Pervaporative Denitrogenation of Pyrrole/n-heptane as Model Oil using MoO3 -PEBAX/PAN membrane

The environmentally obnoxious nature of Nitrogen in fuel oils requires serious attention for its removal. In this work, a novel hybrid matrix membrane was prepared by introducing MoO 3 nanoparticles into poly (ether-block-amide) (PEBAX2533) and was in turn used for the pervaporative separation of pyrrole/n-heptane mixtures. The interactions between the membrane, pyrrole and n-heptane were investigated by swelling experiments. Pervaporative separation performance of PEBAX membranes revealed higher selectivity for pyrrole from its mixture with n-heptane, which further increased with increasing MoO 3 contents in membrane along-with correspondingly increasing the total ux and the separation factor as well. At 2000 µg·g − 1 pyrrole concentration and 30°C temperature, the total ux and the separation factor reached the maximum values of 2.46 kg•m − 2 •h − 1 and 17.58, respectively. Attributed to the outstanding separation performance of PEBAX membranes, this work may provide a useful insight into the viable removal of nitrogenous compounds from gasoline via pervaporation.


Introduction
Due to the increasingly stringent environmental regulations regarding the quality of transportation fuels, ultra-low sulfur, nitrogen, and aromatics containing fuels are highly demanded (Busca &Research 2009, Mojaverian Kermani et al. 2018, Muhammad et al. 2019b). These species lead to the production of NO x , SO x , CO and CO 2 as major pollution contributors upon combustion of fuel oils (Ali et al. 2006, Rahman et al. 2018). Among these, nitrogenous compounds such as indole and carbazole are more stable having lower activity than sulfur compounds, and hence their removal via hydrodenitri cation (HDN) is a challenging task (Oliveira et al. 2004). In addition, these nitrogenous compounds also detrimentally affects the e ciency the industrial various desulfurization processes (Ammar et al. 2020, Laredo et al. 2015, Nunes et al. 2014. Along with their stability, neutral nitrides are easily converted into basic nitrides which then compete with sulfur compounds for active sites of catalysts, causing catalyst deactivation and ultimately reduce catalyst service life (Dorbon &Bernasconi 1989). Apart from this, combustion of these nitrogenous compounds leads to NO X production which in turn trigger acid rain endangering biotic as well as abiotic segments of ecosystem (Abdul-quadir et al. 2018, Marani et al. 2017, Sun et al. 2018). In addition to acid rain, NO X also lead to photochemical pollution and haze and depletion of ozone layer. In combination to NOx production, some of the neutral nitrides in fuel oils like indoles can go through oligomerization during hydrogenation catalysis, can corrode equipment and interfere and deteriorate the stability of re ned products through gum formation and altering color and odor (Adams et al. 1984, McKay et al. 1976).
Awarding to these aspects, it is signi cantly desired to remove basic nitrogen compounds for optimizing the uid catalytic cracking (FCC) process and improving the quality of light fuels.
The removal of nitrogen compounds from gasoline is mainly achieved by HDN utilizing a metal oxide/sul de-based catalyst and zeolites (Escola et al. 2012, Li et al. 2009, Mapiour et al. 2010, Yu et al. 2010). However, due to high operating temperatures and pressures, HDN requires exorbitant devices, high hydrogen consumption, high production cost, and reduced heptane number of the nal gasoline. On the contrary, pervaporation is a promising and effective technique for the removal of nitrogen attributed to its low operating cost, high separation e ciency, simpli ed operation procedure, and adaptability to changes in processing streams (Mortaheb et al. 2012).
Pyrrole is major nitrogen compound in FCC gasoline. The solubility parameter is the square root of cohesive energy per molar volume and is calculated by group contribution method (Balko et al. 2002).
The solubility parameters of pyrrole, n-heptane and Poly (ether-block-amide) (PEBAX2533) are 19.5 J 1/2 •cm −3/2 ,15.3 J 1/2 •cm −3/2 and 19.1 J 1/2 •cm −3/2 respectively, suggesting similar solubility parameters of pyrrole and polymeric material than n-heptane. PEBAX2533 are comprised by a series of block copolymers of exible polyether (PE) and blocks of rigid polyamide (PA) (Lin et al. 2006). PEBAX copolymer possesses high mechanical strength and toughness with high concentration of PA chain segments, thus improving the ratio of PE chain segments in PEBAX which can award the polymer with better a nity to organic compounds (Lin et al. 2006). Among these, PEBAX2533 contain the highest concentration of PE, and is deemed a promising membrane material due to its high hydrophobicity, remarkable physical and mechanical strength, and good thermal stability, has been ubiquitously applied in separation of aromatic compounds (Gao et al. 2020, Lin et al. 2008, Lin et al. 2009). Kun Liu et al. separated mixture of thiophene/n-heptane using PEBAX/PVDF-composited membranes and PEBAX membrane for the separation of n-butyl acetate, n-butanol, and acetic acid from aqueous solutions via pervaporation (Liu et al. 2013). Similarly, Ding He et al. reported enhanced desulfurization performance and stability of PEBAX membrane (Amaral et al. 2014, Huang et al. 2020, Song et al. 2020). Fusheng Pan et al. embedded Ag + @COFs into PEBAX membrane to construct mass transport channels which elevated desulfurization performance (Le et al. 2011).
To our knowledge, no reports on the pervaporative removal of pyrrole from gasoline using composite PEBAX membranes have been established yet. These type of membranes with dense active layer covered on a porous supporting layer, have been extensively adopted in commercial pervaporation process endorsed to their enhanced mechanical properties, resistance to corrosion and swelling (Yildirim et al. 2008). This type of interaction can synchronously enhance permeating ux and selectivity of composite membrane. In this connection, mixed matrix membranes, composed of e cient activated adsorption llers dispersed in a polymeric matrix have been reported with enhanced pervaporation performance (Castro-Muñoz 2019, Merkel et al. 2013). Among the many adsorbent, MoO 3 has been proved nitrogenselective (specially preferential retaining e ciency for pyrrole) from its mixture with other organic compounds (Chen et al. 2010, Huang &Meagher 2001. Thus, herein we report a MoO 3 microspheres introduced PEBAX polymer to obtain MoO 3 lled PEBAX Polyacrylonitrile (PEBAX/PAN) membrane and in turn apply it for the pervaporative removal of pyrrole from n-heptane as model gasoline. The structure morphology of MoO 3 microspheres and hybrid membranes were characterized by scanning electron microscopy (SEM), while MoO 3 concentration, temperature and pyrrole concentration of feed were systematically evaluated affecting the separation performance of membrane in the swelling and pervaporation experiments.
Page 4/25 2. Experimental 2.1. Materials PEBAX2533 was supplied by Arkema Inc. PAN ultra ltration membrane with a molecular weight cut-off of 5 kDa was purchased from Shanghai Lanjing membrane & engineering Co. Ltd. N-heptane, n-butanol, ethyl alcohol and n-propanol were provided by Guangdong xilong chemicals Inc. Guangdong China. Pyrrole and MoO 3 were purchased from Aladdin Inc. China. All reagents were of analytical grade and used without further puri cation. Deionized water was used throughout the experiments where required.

Preparation of membranes
2.2.1 Preparation of PEBAX homogeneous membrane and MoO 3 lled membrane PEBAX 2533 particles were dried to constant weight in an oven at 50 o C, and were then dissolved in a conical bottle containing certain amount of n-butanol. The mixture was heated to 50 o C in a water bath and stirred for 1 h till the formation of homogeneous solution representing 7 wt.% transparent polymer casting solution. This solution was vacuumed for 0.5 h to remove bubbles and was evenly poured onto the clean and at glass plate to form a uniform lm. The solvent was evaporated by heating in a dustfree environment for 24 h, and was then dried thoroughly in a vacuum drier at 50 o C. The lm was stored in a dryer for using.
To prepare the lled lm, 7 wt.% casting solution was made by the same method, followed by the addition of nano MoO 3 equivalent to 2 wt.%, 4 wt.%, 6 wt.% and 8 wt.% of the mass of PEBAX2533 particles. The casting solution was dissolved and stirred for 2 h at 70 o C followed by ultrasonication for 10 min. The solution was once again stirred for 1 h to enhance the dispersion of MoO 3 . Finally, the solution was allowed static under vacuum for 30 min to remove any trapped air bubbles.

Preparation of PEBAX/PAN composite membrane and nano MoO 3 lled-composite membrane
Composite membrane was prepared by treating PAN supporting membrane with anhydrous ethanol and was xed on a clean and dry glass plate. When the membrane was semi-wet, the casting solution was evenly dumped into the middle of the membrane, and was then scraped. The treatment and preservation steps were similar as those mentioned for the preparation of homogeneous membrane.
Filled-composite membrane was prepared by the above-mentioned nano MoO 3 lled casting membrane with different lling ratios followed by scraping it on the PAN support membrane. The treatment and preservation steps were again similar to those mentioned for the preparation of lled membrane.

Characterization of membranes
The morphology of MoO 3 microspheres was observed by SEM (Hitachi S3400, Japan). The pore size of PAN lm was measured by Aperture analyzer. Membrane samples were fractured in liquid nitrogen and then analyzed for surface morphology and dispersion of MoO 3 microspheres by SEM. Fourier transform infrared (FT-IR, 8400S, Japan) spectra and X-ray diffraction (XRD, TD-2500, China) were used to determine the chemical composition and crystallite size of the microspheres, respectively.

Swelling experiments
A cut piece of membrane was dried in a vacuum oven at 50 o C for 48 h until constant mass. It was then immersed in the feed mixture for 5 min and removed with a forceps and wiped with lter paper to remove the surface mixture following by weighing it. Each experiment was performed in triplicates and average values were recorded. Films dried until constant mass were placed in weighing bottles to weigh then recorded as W d . Under dry environment, lms were immersed in capacity bottles containing 40 mL pyrrole/n-heptane with different pyrrole concentration. Films were heated by water bath pot according to different temperatures. The residual liquid on lm surface was sponged with lter papers. The weight of the wet lms was recorded as Ws. The swelling experiments were repeated three times, and average values were reported . The degree of swelling (DS) was calculated using Eq. (1) (Qu et al. 2010): The main contents of swelling experiment were as follows: the relationship between DS and time of lled lm was investigated, when the pyrrole concentration in feed liquid was 0, 1000 μg·g -1 , 2000 μg·g -1 , 3000 μg·g -1 , 4000 μg·g -1 , 5000 μg·g -1 , respectively and at 30 o C feed temperature, the relationship between temperature and swelling performance of lled lms was investigated.

Pervaporation experiments
The schematic of the pervaporation apparatus is shown in Fig. 1 (Sampranpiboon et al. 2000). The liquid circulation system consists of temperature controller, liquid tank and diaphragm circulating pump. The effective membrane area of membrane module customized by Zhejiang University is 19.63 cm 2 . The feed liquid is pumped from the circulating pump to membrane tank at atmospheric upstream pressure, while that of the downstream was maintained at approximately -100 kPa by a vacuum pump. Interception uid is returned to the liquid tank for continuous circulation, and osmotic uid is collected by condensing in liquid nitrogen (-196 o C). Samples were collected twice per hour in three parallel experiments.
Permeation total ux refers to the quality of feed liquid passing through membrane of unit area per unit time, which re ects the capacity of membrane to treat feed liquid and is calculated via Eq. (2) (Sampranpiboon et al. 2000): Here J is the total permeation ux (kgm -2 h -1 ), M and A refer to mass of the osmotic uid (g) and effective area of membrane (m 2 ), respectively, and t is Penetration time.
Separation factor refers to the ratio of pyrrole concentration in osmotic uid and raw feed liquid, which re ects the Separation selectivity of membrane, and is calculated using Eq. (3): Where a is the separation factor, y 1 and y 2 refer to the mass fractions of pyrrole and n-heptane in feed liquid, respectively, and x 1 and x 2 are the mass fractions of pyrrole and n-heptane, respectively.
Pervaporation separation index is used as a characterization parameter for the comprehensive performance of pervaporation membrane and is measured using Eq. (4) (Zi 1996): Where, PSI is pervaporation separation index, J is the permeation ux (kgm -2 h -1 ), and 'a' is the separation factor.

Morphology of membranes
The structure of composite membrane and lled composite membrane was investigated by SEM. Fig. 2b showcasing the surface of the composite membrane suggests pyknotic, imperforated and awless PEBAX2533 active layer. Fig. 2c manifested well monodispersed MoO 3 microspheres in PEBAX2533 layer without any aggregation. The lling amount of MoO 3 microspheres in the lled composite membrane was selected as 4 wt.% (due to its best performance of swelling and pervaporation to be discussed in the proceeding sections of the manuscript). The cross-sectional SEM images of the composite membrane and lled composite membrane shown in Fig. 2d and 2e, respectively, suggested a dense PEBAX2533 active layer with a uniform thickness of about 22 μm, rmly adhered on the PAN support without any blemishes, which con rmed the uniform dispersion of MoO 3 particles. The interface between the two layers was properly interconnected, while PEBAX2533 polymer solution did not penetrate into the PAN support

FT-IR analysis of microspheres
The FT-IR results of composite membrane and composite-lled membrane shown in Fig. 3 indicated peaks at 723 cm -1 and 1100 cm -1 ascribed to characteristic stretching-vibration of -CH 2 -CH 2 -CH 2 -bond and the band of -C-O-in polyether of PEBAX, respectively (Liu et al. 2013). Peak at 1740 cm-1 was caused by the stretching vibration of -C=O-double bond in polyamide, while the one at 3294.19 cm -1 was awarded to the characteristic stretching vibration of -NH-in polyamide (Wu &Xu). Furthermore, the FTIR spectra of the composite membrane was very much identical to that of the lled-composite membrane, without the appearance of new functionalities, which suggested that MoO 3 particles physically lled into the PEBAX2533 active layer.

XRD analysis of the microspheres
The XRD patterns of MoO 3 particles, composite membrane, and composite-lled membrane are shown in Fig. 4. The smaller diagram in the right upper shows the spectrum of MoO 3 . The spectra of composite membrane and composite-lled membrane are basically the same, and only a wide diffraction peak appears in the range of 12 o~3 5 o , which suggested the amorphous nature of PEBAX2533 membrane, while MoO 3 lling did not cause any visible change in its structure due to the physical mixing of the two species. The diffraction peaks corresponding to MoO 3 and PEBAX 2533 occurred closely and hence coincided with each other, which resulted in broader diffraction peak at 12 o~3 5 o for the nal compositelled membrane.

Swelling performance of membranes
The a nity of membrane for a given chemical can be quanti ed by its DS (Liu et al. 2011, Mandal &Bhattacharya 2006. At room temperature, the swelling property and sorption capability of the lled membranes with different ller contents of MoO 3 (in the feed mixture) were studied, with pyrrole concentration of 2000 μg·g -1 and the results are compiled in Fig. 5. It can be seen that DS of lled membranes initially increased followed by a decrease with increasing MoO 3 microspheres ller content.
The DS of lled membrane reached to maximum (an increase of 13.84%) at ller content of 4 wt.%. Further increasing the ller content led to decrease in DS. Below 4 wt.% ller contents, sorption performance increased with increasing ller content due to fact that MoO 3 particles in the membrane were well-distributed and possessed high speci c surface areas (SSA) and de cient electronic structure, facilitating strong adsorption interaction with the feed mixture. This was more favorable for pyrrole having solitary electron pair, thus strongly enhanced the swelling performance of membranes. With increasing ller content, uni cation of particles occurred, which impeded the movement of the polymer chains in the membrane and hence reduced the number of channels for mixture feed to enter membranes. From these results, 4 wt.% MoO 3 was chosen as the optimum ller amount for onward experiments.
The effect of temperature on the swelling properties of lled membranes was investigated at 2000 μg/g concentration of binary mixture and 4 wt.% ller contents. The results shown in Fig. 6 (a) suggested that DS of the lled lm increased gradually with increasing temperature, and achieved maximum value of 31.25% at 70 o C. This was attributed to the increased Brownian movements of molecules in the binary mixture as well as those in polymer chain, hence facilitated better contacts and led to enhanced swelling behavior. Furthermore, the swelling resulted in increased volume of the membranes, thus increasing the free space.
We further analyzed the effect of pyrrole concentration on DS of 4 wt.% lled lm and the results are picturized in Fig. 6 (b), which suggested that a dramatic increase in the DS with increasing time and reached equilibrium after 30 min. The swelling equilibrium increased with increasing pyrrole concentration. At 5000 μg·g -1 feed liquid concentration, DS of the lled lm reached 24.95%. These results revealed that membranes preferentially adsorbed pyrrole from its mixture with n-heptane, which can be explained by the solubility parameter theory. The solubility parameter of PEBAX2533 is 19.5 J 1/2 .cm -3/2 , which is much closer to that of pyrrole (19.1 J 1/2 .cm -3/2 ) than that of n-heptane (15.3 J 1/2 .cm -3/2 ). This high selectivity of PEBAX/PAN membranes could of great potential for practical applications involving gasoline denitri cation. Solution-diffusion mechanism is widely used to describe the principle of pervaporative separation (Wang et al. 2016). To better illustrate the effect of incorporated nano MoO 3 on solution and diffusion process, the sorption of penetrant molecules in the membrane was evaluated by sorption experiments. The sorption capacity of membranes in the feed solution (2000 ppm pyrrole/n-heptane as model gasoline feed) at 30 °C was evaluated and the results are shown in Fig. 6 (b). The sorption amount of pyrrole in the membranes is much lower than that of n-heptane, because of its low concentration. The incorporated MoO 3 shows minor in uence on the sorption capacity of the membranes towards pyrrole and n-heptane.
The pyrrole/n-heptane sorption selectivity of PEBAX pristine membrane and PEBAX-MoO 3 hybrid membranes suggested that MoO 3 and PEBAX have similar a nity towards pyrrole. However, the diffusion selectivity increased by more than 50% after incorporating MoO 3 in PEBAX. Therefore, the diffusion process demonstrated a decisive effect on the enhanced permeation ux and separation factor of the hybrid membrane. The hybrid membrane can be divided into three interfaces: polymer matrix, ller and polymer-ller interface. Among the three sections, MoO 3 are impermeable to the penetrant molecules (Salmeron et al. 1982). The crystallinity of the PEBAX-MoO 3 hybrid membrane is similar to that of PEBAX pristine membrane. The fractional free volume of PEBAX-MoO 3 hybrid membrane is even smaller than that of PEBAX pristine membrane. However, the diffusion coe cients of pyrrole and n-heptane exhibit the opposite trends compared to the change of free volume properties. Thus, one can conclude that the increase in diffusion coe cient mainly arise from the PEBAX-MoO 3 interface. The interaction between the basal plane of MoO 3 and pyrrole molecules lies within the scope of reversible chemical complexation (King 1987), indicating that MoO 3 particles can serve as a facilitated transport carrier for pyrrole molecules via "hopping" from one carrier to another (Pinnau &Toy 2001). The large surface area of MoO 3 particles provides a continuous transport pathway for pyrrole molecules, which is quite different from the traditional isolated facilitated transport sites. The reversible reaction between pyrrole and MoO 3 particles enhances the transport rate of pyrrole in the membrane, thus endows the hybrid membrane with higher diffusion selectivity (Hong et al. 2000). After incorporating MoO 3 particles into PEBAX matrix, the continuous facilitated transport pathway on MoO 3 particles obviously enhances the diffusion coe cient of pyrrole in comparison with that of n-heptane, leading to increased diffusion selectivity, which consequently endows the membrane with high pyrrole selectivity.
These results indicated that the membrane with 4 wt.% ller has the best pervaporation performance, optimal total ux and separation factor of pyrrole, which is identical to the conclusion of the swelling experiments. We further applied PSI to comprehensively evaluate the parameter responsible for membrane separation performance. The data in Fig. 8 show that PSI of 4 wt.% ller content based membranes was maximum i.e. 29.53%, which once again con rmed the results reported in earlier sections of this study.

Effect of feed-liquid temperature
The effect of feed-liquid temperature on pervaporation performance of 4 wt.% lled-composite membranes are shown in Fig. 9. At constant pyrrole concentration of 2000 μg·g -1 , the permeation total ux increases continuously with increasing feed-liquid temperature from 30 to 70 o C, while the separation factor decreases gradually. During the pervaporation process, the pressure difference is the primary driving force due to the fact that upstream pressure is atmospheric while the downstream pressure is kept at about 101 kPa. Additionally, increasing operating temperature exacerbates the movement of molecules, accelerating the diffusion rate, and intensi es the movement of polymer chain segments. Molecules of each component can quickly and effectively diffuse through the larger free volume of the polymer chain gap, which results in increasing the permeation ux (Qi et al. 2006). However, the separation factor decreases with increasing temperature, which may be due more sensitive nature of nheptane permeation to temperature.
The sensitivity of component ux to temperature can be further re ected by the activation energy, which in turn re ects the extent to which the temperature affects the permeability of individual component (Saini et al. 2017). According to Arrhenius Eq. (5) (Pan et al. 2018), as shown in Fig. 10, the osmotic activation energy of pyrrole and n-heptane are 6.61 kJ·mol -1 and 15.59 kJ·mol -1 , respectively. Compared with pyrrole, the effect of temperature change on n-heptane ux is more signi cant. As the feed temperature rises, the permeation ux of n-heptane increases more pronouncedly, while the a nity of membrane for pyrrole weakens, and n-heptane molecules pass through the membrane more easily, which decreases the separation factor. This conclusion is consistent with the results of Fig. 10.
Where, J i is the permeation ux of component i, kgm -2 h -1 . A i is the pre-exponential factor of component i, kgm -2 h -1 . E i is the permeation activation energy of component i, kJ·mol -1 . T is the absolute temperature; K.
R is the molar gas constant, 8.314 J×mol×K -1 .

Effect of feed concentration
The effect of pyrrole feed concentrations on total ux and separation factor are shown in Fig. 11. At 70 o C feed-liquid temperature and increasing pyrrole concentration in feed from 1000 to 5000 ppm, the permeation ux and separation factors increased continuously. The total ux increased from 2.23 to 3.12 kgm -2 h -1 corresponding to an increase of 39.91%, while the separation factors enhanced from 15.65 to 19.76. The enhanced total uxes further con rmed the results of swelling experiments. Previous swelling experiments showed that higher concentration could increase the DS of membranes, thus weakening the interaction between the polymer chains, which in turn provides more free space in the membrane for molecules for feed liquid in penetrating the membrane, thus increasing the permeation ux. As pyrrole is a weak polar molecule with Lewis-base, thus MoO 3 particles exhibit strong a nity for it (Mandal &Bhattacharya 2006, Yang et al. 2012. Upon increasing pyrrole concentration in feed liquid, more pyrrole molecules dissolve and diffuse into the membrane, which desorb downstream, thus leading to increase in separation factor.
3.4. Mechanism of the Pervaporative Denitrogenation of Pyrrole/n-heptane Fig. 12 shows the mechanism of pyrrole molecules being a weak base once interacts with MoO 3 Lewisacid particles in n-heptane solvent via PEBAX/PAN membranes. As discussed above, increase in pyrrole concentration increases the a nity of MoO 3 species hence increases permeation ability. MoO 3 losses its electron while in contact with the pyrrole species by oxidizing to Mo 6+ . The oxidized MoO 3 then reduces by adsorbing to the polymer metrics. Where the pyrrole molecules dissolve and diffuse into the membrane, which desorb downstream, thus leading to increase in separation factor.

Conclusions
In summary, MoO 3 -PEBAX/PAN lled-composite membrane was prepared using MoO 3 as ller. The swelling and pervaporation properties of the prepared membranes in the pyrrole/n-heptane system as model fuel oil were investigated. Results showed that higher concentration of pyrrole facilitated better swelling performance of the membrane due to their higher mutual a nity. The permeation ux and separation factor of the membrane increased with increasing pyrrole concentration but declined with increasing operating temperature. Pervaporation experiments revealed that at 4 wt.% MoO 3 contents, membrane showed the best comprehensive separation performance. When the feed temperature is 30 o C and 5000 µg⋅g −1 feed concentration, the total permeation ux and separation factored reached to 3.12 kg•m − 2 •h − 1 and 19.76 respectively. This study attributed to the ease of synthesis, cost effectiveness and high e ciency of the proposed PEBAX/PAN lled-composite membrane, could be envisaged of potential interest for industrial applications involving denitri cation of fuel oils.

Declarations
Ethics approval and consent to participate The materials used in this study and the strategies followed in designing the system do not harm any living or non-living creature, both directly or indirectly, hence obeys all the ethical values.
Secondly, the research is conducted in the abovementioned laboratory with high accuracy and originality.
The data has also not been submitted to this journal before, or to any other journal in parts or as whole.
Therefore, the consent of participation to the journal by the authors is ethically approved.

Consent for publication
We the authors of the manuscript hereby transfer all the copy rights to the Environmental Science and Pollution Research under the helm of Springer.  Effects of ller contents on swelling degree of membranes (T=30 oC, C=2000 μg×g-1) Figure 6 Effects of feed temperature on swelling degree of the 4 wt.% lled membrane (C=3000 μg×g-1 ) (a); swelling degree of 4 wt.% lled membrane under different feed concentrations and immersion times (T=30℃) (b).

Figure 7
Effects of ller contents on pervaporation performance of the membrane (T=30℃, C=3000 μg×g-1 ) Figure 8 Pro le of pervaporation separation index as a function of ller content (T=30 oC, C=3000 μg×g-1) Figure 9 Effect of feed temperature on pervaporation separation performance of the 4 wt.% composite-lled membrane (C=2000μg×g-1) Figure 10 Arrhenius plots between the partial ux and operation temperature