Controlled Carbonization Heating Rate for Enhancing CO2 Separation Based on Single Gas Studies

Concerns about the impact of greenhouse gas have driven the development of new separation technology to meet CO2 emission reduction targets. Membrane-based technologies using carbon membranes that are able to separate CO2 efficiently appears to be a competitive method. This research was focused on the development of carbon membranes derived from polymer blend of polyetherimide and polyethylene glycol to separate CO2 rendering it suitable to be used in many applications such as landfill gas purification, CO2 removal from natural gas or flue gas streams. Carbonization process was conducted at temperature of 923 K and 2 h of soaking time. To enhance membrane separation properties, pore structure was tailored by varying the carbonization heating rates to 1, 3, 5, and 7 K / min. The effect of carbonization heating rate on the separation performance was investigated by single gas permeabilities using CO2 , N2 , and CH4 at room temperature. Carbonization heating rate of 1 K / min produced carbon membrane with the most CO2 / N2 and CO2 / CH4 selectivity of 38 and 64, respectively, with the CO2 permeability of 211 barrer. Therefore, carbonization needs to be carried out at sufficiently slow heating rates to avoid significant loss of selectivity of the derived carbon membranes.


Introduction
Concerns about the impact of global warming and climate change have triggered global efforts to reduce the CO 2 concentration in the atmosphere. Increasing concerns on this issue have led to numerous attempts and development of advanced technologies to solve the problem of excessive CO 2 emitted to the atmosphere. Carbon capture and storage is the most significant technology to decrease CO 2 emission to the atmosphere, and involved separation of CO 2 , transportation, and storage [1]. Among the CO 2 separation technologies, membrane-based technology is one of the most promising techniques which offers many advantages such as low capital investment, high process flexibility, and easy operation as well as compact equipment [2]. Baker [3] estimated that the market scale of the membrane-based gas separation will grow from US$ 150 million in 2002 to around US$ 760 million in year 2020. It is expected that the membrane-based gas separation plays an increasingly important role in reducing the global environmental impact and operating costs.
Currently, the dominant membranes used in the industrial gas separation are polymeric membranes. The polymeric membranes are inexpensive and easier to manufacture into large-scale modules [4]. However, at high pressures and temperatures as well as in aggressive environments, polymeric membranes undergo plasticization and compaction, which dramatically reduce membrane separation capabilities and cause irreparable damaged [5]. Carbon membrane with high thermal stability and superior chemical resistance in corrosive environments has great potential to overcome the disadvantages of polymeric membranes [6]. Carbon membrane exhibits excellent separation for gas mixtures even between gases with almost similar molecular size [7]. Carbon membrane can be prepared by controlled carbonization process of the various natural and synthetic precursors such as coal [8], polyfurfuryl alcohol (PFA) [9], polyimide [10,11], polyetherimide [12], and cellulose acetate [13]. Carbon membrane has a unique microporous structure, which allows it to discriminate gas molecules by size and shape [14]. The pore structure of the carbon membrane can be tailored to suit the separation of the particular gas mixtures by fine-tuning the preparation steps. Several factors including precursor selection, membrane preparation, carbonization conditions, and pre-/ post treatment conditions are well recognized as important factors influenced the separation performance of carbon membranes [15,16].
Heating rate is one of the carbonization conditions that is reported to influence the structure and properties of the derived carbon membrane [14]. Carbonization process is typically carried out at heating rate ranging from 1 to 13 K / min. Suda and Haraya [17] studied the effect of carbonization heating rate on the carbon membranes derived from the Kapton polyimide. Gas permeability of the carbon membranes increased with the increase of carbonization heating rates. Moreover, studies conducted by Sazali et al. [11] and Salleh and Ismail [18] showed that carbon membranes produced at lower heating rates produced carbon membranes with better performance than those obtained at higher heating rates.
The objective of this study was to investigate the effect of carbonization heating rates towards the morphology structure, microstructure, and gas separation performance of the polyetherimide (PEI) and polyethylene glycol (PEG) blended (PEIPEG) carbon membranes. Polymer blend solution consisting of PEI and PEG was dip coated onto the porous alumina support. The high molecular weight PEG which is a thermally labile polymer was chosen to blend with PEI to increase the diffusion pathways by the increase of total pore volume [19]. The membrane was characterized by scanning electron microscopy (SEM), X-ray diffraction (XRD), elemental analyzer (EA), and single gas permeation measurement. The distribution of membrane pore size was determined using N 2 adsorption-desorption via the Horvath-Kawazoe method.

Preparation of carbon membranes
Disk supported carbon membrane derived from polymer blend of PEI and PEG was prepared in this study. PEI was used as a main membrane precursor due to its superior strength and chemical resistance. PEI was blended with PEG to enhance the gas permeability as well as to improve the gas separation factor of the derived carbon membranes. Preparation of the alumina support and intermediate layer were already described in previous work [19]. The polymer solution was prepared by dissolving 2 g of PEI and 0.2 g of PEG in 18 g of N-methyl-2-pyrrolidone (NMP). The mixture was stirred at 353 K for 2 h to obtain a clear yellowish solution. The polymer solution was cooled down to room temperature before depositing on a support by a single dip-coating technique. The coated-support was immersed into the polymer solution for 10 s and carbonized in a horizontal tube furnace. Prior to carbonization, the sample was stabilized in a horizontal tube furnace at 573 K for 30 min. Stabilization process was done to promote the intermolecular cross-linking of the polymer chains and provide greater stability to sustain high temperature during carbonization [20]. Carbonization process was performed at the final carbonization temperature of 923 K and 2 h of soaking time with various heating rates (1, 3, 5, and 7 K / min). Finally, the carbon membranes were cooled down naturally to room temperature. The carbonization conditions are summarized in Table 1.

Gas permeability study
Separation performance of the derived carbon membranes was analyzed by single gas permeability study of CO 2 , N 2 , and CH 4 with kinetic diameter of 0.330 nm, 0.364 nm, and 0.380 nm, respectively. A single gas permeability test was conducted at an ambient temperature with feed pressure set at 2 bar using bubble soap flow meter. Fig. 1 shows the schematic diagram of the experimental set-up of the single gas permeability test. The supported carbon membrane was placed in the membrane permeation cell. Before each permeability test, both sides of the membrane permeation cell were evacuated to less than 0.1 bar by a mechanical vacuum pump. The feed gas was introduced at the upper side of the carbon membrane.
The permeability, P of a single gas through the carbon membrane was calculated using Eq. (1): where P is the permeability (cm 3 (STP) cm / cm 2 s cm Hg or barrer), Q is the volumetric flow rate of gas at standard temperature and pressure (cm 3 / s), l is the thickness of the membrane material (cm), Δp is the pressure difference across the membrane (cm Hg), and A is the membrane area (cm 2 ). The volumetric flowrate, Q of the permeate gas was calculated using Eq. (2): where V is the volume of permeate gas (cm 3 ) and t is the time (s). The pressure difference between the feed side and permeate side was kept constant. After the permeability value of each gas was obtained, the ideal selectivity, α, was calculated. The ideal selectivity for pure gas A to B, α A/B is defined as the ratio of the pure gas permeability of A, P A toward the pure gas permeability of B, P B and expressed in Eq. (3):

Membrane characterization
Surface and cross sectional morphologies of the carbon membrane were observed using scanning electron microscope (SEM) model Quanta FEG 450. Elemental analysis of the PEIPEG membrane before and after carbonization were determined using Perkin Elmer elemental analyzer model Series CHNS / O Analyzer 2400. X-ray powder diffraction (XRD) analysis of the carbon membrane was carried out using Philips X-ray diffractometer model PW 1820 with the 2q diffraction angle of 10-60° operated using Cu Ka radiation of 0.154 nm wavelength.
The d-spacing value can be calculated from Bragg equation as expressed in Eq. (4): where λ is the X-ray wavelength in nm, d is the dimension spacing in nm, θ is the diffraction angle (°), and n is an order that is equal to 1. Brunauer-Emmett-Teller (BET) surface area and pore size distribution of the carbon membrane was determined using Autosorb-1C (Quantachrome Instruments) at 77 K.
3 Results and discussion 3.1 Surface morphology SEM images of the top view and cross-sectional view of the carbon membranes derived at various carbonization heating rates are shown in Fig. 2 and Fig. 3  of thermal shrinkage on the carbon layer. It is believed that heat treatment process will affect physical structure arrangement of the membrane [11].

Microstructure analysis
In this study, XRD was used to analyze the microstructure and determine the interlayer distance (d-spacing) of the derived carbon membranes. The average d-spacing calculated using the Bragg equation does not represent the pore dimensions, but gives a measure of the distance between neighboring plane interlayer. This interlayer distance can be considered as a diffusional pathway for gas molecules through the carbon membranes [21,22]. Fig. 4 illustrates the XRD pattern of carbon membranes derived from various carbonization heating rates. All patterns showed broad peaks which indicated the amorphous structure of the carbon membranes.
Peaks were observed at 2q = 20° to 25° and another weak peak at 2q = 42° to 45° was attributed to (002) peak and (100) peak, respectively [23]. The (002) peak is derived from spacing between graphite sheets and (100) peak is related to the distance between carbon atoms within a graphite sheet [24]. As the carbonization heating rates increased from 1 to 7 K / min, the (002) peak shifted to smaller 2θ which resulted in higher d-spacing value. The (002)  increased with the increase of heating rate. At low heating rate, the carbonization process occurred slowly. The pressure built up from the process was small and thus produced carbon membrane with more ordered structure. At high carbonization heating rate, the process occurred rapidly and had caused imperfect alignment of graphitic structures and led to higher d-spacing. Table 2 summarizes percentage of elements contained in the PEIPEG precursor and PEIPEG-1 carbon membrane. After carbonization, the carbon element (C) in the PEIPEG carbon membrane increased about 9 % whereas the hydrogen (H), nitrogen (N), and oxygen (O) elements decreased about 1 -6 %. The H / C, N / C, and O / C ratios also decreased after carbonization. During carbonization, most of the heteroatoms in the polymer precursor are volatilized in the form of methane, hydrogen, hydrogen cyanide, water, carbon monoxide, carbon dioxide, ammonia, and various other gases [25]. The evolution of gas left behind a stiff and cross-linked carbon matrix [26]. Results showed that the PEIPEG precursor had evolved to higher carbon content material after carbonization process. The observed trend was similar to the PBI/PI carbon membrane reported by Hosseini and Chung [27], in which the C content was increased, while O and N contents were decreased after carbonization as compared to its precursor.

Pore size distribution
The N 2 adsorption-desorption exhibited that the PEIPEG-1 carbon membrane has S BET of 544.07 m 2 / g as reported in our previous study [19]. The PEIPEG-1 carbon membrane was further analyzed to investigate the pore size distribution (PSD). The PSD of the PEIPEG-1 carbon membrane as displayed in Fig. 5 was dispersed in the range of 0.3725 to 0.9125 nm and the peak of PSD centered at about 0.7975 nm. This indicated that the PEIPEG-1 carbon membrane was comprised of bimodal structure, which consists of micropores (< 2 nm) and ultramicropores (< 0.7 nm) [28].  These micropores and ultramicropores were created from imperfections between the microcrystalline regions of the polymeric material that are formed during the carbonization process [29]. The micropores contributed to the gas diffusion whereas the ultramicropores functioned for molecular sieving, which contributed to a highly permeable and highly selective carbon membrane [10]. A similar observation was obtained by carbon molecular sieve membranes prepared from novolac resin and boehmite [14].

Permeation properties
The gas permeabilities and ideal selectivities of the derived carbon membranes are tabulated in Table 3. As the carbonization heating rate increased from 1 to 3 K / min, the CO 2 , N 2 , and CH 4 permeabilites of the PEIPEG-1 and PEIPEG-3 were decreased by 84 %, 81 %, and 80 %, respectively. Reduction of the gas permeabilities were due to narrowing of the pore size distribution to smaller size with simultaneous densification of the carbon structure [18]. The gas permeabilities of both the PEIPEG-1 and PEIPEG-3 were in the order of CO 2 > N 2 > CH 4 , which indicated that the gas transport through the carbon membranes was controlled by a molecular sieving. Further increase in the carbonization heating rates to 5 and 7 K / min resulted in a significant increase of the gas permeabilities. Gas permeabilities of the PEGPEI-5 and PEGPEI-7 were in the order of CO 2 > CH 4 > N 2 indicating that the gas transport neither follows molecular sieving nor Knudsen diffusion. This behavior showed that the pores structure was irregular. At low heating rate, the development of pores was uniform because pressure built up from the release of volatiles was small. There was sufficient time for the volatiles to escape through the small pores to the surface before pressure can be built up [30]. On the other hand, at higher heating rate, irregular pores formed due to the excessive increase of pressure as a result of the accumulation of evolved gases in the carbon membranes [9].
The increase of carbonization heating rate from 1 to 7 K / min reduced the carbon membrane selectivities. Carbon membranes derived at carbonization heating rates of 1 and 3 K / min exhibited relatively high ideal selectivity compared to the carbon membranes derived at 5 and 7 K / min. The highest of the CO 2 / CH 4 and CO 2 / N 2 ideal selectivities were achieved by the PEIPEG-1 and determined to be 64 and 38, respectively. Low carbonization heating rate led to a mild release of the volatile compounds from the matrix and produced a uniform pore structure with small pore size in the carbon matrix [9]. Subsequently, the carbon membrane produced at low carbonization heating rate became more selective. However, further increase the carbonization heating rates to 5 and 7 K / min has reduced the ideal selectivities. The loss of the ideal selectivities were possibly due to the presence of larger pores in the membranes structure. As a whole, the PEIPEG carbon membranes produced at lower carbonization heating rates showed higher ideal selectivity compared to those derived at higher carbonization heating rates. The result is in good agreement with Suda and Haraya [17]. Fig. 6 represents the trade-off between the CO 2 permeability versus the CO 2 / CH 4 , and CO 2 / N 2 ideal selectivities of the PEIPEG carbon membranes derived at various carbonization heating rates together with the Robeson upper bound [31]. The PEIPEG-1 has exceeded the upper bound for the CO 2 / CH 4 separation. For the CO 2 / N 2 separation, the PEIPEG-1 nearly approached the upper bound line. This behavior shows a potential of PEIPEG carbon membranes for gas separation application.

Conclusion
Four different PEIPEG carbon membranes were successfully derived by carbonization of polymer blend precursors of PEI and PEG at final carbonization temperature of 923 K and 2 h of soaking time. Characterization analysis showed that the morphology and microstructure of the derived PEIPEG carbon membranes were affected  by the carbonization heating rates. The PEIPEG carbon membranes derived at 1 K / min showed the highest of CO 2 / CH 4 and CO 2 / N 2 ideal selectivity, which were found to be 64 and 38 with CO 2 permeability of 211 barrer, respectively. From the study, it can be concluded that the carbonization process must be conducted at a sufficiently slow heating rate to avoid the formation of defects, which can cause a significant loss of the selectivity of the derived PEIPEG carbon membranes. The derived PEIPEG carbon membranes showed great potential to separate CO 2 , rendering it suitable to be used in many applications such as CO 2 removal from natural gas or flue gas streams. As a whole, this study shows that the carbon membrane structure can be engineered by controlling the carbonization parameters depending on its applications in the industry.