tween content, CO₂ permeability and CO₂/N₂ selectivity simultaneously increased. The best separation performance was achieved for the PEBA2533/Tween80-65 gel membrane, which had CO₂ permeability of 289 Barrer and CO₂/N₂ selectivity of 40.70, at 0.6 atm and 25 °C. This suggests a potential application in gas separation membranes, for instance for CO₂ capture.

---

1. Introduction

Carbon dioxide capture from power plant flue gas and subsequent sequestration is expected to play a key role in mitigating global climate change. Membrane separation is one of the potential methods to remove CO₂ from flue gas.^1–3 The membrane separation process can offer many advantages, such as low energy consumption, mechanical simplicity, ease to scale up and smaller footprint.⁴ The membrane separation technology for the separation of CO₂ mainly contains polymeric membranes, mixed matrix membranes, and supported ionic liquid membranes. The polymeric membranes have the advantages of low cost and simple flow configuration, but they cannot be endured high temperature corrosive environments.⁵ Additionally, the tradeoff between permeability and selectivity is one of the biggest problems faced by polymeric membranes, which greatly limits their further application in the chemical industries.⁶ The mixed matrix membranes combine the merits of polymer membrane of easy operation ability and inorganic membrane of superior permeability and selectivity.⁷ But their performance can be limited due to inevitably non-selective voids generating at the interface of polymer and inorganic particles due to the incompatibility of them.⁸ The supported ionic liquid membranes have the advantages of high selectivity and permeability. However, operating conditions, such as high temperature and moderate pressure, limit their application.⁹

In recent years, gel membranes started to emerge as an alternative approach in gas membrane separation technology for separation of CO₂. They can be formed from a polymer solution in a mixture of solvents, in which the “good” solvent is prone to evaporate while the “poor” solvent is non-volatile.¹⁰ The evaporation of the good solvent results in the polymer phase separation. The gel membranes can improve problem of incompatibility of the mixed matrix membranes, because the non-volatile solvent has better compatibility with polymer than inorganic particles. Furthermore, the gel membranes have better stability than the supported ionic liquid membranes. There are several reports on the application of gel membranes for gas separation, such as ionic liquids/polymer gel membranes.^10–12 However, ionic liquids are expensive, and their synthesis usually requires several synthetic steps, thus they are limited in the commercial application.¹³ Recently, Zhang et al.¹⁴ found that hydrocarbon surfactants such as Tween and Triton have a strong ability to dissolve CO₂. Similar to the ILs, these hydrocarbon surfactants have low volatility and good thermal stability. At the same time, they are non-toxic, low-cost, and have excellent compatibility with most polymers. The hydrocarbon surfactant is a kind of desirable functional carrier to build gel membranes.

Poly(ether-block-amide) (PEBA) is a family of copolymers, consisting of polyamide hard segments and polyether soft segments in the polymer chain. PEBA not only has favorable membrane-forming properties but also good chemical resistance to acid, basic and organic solvents and high thermal and mechanical stabilities.¹⁵ PEBA is a thermoplastic elastomer material and its structure is described as follows:

where PA is an aliphatic polyamide “hard segment” and PE is an amorphous polyether “soft segment”. The soft segment is either poly(ethylene oxide) or poly(tetramethylene oxide).¹⁶ The PA provides mechanical strength and the PE offers high permeability.¹⁷

In this study, the thermoplastic copolymer PEBA2533 has been coupled with three kinds of Tweens (Tween20, Tween21 and Tween80), to prepare polymeric gel membranes. Properties of three Tweens are shown in Table 1.¹⁸ Tween20 and Tween21 have the same hydrophobic groups (alkyl chain), but different hydrophilic groups. In contrast, Tween20 and Tween80 have the same hydrophilic groups, but different hydrophobic groups (alkyl chain). The motivation of this work is to study the effect of Tween with different hydrophobic and hydrophilic groups on gas transport properties. Additionally, the influences of Tween (20, 21 and 80) contents on gas permeation performances of the gel membranes were evaluated.

Table 1 Molecular structure, selected properties and viscosity coefficients of Tween20, Tween80 and Tween21

Surfactant	Structure	Molecular formula	Molecular weight	viscosity^a/(cp)
a Test condition: 80 r min^−1, rotor is CP52 at 25 °C.
Tween20		C58H114O26	1227.51	432.6
Tween21		C26H50O10	522.67	670.4
Tween80		C64H124O26	1309.65	372.7

---

2. Experimental section

2.1 Materials

The PEBA (grade 2533) copolymer (comprise 80 wt% of poly(tetramethylene oxide) [PTMO] and 20 wt% Nylon-12 [PA12]) was supplied by Arkema. N-Butanol was purchased from National Pharmaceutical Group Chemical Reagent Co. Ltd., China, and used as a solvent without further purification. The hydrocarbon surfactants used in this study were Tween20, Tween80 and Tween21, which were obtained from National Pharmaceutical Group Chemical Reagent Co. Ltd., China. Their properties were summarized in Table 1. CO₂ and N₂ were supplied by Wuxi Xinnan Chemical Gas Co. Ltd., China, and were of at least 99.99% purity. Gases were used without further purification.

2.2 Membrane preparation

PEBA2533 was dissolved at a concentration of 10 wt% in n-butanol for 3 h at 353 K. Different amounts of Tween20, Tween80 and Tween21 were added to the polymer solutions and dissolved instantaneously. The solutions were then stirred for another 1 h to ensure homogeneous mixtures.

Membranes were prepared by controlled solvent evaporation of the above-mentioned solutions. The polymer solutions were cast onto clean Teflon plates and dried at room temperature for at least 48 hours under a fume hood. In order to guarantee a complete removal of n-butanol, they were further dried at vacuum drying oven at 40 °C for another 12 h. The obtained gel membranes were designated as PEBA2533/X-Y, where X is Tween20, Tween80 and Tween21, Y is the wt% of Tween out of the total mass of polymer/Tween gel membranes. The thickness of prepared membranes was about 130–140 μm.

2.3 Fourier transform infrared spectroscopy

The chemical structures of Tween, PEBA2533 and PEB2533/Tween membranes were characterized by a FTLA 2000 type Fourier transform infrared (FT-IR) spectrometer. The samples for FT-IR measurement were obtained by spreading a thin film of their solutions on a potassium bromide flake.

2.4 Scanning electron microscope

Scanning electron micrographs of PEBA2533 and PEBA2533/Tween membranes were performed on a Hitachi S4800 scanning electron microscope (SEM) instrument. All the samples were coated with a thin layer of gold to prevent charging.

2.5 Swelling performance tests

The dried pure PEBA2533 membranes were weighed and immersed into Tween in a sealed vessel at different temperatures until the equilibrium was reached. At regular intervals, the swollen membranes were wiped out carefully with filter paper to remove superficial liquid and weighted quickly. The degree of swelling (DS) was calculated using the following eqn (1):where m₀ and m_t are the weights of dry and swollen membranes, respectively.

2.6 Differential scanning calorimetry

Differential scanning calorimetry (DSC) curve of PEBA2533/Tween membranes were obtained through a Perkin-Elmer (DSC 8000). All PEBA2533/Tween samples were subjected to a heating/cooling/heating cycle in the range from −79 °C to 170 °C at a scan rate of 10 °C min⁻¹ and the experiments were conducted using a nitrogen purge gas stream with the scavenging rate of 20 mL min⁻¹. The DSC curves were obtained from a second heating circle for removing heat-history.

2.7 Mechanical properties

The samples were cut into some 2 × 10 mm of dumbbell bars using a standard cutting knife. Stretching tests of PEBA2533/Tween membranes were performed at room temperature using an electronic universal testing machine (Shenzhen, China) with a crosshead speed of 10 mm min⁻¹. The membranes were evaluated by three parameters σ (MPa), E (MPa) and ε (%), where σ is the tensile stress, and ε is the Maximum deformation. E represents the Young's modulus.

2.8 Gas permeability measurements

The permeability of PEBA2533/Tween membranes were measured using a time-lag measurement apparatus. Details concerning the experimental device were described in the literature.^19,20 Gas permeation rate tests on different PEBA2533/Tween membranes were carried out at 25 °C and at a feed pressure of 0.6 atm. After thorough evacuation of the membrane to remove all previously dissolved species, the membrane is exposed to the feed gas and from that moment the pressure in the fixed permeate volume is monitored. A total membrane area of 13.80 cm² was used.

The permeability (P), diffusivity (D), solubility (S) and selectivity (α) for gases CO₂ and N₂ were determined under steady state by the following equations:^21,22

where V is the constant permeate volume, R is the universal gas constant, l is the film thickness, T is the absolute temperature, A is the effective membrane area, ΔP is the pressure difference of membranes on both sides, t is the time for permeate pressure increase from P_p1 to P_p2, θ is called time-lag.

CO₂ sorption measurements of the PEBA2533 and PEBA2533/Tween membranes were conducted by the module shown in Fig. S1.†²³ More detailed steps were shown in ESI.† The sorption uptake of CO₂ can be calculated according to eqn (6):

where P_e is adsorption equilibrium pressure, Pa; V_R and V_m are the void volumes of the sample chamber and the reference chamber, respectively; T₀ and P₀ are the standard temperature and pressure, respectively.

3. Results and discussion

3.1 Fourier transform infrared

The FT-IR spectra of Tween, PEBA2533 and PEBA2533/Tween membranes are shown in Fig. 1. In the spectra of PEBA2533, the strong band at 1100 cm⁻¹, 1640 cm⁻¹ and 3300 cm⁻¹ are assigned to the C–O stretching vibrations, the C=O stretching vibrations and the N–H stretching vibrations, respectively. In the spectra of Tween, the broad and strong band for Tween from 3100 to 3500 cm⁻¹ represents the stretching vibrations of O–H. The band at 1100 cm⁻¹ is assigned to asymmetric C–O stretching vibrations; The C=O stretching vibrations appears at 1730 cm⁻¹. Compared with the spectra of PEBA2533 and Tween, no new absorption peak could be observed in the spectra of PEBA2533/Tween membranes, suggesting that Tween were physically blended within the polymer matrix. Moreover, the peaks at around 1100 cm⁻¹ and 1640 cm⁻¹ of the PEBA2533 appear weaker and show redshift after the addition of Tween, suggesting that H-bond interaction happens between Tween and PEBA2533, and that they have a good compatibility.

Fig. 1 FTIR spectra of Tween, PEBA2533 and PEBA2533/Tween membranes.

3.2 Scanning electron microscope

Fig. 2 shows the scanning electron micrographs of the cross section of PEBA2533 and PEBA2533/Tween-50 membranes. The photographs of PEBA2533/Tween-50 membranes are shown in Fig. S2.† The thickness of all membranes is in a range of 130–140 μm. As shown in Fig. 2(a-2), the cross section of pure PEBA2533 membrane is smooth while the cross sections of PEBA2533/Tween-50 membranes are rough. Rabiee et al. hypothesised that it can be the result of the decline of crystallinity of membranes with Tween addition. The crystallinity reduction causes more amorphous and unsymmetrical structure in membrane body. These structures are like scaffold and physical crosslinks in membranes to keep it uniform.²⁴ Moreover, no appreciable pore could be observed, indicating that defect-free dense membrane is synthesized and a good compatibility between Tween and PEBA2533.

Fig. 2 SEM images of the cross section of PEBA2533 and PEBA2533/Tween-50 membranes (a): PEBA2533; (b) PEBA2533/Tween20-50 membranes; (c): PEBA2533/Tween21-50 membranes; (d): PEBA2533/Tween80-50 membranes.

3.3 Thermal analysis of gel membranes

The thermal properties of gel membranes containing Tween20, Tween21 and Tween80 in PEBA2533 are researched by DSC analysis (second scan). The results are displayed in Fig. 3. The low temperature melting point, T_m (PTMO), is ascribed to melting of crystals of the polyether blocks and occurs about 0–20 °C. The high temperature melting point, T_m (PA), is attributed to melting of polyamide crystals and exists approximately 140–160 °C. From DSC experiment, the heat of fusion (ΔH_m) and the degree of crystallization (f_c) of PEBA2533/Tween membranes were obtained, which were listed in Table 2. The degree of crystallization of each PEBA2533/Tween membrane was commonly calculated from the ratio of the heat of fusion to per gram of PA segments in DSC experiment, i.e., 246 J g⁻¹, which is literature value of the heat of fusion for nylon-12.²⁵ From Table 2, for all PEBA2533/Tween gel membranes, f_c rapidly at the beginning and then gradually with increasing Tween content and a transition phenomenon of f_c vs. Tween content curve reveals and locates at PEBA2533/Tween-50 membrane. It suggests that the Tween is added to weaken the force between the molecular chains and increase the plastic effect of the polymer, inducing the smaller crystallization of the polymer chains. Furthermore, DSC analysis shows a smaller crystallinity of the PA block in PEBA2533/Tween21 compared with PEBA2533/Tween20 and Tween80.

Fig. 3 DSC thermograms of the PEBA2533 and PEBA2533/Tween membranes (a): PEBA2533 and PEBA2533/Tween20 membranes (b): PEBA2533 and PEBA2533/Tween21 membranes (c): PEBA2533 and PEBA2533/Tween80 membranes.

Table 2 Heat of fusion (ΔH_m) and degree of crystallization (f_c) of PEBA2533 and PEBA2533/Tween membranes

Sample	Tween content	ΔHm (J g^−1)	fc (%)
PEBA2533	0 wt%	6.5928	13.4
PEBA2533/Tween20	15 wt%	4.3146	10.32
	35 wt%	2.1454	6.71
	50 wt%	0.7463	3.03
	65 wt%	0.5177	3.00
PEBA2533/Tween21	15 wt%	1.7253	4.13
	35 wt%	1.0045	3.14
	50 wt%	0.3748	1.52
PEBA2533/Tween80	65 wt%	0.1599	0.93
	15 wt%	2.1938	5.24
	35 wt%	1.4246	4.45
	50 wt%	0.6194	2.52
	65 wt%	0.4017	2.33

---

3.4 Swelling performance analysis

To investigate the compatibility of PEBA2533 with Tween, the equilibrium degree of swelling (DS) values of pure PEBA2533 in Tween20, Tween21 and Tween80 at different temperatures are measured and shown in Fig. 4. It shows that at the same temperature, the values in Tween21 are greater than that in Tween20 and Tween80, indicating that Tween21 has better compatibility with PEBA2533 than Tween20 and Tween80. This is because the swelling of a polymer material in a solvent is proportional to its interaction or affinity with the solvent.^26,27 Hydroxyl groups of three Tweens could form H-bond interaction with the PA block in PEBA2533 (shown in FT-IR), enhancing solubility of three Tweens in PEBA2533, resulting in high welling (DS) values. Moreover, smaller molecular weight of Tween21 is more prone to form H-bond interaction with PEBA2533 than Tween20 and Tween80, leading to greater welling (DS) values. Both the degree of swelling of PEBA2533/Tween20 and PEBA2533/Tween80 slowly increase with temperature up to about 100 °C and then obviously increase above 120 °C. In contrast to that, the swelling of PEBA2533/Tween21 increase dramatically with the increase of temperature from the beginning. Above 120 °C, the polymer becomes completely soluble, confirming the good compatibility of the polymer with Tween21. The similar phenomenon is reported by Friess et al.¹¹ when they study the swelling performance of ionic liquid polymeric gel membranes. The complete dissolution of the polymer above 120 °C is in accordance with the strong melting point depression of the membranes with the highest Tween21 content observed by DSC analysis. In this light, swelling of a neat polymer membrane in the Tween21 could be a successful alternative method to prepare Tween21-containing gel membranes. However, this is not suitable to reach high Tween21 content because the membrane is easily deformed or damaged during the swelling as a result of the dissolution of the polymer.

Fig. 4 Effect of temperature on the equilibrium DS of PEBA2533/Tween membranes.

3.5 Mechanical properties

In order to research the effect of different Tweens on mechanical properties, the tensile strength, Young's modulus and maximum deformation of PEBA/Tween gel membranes are measured, as shown in Fig. 5. All membranes are obtained as flexible dense membranes with sufficient mechanical resistance to be handled without difficulties. With the increase of Tween content, a declining trend is observed in Fig. 5(a–c). This phenomenon is attributed to strong plasticizing effect of Tween on the polymer, causing the decline of the tensile strength and the ability to withstand the tensile deformation. Meanwhile, dispersion of the Tween in polymer chains suppresses the formation of crystal, which is confirmed by DSC. The polymer chain moves easier, resulting in lower Young's modulus and resistance to elastic deformation. Addition of Tween also weakens physical cross-linking junction of the polymer and reduces the proportion of polymer in the membrane, causing the decreasing of deformation of polymer and rapid decline of the elongation at break.

Fig. 5 Mechanical properties of membrane samples of PEBA2533/Tween20, PEBA2533/Tween21 and PEBA/Tween80: tensile strength (a), Young's modulus (b) and maximum deformation (c).

As can be seen from Fig. 5(a and b), PEBA2533/Tween20 has a higher Young's modulus and tensile strength than PEBA2533/Tween80 and PEBA2533/Tween21 over the entire composition range. It may be explained that Tween21 has better compatibility with PEBA2533 than Tween20 and Tween80 as can be confirmed by swelling experiment as shown in Fig. 4, causing greater decline of the tensile strength and Young's modulus. But in terms of the maximum deformation, PEBA2533/Tween21 is greater than PEBA2533/Tween80 and PEBA2533/Tween20. For PEBA2533/Tween21 gel membrane, there is an initial increase with the increase of Tween21 content due to the beneficial effect of plasticization of stiff polymer, increasing the flexibility of polymer. Then an obvious decline appears above 15 wt% of Tween. This is because sustained increase of Tween content makes gel membranes too weak to resist strong plastic deformation. Another reason for this phenomenon is that at high Tween21 content the intrachain crystallization becomes relatively more important than interchain crystallization,¹¹ so the gel membranes lose their mechanical strength and elasticity. As a whole, the strength and toughness of gel membranes reduced with the increasing Tween content, but they still have excellent mechanical properties as gas separation membranes.

3.6 Gas transport properties

3.6.1 CO₂ sorption. Fig. 6 shows CO₂ sorption uptake in PEBA2533/Tween gel membranes at temperatures 25 °C and at adsorption pressure of 0.6 atm. For all PEBA2533/Tween gel membranes, when the Tween content is less than 50 wt%, response of Tween content to the CO₂ sorption uptake is nearly linear. Above 50 wt%, it shows a positive deviation from the linear increase. This phenomenon can be explained that polar ethylene oxide (EO) units of Tween (20, 21 and 80) have high affinity to CO₂, enhancing CO₂ solubility in PEBA2533/Tween gel membranes, resulting in increasing CO₂ sorption uptake.¹⁴ Furthermore, at the same Tween content, the CO₂ sorption uptake in different PEBA2533/Tween gel membranes is increased in the order: PEBA2533/Tween21 < PEBA2533/Tween20 < PEBA2533/Tween80. It is probable that both PEBA2533/Tween20 and PEBA2533/Tween80 have a larger CO₂ solubility than PEBA2533/Tween21 due to more ethylene oxide (EO) units that Tween20 and Tween80 contain (shown in Table 1), leading to high CO₂ sorption uptake. As for PEBA2533/Tween20 and PEBA2533/Tween80, PEBA2533/Tween80 has a smaller crystallinity of the PA block (shown in Table 2) and higher swelling values (shown in Fig. 4) than PEBA2533/Tween20, which suggests that Tween80 has better compatibility with PEBA2533 than Tween20, resulting in higher CO₂ sorption uptake in PEBA2533/Tween80.

Fig. 6 Sorption uptakes of CO₂ in PEBA2533 (Tween content is 0 wt%) and PEBA2533/Tween membranes.

Fig. 7 shows the sorption isotherms of CO₂ in PEBA2533/Tween gel membranes at different temperatures. The CO₂ sorption uptake is increased with the increasing pressure and decreasing temperature. In terms of pressure, at low pressure (under 0.1 MPa), the CO₂ sorption uptake is almost linear growth in relation to the gas phase pressure, which can be approximated by the Henry's law. Feng et al. and Yampolskil et al. argued this linear growth indicated that the pressure had little effect on the solubility.^23,28 Analogous phenomenon was reported by Merkel et al. when they studied the sorption of hydrocarbons or other organic vapors in rubbery polymers.²⁹ Bondar et al. found more comprehensive phenomenon that the convex sorption isotherm is characteristic for more condensable gases (such as butane, ethane, and CO₂) and the linear isotherm is characteristic for permanent gases (such as H₂, N₂, O₂, and CH₄) when they studied the solubility of several gases in a series of PEBA copolymers.³⁰

Fig. 7 Sorption isotherms of CO₂ in PEBA2533/Tween membranes.

3.6.2 The gas transport properties of PEBA2533/Tween membranes. The solution-diffusion transport model³¹ was applied for discussing the gas transport properties of PEBA2533/Tween gel membranes. The data have been evaluated by the permeability (P) as the product of a thermodynamic parameter, the solubility coefficient (S) and the diffusivity coefficient (D) according to eqn (4).

The CO₂, N₂ permeability, diffusivity, solubility and CO₂/N₂ selectivity are recorded in Table 3. In terms of diffusivity, for all PEBA2533/Tween gel membranes, both CO₂ and N₂ diffusivity coefficients increase with the increase of Tween content. And these values are greater than pure PEBA2533. This is because the addition of Tween20, Tween21 or Tween80 has a strong plasticizing effect on the polymer. It weakens the force between polymer chains, increases the activity of the polymer molecular chain and improves the diffusion rate of gas molecules in the polymer.³² According to the order of the molecular diameter, d_N2 > d_CO2, CO₂ has a smaller kinetic diameter (0.330 nm) than N₂ (0.364 nm), the CO₂ diffusivity coefficients should be greater than the N₂ diffusivity in theory. But the anomalous phenomenon is shown in Table 3, the N₂ has greater diffusivity coefficients than CO₂. It is likely that the CO₂ is polar gas, which has interaction with polar groups (such as EO) in the polymer, delaying the diffusion of CO₂ in gel membranes, resulting in lower diffusion coefficient for CO₂ than for N₂.³³ For PEBA2533/Tween21 gel membranes, CO₂ and N₂ diffusivity coefficients increase slightly with increasing Tween21 content. In contrast, for PEBA2533/Tween20 and PEBA2533/Tween80, there is strongly increase with Tween20 and Tween80 contents, special for PEBA2533/Tween80. It is probable that Tween80 has longer alkyl substituent than Tween20 and Tween21 shown in Table 1. The longer alkyl substituent can create more free volume within the polymer network due to inefficient packing of the N-alkyl side chains,²⁰ which can enhance gas diffusivity. Additionally, the gas diffusivity is related to the viscosity of the liquid. It is well known that the smaller the viscosity of liquid, the greater gas diffusivity coefficients in liquid.³⁴ The viscosity coefficients of three kinds of Tweens are listed in Table 1. As shown in Table 1, because of smaller viscosity of Tween80 than Tween20 and Tween21, the PEBA2533/Tween80 gel membranes have larger gas diffusivity coefficients. There is a small decrease in D_CO2/N2 with the increase of the Tween content in all PEBA2533/Tween gel membranes, which is ascribed to interaction of CO₂ with EO groups, leading to retard CO₂ migration.³⁵ Ultimately the diffusion selectivity of CO₂/N₂ decreases.

Table 3 CO₂ and N₂ permeability, diffusivity, solubility coefficients measure at 25 °C and CO₂/N₂ gas selectivity for PEBA2533 and PEBA2533/Tween gel membranes^a

Sample	Tween content	PCO2	DCO2	SCO2	PN2	DN2	SN2	αCO2/N2	DCO2/DN2	SCO2/SN2
a Permeability coefficient (Barrer), 1 Barrer = 10^−10 cm^3 (STP) cm per cm^2 per cm Hg per s. Diffusivity coefficient [cm^2 s^−1] × 10^8. Solubility coefficient [cm^3 (STP) per cm^3 per cm Hg] × 10^2.
PEBA2533	0 wt%	169	74.35	2.27	7.5	79.21	0.095	22.53	0.94	24.01
PEBA2533/Tween20	15 wt%	179	77.15	2.32	6.9	81.65	0.085	25.94	0.94	27.46
	35 wt%	210	82.53	2.54	7.4	90.36	0.082	28.38	0.91	31.07
	50 wt%	223	85.30	2.61	7.2	98.30	0.073	30.97	0.87	35.69
	65 wt%	267	96.34	2.77	7.3	123.22	0.059	36.58	0.78	46.78
PEBA2533/Tween21	15 wt%	177	77.09	2.30	7.1	80.30	0.082	24.93	0.96	25.97
	35 wt%	196	80.68	2.43	7.0	87.92	0.080	28.01	0.92	30.51
	50 wt%	200	80.92	2.47	6.8	94.36	0.070	29.41	0.88	35.39
PEBA2533/Tween80	65 wt%	221	86.53	2.55	6.9	110.71	0.057	32.03	0.80	44.68
	15 wt%	200	81.31	2.46	6.8	84.96	0.080	29.41	0.96	30.73
	35 wt%	219	84.93	2.58	6.3	99.88	0.063	34.76	0.85	40.88
	50 wt%	240	88.90	2.70	6.5	114.12	0.057	36.92	0.78	47.40
	65 wt%	289	100.36	2.88	7.1	139.16	0.051	40.70	0.72	56.44

---

For all PEBA2533/Tween gel membranes, the solubility of CO₂ gradually increases with the increase of Tween content, and N₂ decreases. The solubility coefficient of CO₂ increases because CO₂ is an electron acceptor and EO is an electron donor, the Lewis acid–base interaction between CO₂ and EO can enhance the dissolution of CO₂ in the Tween. The solubility coefficient of N₂ decreases due to the strong hindering effect of EO groups to N₂. Among three kinds of polymeric gel membranes, both Tween20 and Tween80 have 20 EO units whereas Tween21 has only 4 EO units. As a result, at the same Tween content, the PEBA2533/Tween20 and PEBA2533/Tween80 have greater solubility coefficient of CO₂ than PEBA2533/Tween21. On the other hand, different hydrophobic groups of Tween20 and Tween80 have influence on gas transport properties. To get more information about it, Henry's constants¹⁴ and enthalpies and entropies of CO₂ dissolution in the two surfactants of Tween20 and Tween80 are listed in Table 4. It is well known that the smaller the Henry's constant is, the larger the solubility becomes.¹⁴ Therefore, PEBA2533/Tween80 gel membranes have greater solubility than PEBA2533/Tween20 due to smaller the Henry's constant for PEBA2533/Tween80 gel membranes. The CO₂/N₂ solubility selectivity is far greater than CO₂/N₂ diffusivity selectivity, indicating that there is a “solubility controlled” transport in all PEBA2533/Tween gel membranes.

Table 4 Experimental Henry's constants at 25 °C and enthalpies and entropies of dissolution for CO₂ dissolved in surfactants

Surfactant	Henry's constant/bar (25 °C)	ΔH (kJ mol^−1)	ΔS (J mol^−1 K^−1)
Tween20	10.7	−11.9 ± 1.1	−39.9 ± 3.3
Tween80	10.1	−10.2 ± 0.8	−33.6 ± 2.5

---

CO₂ permeability obviously increases with the Tween contents, while N₂ permeability doesn't change much. The CO₂ permeability is greater than N₂ permeability. The degree of increase of gas permeability with increasing Tween80 content is much more pronounced compared to the addition of Tween20 and Tween21. The CO₂ permeability value achieves 289 Barrer at 65 wt% of Tween80. This significant increase can be ascribed to two separate effects. First, increasing Tween content means increasing overall concentration of EO units. This can enhance the CO₂ solubility and the free volume available for permeation due to the plasticizing effect, analogous to the results reported by Car et al.^36–38 Second, Tween80 has longer alkyl substituent, which results in higher CO₂ diffusivity coefficient (shown Table 3). The permeability in membranes closely related to their diffusivity and solubility, which can be seen in formula (5). Thus the CO₂ permeability of PEBA2533/Tween80 gel membranes is greater than PEBA2533/Tween20 and PEBA2533/Tween21 gel membranes.

The interaction between CO₂ and the polymer makes the solubility selectivity of CO₂/N₂ increase significantly, while the diffusion selectivity doesn't change much, leading to improve the permeability selectivity. The addition of Tween results in the permeability selectivity increasing from 22.53 to 36.58 for PEBA2533/Tween20, 22.53 to 32.03 for PEBA2533/Tween21 and 22.53 to 40.70 for PEBA2533/Tween80 gel membranes.

In conclusion, it should be noted that the number of hydrophilic group in Tween has a significant impact on the gas solubility in all PEBA2533/Tween gel membranes. The more the hydrophilic groups in Tween, the greater the solubility of CO₂ in membrane. On the other hand, the length of alkyl substituent of hydrophobic group can affect gas diffusivity in polymer gel membranes. The longer the alkyl chains in Tween, the greater the diffusivity of CO₂ in membrane. Furthermore, among three types of PEBA2533/Tween gel membranes, PEBA2533/Tween80 membranes have more excellent diffusion performance and solubility properties of CO₂ than PEBA2533/Tween20 and PEBA2533/Tween21 gel membranes, resulting in higher permeability coefficient in PEBA2533/Tween80.

3.6.3 The gas transport properties of PEBA2533/Tween20-50, PEBA2533/Tween21-50 and PEBA2533/Tween80-50 membranes at different temperatures. Table 5 lists CO₂, N₂ diffusion, solubility, permeability coefficients and CO₂/N₂ selectivity in PEBA2533/Tween20-50, PEBA2533/Tween21-50 and PEBA2533/Tween80-50 gel membranes at various temperatures. The temperature dependence of diffusivity, solubility and permeability coefficients can be described by the following equations.

D = D₀ exp(−E_d/RT) (7)

where E_d is the activation energy of diffusion and D₀ is a constant.³⁹

S = S₀ exp(−ΔH_s/RT) (8)

where S₀ is a constant and ΔH_s is the partial molar enthalpy of sorption.

P = P₀ exp(−E_p/RT) (9)

where P₀ is a pre-exponential factor, E_p is the apparent activation energy for permeation, which is equal to the algebraic sum of E_d and ΔH_s. Values of ΔE_p, E_d and H_s for CO₂ and N₂ of different gel membranes are presented in Table 6.

Table 5 Gas transport properties of PEBA2533/Tween gel membranes at different temperatures: PEBA2533/Tween20-50 (a), PEBA2533/Tween21-50 (b) and PEBA2533/Tween80-50 (c)^a

Sample	Temperature	P		D		S		Selectivity
		CO2	N2	CO2	N2	CO2	N2	α	αD	αS
a Permeability coefficient (Barrer), 1 Barrer = 10^−10 cm^3 (STP) cm per cm^2 per cm Hg per s. Diffusivity coefficient [cm^2 s^−1] × 10^8. Solubility coefficient [cm^3 (STP) per cm^3 per cm Hg] × 10^2.
PEBA2533/Tween20-50	15 °C	209	5.5	78.06	94.10	2.68	0.058	38.00	0.83	45.81
	25 °C	223	7.2	85.30	98.30	2.61	0.073	30.97	0.87	35.69
	35 °C	287	12.7	119.29	134.46	2.41	0.094	22.60	0.89	25.47
PEBA2533/Tween21-50	15 °C	189	5.6	75.38	92.55	2.51	0.061	33.75	0.81	41.44
	25 °C	200	6.8	80.92	97.36	2.47	0.070	29.41	0.83	35.39
	35 °C	214	9.8	99.59	119.17	2.15	0.082	21.84	0.84	26.13
PEBA2533/Tween80-50	15 °C	229	6	83.09	107.99	2.76	0.056	38.17	0.77	49.60
	25 °C	240	6.5	88.90	114.12	2.70	0.057	36.92	0.78	47.40
	35 °C	315	14.5	128.44	146.38	2.45	0.099	21.72	0.88	24.76

---

Table 6 Heats of sorption (kJ mol⁻¹), ΔH_s, permeation activation energies, E_p, and diffusion activation energies, E_d, for CO₂ and N₂ of PEBA2533/Tween membranes

Sample		Ep	Ed	ΔHs
PEBA2533/Tween20-50	CO2	11.625	15.546	−3.921
	N2	30.755	13.058	17.697
PEBA2533/Tween21-50	CO2	4.579	10.220	−5.641
	N2	20.613	10.853	9.760
PEBA2533/Tween80-50	CO2	11.672	15.946	−4.274
	N2	32.260	11.143	21.117

---

It is observed that the diffusion of CO₂ and N₂ are on the rise with the increasing temperature in Table 5. Generally, gas diffusion coefficients can appreciably increase with increasing temperature if the polymer does not suffer from thermally induced morphological rearrangement over the temperature range.⁴⁰

For the gas solubility, dissolution of gas molecule in a polymer matrix can be regarded as two thermodynamic processes: (1) condensation of the gas molecule to a condensed density; (2) forming a molecular scale gap in the polymer to accommodate the gas molecule. As a result, the enthalpy of sorption can be written as:

ΔH_s = ΔH_condensation + ΔH_mixing (10)

where ΔH_condensation and ΔH_mixing are the enthalpy changes related with the first and second thermodynamic processes, respectively.⁴¹ For non-condensable gas molecule such as N₂, ΔH_s is determined by ΔH_mixing due to ΔH_condensation being very small. The interaction between polymers and N₂ is very weak, and ΔH_mixing is positive, causing the N₂ solubility improves with the increasing temperature. But for condensed gas molecule such as CO₂, ΔH_condensation is large and negative. The ΔH_s is determined by ΔH_condensation, causing a negative ΔH_s value. Another reason for the negative ΔH_s is that a strong interaction between PTMO chains and CO₂ molecules improved inherent condensability.⁴² Therefore, the solubility coefficient of CO₂ decreases with the increasing temperature (shown in Table 5). Since the diffusivity generally has a stronger function of temperature than the solubility, the gas permeability usually increases with the increasing temperature.

The effect of temperature on the solubility selectivity is primarily determined by the chemical nature of the gas molecules and polymer–gas interactions. For most gases, the solubility decreases as temperature increases. The α_CO2/N2 significantly decreases, due to obviously decreased solubility selectivity and unchanged diffusivity selectivity, as shown in Table 5. It is similar to the case reported by Li et al.⁴³

3.7 Comparison of CO₂/N₂ separation performance among polymer membranes

Fig. 8 shows the trade-off relations of CO₂ and N₂ for universal polymeric membranes. For numerous polymers, the α_CO2/N2 values decreased with the increasing of P_CO2 values, namely, Robeson's traded-off relation.⁴⁴ For example, 80% reduction of α_CO2/N2 for poly(ether imide) segmented copolymers with 68.6 wt% of PEO content.⁴⁵ Similar to PEBA1657 copolymer, 75% decrease of PEO-filled polyacrylonitrile membranes with 42.2% of graft yield.⁴⁶ Liliana et al.⁴⁷ have reported that PIL/IL blend membranes had P_CO2 of 5.09 Barrer and 313 Barrer with P_CO2/P_N2 of 22.2 and 28.9, respectively. Wilfredo et al.⁴⁸ have obtained PEBA/PEG-DME membranes, which had P_CO2 from 78 to 606 Barrer. However, P_CO2/P_N2 decreased from 49 to 43. These results together with those previously reported^10,37,48 are shown in Fig. 8. The anti-trade-off phenomenon in gas separation for CO₂/N₂ by the PEBA2533/Tween gel membranes were observed, which displays that the α_CO2/N2 increases with an elevation in the P_CO2 values. Specifically, with the increase of Tween20, Tween21 and Tween80 from 0 to 65 wt%, α_CO2/N2 was improved from 22.53 to 36.58 for PEBA2533/Tween20 membrane, 22.53 to 32.03 for PEBA2533/Tween21 membrane and 22.53 to 40.70 for PEBA/Tween80 membrane, respectively.

Fig. 8 Robeson plot for CO₂/N₂ upon variation of the Tween addition in the membranes.

4. Conclusions

The polymeric gel membranes, PEBA2533/Tween, are successfully prepared using solvent casting method. The polymer shows a better compatibility with Tween21 than Tween20 and Tween80. DSC analysis shows decrease of crystallinity of the polyamide (PA) blocks in PEBA2533/Tween20, PEBA2533/Tween21 and PEBA2533/Tween80.

The incorporation of Tween into PEBA2533 not only increases the CO₂ permeability, but also enhances the separation factor of CO₂/N₂. Increasing the amount of N-alkyl groups in Tween improves CO₂ permeability and the separation factor of CO₂/N₂. The best separation performance is achieved for the PEBA2533/Tween80-65 gel membranes, which have CO₂ permeability of 289 Barrer and CO₂/N₂ selectivity of 40.70. The α_CO2/N2 increases with an elevation in the P_CO2 values, which have been discovered to solve the trade-off phenomenon.

Acknowledgements

This research was financially supported by the National Nature Science Foundation of China (Grant no. 21106053), the Fundamental Research Funds for the Central Universities (JUSRP311A01) and the Industry-Academia Cooperation Innovation Fund Projects of Jiangsu Province (BY2012058).

References

1. H. Lin and B. D. Freeman, J. Membr. Sci., 2005, 739, 57–74.
2. S. D. Karsari, D. Yang, G. D. Jiang, S. J. Zhang, J. J. Wang, A. G. Russell, Q. Wei and M. Fan, RSC Adv., 2013, 3, 22739–22773.
3. M. G. Buonomenna, RSC Adv., 2013, 3, 5694–5740.
4. Z. J. Ma, Z. H. Qiao, Z. Wang, X. C. Cao, Y. P. He, J. X. Wang and S. C. Wang, RSC Adv., 2014, 4, 21313–21317.
5. K. Golzar, S. A. Iranagh, M. Amani and H. Modarress, J. Membr. Sci., 2014, 451, 117–134.
6. Y. Y. Zee, L. C. Thiam, W. Z. Peng, A. R. Mohamed and S. P. Chai, J. Nat. Gas Chem., 2012, 21, 282–298.
7. M. G. Buonomenna, W. Yave and G. Golemme, RSC Adv., 2012, 2, 10745–10773.
8. H. Q. Yang, Z. H. Xu, M. H. Fan, R. Gupta, R. B. Slimane, A. E. Bland and I. Wright, J. Environ. Sci., 2008, 20, 14–27.
9. P. Uchytil, J. Schauer, R. Petrychkovych, K. Setnickova and S. Y. Suen, J. Membr. Sci., 2011, 383, 262–271.
10. P. Bernardo, J. C. Jansen, F. Bazzarelli, F. Tasselli, A. Fuoco, K. Friess, P. Izák, V. Jarmarová, M. Kačírková and G. Clarizia, Sep. Purif. Technol., 2012, 97, 73–82.
11. J. C. Jansen, K. Friess, G. Clarizia, J. Schauer and P. Izak, Macromolecules, 2011, 44, 39–45.
12. S. U. Hong, D. Park, Y. Ko and I. Baek, Chem. Commun., 2009, 7227–7229.
13. H. Iken, F. Guillen, H. Chaumat, M. R. Mazières, J. C. Plaquevent and T. Tzedakis, Tetrahedron Lett., 2012, 53, 3474–3477.
14. J. L. Zhang, B. X. Han and Y. J. Zhao, Chem. Commun., 2011, 47, 1033–1035.
15. L. Liu, A. Chakma and X. S. Feng, Chem. Eng. J., 2004, 105, 43–51.
16. Y. W. Cen, C. S. Bickel and R. N. Lichtenthaler, J. Membr. Sci., 2002, 206, 341–349.
17. G. D. Fu, G. L. Li, K. G. Neoh and E. T. Kang, Prog. Polym. Sci., 2011, 36, 127–167.
18. Y. Hou and R. E. Baltus, Ind. Eng. Chem. Res., 2007, 46, 8166–8175.
19. T. Uragami, H. B. Hopfenberg, W. J. Koros, D. K. Yang, V. T. Stannett and R. T. Chern, J. Polym. Sci., 1986, 24, 779–792.
20. J. E. Bara, S. Lessmann and C. J. Gabriel, Ind. Eng. Chem. Res., 2007, 46, 5397–5404.
21. A. M. Shishatskii, Y. P. Yampolski and K. V. Peinemann, J. Membr. Sci., 1996, 112, 275–285.
22. J. Qiu, J. M. Zheng and K. V. Peinemann, Macromolecules, 2007, 40, 3213–3222.
23. L. Liu, A. Chakma and X. S. Feng, Chem. Eng. Sci., 2006, 61, 6142–6153.
24. H. Rabiee, M. Soltanieh, S. A. Mousavi and A. Ghadimi, J. Membr. Sci., 2014, 469, 43–58.
25. A. Xenopoulos and B. Wunderlich, J. Polym. Sci., Part B: Polym. Phys., 1990, 28, 2271–2290.
26. Q. Zhao, J. W. C. Dunlop, X. L. Qiu, F. H. Huang, Z. B. Zhang, J. Heyda, J. Dzubiella, M. Antonietti and J. Y. Yuan, Nat. Commun., 2014, 5, 4293.
27. Q. F. An, J. W. Qian, Q. Zhao and C. J. Gao, J. Membr. Sci., 2008, 313, 60–67.
28. Y. Yampolskii, I. Pinnau and B. D. Freeman, Materials Science of Membranes for Gas and Vapor Separation, John Wiley & Sons Ltd., West Sussex, 1st edn, 2006, ch. 2, pp. 67–68.
29. T. C. Merkel, V. I. Bondar, K. Nagai, B. D. Freeman and I. Pinnau, J. Polym. Sci., Part B: Polym. Phys., 2000, 38, 415–434.
30. V. I. Bondar, B. D. Freeman and I. Pinnau, J. Polym. Sci., Part B: Polym. Phys., 1999, 37, 2463–2475.
31. J. G. Wimans and R. W. Baker, J. Membr. Sci., 1995, 107, 1–21.
32. K. Friess, J. C. Jansen, F. Bazzarelli, P. Izák, V. Jarmarová, J. Schauer, G. Clarizia and P. Bernardo, J. Membr. Sci., 2012, 415–416, 801–809.
33. M. Amani, S. A. Iranagh, K. Golzar, G. M. M. Sadeghi and H. Modarress, J. Membr. Sci., 2014, 462, 28–41.
34. B. E. Poling, J. M. Prausnitz and J. P. Connell, The properties of gases and liquids, McGraw-Hill, New York, 5th edn, 2001.
35. R. A. Zoppi and C. G. A. Soares, Adv. Polym. Technol., 2002, 21, 2–16.
36. A. Car, C. Stropnik, W. Yave and K. V. Peinemann, J. Membr. Sci., 2008, 307, 88–95.
37. W. Yave, A. Car, K. V. Peinemann, M. Q. Shaikh, K. Rätzk and F. Faupel, J. Membr. Sci., 2009, 339, 177–183.
38. S. R. Reijerkerk, M. H. Knoef, K. Nijmeijer and W. Matthias, J. Membr. Sci., 2010, 352, 126–135.
39. B. D. Freeman and I. Pinnau, ACS Symp. Ser., 1999, 733, 1–27.
40. J. C. Chen, X. S. Feng and P. Alexander, Sep. Sci. Technol., 2010, 39, 149–164.
41. K. Ghosal and B. D. Freeman, Polym. Adv. Technol., 1993, 5, 673–697.
42. J. H. Kim and Y. M. Lee, J. Membr. Sci., 2001, 193, 209–225.
43. J. Li, S. Wang and K. Nagai, J. Membr. Sci., 1998, 138, 143–152.
44. L. M. Robeson, J. Membr. Sci., 1991, 62, 165–185.
45. K. Okamoto, M. Fujii, S. Okamyo, K. Tanaka, H. Kita and M. Fujii, Macromolecules, 1995, 28, 6950–6956.
46. J. H. Kim, S. Y. Ha, S. Y. Nam, J. W. Rhim, K. H. Baek and Y. M. Lee, J. Membr. Sci., 2001, 186, 94.
47. L. C. Tome, D. Mecerreyes, C. S. R. Freire, L. P. N. Rebelo and I. M. Marrucho, J. Membr. Sci., 2013, 428, 260–266.
48. W. Yave, J. Car and K. V. Peinemann, J. Membr. Sci., 2010, 350, 124–129.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra14877k

---