Misagh Ahmadi,Sara Masoumi,Shadi Hassanajili*,Feridun Esmaeilzadeh

Chemical and Petroleum Engineering School,Shiraz University,71348-51154 Shiraz,Iran

A B S T R A C T Integrally skinned asymmetric gas separation membranes of polyethersulfone(PES)/polyurethane(PU)blend were prepared using supercritical CO2(SC-CO2)as a nonsolvent for the polymer solution.The membrane consisted of a dense and a porous layer,which were conjoined to separate CO2 from CH4.The FTIR,DSC,tensile and SEM tests were performed to study and characterize the membranes.The results revealed that an increase in SC-CO2 temperature causes an increment in permeance and a decrease in membrane selectivity.Furthermore,by raising the pressure,both permeance and selectivity increased.The modified membrane with SC-CO2 had much higher selectivity,about 5.5 times superior to the non-modified membrane.This higher selectivity performance compared to previous works was obtained by taking the advantages of both using partial miscible blend polymer due to the strong polar–polar interaction between PU PES and SC-CO2 to fabricate the membrane.The response surface methodology(RSM)was applied to find the relationships between several explanatory variables and CO2 and CH4 permeance and CO2/CH4 selectivity as responses.Finally,the results were validated with the experimental data,which the model results were in good agreement with the available experimental data.

1.Introduction

Carbon dioxide is an acidic gas and if combined with water,it shows severe corrosive behavior damaging pipelines and equipment.The presence of carbon dioxide in natural gas reduces its thermal value and leads to the loss of the pipeline's capacity.There are myriad ways to remove carbon dioxide;absorption processes such as Ben field process,low temperature processes,surface absorption methods such as pressure swing adsorption(PSA),temperature swing adsorption(TSA)and membrane methods[1].Each of the before-mentioned processes has their own strengths and weaknesses.However,the membrane process is the most cost-effective method,especially when the flow rate and the amount of present carbon dioxide are high.For commercial utilization of polymeric membranes in separation processes,those polymers with higher permeability and selectivity for the particular material to be separated,are often preferred.Moreover,the membrane material should have consistency,chemical resistance,and good thermal and mechanical stability under operational conditions.It should also be cost-effective and able to keep its separation properties under difficult and complex conditions and environments.Glass polymers have low permeability and high selectivity,but rubber polymers show high permeability and low selectivity in gas separation processes,particularly carbon dioxide separation.Therefore,combining these two groups of polymers may result in higher permeability along with high selectivity[2].In recent decades,several investigators worked on blended membranes due to their superior performances[3,4].

Hosseini et al.[5]applied a chemical modification on the blend membrane prepared from Matrimid and polybenzimidazole(PBI),which are completely miscible in a whole range of compositions at the molecular level.It was ascertained that the Matrimid/PBI(25 wt%/75 wt%)blend membrane cross-linked with p-xylene diamine represented the best performance with a H2/CO2selectivity of about 26,which is due to high chain packing density and limitations in segmental mobility of polymer chains.

Sales et al.[6]exploited polyurethane(PU)and polyurethane–poly(methyl methacrylate)(PMMA)blend membranes in gas separation studies.They investigated the effects of blend composition,temperature,and pressure on the permeability,diffusivity and solubility of CO2,H2,O2,CH4,and N2.It was observed that the permeability and diffusion coefficients of CO2,H2,O2,CH4,and N2decrease in PU/PMMA blend membranes with increasing PMMA content.

Rahimpour et al.[7]prepared the high performance blend membranes with commercial polyethersulfone and synthesized poly(amide–imide)(PAI)by a wet phase inversion technique in the presence of polyvinylpyrrolidone as the pore former.It was found that the hydrophilicity of PES membrane is improved by blending PES with PAI.

Lakra et al.[8]synthesized polyethersulfone(PES)-biopolymer blend membranes such as cellulose(CS)and chitosan(CH).They studied the solute rejection of organic acid and reducing sugars in alkaline/acid hydrolyzed solutions of rice husk.The results provided an insight that the biopolymer blend membranes improved flux without compromising the rejection efficiency.

In general,constructing composite polymeric membranes used for gas separation have their own special advantages and disadvantages.The benefits include lower cost of the separation process in comparison with processes using amine or other chemical methods.However,one of the problems usually arises in separation processes applying polymeric membranes is lower mechanical strength against large volume ofgas,which would eventually lead to membrane rupture.To overcome this problem,researchers try to increase the thickness of the composite membrane so that they can enhance the membranes mechanical strength and prevent their rupture.Even though,on the one hand,increasing the thickness of the dense layer composite membrane causes a reduction in separation efficiency;on the other hand,a reduction in the membrane thickness increases membrane separation efficiency,but it causes a sharp decrease in the membrane mechanical strength.Recently,to solve this problem,a synthesis method with supercritical fluid is implemented.In this method,not only controlling the membrane thickness is adjustable,but also it creates a porous layer in membrane structure which causes a great increase in permeability and selectivity[9].

Torres-Trueba et al.[10]built integrally skinned asymmetric polysulfone membranes prepared by a promising method using chloroform and supercritical CO2on one side of an already formed polysulfone dense film.This film eschews preparing a typical ternary casting solution.The results proved the success of the method persuading a very controlled asymmetry in a dense film.Also,it was shown that the thickness of the porous layer increases while the dense skin layer decreases as the chloroform/polysulfone mass ratio increases.

Cong et al.[11]presented a review paper about polymer–inorganic nanocomposite membranes.These types of membranes possess properties of both organic and inorganic membranes such as good permeability,selectivity,mechanical strength,thermal and chemical stability.Nanocomposite membranes with inorganic nanofillers embedded in a polymer matrix have potentials to provide economical,high performance gas-separation membranes.

Adib et al.[12]constructed integrally skinned asymmetric polyethersulfone/silica membranes using the supercritical CO2as the nonsolvent.It was proved that by changing the temperature and pressure,a very-controlled thickness in the dense film,and pore dimension in the porous layer can be achieved.Further,they demonstrated that the silica nanoparticles can increase the permeation flux of CO2,hence the membrane selectivity.

Shaver and co-workers[13]worked on Poly(2,6-dimethyl-1,4-phenylene oxide)blends with a poly(arylene ether ketone)to use in gas separation membranes.They showed that the crosslinked blends improve gas selectivities over their linear counterparts.

In order to present an optimized model,describing the behavior of system,recently many researchers applied the response surface methodology(RSM)to probe the relationships between several explanatory variables and one or more responses[14–16].The objective of this method is to simultaneously optimize the levels of variables to attain the best system performance.Salahi et al.[17]employed nano-porous sheet membranes of polyacrylonitrile(PAN)in order to treat the oily wastewater in a desalter plant.They utilized response surface methodology(RSM)based on central composite to design experiments,model and optimize permeation flux as a response.

In the present research,asymmetric polyethersulfone/polyurethane blend membranes consisting a dense layer on the top and a porous substrate layer were prepared to study CO2and CH4permeances.The dense layer was prepared by mixing solution method while the porous layer was formed by supercritical CO2as the nonsolvent.The porous layer was fabricated by pouring a certain amount of dimethylacetamide as a solvent on the dense layer,next it was contacted with SC-CO2in required temperature and pressure,then allowing the supercritical CO2expansion to occur.The effects of various parameters including the temperature and pressure of SC-CO2and the pressure of permeance measuring device on CO2and CH4gas permeance and the selectivity of membrane were assessed.The experimental pressures to construct the porous layer of membrane are 14,15 and 16 MPa and the temperatures are 35 °C,40 °C and 45 °C.The prepared membranes were characterized by Fourier transform infrared spectroscopy(FTIR),differential scanning calorimetry(DSC),the mechanical and scanning electron microscopy(SEM)tests for a better understanding of the effect of blending PU and PES polymers.The selectivity performance of modified PES/PU membrane with SC-CO2and non-modified one was compared.Moreover,to find the relationships between operating conditions and the responses of CH4and CO2permeation fluxes and their membrane selectivities,response surface methodology(RSM)which is an important branch of experimental design was employed.RSM obtained the relationships between several explanatory variables and CO2and CH4permeances and CO2/CH4selectivity as responses.Eventually,the results were validated with the experimental data.

2.Experiment

2.1.Materials

Polyethersulfone E6020P(MW=51000 g·mol-1)and polyurethane were prepared from BASF(Germany)and Coim Co.(Italy),respectively.N,N-dimethylacetamide(DMAc)was used as solvent provided by Merck(Germany),CO2gas(purity 99.99%)was purchased from Aboughadare(Shiraz,Iran),and CH4(purity 99.9%)was prepared from Air Products Co.

2.2.Preparation of membranes

The used dense blend membrane was prepared using solution casting method based on the conditions suggested by the RSM software which is expressed in Table 1 in design expert section.In order to synthesize the membranes,different concentrations of PES and PU in DMAc as solvent were mixed(see Table 2).The prepared solution was riled and heated up to 75°C for 24 h.Finally,the blend solution was shed on a glass plate and dried in an oven for 24 h at 80°C.Afterwards,the polymer film was located in a vacuum oven at 80°C for another 2 h to complete the removal of the remaining solvent.

Table 1Supercritical carbon dioxide densities at different pressures(T=45°C)

Table 2The composition of the prepared membranes

2.3.Preparation of integrally skinned membrane by supercritical fluid

After the preparation of the polymer membrane,a predetermined amount of DMAc as solvent(adjusted as the DMAc/PES mass ratio)was added to a prepared film membrane area which was located on a glass plate.The membrane was then put into a high pressure vessel,and CO2was sent into it until the desired pressure was achieved.The system formed a ternary mixture of polymer/solvent/non solvent and it should be remained in the vessel for 60 min.Supercritical CO2makes the solvent abandon the membrane,and the thin,porous layer was created on the surface.Then,the system was slowly depressurized for about 1 h at the experimental temperature,eventually,a dried asymmetric PES/PU blend membrane was obtained(see Fig.1).

Fig.1.Integrally skinned asymmetric membrane obtained by supercritical fluid.

2.4.The supercritical laboratory device for manufacturing membrane

Fig.2 shows the device used in membrane construction by supercritical carbon dioxide.Since the system pressure was provided by a pump,a liquefaction vessel was placed prior to the pump in order to convert CO2gas into liquid.The compressed liquefied carbon dioxide then entered into a spiral tube located in a hot-water tank with constant temperature.The process continued until CO2reached its supercritical temperature.The supercritical CO2entered a surge drum in order to avoid pressure fluctuations,and it finally went to the membrane forming chamber whose temperature was sighted and set with a temperature controller and an indicator.

2.5.Membrane characterization

2.5.1.Fourier Transform Infrared spectroscopy(FTIR)

ABIO-RED(USA)FTIR spectrometer was used to evaluate the Fourier transform infrared(FTIR)spectra of the prepared thin samples of membranes in the frequency range of 400–2000 cm-1.

2.5.2.Differential scanning calorimetry(DSC)

The thermal properties of prepared samples were carried out by a Mettler Toledo DSC822e calorimeter.The temperature analysis is non-isothermal for measuring Tgaccording to the ASTMD 99–3418 from 0 to 300 °C with the heating rate of 5 °C·min-1.

2.5.3.Scanning electron microscopy(SEM)

The morphologies and surfaces of the synthesized membranes were scanned by a Philips XL30(Philips,Netherlands)scanning electron microscope(SEM).The membranes were fractured in liquid nitrogen and coated with gold by a device manufactured in Bal-Tec Co.of Switzerland.

2.5.4.Porosity measurement

The membrane porosity was measured according to the liquid displacement[18].Since ethanol penetrates easily into the pores of membrane without inducing swelling or reducing the matrixes,ethanol was elected as the displacing liquid.The membrane was plunged into a known volume of ethanol(V1)in a graduated cylinder until it saturated.The total volume of ethanol and ethanol impregnated membrane was noted as V2.After bringing out the impregnated membrane from the cylinder,the residual volume was noted as V3.The total volume of the membrane was evaluated from the below expression:

where V2-V1is the initial volume of the membrane,V1-V3is the penetrated ethanol into the membrane.

The porosity of the membrane was calculated by the following equation:

The estimated porosities were exposed in Table 3.

Fig.2.The schematic diagram of the supercritical laboratory device for manufacturing membrane.

Table 3Membrane porosities in used operating conditions

2.5.5.Tensile test

A SANTAM-STM(150)Load Cell-S 1000 tensile test machine was applied to test the mechanical properties of the prepared membranes.All the samples should be completely dried and cut in the standard shape before being tested.

2.5.6.Gas permeance measurement device

The gas permeance experiment is the main section of this study in which the permeances of CO2and CH4were measured.The individual experiments were fulfilled for methane and carbon dioxide,respectively;and repeated three times to achieve high accuracy.It should be noticed that between these two stages,1 h was required to let the previous gas get out.All the experiments were carried out at constant pressures of 0.6,0.8 and 1.0 MPa and the temperature of 29°C.A circular membrane disk with an effective permeation area of 12.5 cm2was located in a permeation cell,which its permeate side was preserved in the atmospheric pressure.

The gas permeation flux can be specified from Eq.(3)[19]:

where P/l is the pressure-normalized flux(10-6cm3(STP)·cm-2·s-1·(cmHg)-1,GPU),Qiis the flow rate of the permeate gas“i”passing through the membrane(cm3(STP)·s-1),ΔPiis the pressure drop of component“i”across the membrane(cm Hg,1 cm Hg=1333.22 Pa)and A is the effective membrane area(cm2).

GPU is the gas permeance unit and is a common used unit in reporting the pressure-normalized flux.

The membrane selectivity for the two gases can be computed based on the following equation[20]:

Throughout this paper,i and j are defined for CO2and CH4.

3.Design Expert

In order to extract mathematical model, find the most impact factor,the response surface methodology was used.It can determine the number of experiments and the surface of each variable in all experiments.This method uses the number of variables and their determined maximum and minimum limits to design the experiment matrix.As the number of variables is large,this method is preferable in comparison to the voluminous method such as full factorial.

Central composite design,Box–Behnken,and Doehlert matrix[21]are the three main methods of response surface methodology;among these three methods,the central composite design was elected.A central composite design with four independent variables was applied to obtain the effect of variables on CH4and CO2permeation flux and their membrane selectivity.The four factors,or independent variables,were coded at three levels between-1 and+1,where-1 corresponds to the minimum and+1 corresponds to the maximum value of each variable,as noted in Table 4.

Table 4Codes,ranges and levels of independent variables in RSM design

The test factors were coded according to Eq.(5)[22]in the regression equation:

where xiis the coded value of the ith independent variable,Xiis the natural value of the ith independent variable,Xixis the natural value of the ith independent variable at the center point andΔXiis the step change value.

The response is depended on the selected factors and linear and quadratic terms,as presented below[23]:

In this equation,Y is the response,xiand xjare the independent variables,C0is the constant coefficient,Cj,Cjjand Cijare the coefficients of linear,quadratic and interaction effects,respectively,and eiis the error.

The proposed models for responses are presented in the Results and Discussion section.In order to check the performance and accuracy of each model,the charts and three-dimensional diagrams will be exhibited to compare the actual values with the predicted result's values.

4.Results and Discussion

4.1.Membrane characterization

4.1.1.Fourier Transform Infrared spectroscopy(FTIR)

The results related to the FTIR test of pure polyurethane,polyethersulfone and PES/PU blend membrane were shown in Fig.3.Results showed that there are different bands for PU related to the C=Ostretching(1600–1800 cm-1),C--O--C stretching(1000–1200 cm-1)and υ(C--N)+δ(N--H)of urethane groups(1241 cm-1).The band at carbonyl stretching region(1600–1800 cm-1)is resolved into two spectral components:one at 1682 cm-1assigned to the hydrogen bonding between hard segments and the other at 1718 cm-1which corresponds to the stretching vibration of free carbonyl group of hard segments[24].According to this figure,for pure PES,there are two picks in the wave number of about 1150 cm-1and 1321 cm-1which are assigned to the symmetric and asymmetric stretches of a sulfone group,respectively.The C--O--C stretching peaks appear at 1324 cm-1and 1239 cm-1.Aromatic bonds belong to C=C and benzene ring is located at 1485 cm-1and 1578 cm-1,respectively[25].Comparison of the FTIR results reveals that the obvious pick 1730 cm-1of the PES/PU spectra corresponded to C=O free bond of polyurethane and its presence in the blend[26].It can be found from Fig.3 that absorption intensity,and the wave number of polyethersulfone are changed under the effect of polyurethane.Therefore,we concluded that polyurethane and polyethersulfone have intermolecular interactions[25].The high porosity and high free volume fraction in the PES/PU blend membrane structure cause a higher transition and a lower absorption at Fourier transform infrared spectroscopy compared with the pure polyethersulfone membrane[27].

Fig.3.FTIR spectra of the membranes prepared from PU,PES and PES/PU(1.67%PU).

4.1.2.Differential Scanning Calorimetry(DSC)

The compatibility of polyethersulfone and polyurethane in PES/PU blend can be achieved by the DSC experiment and measuring the glass transition temperature(Tg).Tgin the blend polymers is very sensitive to blend composition and also indicates partial miscibility.When two polymers are partially miscible,each phase may contain a significant fraction of the other component[28].Fig.4 presents DSC thermograms of PU and PES/PU blends with various PU concentrations.Obviously,two Tgtransitions can be seen in PES/PU blends(see Table 5).It is found that the Tgof PU in the blends which is-40°C in the pure membrane shows shifting towards right may be due to the strong interaction between PU and PES.At the same time,a slight shift down in Tgcan be observed for PES phase in the blends.A lower Tg,a lower thermal energy,is needed to overcome the interactions between polymer chains;consequently,the mobility of polyethersulfone chains increases in the presence of polyurethane.

Fig.4.DSC thermograms of PES/PU membranes containing different mass percents(1.72%,3.44%and 5.17%)of PU.

Table 5Glass transition temperature of PU,PES and PES/PU blend membranes

4.1.3.Tensile test

The PES/PU blend specimens were made for this experiment and the force with the speed of 200 mm·min-1was imposed on them.The results of tensile strength experiments for pure PES and polyurethane compositions of 1.72%,3.44%and 5.17%in PES/PU blend membranes were shown in Table 6.If the tensile strength and the strain at break are low,the specimen breaks earlier which is an inappropriate factor for membrane.According to Table 6,the pure polyethersulfone specimen possesses low tensile strength and strain at break;while the PES/PU specimens have higher ones by adding PU into the PES;consequently,the elongation at break increases significantly due to the higher elasticity of the PU in comparison with the PES.

Table 6The results of the tensile strength and strain at break of different PES/PU blends(T=25°C,P=0.1 MPa)

4.1.4.Scanning Electron Microscopy(SEM)

The SEM results of cross-section of the integrally skinned asymmetric PES/PU blend membrane at the 14 MPa and the temperatures of 35°C,40 °C and 45 °C were shown in Fig.5.By increasing the temperature,the solubility of solvent in SC-CO2slakes and the amount of solution in the membrane surface decrease,which leads to reduce the density of the dense top layer and as a result,CO2membrane selectivity was dropped.According to Fig.5,with increasing the temperature,the pores of the porous layer become larger and more irregular.

As can be observed in Table 7,an increase in temperature at constant pressure causes a decrease in SC-CO2density;consequently,the diffusivity of SC-CO2into the membrane descends.

The densities presented in Table 7 were evaluated from the following equation[30]:

Table 7Supercritical carbon dioxide densities at different temperatures(P=14 MPa)

where T,P and ρ are the temperature(K),pressure(MPa)and density(kg·m-3),respectively.Therefore,we can see in Fig.5 that the porosity of the membrane decreases and the pores become larger and more irregular[31].

The SEM results of cross-section of the integrally skinned asymmetric PES/PU blend membrane at 45°C and pressures of 14,15 and 16 MPa were shown in Fig.6.By increasing the pressure at constant temperature,the density of SC-CO2increases(see Table 7);therefore,the solubility of solvent in SC-CO2grows up.This factor makes more solvent near the top surface of the dense layer diffuse rapidly into SC-CO2,thus the polymer concentration increases at the interface.Consequently,the solidification rate of the polymer at the interface becomes faster,which makes the skin layer thicker and improves the selectivity of the membrane.According to Fig.6,it can be observed that a rise in pressure can form more symmetric and regular pores in the porous layer of membrane,which can improve the permeability[32].

The SEM results of cross-section of the integrally skinned asymmetric PES/PU blend membrane at the compositions of 1.72%,3.44%and 5.17 wt%,the temperature of 45°C and the pressure of 14 MPa were shown in Fig.7.As illustrated in this figure,by raising the percentage of PU,the skin layer thickness increases due to the increase in the solution viscosity[27].On the other hand,by increasing the viscosity,the shrinkage stress of solidified polymer cannot be released by the creep relaxation of polymer,therefore,the skin layer ruptures[33].As it is observed from Fig.7,the dense layer in PES/PU blend membrane with 5.17%PU includes large pores which causes a reduction in the selectivity of membrane and an enhancement in permeability.Referring again to Figs.4–6,all membranes show that the cross section comprised a skin layer with sponge-like substructure that supports the top layer.According to Ismail et al.[34],asymmetric membranes usually contain few defects on their top layer,which are attributed to the incomplete coalescence of the nodule zones of the composed skin layer.As it is shown in Fig.8(SEMimages of the top surface),at the optimal condition(T=45°C,P=14 MPa and 1.72%PU)with respect to permeation and selectivity results,a nearly defect-free surface is achieved for PES/PU membrane(Fig.8A).In other hand,the lowest CO2/CH4selectivity is observed for the membrane with the concentration of 5.17%PU(T=45°C,P=14 MPa)due to the creation of much more imperfections(defects)in the skin layer(Fig.8B).

Fig.6.Cross section SEM images of PES/PU membranes at different pressures A)14 MPa,B)15 MPa and C)16 MPa(T=45°C,1.67%PU).

Fig.7.Cross section SEM images of PES/PU blend membranes containing A)1.72,B)3.44 and C)5.17 wt%PU(T=45°C,P=14 MPa).

Fig.8.SEM images of the top surface of PES/PU membranes.A)T=45 °C,P=14 MPa and 1.72%PU and B)T=45 °C,P=14 MPa and 5.17%PU.

Fig.9.Temperature effects of SC-CO2 on CO2 permeability(P=14 MPa,1.72%PU GPU:gas permeation unit(1 GPU=10-6 cm3(STP)·cm-2·s-1·(cm Hg)-1)).

Fig.10.Temperature effects of SC-CO2 on ideal selectivity of CO2 over CH4(P=14 MPa,1.67%PU).

4.2.Effect of temperature changing of supercritical CO2

By considering Figs.9 and 10 as previously mentioned,by an increment in the temperature of SC-CO2,CO2permeance increases while the membrane selectivity reduces.According to the experimental results for the PES/PU blend membranes,which were conducted at different measuring device pressures of 0.8,1.0 and 1.2 MPa,increasing pressure at 1.72%of PU can improve permeability for CO2and CH4.

4.3.Effect of pressure changing of supercritical CO2

By considering Figs.11 and 12,with raising the pressure of SCCO2,both CO2permeability and CO2/CH4selectivity increase.According to the experimental results for the PES/PU two-layer membranes which were conducted at different measuring device pressures of 0.8,1.0 and 1.2 MPa,the increase in pressure at special percent composition can improve permeability for CO2and CH4.

4.4.Effect of changing the polymer blend composition

By considering Figs.13 and 14,it can be observed that with an increment in the percentage of PU,the permeability of CO2improves,while the membrane selectivity diminishes.According to the experimental results for the PES/PU two-layer membranes which were conducted at different measuring device pressures of 0.8,1.0 and 1.2 MPa,the increase in pressure at special percent composition can improve permeability for CO2and CH4.

Fig.11.Pressure effects of SC-CO2 on CO2 permeance(T=45 °C,1.67%PU GPU:gas permeation unit(1 GPU=10-6 cm3(STP)·cm-2·s-1·(cm Hg)-1)).

Fig.12.Pressure effects of SC-CO2 on ideal selectivity of CO2 over CH4(T=45°C,1.67%PU).

Fig.13.Effects of polymer blend compositions on CO2 permeance(P=140 MPa,T=45 °C GPU:gas permeation unit(1 GPU=10-6 cm3(STP)·cm-2·s-1·(cm Hg)-1)).

Fig.14.Effects of polymer blend compositions on ideal selectivity of CO2 over CH4(P=14 MPa,T=45°C).

4.5.The comparison between modified PES/PU membrane by SC-CO2 and non-modified one

As it can be concluded from Figs.10,12 and 14,the best selectivity of CO2/CH4may be achieved at low SC-CO2temperature,high SC-CO2pressure,low wt%PU in PES/PU blend and high measuring device pressure.Therefore,an experiment was prepared at 35°C,16 MPa and 1.72 wt%PU in PES/PU membrane in measuring device pressure of 1.2 MPa to compare the modified PES/PU membrane with non-modified one.Table 8 shows clearly the improvement in selectivity of modified membrane about 5.5 times superior to non-modified one.

Table 8The comparison between modified PES/PU membrane by SC-CO2 and non-modified one

4.6.Design expert

The design expert is such that even without repeating the experiment,the reliable statistical results are obtained.Therefore,this method facilitates the research process,reduces time and costs.With the results of design expert,a total of 25 experiments were designed,which were inserted in Table 9.

4.6.1.Model development

In order to explore the relationship between the explanatory variables and response variable,the response surface methodology(RSM)was applied.The results were analyzed using statistical method based on RSM with three levels and four variables.These four independent variables include supercritical temperature and pressure,blend composition and pressure of permeability measuring device.The central composite design was used to evaluate the quadratic models for the permeation flux and selectivity.The proposed equations for predicting CO2and CH4permeances and the selectivity of CO2over CH4are held as follows:

where T,P,P2and C are the temperature and pressure of supercritical CO2,pressure of the permeability measurement and PES/PU composition,respectively.

The proposed model should be validated with the experimental data;therefore,the R-squared(R2)and the adjusted R-squared(adj R2)of the proposed model were presented in Table 10.Based on the reported values,there is a good accordance between the experimental data and estimated values by the model.

4.6.2.ANOVA analysis

The analysis of regression and variance(ANOVA)was carried out to demonstrate the effects of variables as linear,quadratic or interaction coefficients on the response.Analysis of variance for CO2and CH4permeances and the selectivity are tabulated in Tables 11–13.F-value shows the influences of variations of independent variables on the response of dependent variable.As one parameter has a higher F-value,it has more effect on the response,consequently,the SC-CO2temperature(F-value=587.92)has the most influence on CH4permeance,the SC-CO2pressure(F-value=484.78)on CO2permeance and the SC-CO2temperature(F-value=317.68)on the selectivity.The Prob＞F of less than 0.05 represents that the model terms are significant,whereas the values greater than 0.1000 are not significant.

Table 9The experimental data of permeances and selectivities used in design expert analysis

Table 11ANOVA for the regression model and respective model terms for CH4 permeation flux

Table 12ANOVA for the regression model and respective model terms for CO2 permeation flux

Table 13ANOVA for the regression model and respective model terms for selectivity

4.6.3.Adequacy of the models

Generally,it is momentous to acknowledge the model to make sure that it exhibits a competent approximation to the actual data.Diagnostic plots such as the predicted versus actual values are analyzed to judge the model satisfactoriness.The data points stay close to the straight line and demonstrate a sufficient agreement between the real data and predicted values calculated from the models(Fig.15).

4.6.4.Effect of process variables on the CO2permeance

A plot was drawn to present the effects of two variables simultaneously on the response while the other one is constant.Fig.16 presents the effects of pressure and temperature of SC-CO2on the CO2permeance.By comparing the slope of the SC-CO2pressure and temperature,it is noticed that the SC-CO2pressure has the greatest impact on CO2permeance which was previously proved.

5.Conclusions

Integrally skinned asymmetric polyethersulfone/polyurethane membranes were prepared by a novel approach to the dry/wet phase inversion processes using DMAc as a solvent.Supercritical CO2was applied to make a porous layer on one side of an already formed PES/PU blend dense layer.FTIR study of PES/PU blend membrane indicated that polyethersulfone and polyurethane have an appropriate compatibility at low weight percent of PU.The DSC test confirmed a glass transition temperature for PES/PU blend,which means a good compatibility between two polymers.The effects of experimental operating conditions such as the SC-CO2pressure,temperature and the composition of polyurethane on gas transport and membrane morphology were assessed.The SEM results showed that by increasing the temperature which was varied from 35°C to 45°C,it is possible to induce a very-controlled asymmetry in a dense film and pore sizes,therefore,the permeance and selectivity performance of the membranes enhance.By raising the pressure,which was varied from 14 to 16 MPa,the pore shapes become more symmetrical and more regular,which can improve the permeance and selectivity.The best ideal selectivity of 60.25 was obtained for CO2/CH4at 35°C,16 MPa and 1.72 wt%PU in PES/PU membrane in measuring device pressure of 1.2 MPa,which has a significant improvement compared with the previous works.This higher selectivity performance compared to previous works was obtained by taking the advantages of both using well miscible blend polymer and SC-CO2to fabricate membrane.

Fig.15.Comparison between the predicted and actual values of(A)PermCH4,(B)PermCO2,(C)selectivity of CO2/CH4.

Fig.16.Effects of pressure and temperature on CO2 permeance.