Rheological evaluation of the fabrication parameters of cellulose acetate butyrate membrane on CO2/N2 separation performance

The rise in emission of greenhouse gases (GHGs) mainly carbon dioxide (CO2) in recent years due to rapid development of modern civilisation, has been listed as the primary contributor to global warming. To address this global issue, membrane technology was applied and developed intensively because of its superior performance in terms of efficiency and economic advantages. In this study, the cellulose acetate butyrate (CAB) polymer was selected as the polymer matrix material since it exhibited excellent film-forming properties. In addition, the wet-phase inversion technique was adopted to synthesise the membrane based on different casting conditions. The optimum outcomes of the fabrication conditions were then characterised with the scanning electron micrograph (SEM) to determine the best CAB membrane for CO2/N2 separation. The results showed that CAB-70000 fabricated with 4 wt% of CAB polymer concentration, casting thickness of 250 µm, solvent evaporation time of 5 minutes, and 30 minutes of solvent exchange for isopropyl alcohol and n-hexane, exhibited the best gas separation performance. Further, CAB-70000 showed an average selectivity of 6.12 ± 0.09 and permeance up to 227.95 ± 0.39 GPU for CO2 and 37.28 ± 0.54 GPU for N2, respectively. In summary, this study is expected to show a detailed outline of the future direction and perspective of the novel CAB polymeric membrane that is suitable to be applied in the industry, and serves as an insight for researchers and manufacturers working in the related field of gas separation.


3
To address this global issue, membrane technology was applied and developed intensively because of 4 its superior performance in terms of efficiency and economic advantages. In this study, the cellulose 5 acetate butyrate (CAB) polymer was selected as the polymer matrix material since it exhibited excellent 6 film-forming properties. In addition, the wet-phase inversion technique was adopted to synthesise the 7 membrane based on different casting conditions. The optimum outcomes of the fabrication conditions 8 were then characterised with the scanning electron micrograph (SEM) to determine the best CAB 9 membrane for CO2/N2 separation. The results showed that CAB-70000 fabricated with 4 wt% of CAB 10 polymer concentration, casting thickness of 250 µm, solvent evaporation time of 5 minutes, and 30 11 minutes of solvent exchange for isopropyl alcohol and n-hexane, exhibited the best gas separation 12 performance. Further, CAB-70000 showed an average selectivity of 6.12 ± 0.09 and permeance up to 13 227.95 ± 0.39 GPU for CO2 and 37.28 ± 0.54 GPU for N2, respectively. In summary, this study is 14 expected to show a detailed outline of the future direction and perspective of the novel CAB polymeric 15 membrane that is suitable to be applied in the industry, and serves as an insight for researchers and 16 manufacturers working in the related field of gas separation. 17

Introduction 18
There is a trend of rapid increase in world population, which is expected to hit 10 billion by 2050 (Lalia 19 et al., 2013). In this regard, higher demand in energy will be required for the 21st century to meet the 20 urgent needs. It is predicted that the energy demand will increase by 57 per cent in 2030 (Conti et al., 21 2016). As a major contributor to the world energy supply, fossil fuel solely contribute around 40 per 22 cent of the total carbon dioxide (CO2) emission into the environment, which is mainly attributed to the 23 massive coal combustion activities (Carapellucci and Milazzo, 2003). Global warming has become a 24 genuine problem due to the excessive discharge of pollutants emitted from the combustion activities in 25 the primary industries (Yang et al., 2008).

26
In the past few decades through their efforts, researchers have contributed in combating this global 27 issue to limit and minimise the impact of greenhouse gases (GHGs). They have outlined three feasible 28 options. The first comprises of saving energy used intensively with methods that are more efficient. The 29 second option is to minimise the usage of carbon-based material source or replace it with renewable 30 energy, and the third is to improve the effectiveness of CO2 sequestration with more advanced 31 technology development (Yang et al., 2008). For the past few years, membrane separation technology 32 has been utilised intensively for both water treatment and gas separation purpose (Yang et al., 2008, weather and chemical resistant (Feng et al., 2015, Basu et al., 2010, Kunthadong et al., 2015. The CAB 45 was first investigated and studied by Sourirajan back in 1958, then followed by Manjikian and others in 46 reverse osmosis (RO) separation (Wang et al., 1994). They reported that the CAB membrane owned 47 high solute separation with tolerable membrane flux result, and also provided ease of fabrication as 48 some pre-treatment was negligible (Ohya et al., 1980, Wang et al., 1994. However, limited studies have 49 been conducted on the effects of the acetyl group content on CAB membranes in the CO2/N2 gas 50 separation field. Further, no reports or systematic studies have been performed on the effects of 51 membrane production procedure and fabrication parameters. This includes membrane-casting thickness, 52 solvent exchange time for both isopropyl alcohol and n-hexane with different CAB molecular weights 53 as well as the polymer matrix material structure and performance of CAB membranes. Therefore, the 54 primary objective of this study is to investigate the effects of membrane production procedure and 55 fabrication parameters. Discussions on how the mentioned parameters can affect the membrane in terms 56 of morphology and gas separation performance are presented in this report. The separation performance 57 of the synthesised CAB membrane was selected to evaluate the specified parameters towards CO2/N2.

61
The cellulose acetate butyrate (CAB, Mn~12000, 65000, 70000) in powder form was purchased from 62 Sigma-Aldrich (Malaysia) for membrane preparation. Solutions required for membrane preparation i.e., 63 chloroform, isopropyl alcohol, and n-hexane were purchased from Merck (Malaysia). Distilled water 64 was used for the phase-inversion steps, specifically for immersion precipitation for membrane formation.

66
The CAB membrane was prepared using the wet-phase inversion method, followed by solvent exchange 67 to dry the membrane. A dope solution consisting of 4 wt% CAB (Mn=70000) powders and 96 wt% 68 chloroform was prepared following the condition of each parameter. The solution was stirred for 24 69 hours, and then sonicated for 20 minutes to eliminate the gas bubbles in the solution (Ahmad et al., 2014, 70 Feng et al., 2015. The solution was then poured into space within the casting bars with glass plate 71 underneath. An automatic film applicator (Elcometer 4340, E.U.) was then used for the casting of the 72 membrane. Referring to our previous work, 5 minutes of solvent evaporation time was allowed 73 following each parameter's condition before immersing the membrane in distilled water (27 °C) for a 74 duration of 24 hours (S. Minhas, 1992, Lee et al., 2017. The solvent exchange was performed on the as-75 spun membrane first with 60 minutes immersion period in isopropyl alcohol and then another 60 minutes 76 immersion period in n-hexane. The resultant membrane was then dried at ambient temperature to 77 eliminate the remaining volatile liquid in between two glass plates filled with filter paper for 24 hours 78 before use (S. Minhas, 1992, Jawad et al., 2015a  Meanwhile, for the effect of solvent exchange time, the membranes were prepared following the 87 fabrication method as described in section 2.2. The solvent exchange duration studied was 15 minutes 88 (CAB-15Iso), 30 minutes (CAB-30Iso), and 60 minutes (CAB-60Iso) for isopropyl alcohol, followed 89 by 60 minutes of n-hexane.

104
The procedure for gas permeation measurement was discussed in our previous published work (Lee et   The CAB membrane structures including surface and cross-sectional, were observed via SEM (Hitachi 111 TM3000, Tokyo, Japan). Each membrane sample was cut into small pieces, and then kept on a plastic 112 petri dish in the cryogenic freezer at a temperature of up to -80°C for 24 hours to give a consistent and    Fig. 1c. Alternatively, a rough surface was formed for CAB-300 (250 µm), as seen 131 in Fig. 1e. The change in the structure was due to the different rates of demixing that occurred as the 132 phase precipitation proceeded when high casting thickness was applied, causing the deposition speed of 133 the membrane to reduce during the membrane formation phase. The slow deposition rate avoids rapid 134 exchange of non-solvent and solvent within the membrane. As a result, the surface structure of the CAB 135 membrane was built-up based on the sufficient phase precipitation period given (Ahmad et al., 2013, 136 Thomas et al., 2014).

137
The cross-sectional micrographs of the fabricated CAB membrane at casting thickness of 200 µm 138 , 250 µm (CAB-250), and 300 µm (CAB-300) were revealed in Figs. 1b, d, and f, 139 respectively. From the micrographs, dense structures were depicted from all the cross-sectionals of the 140 CAB membranes. The dense structure formation was due to the densification of the membrane during 141 the immersion period, whereby the remaining solvent imbedded in the polymer matrix was replaced by 142 distilled water. As the volatility of the solvent was generally higher than distilled water the membrane 143 thickness changed from 12.42 ± 0.05 µm to 11.32 ± 0.06 µm and 12.89 ± 0.10 µm for CAB-200, CAB-144 250 and CAB-300, respectively. The reduction of membrane thickness from 12.42 ± 0.05 µm (CAB-145 200) to 11.32 ± 0.06 µm (CAB-250) was due to thicker casting thickness applied during membrane 146 fabrication, which allows more solvent embedded in the polymer matrix to be replaced by non-solvent 147 (H20) during the immersion period, resulting in a denser and thinner membrane thickness for CAB-250 148 (Ahmad et al., 2013). In contrast, a thicker membrane was obtained when increasing the membrane 149 thickness further to 300 µm for CAB-300 (12.89 ± 0.10 µm). This is correlated to the increase resistance 150 of inward diffusion of non-solvent, due to higher casting thickness applied, causing a delay transition 151 demixing in the film membrane (Tiraferri et al., 2011). (143.03 ± 0.62 GPU) and CAB-300 (12.93 ± 0.34 GPU). This was because of the reduction in its 156 membrane thickness (11.32 µm, Fig. 1d) and its selective smooth surface structure, which allowed the 157 solution diffusion mechanism to occur efficiently. Therefore, the CO2 permeance of CAB-250 increased 158 (Jawad et al., 2015a). Meanwhile, the CO2 permeance of CAB-300 reduced to 12.93 ± 0.34 GPU, 159 indicating that a higher casting thickness beyond 250 µm can exert extra resistance towards gas diffusion 160 within the membrane, which in turn affects the efficiency of gas permeation due to the thick dense 161 membrane synthesised (Fig. 1f).

175
The acceptable result obtained for CAB-250 was due to the membrane structure formation, which 176 eventually increased the CO2 permeance against the N2 permeance attained. However, the selectivity 177 reduced to 1.15 ± 0.01 GPU when the higher casting thickness (300 µm) was implemented for CAB-178 300. Even though the thickness of the membrane was essential for effective gas separation, however, 179 excessive membrane thickness restricted the gas diffusion within the membrane. within the CAB polymer matrix caused less CO2 molecules to interact with the water, therefore allowing 215 more CO2 gas to permeate through the membrane (Jawad et al., 2015b). In the meantime, the high CO2 216 permeance rate for CAB-60Iso (60 minutes) contributed to the thin dense membrane structure, which 217 allowed the CO2 feed gas to pass through the membrane with least resistance pathway as compared to 218 the thick dense membrane (Tiraferri et al., 2011). Thus, the CAB-60Iso (60 minutes) yielded the highest 219 CO2 permeance rate amongst the other membranes (CAB-15Iso and CAB-30Iso). The N2 permeance rates for CAB-15Iso, CAB-30Iso, and CAB-60Iso are depicted in Fig. 7 15Iso, CAB-30Iso, and CAB-60Iso, respectively. The possible explanation for this trend was due to the 224 reduction in the membrane thickness from 13.87 µm to 9.3 µm (Fig. 5). In addition, as isopropyl alcohol 225 was mainly made up from non-polar molecules, the remaining molecules within the CAB structure can 226 easily attract light gas molecules (Katayama and Nitta, 1976  As discussed previously, CAB-60Iso (60 minutes) showed a thin dense membrane formation with 232 high CO2 and N2 permeance rates. However, based on Fig. 8, the CAB-30Iso (30 minutes) yielded the 233 best selectivity performance. This was due to the smooth homogeneous surface and superior cross-234 sectional morphology, which selectively allowed a predetermined amount of CO2 and N2 to pass through 235 the dense membrane. On the contrary, the CAB-15Iso (15 minutes) demonstrated low selectivity (Fig.   236   8). This was due to the presence of a thick irregular surface morphology (Figs. 5a and b), which imposed 237 an undesirable effect on membrane permeance performance due to extra resistance pathway generated 238 (Rahimpour et al., 2008, Yang andWang, 2006). Therefore, CAB-30Iso (30 minutes) was preferred as 239 compared to CAB-15Iso (15 minutes) and CAB-60Iso (60 minutes) because of its excellent morphology 240 and good selectivity performance.

259
As seen from these results, the increased exchange time of n-hexane caused the CAB membrane to 260 become more compact due to membrane densification as time passed (Sabde et al., 1997). In addition, 261 the main reason for the reduction in the membrane thickness was due to the isopropyl alcohol imbedded 262 within the membrane slowly being replaced by n-hexane with time. The replacement of isopropyl 263 alcohol with n-hexane occurred when the molecular affinity of n-hexane was greater than isopropyl 264 alcohol (Hansen, 2007). Referring to the Hansen solubility chart, the solubility for isopropyl alcohol, n-265 hexane, and water are 23.6, 14.9, and 47.9 MPa 1/2 , respectively (Egan andDufresne, 2008, Hansen, 266 2007). Therefore, the molecular affinity is in the order of CAB-water>CAB-isopropyl alcohol>CAB-n-267 hexane. The order of the molecular affinity represents the attraction force between the polymer and the 268 solvent and non-solvent used (Kim and Oh, 2001).

Surface and cross-sectional SEM of CAB membrane dried with 30 minutes of isopropyl alcohol first then followed by; (a-b) 15 minutes (CAB-15H), (c-d) 30 minutes (CAB-30H), and (e-f) 60 minutes (CAB-60H) of solvent exchange time using n-hexane, at casting thickness of 250 µm and 5 minutes solvent evaporation time
According to the CO2 permeance results displayed in Fig. 10, there was clear indication that CAB-60H showed the highest CO2 permeance rate followed by CAB-30H and subsequently, by CAB-15H. As seen in Fig. 10, the CO2 permeance increased significantly from 21.55 ± 0.03 GPU to 227.95 ± 0.39 GPU when the solvent exchange time increased from 15 minutes (CAB-15H) to 30 minutes (CAB-30H). This was because when the exchange time was increased, sufficient time was provided for the exchange of the isopropyl alcohol content with n-hexane and therefore, generating a relatively thinner and compact cross-sectional membrane, which favoured CO2 permeation through the membrane (Jawad et al., 2015b). In addition, the CO2 permeance increased further when the solvent exchange duration was increased from 30 minutes to 60 minutes, as observed from CAB-30H ( to the increase in the number of the remaining polar n-hexane molecules within the membrane structure, resulting in a more active interaction with the CO2 molecules and hence, higher CO2 permeance yield (Jawad et al., 2015b). to 30 minutes (CAB-30H). The reason for this increment was mainly due to the thin dense membrane 273 structure of CAB-30H (9.50 ± 0.10 µm), which allowed the feed of N2 gas to pass through a least 274 resistance pathway. However, the high N2 permeance for CAB-60H (70.49 ± 0.33 GPU) was due to 275 stress of surface tension caused by high capillary forces because of the evaporation of residual n-hexane 276 within the membrane, which led to the collapse in the structure (Matsuyama et al., 2002).

Fig 11. N2 permeance for membrane dried with 15 minutes (CAB-15H), 30 minutes (CAB-30H), and 60 minutes (CAB-60H) of n-hexane, at casting thickness of 250 µm and 5 minutes solvent evaporation time
As seen in Fig. 12, the CAB-30H membrane showed the highest gas selectivity, which was 279 achieved at 6.12 ± 0.09. This result further proved that to have a high gas separation performance a 280 smooth surface with regular thin dense membrane morphology was preferable (Figs. 9c and d) (Huang 281 and Feng, 1995, Jansen et al., 2005, Matsuyama et al., 2002, Lui et al., 1988

298
According to Coltelli et al. (2008), the acetyl group has been deduced to have prominent effect on the 299 membrane gas separation performance, as excessive acetyl composition in the membrane could promote 300 plasticisation within the membrane (Coltelli et al., 2008, Ismail andLorna, 2002). Thus, different CAB 301 molecular weights with different acetyl, butyryl, and hydroxyl groups were investigated, as 302 demonstrated in Fig. 13.

303
As depicted in Figs. 13a and c, a porous structure was observed for both CAB-12000 (Mn=12000) 304 and CAB-65000 (Mn=65000), while CAB-70000 (Mn=70000) showed a smooth surface (Fig. 13e). The 305 reason the membrane surface changed from porous to smooth was due to the high molecular weights of 306 CAB, which caused the increase in the number of entanglements between the macromolecular chains in 307 the solution (Jansen et al., 2006). Therefore, the high molecular weights of CAB favoured the gelation 308 of the polymer rich phase after the phase-inversion occurred and hence, suppressed the formation of the 309 porous structure during the early stages (Jansen et al., 2005).

22
Based on Figs. 13b, d, and f, the thickness of CAB-12000, CAB-65000, and CAB-70000 were 311 10.96 ± 0.10, 16.05 ± 0.17, and 9.50 ± 0.10 µm, respectively. The increment in the CAB molecular 312 weights further influenced the membrane thickness through the rheological properties of the casting 313 solution (Jansen et al., 2005). This was due to the high molecular weights of the CAB polymer being 314 utilised for membrane fabrication, which gave the rapid gelation (Jansen et al., 2005). After the rapid 315 gelation, the porous structure was greatly suppressed and further evaporation of solvent and non-solvent 316 from the polymer matrix resulted in gradual shrinkage of the structure (Jansen et al., 2005). Therefore, 317 the thickness of CAB-70000 (9.50 ± 0.10 µm) was thinner than CAB-12000 (10.96 ± 0.10 µm) and in CO2 permeance was caused by the acetyl groups rigidity and steric effects (Wan et al., 2003).

333
Therefore, this allowed the higher intrinsic solubility of CO2 due to the greater number of acetyl-acetyl 334 interactions that existed (Koros et al., 1988b, Scholes et al., 2012. In addition, increasing the CAB 335 molecular weight from 65000 to 70000 had increased the permeance rate drastically from 74.37 ± 1.25 336 GPU to 227.95 ± 0.39 GPU. Even though, CAB-70000 (12-15 wt%) has the lowest acetyl-acetyl 337 interactions due to low acetyl group composition compared to other CAB polymers. The significant 338 increase in the CO2 permeance was due to the thin dense membrane exhibited for CAB-70000, as thin 339 dense membrane usually impose less flux resistance for the membrane (Pandey and Chauhan, 2001).

340
Therefore, the permeance of CO2 was highest among all as the membrane thickness was the thinnest. As portrayed in Fig. 15 (Wan et al., 2004, Ong et al., 2012.

363
The XPS characterisation was adopted in this study to analyse the quantitative element composition of 364 the CAB membrane fabricated. The quantitative element composition of the membrane surface can be 365 determined from the spectrum obtained. Consequently, CAB-12000, CAB-65000, and CAB-70000 were 366 analysed through XPS analysis. The surface chemical quantitative compositions are depicted in Table   367 1 and Fig. 17, respectively.

385
As indicated in Fig. 16, the butyryl group played a crucial role in manipulating the selectivity 386 performance of the membrane, because it can increase the CO2 diffusion due to the increase of the non-387 polar butyryl chain within the structure of the membrane (Wan et al., 2004). As a result the membrane 388 became more hydrophobic in nature, and hence, promoted better CO2 permeance flux (Ong et al., 2012).

Fig 17. Element composition of XPS spectrum of CAB-12000, CAB-65000, and CAB-70000
The CO2/N2 separation performance of this current study were summarised and compared with 390 other research works, as shown in incorporating the polymer matrix with inorganic filler to produce the hybrid system of mixed matrix 400 membranes (MMMs) (Aroon et al., 2013, Chung et al., 2007, Ismail et al., 2009, Goh et al., 2011.

407
The optimisation of membrane morphology conducted with respect to the different parameters was 408 found to be successful for the preparation of the highly selective CAB gas separation membrane. The membrane formation and morphology were closely related to the rheological behaviour of the casting 410 solution. The results have shown that membrane casting thickness, solvent exchange duration for both 411 isopropyl alcohol and n-hexane, and the molecular weights of the CAB polymer had a significant role 412 in manipulating the CO2/N2 gas separation performance as well as the morphology of the membranes.

413
Under optimised conditions, the best membrane was found to be the CAB-70000, which was fabricated 414 with 4 wt% polymer concentration, 250 µm casting thickness, 5 minutes solvent evaporation time, 30 415 minutes solvent exchange with isopropyl alcohol followed by another 30 minutes of solvent exchange 416 with n-hexane. Moreover, the CAB-70000 had the best gas separation performance with an average 417 selectivity of 6.12 ± 0.09 and permeance up to 227.95 ± 0.39 GPU for CO2 and 37.28 ± 0.54 GPU for 418 N2, respectively. The superior CO2/N2 separation performance of the membrane was mainly contributed 419 by the quality formation of the smooth surface, with thin dense and defect-free membrane structure.

420
Further, it has been suggested that to improve the performance of the CAB membrane, inorganic 421 nanoparticle fillers such as carbon nanotubes (CNTs) be incorporated to produce mixed matrix 422 membrane (MMM).