Gas separation properties of Troeger’s base-bridged polyamides

2.1 Materials

4,4′-(Hexafluoroisopropylidene)bis(benzoic acid) (HFPA) (98%), isophthalic acid (98%), 2,6-naphthalenedicarboxylic acid (99%), 4,4′-oxy(benzoic acid) (99%), 2-methy-4-nitroaniline (99%), trifluoroacetic acid, Pd/C, triphenyl phosphate (TPP) were purchased from Aldrich (USA). N, N-dimethylacetamide (DMAc), 1-methyl-2-pyrolidine (NMP), pyridine, NaOH, and CaCl₂ were purchased from E. Merck, India. Methanol was purchased from SD Fine Chemicals, India. DMAc and NMP were dried using NaOH and distilled from P₂O₅ before use. Pyridine was purified by distillation after being stirred with NaOH. 2,8-Diamino-4,10-dimethyl-6H,12H-5,11-methanodibenzo[1,5]-diazocine was prepared according to the reported procedure (28).

2.2 Characterization methods

The carbon, hydrogen and nitrogen contents of the compounds were analyzed by the pyrolysis method using a Vario EL (Elementar, Germany) elemental analyzer. The attenuated total reflection mode Fourier transform infrared (ATR-FTIR) spectra of the polymer membranes were recorded using a NEXUS Nicolet Impact-410 spectrophotometer. ¹H NMR spectra of the polymers were recorded using a Bruker 600 MHz instrument (Switzerland) by dissolving the desired quantity of the samples in different deuterated solvents, e.g. DMSO-d₆ or pyridine-d₅ at 30°C. The inherent viscosities (η_inh) of all the polymers were measured using an Ubbelohde viscometer at a polymer concentration of 0.5 g dl⁻¹ in DMAc at 30°C. The wide angle X-ray diffractometry (WAXD) of the polymers were conducted in reflection mode at room temperature using a Rigaku, Ultema III X-ray diffractometer. The Cu Kα (0.154 nm) radiation source was used and a scanning range of 5–40° (2θ) was employed with an intensity beam of 40 kV and 40 mA. The value of the amorphous peak maxima (d_spacing) was calculated utilizing Bragg’s law (d=nλ/2 sinθ). TA instrument Q20 differential scanning calorimetry (DSC) was performed at a heating rate of 20°C/min to determine the glass transition temperature (T_gs) of the polymer. The T_g of the polyamides were determined from the midpoint of the step transition in a second heat scan under a nitrogen atmosphere. A TA instruments TA Q50 instrument was used to perform the thermo gravimetric analysis (TGA) of the polymers. The polymer samples were dried at 160°C for 24 h under vacuum before performing TGA. The dried samples were heated from 50 to 800°C at a heating rate 10°C/min under synthetic air (N₂:O₂ is 80:20). Mechanical properties of the membranes were measured using a Universal Testing Machine UTM H5KS Tinius Olsen, equipped with 50 N load cell and a strain rate of 5% min⁻¹ of the specimen length at 30°C. Rectangular specimens of 10 mm width, with an effective length of 25 mm (distance between the clamps) and a thickness of around 63–77 μm were used for the measurement. At least four specimens of each sample were used for each measurement, and the average values were reported. The dielectric constant (ε) values were calculated using a HIOKI 3532-50 LCR Hi Tester. The capacitance values of the polymer membranes were measured using the parallel plate capacitor method, at 1 MHz at 30°C and relative humidity 45. The densities (ρ) of the PA membranes were determined by Archimedes’ principle using a Wallace High Precision Densimeter-X22B (UK) at 30°C. The samples were sequentially weighed in air and high purity isooctane at 30°C. The fractional free volumes (FFV) of the polymers were determined by the following equation [1]:

where, V is the specific volume (V=1/ρ) of the polymer determined from the measured density. The van der Waals volume (V_w ) was estimated by the Hyperchem computer program, version 8.0 (29).

2.3 Synthesis of TB-based polyamides

The polyamides (PA “I”–“IV”) were synthesized by the polycondensation reaction of an equimolar amount of the Troeger base-bridged diamine with four different aromatic diacids, using NMP as a solvent and in the presence of TPP and pyridine as a condensing agent. As a representative, the polymerization of one of the polyamide PA “I” is described here. A three-neck round bottom flask equipped with a nitrogen inlet and a refluxed condenser was charged with 0.44 g TB (1.57 mmol), 0.616 g HFPA (1.57 mmol) and 0.36 g calcium chloride. To that flask 5 ml NMP, 1.4 ml pyridine and 1.4 ml triphenyl phosphate (TPP) (5.34 mmol) were added. The mixture was heated to 110°C for 6 h under constant stirring. The highly viscous reaction mixture was cooled to room temperature and slowly added to 300 ml methanol with constant stirring. A fibber like precipitate was obtained. The white fibrous polymer was filtered and washed with methanol followed by distilled water (hot and cold) to remove any adsorbed solvent and CaCl₂. The polymer was dried overnight at 120°C under vacuum; then PAs were prepared by a similar polyamidation reaction of TB with three other dicarboxylic acids.

The analytical details of the polymers are given below:

Anal. calcd for C₃₄H₂₆F₆N₄O₂ (636.59 g mol⁻¹): C, 64.15%; H, 4.12%; N 8.80%; O 5.03%; F, 17.91%. %; found: C, 64.21%; H, 4.14%; N 8.84%; O 5.01%; F, 17.87%. FTIR (KBr, cm⁻¹): 3302 (N-H stretching), 2947 (aromatic C-H stretching), 1656 (>C=O stretching). ¹H-NMR: δ_H(ppm) (600 MHz; pyridine-d₅; Me₄Si): 11.06 (s, 2H, amide protons), 8.28 (d, J=7.8 Hz, 4H, H⁶), 7.76 (m, 4H, H^1, H²), 7.51 (d, J=7.2 Hz, 4H, H⁷), 4.59 (d, J=12 Hz, 2H, H³), 4.31 (s, 2H, H⁴), 4.15 (d, J=12 Hz, 2H, H⁵), 2.43 (s, 6H, H⁸).

Anal. calcd for C₃₁H₂₆N₄O₃ (502.56 g mol⁻¹): C, 74.09%; H, 5.21%; N 11.15%; O 9.55%. %; found: C, 74.03%; H, 5.26%; N 11.13%; O 9.57%. FTIR (KBr, cm⁻¹): 3286 (N-H stretching), 2946 (aromatic C-H stretching), 1637 (>C=O stretching).

Anal. calcd for C₂₅H₂₂N₄O₂ (410.47 g mol⁻¹): C, 73.15%; H, 5.40%; N 13.65%; O 7.80%. %; found: C, 73.17%; H, 5.36%; N 13.66%; O 7.85%. FTIR (KBr, cm⁻¹): 3295 (N-H stretching), 2938 (aromatic C-H stretching), 1653 (>C=O stretching).

Anal. calcd for C₂₉H₂₄N₄O₂ (460.53g mol⁻¹): C, 75.63%; H, 5.25%; N 12.17%; O 6.95%. %; found: C, 75.60%; H, 5.23%; N 12.21%; O 6.98%. FTIR (KBr, cm⁻¹): 3286 (N-H stretching), 2943 (aromatic C-H stretching), 1645 (>C=O stretching).

2.4 Membrane fabrication

PA membranes were prepared by solution casting. The homogenous polymer solution was prepared by dissolving the polymers in DMAc at a concentration of 10% (w/v). The polymer solutions were filtered and poured into Petri dishes. The Petri dishes were placed in an oven at 80°C overnight, followed by slow heating to 150°C and kept for 6 h to remove most of the solvent. The membranes were further vacuum dried at 150°C for 48 h. The membranes were removed from the Petri dishes by immersing them in boiling water. The membranes obtained were dried under vacuum at 120°C for 8 h to remove any absorbed moisture. The thickness of the membranes ranged from 63 to 77 μm.

2.5 Gas permeation measurements

The gas permeability of the PA membranes were investigated using permeation test system PTS 50F-16M manufactured by Indian High Vacuum Pumps (India). The membranes were degassed in the permeation cell under high vacuum at 35°C for at least 24 h. The pure gases (excel grade) were pressurized (3.5 bar) on the membrane in the following order to avoid plasticization: CH₄, N₂, O₂ and CO₂. Gas permeability coefficient (P), diffusion coefficients (D), solubility coefficients (S), ideal diffusibility selectivity (α_D) and ideal solubility selectivity (α_S) were calculated according to the reported procedure (4, 10).

3.1 Synthesis of polymers

A series of new aromatic polyamides PA “I”–“IV” were prepared by phosphorylation polyamidation reaction of TB with different commercial diacids using TPP and pyridine as condensing agent as shown in Scheme 1 (30). The polymerization yield was nearly quantitative. The inherent viscosity (Table 1) of the polymers in DMAc ranged from 0.85 to 1.13 dl/g. Membranes prepared from solution casting were tough and flexible. The elemental (C, H, N) composition of the polymers were checked by the pyrolysis method and were in good agreement with the calculated values from the polymer repeat unit structures. The formation of the polymers were confirmed by FTIR-ATR and ¹H NMR spectroscopic methods. The FTIR-ATR spectra of these polymers exhibited the characteristic absorptions bands at 3280–3320 and 1636–1659 cm⁻¹ that correspond to characteristics >N-H and >C=O stretching (31). The structures of the PA “I” were also verified by ¹H NMR spectra (Figure 1). The peaks appeared at 11.06 ppm corresponds to the amide proton that supports the formation of the amide linkages. All the polyamides are soluble in polar aprotic solvents, such as NMP, DMAc and DMF, at room temperature. However, except PA “I” (containing 6F group), other polymers showed poor solubility in pyridine solvents, due to the strong inter chain interactions and efficient packing of the polymer chains of amide linkages.

Scheme 1:

Synthesis of the polyamides (PA “I”–“IV”).

Figure 1:

¹H-NMR spectrum of PA “I” in pyridine-d₅ (*signals at 8.74, 7.59, and 7.22 ppm are for pyridine).

3.2 Molecular packing

WAXD diffraction patterns were measured to investigate the polymers chain packing. Figure 2 represents the WAXD diffraction patterns of the polyamides and the packing parameters of WAXD are summarized in Table 1. The broad spectra suggests the amorphous nature of the polyamides. The center of the broad peak in the diffraction pattern of WAXD was used to calculate average intersegemental spacing or d_spacing between the chains, calculated by Bragg’s equation (16), (32). All PA membranes had two broad diffraction peaks, ranging from low angel side to high angel side. For PA “I”, PA “III” and PA “IV” peaks located at low angel 2θ=16.7°, 19.7 and 19.8° can be assigned to the main inter chain distance in the amorphous domain. The peaks located at the high angel side 2θ=22.6°, 24.6 and 25.7° can be assigned to the π-π stacking of aromatic rings or the interchain distance in the ordered domain or the lose stacks in the polymer chains. PA “II” showed only a broad amorphous halo peak in WAXD at 2θ=19.1°, representing typical amorphous morphology (33). Different diffraction patterns in WAXD provides an idea of free volume distribution and morphologies apart from the free volume. Different free volume morphologies arise from the structural differences of the PAs which attributed different gas permeability (34).

Figure 2:

WAXD diffraction curves of the polyamide membranes.

3.3 Thermal properties

The glass transition temperature (T_g) of the polymers (Table 2) were found to be in the range of 283–290°C, depending on the diacid components (Figure 3). The T_g values were according to the decreasing order of stiffness of the polymer chains. The lowest T_g was observed for the PA derived from the diacid with flexible isopropylidene units (6F). The polymers with rigid substituents restricted the free rotation of the polymer chain and showed a higher T_g compared to that of the other polymers. The rigid naphthalene containing polymer backbone exhibited the highest T_g value (290°C). The thermal stabilities of the polymers were evaluated by thermogravimetric analysis (TGA) in an air atmosphere (Figure 4). Their decomposition temperatures (10% weight loss) for the PAs were in the range of 343–431°C, and the maximum decomposition temperatures were in the range of 431°C.

Figure 3:

DSC curves (2^nd heat scan) of the polymers (heating rate: 20°C min⁻¹).

Figure 4:

TGA thermograms of the polymers in air (heating rate: 10°C min⁻¹).

The mechanical properties of the PA membranes are listed in Table 2. The membranes showed a tensile strength in the range of 66–95 MPa, an elongation at break in the range of 5.5–7.9% and a tensile modulus in the range of 1.9–2.6 GPa.

3.4 Gas permeation properties

The gas permeability of the PA membranes was tested for four gases CH₄, N₂, O₂ and CO₂ at 35°C with a feed pressure of 3.5 bar. The gas permeability coefficients (P) and ideal selectivities (α) are shown in Table 3. p-Values were measured from the steady-state region of the downstream pressure-time curve by using the constant-volume/variable-pressure technique. The order of gas permeabilities for all PA membranes follow the following order: P(CO₂)>P(O₂)>P(N₂)>P(CH₄). This order is the reverse order of their kinetic diameters (σ) of the gases (Å): 3.3 for CO₂, 3.46 for O₂, 3.64 for N₂ and 3.8 for CH₄, representing that the gas permeabilities decrease with the increase of kinetic diameter of gases, and the same trend was observed in other glassy polymers (36). For each gases, the gas permeability increases in the order of PA “IV”<PA “III”<PA “II”<PA “I”. The CO₂ permeability of the membranes were in the range of 86–109 Barrer. These values are much higher than commercial polyimide Matrimid^® (P_CO2=8.7). The methyl groups present in the TB unit of the polymer help to increase the distance between the polymer chains and to increase the permeability (18). Gas permability of the membranes strongly depends on average interchain spacing (d_spacing) and FFV as can be seen from Table 3. Chemical structures of the polymer have a significant influence on permeability. PA “I” containing large bulky 6F groups exhibited a relatively higher permeability. Interchain spacing is used to measure the openness of the polymer backbone, i.e. FFV and is comparative to the permeability. The presence of TB units in polymer backbone improves the permeability, compared to the analogous 6F group (same acid part HFPA) containing polymer membranes like PA 9B and PA “S2” (37), (38). The permeability of PA “II” is greater than those PA “III” and PA “IV”, due to the incorporation of flexibility into the polymer chain through ether linkages. The PA “III” showed higher permeability than PA “IV” due to the unsymmetrical structure of the acid moiety, except CO₂ gas. The PA “IV” exhibited lowest permeability for all the gases due to higher chain packing density, as evident from its FFV and d_spacing values (Table 1). For amorphous polymer systems, the gas permeability can be approximately correlated with FFV by the following equation P=Ae^(−B/FFV) where, the parameters A and B characteristic of each gas (10), (11). The PAs in the present investigation also satisfactory follow the same trend (Figure 5). According to the free volume theory, the higher is the free volume in polymer matrix the higher will be the available space for a gas molecule to diffuse. FFV and diffusitivity both increases by the introduction of side methyl groups in m-position of the amide bond. This group hinders the rotation of the polymer chains that causes less efficient chain packing in the polymer matrix. Despite having very high FFV of PA I, the gas permeability of this polymer was considerably lower than many other PIMs, e.g. TBDA1-SBI-PI (Table 3) (24). However, gas permeability depends on many factors, e.g. type of PIMs, membrane preparation technique, measurement technique, and applied pressure and temperature.

Table 3:

Gas permeability coefficients (P) in Barrer and permselectivities (α) of the polyamides, at 35°C, 3.5 bar (data from film aged for 10 days) and their comparison with other commercially available and reported polymers.

Polymer	Polymer structure	P(CO2)	P(O2)	P(N2)	P(CH4)	α(CO2/CH4)	α(O2/N2)	Ref.
PA “I”		109.1	21.6	2.87	2.17	50.3	7.52	a
PA “II”		92.2	18.3	2.54	1.97	46.8	7.20	a
PA “III”		84.7	16.8	2.33	1.75	48.4	7.21	a
PA “IV”		86.4	14.1	2.16	1.61	53.7	6.52	a
PI-TB-2		55.0	14.0	2.50	2.10	26.0	5.48	(16)b
TBDA1-6FDA-PI		155	28.0	6.50	3.30	46.9	3.90	(20)c
TBDA1-SBI-PI		895	190	350	45.0	19.7	5.40	(24)d
PI-TB-3		218	42.0	9.50	6.70	32.7	4.50	(25)e
PI-TB-4		13.5	3.50	1.30	1.00	13.5	2.80	(25)e
PI-TB-5		19.6	4.90	1.90	1.70	11.5	2.60	(25)e
Matrimi-d®		8.70	1.90	0.27	0.24	36.0	7.00	(35)
Ultem®		1.33	0.41	0.05	0.03	36.90	8.00	(4)
Extem®		3.28	0.81	0.13	0.13	25.20	6.20	(4)
Polysulf-one		5.6	1.2	0.18	0.18	22.0	6.00	(4)

1. Gas permeability coefficient (P) in Barrer. 1 Barrer=10⁻¹⁰ cm³ (STP) cm/cm² s cm Hg.

2. ^aThis study. ^bMeasured at 1 atm, 35°C. ^cMeasured at 1 bar, 35°C. ^dMeasured at 1 bar, 35°C. ^eMeasured at 1 atm, 35°C.

Figure 5:

The dependence of gas permeability vs. reciprocal of fractional free volume of polyamides for CO₂, O₂, N₂ and CH₄ gases.

The relative order of diffusion coefficient follows the trend: D (O₂)>D (CO₂)>D (N₂)>D (CH₄). D (O₂)>D (CO₂) cannot be explained by the kinetic diameter values of these two gases. However, this trend is in order of effective molecular diameter (d_ef ) of the gas molecules as reported by Teplyakov and Mears (39). As d_ef of O₂ (3.02 Å) is higher than CO₂ (2.89 Å), the diffusion coefficient of O₂ is higher. From Table 4, it is observed that the D values for every gases through the polymer membranes follow the order PA “I”>PA “II”>PA “III”>PA “IV”.

Solubility coefficient of the studied gases follow the order S (CO₂)>S (O₂)>S (N₂)>S (CH₄). The solubility of gases in a membrane is related to their critical temperatures (T_c), as condensability depends on T_c. Higher solubility of CO₂ into the polymer membranes reflected from the T_c (31.04°C) compares to other gases [T_c CH₄, N₂, O₂=–82.3, –146.9, –118.6°C]. The TB unit in the polymer backbone may account for the enhanced solubility of CO₂. TB contains two extra polar nitrogens (tertiary amine) that could interact strongly by Lewis acid-base interaction with CO₂ molecules having a string quadrupole moment (20). The polar character of the amide bond also influenced the solubility coefficient (12). Analysis of the data in Table 4 shows that permeability differences in the PA membranes mainly comes from solubility coefficient values, rather than diffusion coefficients. The presence of the polar group in the polymer chain enhances intermolecular interactions, results in a high solubility coefficient.

Analysis of the data presented in Table 3 exhibits great improvements in gas separation results, in particular, for the gas pair CO₂/CH₄ and O₂/N₂. The PAs showed improved selectivity as compared to the commercial Matrimid^® membrane (35). The ideal gas selectivity follows the following order PA “II”<PA“III”<PA“I”<PA“IV”. Except the membrane PA “I”, the “trade-off” relationships between P and α follows the same order. The presence of 6F group in PA “I” restricted the segmental motion more, which in turn increases the selectivity. The selectivity coefficients of CO₂/CH₄ and O₂/N₂ were in the range of 48.4–53.6 and 6.5–7.5. High selectivity for CO₂/CH₄ and O₂/N₂ gas pairs were due to amide bonds which introduce intramolecular rigidity by densifying the polymer chain (12).

Diffusivity and solubility coefficients for the PAs are shown in Table 4. As expected, compared to diffusivity values, solubility values are much higher. The gas diffusivity selectivities of uncondensable gas pairs (O₂/N₂) of PAs were in the range of 3.6–4.8, and its solubility.

Selectivites were in the range of 1.5–1.8. So, judging from the result for O₂/N₂ gas pairs permselectivity values, the main contribution factor could be diffusivity selectivity, as O₂ and N₂ gas molecule has less affinity with the polymer matrix. The increased selectivity of PA “IV” could be due to the restricted segmental motions. The gas diffusivity selectivities of (CO₂/CH₄) gas pairs of PAs were in the range of 4.09–4.48, and the solubility selectivites were in the range of 10.8–12.1. The analysis of the data reveals that permselectivity of (CO₂/CH₄) has mainly originated from solubility selectivites, as CO₂ has a strong interaction with the polymer matrix. The increment of the separation performances of (O₂/N₂) gas pairs indicates that separation of this gas pair depends on size selectivity based principle as they have small differences in kinetic diameters between the gas molecules.

Introduction of TB units into the polymer matrix significantly increases in selectivity with moderate permeability. The PAs under this investigation showed a decrease in permeability for all gases in comparison to the TB-bridged polyimides reported by Jin et al. (20). However, the present PAs showed an increase in selectivity for CO₂/CH₄ and O₂/N₂ gas pairs. According to Jian et. al. permeability of TB polymers mostly comes from solubility (S), as it is directly related to the polymer structure (22). The experimental data from Table 4 also indicate the same result. In CO₂/CH₄ separation, solubility selectivity mainly dominated due to the Lewis acid-base interaction between the CO₂ molecule and the TB unit. McKeown et al. also reported the same behavior for the TB polymer. They described selectivity being primarily due to the much higher solubility of CO₂ (solubility selctivity) rather than selectivity based on diffusivity (21).

Matsumoto et al. (40) and Miyata et al. (41) reported a linear relationship between the dielectric constant (ε) and CO₂ and CH₄ permeabilities for several polymers. Miyata et al. also showed a linear relationship between permeability and the diffusion coefficient with ε in 6FDA based polyimides (41). The PAs under this investigation showed a linear relationship between P and ε (Figure 6). From the Table 4, it can be seen that the ε of the PAs varied from 2.07 to 2.39 in 1 MHz. The order of ε was PA “I”<PA “II”<PA “III”<PA “IV”. PA “I” containing [>C(CF₃)] showed the lowest ε value due to highest fluorine content (13).

Figure 6:

The dependence of gas permeability vs. dielectric constant of the polyamide membranes for CO₂, O₂, N₂ and CH₄ gases.

3.5 Comparison with other commercially available polymers and previously reported polymer membranes

The gas separation performance of the PAs is plotted graphically in Figures 7 and 8, respectively. In the graph, permselectivity for the important gas pairs, CO₂/CH₄, O₂/N₂ plotted as ordinate, and P(CO₂), P(O₂) plotted as abscissa. The solid line in the graph represents the Robeson’s 2008 upper bound. In the permeability-selectivity plots, the data points for all the PAs in this study along with some recently reported structurally similar polymers (16), (20), (24), (25), (42) and a few typical commercial membranes (e.g. Matrimid^®, Ultem^®, Extem^® and Polysulfone) are included for better comparison. These plots can be used as a universal performance indicator for the separation of two gases. It can be seen from these plots that the PAs in this investigation showed an improvement in gas separation performance compared to many commercial reported polymers. The results further indicate that methyl substitution and amide bonds in the polymer backbone effectively improved the selectivity of the resulting polymer membranes (25). The performance of these membranes for gas separation properties fall in the attractive region of Robeson’s upper bound. The permeability/selectivity trade-off data of the membrane for the gas pairs exceed those conventional polymers, and outperformed other reported polymers (24).

Figure 7:

Relationship between gas permeability (P) and gas pairs selectivity (α) for O₂/N₂ separation in Robeson upper bound plots.

Data points of structurally similar polymers and other commercially available polymers also reported.

Figure 8:

Relationship between gas permeability (P) and gas pairs selectivity (α) for CO₂/CH₄ separation in Robeson upper bound plots.

Data points of structurally similar polymers and other commercially available polymers also reported.