Effect of membrane-casting parameters on the microstructure and gas permeation of carbon membranes

Abstract

In this work, membrane-casting parameters including solvents and drying methods were investigated to adjust the microstructure and gas permeation of poly(phthalazinone ether sulfone ketone) (PPESK) based carbon membranes. The structure and properties of the membrane samples were characterized by Fourier transform infrared spectroscopy, X-ray photoelectron spectroscopy, X-ray diffraction, transmittance electron microscopy, single gas and mixed gas permeation. Results have shown that the membrane-casting parameters greatly influence the physico-chemical properties of the polymeric membranes, so as to affect the structure and gas permeation of their derived carbon membranes. Under the same drying conditions, the selection of a solvent with a high boiling point is beneficial to the thermal degradation of the polymeric membrane during pyrolysis. In addition, the adoption of a solvent with a close solubility parameter to PPESK is favorable to improving the permeability of carbon membranes. Compared to common warm air drying, cold drying is more favorable for the improvement of the thermal stability of the precursor membranes. With variation of the pyrolysis temperature from 650 °C to 850 °C, the best selectivities of carbon membranes are obtained at 850 °C for refrigerate-drying and at 750 °C for freeze-drying, respectively. All the gas separation data of the present carbon membranes made by cold drying surpass the Robeson's upper bound.

---

1. Introduction

During the last three decades, the concerns of environmental pollution and energy dissipation have trigged an increased interest in the study of carbon molecular sieve membranes (or carbon membranes for short, CMs).^1–4 CMs offer prominent gas separation performance, excellent thermal and chemical stability, with a large variety of promising applications, including removal of sour gas from natural gas, separation of CO₂ from a mixture with CH₄ such as land fill gas, production of O₂ enriched air, separation of light hydrocarbon C₃H₆/C₃H₈, separation of H₂ from CH₄, and so on.^5,6 However, the expensive fabrication costs, about 1–3 orders higher in magnitude than polymeric membranes, have prohibited industrial application of CMs. Therefore, it is imperative to improve the separation performance in order to compensate for the high cost. In the pursuit of increasing gas separation permeation and reproducibility, one must optimize all the fabrication procedure parameters based on systematic experimental results and the underlying rules of the microstructure-performance relationship of membrane materials.^7–9 Although great effort has been paid to the precursor selection, molecular structure modification, pyrolysis conditions, precursor composition, post treatment, etc., in the preparation of CMs,^6,10,11 membrane-casting parameters are usually ignored by researchers. To date, only a few works can be found that study the influence of membrane-casting on the CMs morphology. Tin et al.^12,13 adjusted the gas permeation of CMs using nonsolvent treatment. Shao et al.¹⁴ found that the variation of solvent types of polymeric membranes could tailor the amorphous or crystalline morphology of their derived CMs. In addition, it has been proved that the drying method affects the structure and gas permeation of polymeric membranes.^15–17 Due to the descendent relationship between polymeric precursor and its derived CMs,^18–20 the variation of drying method can definitely tune the structure and properties of CMs by inheriting the original carbonaceous framework of polymeric molecular structure after pyrolysis. However, there is no report concentrated on the effect of drying protocols during the membrane-casting of CMs to the best of our knowledge.

Previously, our work has indicated that poly(phthalazinone ether sulfone ketone) (PPESK) is an excellent precursor for CMs, from which prominent gas separation properties could be achieved for permanent gases by the optimization of molecular structure,^21,22 preoxidation temperature²³ and filler incorporation.²⁴ Here, the membrane-casting parameters, i.e., solvents and drying methods, will be undertaken to investigate the evolution of structure and gas permeation of PPESK-based CMs during pyrolysis.

2. Experimental

2.1. Materials

The precursor PPESK copolymer with a sulfone-to-ketone ratio of 1 : 1 was provided by Dalian Polymer New Materials Co. Ltd (PR China). The schematic molecular structure can be found in our previous report.²¹ According to the principle of “like dissolves like”, several solvents with the solubility parameters approaching to that of PPESK 20.15 (J mL⁻¹)^1/2 were utilized, including N-methyl-pyrrolidone (NMP), dimethylacetamide (DMAc), chloroform (CHCl₃) and acetylene tetrachloride (C₂H₂Cl₄).

2.2. Membrane preparation

A series of membrane solutions were formed by dissolving PPESK into NMP, DMAc, CHCl₃ and C₂H₂Cl₄, respectively. The typical membrane-forming process was performed by casting the membrane solution on a glass plate with doctor blade at 40 °C in a dust-free room beneath the relative humidity of 40%. Fresh cast membranes were first heated at 60 °C for 12 h, followed by warm air drying at 80 °C for 12 h.

On the basis of the viscosity of membrane solutions aiming at define membrane-casting subsequently, appropriate concentrations of PPEK-solution were set at 15% for NMP, 20% for DMAc, 8% for CHCl₃ and 8% for C₂H₂Cl₄. In addition, drying methods of polymeric membranes made by CHCl₃ were also investigated by cold drying (i.e., freeze at −20 °C or refrigerate at 10 °C) for 24 h. The as-obtained polymeric membranes were labelled with PM-NMP, PM-C₂H₂Cl₄, PM-DMAc, PM-Fre and PM-Ref, correspondingly referring to their used solvents or drying method, NMP, C₂H₂Cl₄, DMAc, freeze-drying and refrigerate-drying.

Prior to pyrolysis, the polymeric membranes were thermostabilized at 460 °C for 0.5 h with 200 mL min⁻¹ purge air and a heating rate of 2 °C min⁻¹. The stabilized membranes were denoted as SM-NMP, SM-C₂H₂Cl₄, SM-DMAc, SM-Fre and SM-Ref, reflecting their original polymeric membranes. Afterwards, the stabilized membranes were pyrolyzed at 650 °C for 1 h in a tubular furnace with the sweeping of 200 mL min⁻¹ flowing ultra-purified argon at a heating rate of 1 °C min⁻¹ from ambient temperature. Once the furnace cooled to room temperature, the resultant carbon membranes were collected and preserved in a desiccator to avoid any further possible effect from the air. Likewise, the carbon membranes were also designated in the same manner as stabilized membranes, i.e., CM-NMP, CM-C₂H₂Cl₄, CM-DMAc, CM-Fre and CM-Ref.

Pyrolysis at temperatures of 750 °C and 850 °C were also conducted for CMs made from solvent CHCl₃. For convenience, the as-prepared CMs were denoted as CMf-t or CMr-t, where the letters “f (or r)” and “t” correspondingly refer to drying methods (i.e., f = freeze, r = refrigerate) and pyrolysis temperature (t °C). It notices that some labels are actually referred to the same samples, i.e., CM-Fre = CMf-650, CM-Ref = CMr-650.

2.3. Gas permeation measurement

The ideal permeation property of five single-component gases (i.e., O₂, N₂, CO₂, CH₄ and H₂) for CMs was determined by a conventional variable volume-constant pressure method at 30 °C. The ultra-purified single gases (>99.999%) were supplied from gas cylinders. The depiction of measurement setup and test procedure has been given in our previous report in detail.^21,24 For each measurement, at least three samples were taken to assure the accuracy. The reported permeation data are the averaged value with a measurement precision of 10%. For the details of the test procedure, please refer to our previous report.²⁵

The realistic separation performance of three gas pairs, i.e., O₂/N₂ (21/79 mol%), CO₂/N₂ (50/50 mol%) and CO₂/CH₄ (50/50 mol%), was also evaluated by analyzing the component fractions of permeated gases through a gas chromatograph (Techcomp LTD, Model 7890) equipped with a molecular sieve 5 Å column (3 m × 3 mm I.D.) and a thermal conductivity detector under the carrier gas of argon.

The permeability for a permeated gas (both single gas and mixed gas) was calculated by the following eqn (1).

where P, F, A, L and ΔP refer to the permeability in the unit of Barrer (1 Barrer = 10⁻¹⁰ cm³ (STP) cm cm⁻² s⁻¹ cmHg), the gas permeating flux through membrane, effective membrane area, membrane thickness and trans-membrane pressure differential, respectively.

The ideal selectivity for two single gases “i” and “j” was obtained by the ratio of their permeabilities (eqn (2)).

The realistic selectivity for a two-component gas mixture composed by component “i” and “j”, was obtained by the ratio of their mole fractions in permeate stream (y) and feed stream (z) (eqn (3)).

2.4. Characterization methods

The thermal stability of precursors was studied using a TGA/SDTA851e thermogravimetric analyzer (TGA, Mettler-Toledo, Switzerland) in flowing nitrogen at 15 °C min⁻¹ from ambient temperature to 800 °C.

The functional groups of CMs were analyzed by a 20DXB Fourier transform infrared spectrometer (FT-IR) by KBr pellet method in the wavenumber range of 400–4000 cm⁻¹.

The microstructure of CMs was observed by a Philips Tecanai GZ²⁰ transmission electron microscopy (TEM), at the operation voltage of 200 kV.

The element species in membrane samples were detected by X-ray photoelectron spectroscopy (XPS) with a Perkin-Elmer PhI-5300 spectrometer (MgKα radiation, operated at 12.5 kV and 250 W, the incident angle 45°). For each measurement run, the background pressure in the analysis chamber was maintained at 1.3 × 10⁻⁸ Pa.

Elemental contents of selected samples were measured using a GmbH VarioEL Ш element analyzer (Germany). Atomic percentages of element carbon, hydrogen, nitrogen, oxygen and sulfur were determined by normalization.

X-ray diffraction (XRD) patterns were recorded for samples on a D/Max-2400 X-ray diffraction spectrometer (CuKa radiation with a wavelength of 1.54 Å, operated at 40 kV and 100 mA) in the 2theta range of 5–60°. The micro-structural parameters for interlayer distance (d₀₀₂) and micro-domain thickness (L_c) could be obtained from the well-known Bragg's equation and Scherrer equation, respectively.

3. Results and discussion

3.1. Membrane appearance

The dimensional shrinkage and flexibility level of carbon membranes with the correlation of their solvents are listed in Table 1. The shrinkage of membrane thickness and area of polymeric membranes after pyrolysis varies in the range of 3.56–7.47% and 23.77–44.18%, respectively. Both of the two shrinkages follow the order of PM-C₂H₂Cl₄ > PM-DMAc > PM-NMP, which is in the opposite order of the boiling point of used solvents. The largest dimensional shrinkage for PM-C₂H₂Cl₄ could be ascribed to its largest porosity and relative membrane volume resulted from the lowest boiling point of C₂H₂Cl₄ in comparison to DMAc and NMP, as has been found by Matsuyama, et al.¹⁵ This is not the case for membranes made by cold drying. Compared with common warm air drying, cold drying tends to bring about smaller area shrinkage, higher Td₅ value and char yield. The reason probably lies in the fact that the low boiling point (i.e., 61 °C) of solvent CHCl₃ is prone to yield high degree of drying and low equilibrium vapor pressure. Likewise, the slightly larger shrinkage of PM-Ref than that of PM-Fre after pyrolysis is owing to the difference in the amount of residual solvent. In addition, carbon membranes cast from CHCl₃ present better flexibility when handling. In fact, rare work can be found on flexibility of CMs and further research is needed to analyze the exact reason in depth. In summary, the solvent and drying method have a dramatic impact on the apparent features of polymeric membranes and their derived carbon membranes.

Table 1 Physical property of solvents and polymeric membranes

Precursor code	Solvent properties					After pyrolysis at 650 °C			Thermal stability of precursors
	Molecular weight	Density (g cm^−3)	Boiling point (°C)	Solubility parameter, Sp, (J mL^−1)^1/2	Sp differential between solvent to PPESK, ΔSp, (J mL^−1)^1/2	Area shrinkage (%)	Thickness shrinkage (%)	Flexibility	Td5 (°C)	Char yield (%)
PM-DMAc	87	0.937	165	22.2	2.05	38.76	3.84	Well	177.3	49.18
PM-NMP	99	1.026	203	20.3	0.15	38.32	3.80	Best	179.6	52.51
PM-C2H2Cl4	168	1.587	146	21.3	1.15	44.18	7.47	Brittle	157.1	43.95
PM-Fre	119	1.500	61	19.3	0.85	37.35	3.56	Best	213.3	53.46
PM-Ref						23.77	5.51	Well	208.8	53.70

---

3.2. Thermal stability of precursor

The thermal decomposition temperature at 5% weight loss (Td₅) and char yield at pyrolysis temperature of 650 °C were used as the criteria in determining the thermal stability of precursors. As listed in Table 1, the Td₅ values and char yields of precursor membranes are respectively in the range of 157.1–213.3 °C and 43.95–53.70%. Those values are comparable to thermally stable polymer polyimide,^25,26 demonstrating the high-temperature-resistant property for PPESK. The Td₅ values of present precursor membranes follow the order of PM-C₂H₂Cl₄ < PM-DMAc < PM-NMP < PM-Ref < PM-Fre. The char yields of those samples are in the order of PM-C₂H₂Cl₄ < PM-DMAc < PM-NMP < PM-Fre < PM-Ref. The two orders are in agreement with the sequence of boiling points of solvents. The reason is that higher boiling point of solvent could yield higher amount of residual solvent in membranes, which would bring about plasticization and result into larger continuous weight loss around the glass transition temperature of membranes.²⁷ In comparison with common warm air drying, cold drying tends to give rise to higher char yields and Td₅ values, implying the higher thermal stability of the two membranes (PM-Fre and PM-Ref). That is ascribed to the less solvent remaining and more compact membrane structure as discussed in previous dimensional shrinkage section. There is a referenced empirical rule that more compact carbon structure usually comes into the higher selectivity of CMs. Therefore, in conjunction with the concerns of potentially good separation performance and membrane flexibility, CHCl₃ is regarded as one of the most suitable solvents to fabricate PPESK-based CMs under cold drying method. However, it also needs to consider the appropriate cost and safety concerns of solvent for industrial application.

3.3. Evolution of elemental compositions

In order to insight into the solvent effect on microstructure and surface property, the elemental compositions of membrane samples were analysed in the following.

Fig. 1 gives the full scanning XPS spectra of polymeric membranes, stabilized membranes and CMs made by solvent DMAc and C₂H₂Cl₄. It shows the existence of the primary elements, i.e., carbon, nitrogen, oxygen, sulfur and chlorine, except for hydrogen. For comparison, the elemental contents could also be obtained by normalizing with their relative peak intensities from XPS spectra. It finds that the contents of oxygen and nitrogen for PM-DMAc and PM-C₂H₂Cl₄ vary in the same trend through the evolution from polymeric membranes, to stabilized membranes and CMs. Whereas, the contents of oxygen and sulfur first increase from polymeric to stabilized membranes, and then drop after pyrolysis. The increase of oxygen content is due to the formation of cross-linking network structure after stabilization, during which oxygen is introduced into the molecular chains from the air. The forward reduction of oxygen content is due to the removal of oxygen-containing functional groups from membrane matrix during thermal degradation reactions. The nitrogen content almost increases monotonously throughout the entire heat treatment process. Although there are somewhat content differences for sulfur and oxygen between the detections from XPS (Fig. 1a and b) and elemental analysis (Table 2) due to their different inspective principles and benchmarks, their overall change trends are rational in the same for the contents of carbon and nitrogen.

Fig. 1 XPS spectra of the original polymeric, stabilized and carbon membranes prepared with solvent (a) DMAc and (b) C₂H₂Cl₄.

Table 2 Elemental composition of stabilized membranes and carbon membranes

Sample	Elemental content (%)
	C	H	N	S	O
SM-NMP	73.69	2.78	6.83	1.89	14.81
SM-CHCl3	72.05	3.14	6.53	1.98	16.30
SM-C2H2Cl4	71.88	2.97	6.53	2.11	16.51
SM-DMAc	72.60	3.12	6.67	1.92	15.70
CM-NMP	84.40	3.10	5.23	0.53	6.75
CM-CHCl3	83.30	2.70	5.30	0.50	8.20
CM-C2H2Cl4	82.66	3.24	5.46	0.45	8.19
CM-DMAc	84.41	3.21	5.20	0.54	6.65

---

As listed in Table 2, the elemental compositions of carbon, hydrogen, nitrogen, sulfur and oxygen are roughly close to each other among different stabilized membranes and different CMs. For stabilized membranes, SM-NMP and SM-DMAc have slightly higher carbon and nitrogen contents along with lower sulfur and oxygen contents than the other systems. After pyrolysis, the carbon and hydrogen contents for all membrane samples remarkably increase, together with the reduction on nitrogen, sulfur and oxygen contents. These changes can be assigned to the thermal degradation of precursors, resulting into the release of gases or volatiles such as CO, CO₂, SO₂, etc. It is in good accordance with the result of previous TGA. Among the carbon membranes, CM-NMP and CM-DMAc possess higher carbon content and lower nitrogen content, while CM-CHCl₃ and CM-C₂H₂Cl₄ possess higher oxygen content. The above results justify the significant effects of solvent on the element contents and their changes in membranes.

3.4. Structure analysis

Fig. 2 shows the XRD patterns of CMs in this work to study the solvent effect. Two broad peaks can be found around ∼25° and 43°, reflecting the corresponding (002) and (100) plane diffractions of carbon materials. The interlayer distance d₀₀₂ values are 0.379 nm for CM-NMP, 0.392 nm for CM-C₂H₂Cl₄ and 0.413 nm for CM-DMAc. We can see that the d₀₀₂ values are in the same order of the intervals between the solubility parameters of solvents and PPESK (ΔS_p), DMAc > C₂H₂Cl₄ > NMP. The reason is that smaller ΔS_p between precursor and its solvent tends to form a more crystalline structure in resultant CMs.¹⁴ Because of the relationship between the d₀₀₂ values and the microstructure (e.g., degree of crystallinity or graphitization) for carbon materials, careful selection of solvent could be expectantly produce desired microstructure and gas separation performance of final PPESK-based CMs.

Fig. 2 XRD patterns of carbon membranes prepared at 650 °C by different solvents.

Fig. 3 presents the XRD patterns for CMs made by cold drying method as the function of pyrolysis temperature. In conjunction with Fig. 2, it shows that the sequence of d₀₀₂ values for CMs prepared at 650 °C is CMr-650 > CM-DMAc > CMf-650 > CM-C₂H₂Cl₄ > CM-NMP, along with approaching L_c values. It suggests that cold drying method is generally favourable for retarding the graphitization progress of CMs during pyrolysis. Based on the fact that the present d₀₀₂ values are much higher than perfect graphite (3.35 Å), it elucidates the amorphous carbon structure of all the CMs in this work.

Fig. 3 XRD patterns of carbon membranes made from (a) freeze-drying and (b) refrigerate-drying, with variable pyrolysis temperatures.

With the pyrolysis temperature elevating from 650 °C to 850 °C, the d₀₀₂ values of CMs prepared by freeze-drying first increase then decrease as shown in Fig. 3a. The largest d₀₀₂ value at the pyrolysis temperature of 750 °C reflects the less compact microstructure of CMf-750. Furthermore, it suggests that the thermal reactions are dominated by thermal decomposition at the early stage of pyrolysis (i.e., beneath 750 °C), and then by thermally aromatic polycondensation at the terminal stage. As a result of the structural change, the gas permeation of CMs would vary as well.

As for CMs by refrigerate-drying (Fig. 3b), the change tendency of the microstructure is obviously different from those by freeze-drying. With the pyrolysis temperature increasing from 650 °C to 850 °C, the d₀₀₂ value monotonously decreases from 0.439 nm to 0.379 nm, along with the sharp enhancement of the peak intensity at 43–45°. It implies that the microstructure of CMs becomes more compact with increasing pyrolysis temperature. Apart from that, the growth of microdomain size L_c also verifies this judgement. The difference between the CMs made by freeze-drying and refrigerate-drying is ascribed to the amount of residual solvent in precursor membranes. That is, freeze-drying tends to give higher content of residual solvent in polymeric membranes and less compact structure in CMs prepared at 650 °C and 750 °C. This is in good agreement of the previous result of TGA and appearance analysis. However, the porosity of CMs would be collapsed at extremely high pyrolysis temperature (e.g., 850 °C) due to the occurrence of thermal rearrangement reactions during pyrolysis, as has been found previously.^21,22 Consequently, the compactness relationship between the CMs made at 850 °C by the two cold drying methods is changed.

3.5. Functional groups on the membrane surface

Fig. 4 is the FTIR spectra of CMs prepared by freeze-drying and refrigerate-drying. From Fig. 4a, it can be found that the functional groups on membrane surface of CMf-750 have changed remarkably in comparison to those of CMf-650. Some peak intensities are enhanced, such as the stretching vibration of = C–H (3054 cm⁻¹) and C≡N (2224 cm⁻¹), stretching vibration of C=C in backbone (1601 cm⁻¹), deformation vibration of N–H (1447 cm⁻¹), stretching vibration of aromatic C–N (1324 cm⁻¹ and 1325 cm⁻¹). Moreover, some peaks representing the hydrogen substituted groups beneath 900 cm⁻¹ become more obvious. Meanwhile, several peaks disappear as the pyrolysis temperature is elevated from 650 °C to 750 °C, such as the bending vibration of O=S=O bond (1168 cm⁻¹ and 1115 cm⁻¹) and the stretching vibration of S–O bond (1060 cm⁻¹). However, the characteristic vibration peak for crosslinking bond at 750 cm⁻¹ is intensified for CMf-750. It illustrates that some characteristics of the rudimentary crosslinking structure is maintained although severed thermal degradation reactions have destroyed the sulfur-containing groups. This would be beneficial for the formation and development of porous structure in CMf-750. When the pyrolysis temperature further increases to 850 °C, only a few peaks can be identified on the membrane surface, i.e., N–H stretching vibration (3429 cm⁻¹), C=C backbone stretching vibration (1637 cm⁻¹), S–O stretching vibration (1073 cm⁻¹), together with a very weak vibration peak at 750 cm⁻¹. It suggests that most of the functional groups have been decomposed from the backbone of molecular chain, even the complete collapse of crosslinking structure. This analysis agrees well with the previous XRD result.

Fig. 4 FTIR spectra of carbon membranes made by (a) freeze-drying and (b) refrigerate-drying.

In Fig. 4b, it shows that the evolution of the chemical structure of refrigerate-drying-based membranes differs significantly from that of freeze-drying-based membranes with pyrolysis temperatures. Comparing the spectrum of CMr-750 with that of CMr-650, the N–H (3441 cm⁻¹) in Ar–NH–R groups and C=O (1620 cm⁻¹) in aromatic secondary amide remains intact, and the peaks of C–O–C (1050–1300 cm⁻¹) almost disappear. In addition, the peak reflecting the out-plane vibration of C–H (742 cm⁻¹) becomes more broad and weak. As the pyrolysis temperature increases to 850 °C, there are only a few peaks that can be made out in the spectrum, including N–H stretching vibration (3442 cm⁻¹), C=C backbone stretching vibration (1630 cm⁻¹), C–N stretching vibration (1384 cm⁻¹) and S–O stretching vibration (1110 cm⁻¹), and a very weak peak at 750 cm⁻¹. This implies that the crosslinking structure in the refrigerate-drying-based membranes is more stable than in freeze-drying-based membranes. Thus, it indicates that the functional groups on membrane surface and microstructure could be effectively adjusted by varying the drying method and pyrolysis temperature.

3.6. Morphology observation

Fig. 5 shows the TEM images of CMs obtained from cold drying methods. It can be clearly seen that the two samples exhibit similar amorphous wormy microstructure but distinctly different morphological configuration and microdomain size. This supports the conclusion drawn from XRD on the structural difference between those samples. Meanwhile, it indicates the divergence of drying method in controlling the microstructure of the two CMs. Consequently, the gas permeation of CMs could be effectively tuned with variable drying methods.

Fig. 5 Transmission electron microscopy inserted with the locally magnified images of carbon membranes made by (a) freeze-drying and (b) refrigerate-drying.

3.7. Dependence of solvents

Fig. 6a and b, respectively, give the plots of single gas permeation and ideal selectivity for the as-obtained CMs.

Fig. 6 Single gas permeation of carbon membranes. (a) Gas permeability, and (b) ideal selectivity.

All the gas permeabilities follow the order of H₂ > CO₂ > O₂ > N₂ > CH₄, which is the opposite order of their kinetic molecular diameters. It demonstrates that gas permeates through the present CMs via the molecular sieving mechanism. Moreover, the gas permeabilities for CMs prepared from different solvents follow the order of CM-NMP < CM-C₂H₂Cl₄ < CM-DMAc. This result agrees well with our previous structural analysis. The reason is that more inferior crystallite or graphitization degree in microstructure of CMs gives higher gas permeability.²⁸

For mixed gases (Fig. 7), the gas permeability of CMs presents a similar trend to the pure gases. Among the five CMs, CM-DMAc has the highest O₂, CO₂, N₂ and CH₄ permeability. The reason is that CM-DMAc possesses the lowest compactness degree as has been indicated by the d₀₀₂ values from XRD analysis. Meanwhile, the selectivities for all of the present CMs follow the order of O₂/N₂ < CO₂/N₂ < CO₂/CH₄ due to their molecular sieving mechanism during gas permeation. In addition, CM-NMP has the highest selectivity for O₂/N₂ and CO₂/N₂ gas pairs among the samples. On the whole, CM-Ref and CM-Fre possess moderate permeability, and O₂/N₂ and CO₂/N₂ selectivities.

Fig. 7 Mixed gas separation performance of carbon membranes for (a) O₂/N₂, (b) CO₂/N₂ and (c) CO₂/CH₄.

3.8. Dependence of drying method with pyrolysis temperature

Fig. 8 shows the gas permeation as the function of pyrolysis temperature for CMs made by cold drying method in the well-known Robeson's plots. It can be seen that all of the gas separation data of the as-made CMs have surpassed the Robeson's 1991 upper bound. In particular, it is noteworthy that the gas permeation of present CMs is more attractive for exceeding the Robeson's 2008 upper bound of O₂/N₂ and CO₂/CH₄ systems. Therefore, the as-prepared CMs are very promising in the separation fields of O₂/N₂ and CO₂/CH₄.

Fig. 8 Gas permeation of carbon membranes made by cold drying methods for (a) O₂/N₂, (b) CO₂/N₂ and (c) CO₂/CH₄.

When the pyrolysis temperature increases from 650 °C to 750 °C, the permeability of single gases elevates at the loss of selectivity for freeze-drying-based CMs. In return, the permeability decreases and selectivity increases to a certain degree as the pyrolysis temperature further increases from 750 °C to 850 °C. This agrees well with the evolution of microstructure in CMs from the analysis of TGA and FTIR, namely, the first creation of porosity as the pyrolysis temperature ranging from 650 °C to 750 °C then the shrinkage of pores ranging from 750 °C to 850 °C due to thermal condensation reactions. In comparison with single gas permeation, the permeation of mixed gases through present CMs shows slightly lower permeability and higher selectivity due to the competitive effect, as has been found in literature.^25,26

In Fig. 8, it also shows that both single and mixed gas permeability of CMs by refrigerate-drying monotonously decreases along with the improvement of selectivity (e.g., the O₂/N₂ selectivity ranging from 4.6 to 10.3) as the pyrolysis temperature increases. For CMr-850, it could not be detectable for the trace permeability of CH₄. The reason is that the cease point of thermal decomposition locates about 650 °C, and thermal condensation starts to play the dominant role until to 850 °C, resulting into the structure becoming more compact.

In summary, it concludes that the difference in microstructure and gas permeation for CMs made by the two cold drying methods stems from their initial evaporation and solidification of polymeric membranes during membrane-casting and the subsequent inheritance to their derived stabilized membranes and CMs.

4. Conclusions

On the basis of above investigation on membrane-casting process parameters for PPESK-based carbon membranes, the main conclusions are as follows:

The boiling point of solvent is tightly connected with the shrinkage of polymeric membranes after pyrolysis. Increasing the boiling point of the solvent is favourable for the reduction of the membrane shrinkage and the enhancement of thermal stability. It will intensify the textual porosity and gas permeability with the solubility parameters of solvent approaching to that of PPESK by varying solvent types. The adoption of cold drying, including freeze-drying and refrigerate-drying, is preferential for elevating the thermal stability of PPESK membranes and reducing the dimensional shrinkage after pyrolysis. Of the two cold methods, refrigerate-drying is more favourable to improve the gas permeability. Last, the O₂/N₂ selectivity of carbon membranes prepared by refrigerate-drying increases from 4.6 to 10.3 with the pyrolysis temperature increasing from 650 °C to 850 °C. The highest oxygen permeabilities of 270 Barrer (single gas) and 257 Barrer (O₂/N₂ mixture) are achieved for carbon membranes prepared by freeze-drying at the pyrolysis temperature of 750 °C.

Acknowledgements

This work was financially supported by National Natural Science Foundation of China (No. 21176036, 21376037 and 21436009), National High-tech R&D Program (863 program, 2012AA03A611), the State Key Laboratory of Fine Chemicals of China (KF1107) and the Fundamental Research Funds for the Central Universities (DUT14RC(3)119).

References

1. W. J. Koros and R. Mahajan, Pushing the limits on possibilities for large scale gas separation: which strategies, J. Membr. Sci., 2000, 175, 181.
2. D. R. Paul, Creating new types of carbon-based membranes, Science, 2012, 335, 413.
3. X. Z. He and M. B. Hägg, Hollow fiber carbon membranes: from material to application, Chem. Eng. J., 2013, 215–216, 440.
4. S. C. Rodrigues, R. Whitley and A. Mendes, Preparation and characterization of carbon molecular sieve membranes based on resorcinol-formaldehyde resin, J. Membr. Sci., 2014, 459, 207.
5. K. M. Steel and W. J. Koros, Investigation of porosity of carbon materials and related effects on gas separation properties, Carbon, 2003, 41, 253.
6. S. M. Saufi and A. F. Ismail, Fabrication of carbon membranes for gas separation—a review, Carbon, 2004, 42, 241.
7. D. Q. Vu and W. J. Koros, High pressure CO₂/CH₄ separation using carbon molecular sieve hollow fiber membranes, Ind. Eng. Chem. Res., 2002, 41, 367.
8. A. F. Ismail and L. I. B. David, A review on the latest development of carbon membranes for gas separation, J. Membr. Sci., 2001, 193, 1.
9. K. M. Steel and W. J. Koros, An investigation of the effects of pyrolysis parameters on gas separation properties of carbon materials, Carbon, 2005, 43, 1843.
10. P. S. Tin, Y. Xiao and T.-S. Chung, Polyimide-carbonized membranes for gas separation: structural, composition, and morphological control of precursors, Sep. Purif. Rev., 2006, 35, 285.
11. W. N. W. Salleh, A. F. Ismail, T. Matsuura and M. S. Abdullah, Precursor selection and process conditions in the preparation of carbon membrane for gas separation: A review, Sep. Purif. Rev., 2011, 40, 261.
12. P. S. Tin, T.-S. Chung, S. Kawi and M. D. Guiver, Novel approaches to fabricate carbon molecular sieve membranes based on chemical modified and solvent treated polyimides, Microporous Mesoporous Mater., 2004, 73, 151.
13. P. S. Tin and T.-S. Chung, Advanced fabrication of carbon molecular sieve membranes by nonsolvent pretreatment of precursor polymers, Ind. Eng. Chem. Res., 2004, 43, 6476.
14. L. Shao, T.-S. Chung, G. Wensley, S. H. Goh and K. P. Pramoda, Casting solvent effects on morphologies, gas transport properties of a novel 6FDA/PMDA–TMMDA copolyimide membrane and its derived carbon membranes, J. Membr. Sci., 2004, 244, 77.
15. H. Matsuyama, M.-M. Kim and D. R. Lloyd, Effect of extraction and drying on the structure of microporous polyethylene membranes prepared via thermally induced phase separation, J. Membr. Sci., 2002, 204, 413.
16. Y. H. See-Toh, F. C. Ferreira and A. G. Livingston, The influence of membrane formation parameters on the functional performance of organic solvent nanofiltration membranes, J. Membr. Sci., 2007, 299, 236.
17. G. Illing, K. Hellgardt, R. J. Wakemana and A. Jungbauer, Preparation and characterization of Polyaniline based membranes for gas separation, J. Membr. Sci., 2001, 184, 69.
18. X. He and M.-B. Hägg, Optimization of carbonization process for preparation of high performance hollow fiber carbon membranes, Ind. Eng. Chem. Res., 2011, 50, 8065.
19. W. N. W. Salleh and A. F. Ismail, Effects of carbonization heating rate on CO₂ separation of derived carbon membranes, Sep. Purif. Technol., 2012, 88, 174.
20. M. Kiyono, P. J. Williams and W. J. Koros, Effect of pyrolysis atmosphere on separation performance of carbon molecular sieve membranes, J. Membr. Sci., 2010, 359, 2.
21. B. Zhang, T. Wang, S. Zhang, J. Qiu and X. Jian, Preparation and characterization of carbon membranes made from poly(phthalazinone ether sulfone ketone), Carbon, 2006, 44, 2764.
22. T. Wang, B. Zhang, J. Qiu, Y. Wu, S. Zhang and Y. Cao, Effects of sulfone/ketone in poly(phthalazinone ether sulfone ketone) on the gas permeation of their derived carbon membranes, J. Membr. Sci., 2009, 330, 319.
23. B. Zhang, T. Wang, S. Liu, S. Zhang, J. Qiu, Z. Chen and H. Cheng, Structure and morphology of microporous carbon membrane materials derived from poly(phthalazinone ether sulfone ketone), Microporous Mesoporous Mater., 2006, 96, 79.
24. B. Zhang, T. Wang, Y. Wu, Q. Liu, S. Liu, S. Zhang and J. Qiu, Preparation and gas permeation of composite carbon membranes from poly(phthalazinone ether sulfone ketone), Sep. Purif. Technol., 2008, 60, 259.
25. B. Zhang, Y. Shi, Y. Wu, T. Wang and J. Qiu, Towards the preparation of ordered mesoporous carbon/carbon composite membranes for gas separation, Sep. Sci. Technol., 2014, 49, 171.
26. B. Zhang, Y. Wu, Y. Lu, T. Wang, X. Jian and J. Qiu, Preparation and characterization of carbon and carbon/zeolite membranes from ODPA–ODA type polyetherimide, J. Membr. Sci., 2015, 474, 114.
27. R. Guan, H. Dai, C. Li, J. Liu and J. Xu, Effect of casting solvent on the morphology and performance of sulfonated polyethersulfone membranes, J. Membr. Sci., 2006, 277, 148.
28. D. X. Luu, E.-B. Cho, O. H. Han and D. Kim, SAXS and NMR analysis for the cast solvent effect on SPEEK membrane properties, J. Phys. Chem. B, 2009, 113, 10072.

---