Gholamhossein Sodeifian*,Mojtaba RajiMorteza AsghariMashallah Rezakazemi,Amir Dashti

1Department of Chemical Engineering,Faculty of Engineering,University of Kashan,Kashan 87317-53153,Iran

2Faculty of Chemical and Materials Engineering,Shahrood University of Technology,Shahrood,Iran

Keywords:Membrane Polyurethane Gas separation Permeability Selectivity Mathematical model

A B S T R A C T SAPO-34 nanocrystals(inorganic filler)were incorporated in polyurethane membranes and the permeation properties of CO2,CH4,and N2gases were explored.In this regard,the synthesized PU-SAPO-34 mixed matrix membranes(MMMs)were characterized via SEM,AFM,TGA,XRD and FTIR analyses.Gas permeation properties of PU-SAPO-34 MMMs with SAPO-34 contents of 5 wt%,10 wt%and 20 wt%were investigated.The permeation results revealed that the presence of 20 wt%SAPO-34 resulted in 4.45%,18.24%and 40.2%reductions in permeability of CO2,CH4,and N2,respectively,as compared to the permeability of neat polyurethane membrane.Also,the findings showed that at the pressure of 1.2 MPa,the incorporation of 20 wt%SAPO-34 into the polyurethane membranes enhanced the selectivity of CO2/CH4and CO2/N2,14.43 and 37.46%,respectively.In this research,PU containing 20 wt%SAPO-34 showed the best separation performance.For the first time,polynomial regression(PR)as a simple yet accurate tool yielded a mathematical equation for the prediction of permeabilities with high accuracy(R2＞99%).

1.Introduction

At the moment,many types of research have been focused on climate changes and global warming[1-6].Today,because of CO2emission from power plants,separation of CO2from flue gas streams is important more than ever[7-12].Nowadays,the membrane separation technology has got a growing interest because of its potential energy saving and environmental impact reducing capabilities[13-15].Currently,theworld consumption of natural gasis annually over 3.1 trillion cubic meters(110 trillion standard cubic feet).Raw natural gas usually consists of CH4(30%-90%),with different light hydrocarbons,like C2H6and C3H8and other heavier hydrocarbons.Separation science like adsorption,absorption and membrane technology[16-18]is emerging as new tools in the separation of CO2or other gases.

Membrane technology has emerged as a promising approach for gas separation[19-22].Among gasseparation fields,usingpolymeric membranes for CO2separation has attracted much attention due to the wide variety of its potential applications,such as separation of CO2/H2in gas refinement,CO2/N2in carbon capture,CO2/CH4in natural gas refinement,and CO2/O2in food packaging[23].The natural gas has the impuritieslikeCO2andH2Swhichresultsintheirunpleasingandunfavorable effectsincluding reducing thethermalvalueof naturalgas,difficulties in gas transfer by pipelines,corrosion during transfer and distribution and theirnegativeenvironmentaleffects.Therefore,inrecentyearsmanyattemptshavebeendonetoremovetheseimpuritiesfromthenaturalgas.New research indicates that PU membranes are great candidates for gas separation membranes due to their suitable mechanical properties,thermal stability along with high gas separation efficiency.PU is a polymer with exceptionally considerable flexibility in structure and properties thatare composed of alternating urethane or urea as hard segments and Polyol(polyether/polyester)as a soft segment.The hard segments of urethane/urea are shaped by extending an extremity Diisocyanate with a low molecular weight Diol/Diamine.The soft segment is comprised of high molecular weight polyether/polyester groups[24].The properties of PUs are simply fitted,enabling restrained changes in the length of the Polyol chain,and also changing the proportions andchemicalcharacteristicsofthecomponentswhichcreatetheflexibleandrigid segments of the polymer chain[25].Inorganic fillers are porous or nonporous.Theideaofaddinginorganicfillerstopolymermatrixinorderto enhance separation and mechanical properties has been explored[26,27].It should be noted that the effect of interaction between the particles and the polymer chain and functional groups on the surface of the inorganic phase must be considered[28-32].Addition of particles can improve the separation properties of MMMs by raising matrix tortuous surfacesanddecliningdiffusionoflargergasmolecules.Toimprovepermeation performance,the notion of introducing inorganic filler to the matrix of the polymer was proposed.Addition of various particles like zeolites,carbon molecular sieves and activated carbons,mesoporous materials,non-porous silica,metallic organic frameworks(MOFs),and graphite[33]into the matrix of polymer was investigated.

Gas permeation characteristics of PES-SAPO-34-HMA were investigated by Elif et al.[34].The results showed that the permeability of all the gases through PES-SAPO-34-HMA membranes was significantly more than those through PES/HMA membranes.New MMMs based on polymerizable room-temperature ionic liquids and SAPO-34 as filler were studied[35].It was shown that increasing of ionic liquids in the MMMs,increases the CO2permeability.

The effect of the addition of SAPO-34 as filler on the gas permeation properties of other polymers such as polyetherimide[36],PEBAX[37],polyethersulfone[38],and polysulfone[39]has been researched.Gas separation studies in the field of adding SAPO-34 particles into the polyurethane matrix have been not reported before.

The effect of silica particles on the permeation properties of polyurethanemembraneswasinvestigated[40].Theobtainedresultsindicated the reduction in permeability of CO2,CH4,O2and N2gases.But enhancementofCO2/CH4,CO2/N2andO2/N2selectivitybyincreasingsilicacontentwasobserved.AdditionofaluminananoparticlestothepolyurethanematrixwasstudiedbyAmeri etal.[41].Theresultsshowedthe reduction in permeability of CO2,CH4,O2and N2gases and that this is due to the presence of alumina nanoparticles in the polyurethane soft segment.Enhancement of CO2permeability and CO2/CH4selectivity by the addition of silica nanoparticles was reported by Hassanajili et al.[42].Effect of nano-zeolites 4 A and ZSM-5 on gas separation properties of polyurethane membranes was investigated[43].Results reported by Sadeghi et al.showed that by increasing silica content in polyurethane based polycaprolactone,the permeability of CO2,CH4,O2and N2gases decreased[44].In another work,preparation and investigation of gas separationproperties of polyurethane-TiO2nanocomposite membranes were studied[45].The results showed a reduction in permeability of all gases in nanocomposite membranes.

Gasseparationprocessbymembranesisanonlinearsystemwithnumerous parameters,huge time delay,strong coupling,and extreme uncertainty[46-51].Thus,predictions of the conventional models are not acceptable[52-57].Peer et al.[58]demonstrated a theoretical model based on Coker,Freeman,and Fleming's study and an artificial neural network for calculating the necessary membrane area in the separation of H2from CO by glassy polyimide membranes.They noted an excellent consistency between experimental data and ANN results.

Inthisresearch,thegasseparationpropertiesofPU-SAPO-34MMMs were explored.The effects of operational parameters such as pressure(0.6,0.8,1.0 and 1.2 MPa)and SAPO-34 loading(5%,10%and 20%)on permeability(CO2,CH4,and N2gases)and selectivity of those membranes were further studied.Siliconaluminophosphate crystals(SAPO-34)are non-zeolite molecular sieves in which silica is replaced by aluminumphosphate,withamicrocavityandcrystallineproperties.Zeolite type SAPO-34,for its high thermal and chemical stability,molecular sieve properties,narrow particle size distribution and the high surface area has been considered for separation applications,catalytic applications,and absorption technologies.By reviewing recent studies,it can be seen that the SAPO-34 molecular sieves exhibit better performance in the process of CO2separation from CH4due to their specific structure and the relatively high ratio of silica to aluminum compared to other adsorbents[59].Therefore,in this research,we choose SAPO-34 as afiller.Also,in this paper,a simple yet accurate PR model,for the first time was developed to predict gas permeabilities during membrane process.

2.Experimental

2.1.Materials and methods

Commercial polyurethane was supplied by Coim S.p.A Co.(Italy,LPR7025).Dimethylformamide(DMF,Merck)was used as a solvent to provide pure and MMMs.Aluminiumisopropoxide(98 wt%,Lobo Chem),phosphoric acid(88 wt%aqueous solution,Merck)and TEOS(tetraethyl orthosilicate,99 wt%solution,Sigma-Aldrich)were used as received.CO2gas(99.99%purity)was purchased from Farafan Gas Co.(Tehran,Iran)and CH4gas(99.995%purity)was acquired from Air Products Co.(Tehran,Iran).

2.2.Synthesis of SAPO-34 pseudo-zeolite

Fine particles of SAPO-34 were synthesized via a hydrothermal method[60].The required resources for the formation of SAPO-34 are provided in Table 1.The premixed(gel-like)solution was prepared by mixing deionized water,and then phosphoric acid and aluminum isopropoxide were slowly added to the solution under vigorous stirring for 5 h,followed by the addition of silica gel and TEOS.Afterwards,the mixture was aged for 24 h under agitation at room temperature.The obtained gel was then placed into a Teflon-lined autoclave wherein the crystallization process was carried out at 190°C for 48 h.The synthesized sample was subsequently washed with deionized water and three times centrifuged until the solution pH value was measured to reach about 7.To remove organic template crystal lattice,the calcination process was performed at 600°C for 5 h.

Table 1 Synthesis conditions of SAPO-34

2.3.Membrane preparation

Thebare PU and PU-SAPO-34 membranes were prepared asfollows.To prepare pure PU membrane,polyurethane solution was provided by dissolving an appropriate polymer concentration(3 wt%)in DMF at atmospheric pressure and 70°C temperature within 3 h.The prepared solution was filtered and cast onto a Petri dish followed by heating at 70°C for 48 h.To prepare the MMMs,at first,SAPO-34 nanocrystals were dispersed into DMF and stirred at room temperature for 12 h.The solution was then sonicated for 1 h,to obtain a fairly homogeneous pseudo-zeolite suspension to which the PU was subsequently added whilestirringthesolutionat70°Cfor3 h;thesolutionwasthenfiltered to remove undissolved contents.The bubble-free polymer solution was cast onto clean Petri dishes to the desired thickness before being incubated at 70°C for 48 h.

2.4.Membrane characterization

Chemical structure of the membranes was analyzed by Fourier transform infrared spectroscopy(NICOLET Magna IR 550)within the scanning frequency range of 400-4000 cm?1.The thermal stability of polyurethane MMMs was investigated using thermogravimetric analysis(TGA)within the temperature range of 30-800 °C at 10 °C·min?1,under an atmosphere of nitrogen(Mettler TOLEDO model SDTA 851).

Atomic force microscopy(AFM)analysis was carried out to study the morphology of the prepared membranes and to investigate the presence of nanocrystals on the membrane surface.Accordingly,the membranes were cut into 1 cm×1 cm specimens ready to be photographed(Full plus,Iran).

The membranes'morphologies and SAPO-34 nanocrystals'distribution were investigated by scanning electron microscopy(SEM)on a Hitachi S-4160(Japan).In order to incorporate membranes with polyurethane,SEM cross-section of polyurethane containing20%SAPO-34 is provided.

To investigate the crystalline structure of the used SAPO-34,X-ray diffraction(XRD)patterns were recorded on a Philips X'pert pro MPD wherein Cu Kαradiation served as the X-ray source.Using the Debye Scherrer Equation[61],SAPO-34 crystal size was estimated.

Toanalyzethegasseparationcapabilityofthepreparedmembranes,the permeation test of CO2CH4and N2gases was performed.In this respect,the constant pressure/variable volume method was used to run the gas permeation tests[62].The flux of the permeated gas was measured by a U-shape flow meter.The typical membrane area in the test cell was 18.34 cm2.The schematic view of the experimental setup isillustratedinFig.1.Thegaspermeabilityofmembraneswasestimated using the following equation[63]:

Fig.1.Apparatus used for measuring gas permeability.

where P is permeability expressed in Barrer(1 Barrer=10?10cm3·cm·cm?2·s?1·(cm Hg)?1),q is the flow rate of the permeate gas passing through the membrane(cm3·s?1),membrane thickness is 1(cm),p1and p2are the absolute pressures of feed side and permeate side,respectively(cm Hg)and A is the effective membrane area(cm2).The ideal selectivity,A/B(the ratio of single gas permeability)of membranes was calculated by pure gas permeation experiments.

In this study,error for gas permeation results was about 5%for all the samples and permeability test for all samples repeated twice.Also,the thickness of the membranes in this study was between 60 and 70 μm.

3.Results and Discussion

3.1.FTIR analysis

FTIR spectra can be used to trace molecular interactions,and corresponding changes,among differentcomponentsin the membranes.The FTIR results of pure and MMM polyurethane membranes are shown in Fig.2.The polymer used in this study exhibited an N--H stretch absorption at about 3300 cm?1,a symmetrical CH2absorption at about 2870 cm?1,and an asymmetrical CH2absorption at about 2950 cm?1.Also,the aromatic C--C absorption could be identified at about 1600 cm?1.The peak at about 1540 cm?1was attributed to the second-type amide.In PU,two peaks were attributed to carbonyl groups(C＝O),namely peak at a lower frequency(around 1700 cm?1)and one at a higher frequency(around 1740 cm?1)referring to bonded and free carbonyl groups,respectively.

ToinvestigatetheeffectsofSAPO-34onthephaseseparationofhard andsoftsegmentsofthepreparedmembrane,theattributedpeaktothe carbonyl group(C＝O)should be considered[64].Upon incorporation oftheSAPO-34nanocrystalsintothepolymermatrix,thecorresponding peaks to bonded carbonyl exhibit a small shift toward lower wavelengths.Also,byadding SAPO-34 nanocrystals,theseverity ofhydrogen bonded carbonyl groups lessens.The corresponding peak to free carbonyl shifts toward the hydrogen-bonded carbonyl bond,eventually fully disappeared as the SAPO-34 content is increased.Hence,in fact,the ester groups of the soft segment in Polyol of polyurethane interact with the OH groups on SAPO-34 nanocrystals which less interact with the NH groups on polyurethane in the hard segment[45].The reduced permeability of gases by adding SAPO-34 nanocrystals confirms the likelihood of SAPO-34 distribution in the soft segment of the polymer.

Fig.2.FTIR spectra of pure and PU-SAPO-34 MMMs.

Fig.3.SEM image of the synthesized SAPO-34.

3.2.Morphology characterization

TheSEMimageofsynthesizedSAPO-34isshowninFig.3.Theimage shows that SAPO-34 is crystalline and the cubic crystals of SAPO-34 are obviously visible in this image.

SEM micrographs of pure and PU-SAPO-34 MMMs are shown in Fig.4.The SEM images of the PU-SAPO-34 MMMs illustrate suitable dispersion of pseudo-zeolite particles within the membrane matrix.Small aggregations are observed in the polyurethane containing 20 wt%SAPO-34 which can be attributed to the density difference between the polymer and particles.Also,the distribution of SAPO-34 nanocrystals in the prepared membrane is evident.The cross-section images of the PU20 are presented in Fig.4 and can be determined that this membrane is evidently dense.

3.3.Thermogravimetric analysis(TGA)

By using a TGA analyzer,the thermal stability of all membranes was evaluated.The PU membrane decomposition began at about 270°C and terminated at about 440°C.As shown in Fig.5,an increase in the SAPO-34 loading within the MMMs results in a decrease in the mass loss slope.Slope differences indicate two different types of mass reductions,with the first being related to the breakage of urethane bonds,while the second one stems from the thermal decomposition of polyol[44].As is clear in Fig.5,by increasing SAPO-34 content theslopeofmassreductiondecreasesandthermalstabilityofnanocompositemembranesimproved.Also,theamountoftheremainingmassof nanocomposite membranes over 500°C is equal with the theoretical SAPO-34 mass percentage that used in this research.According to these phenomena,itcan be concludedthat duringnanocomposite membrane fabrication,SAPO-34 nanocrystals were present.

Fig.4.SEM micrographs of pure and nanocomposite membranes:bare PU(a),PU10(b),PU20(c),cross-section of PU20(d).

Fig.5.Thermogravimetric analysis for polyurethane MMMs.

3.4.XRD characterization of the synthesized SAPO-34

The XRD results of the synthesized SAPO-34 pseudo-zeolite are presented in Fig.6(a)where in every peak is associated with pseudozeolite SAPO-34.The highest peak,however,is observed at 2θ =9.5°,being in good agreement with the standard X-ray pattern for SAPO-34 particles[60].Peaks appeared at the 2θ values of 12.8°,13.5°,16°,20.6°,22°,23°,25°,25.8°,29.4°,30.5°and 31.6°,the other peaks were also inconsistency with the standard diagram of SAPO-34.Also,the XRD patterns of the pure membrane and PU with a loading of 20 wt%are presented in Fig.6(b).

The XRD results for pure polymer membrane and PU with a loading massof20%areshowninFig.6(b).AccordingtoFig.6,thepurepolymer has a broad peak in the range of 19°to 23°and that appeared peak can beattributedtothestructureofthebulkortoosmallcrystalsdistributed in the polymer or to the distribution of crystalline spheres.As it is seen,the amount of peak intensity has been increased by adding SAPO-34 to thepolymermatrix[65].Also,withtheadditionofSAPO-34nanocrystal to the polymer matrix,a small amount of peaks has been moved to the right.Ontheotherhand,themotionofthepeakstotherightor,inother words,the reduction of d-spacing,represents the collapse of the atomic layers of the crystals,which can indicate the proper distribution of SAPO-34 particles in the polymer structure and this can improve the polymer structure and the permeability of the gas inside the polymer[66].

Fig.6.(a)XRD patterns of the synthesized sample(SAPO-34 nanocrystals).(b)XRD pattern of pure and MMMs.

Regarding the set of XRD patterns for a pure polymer membrane,it is observed that the polymer structure under membrane construction(including heating and ultrasonic operations)is not damaged.

3.5.Atomic force microscopy(AFM)analysis

By using AFM analysis,the type of aggregation,roughness,and an effective surface can be achieved.The 2D and 3D images of AFM of the neat PU and PU containing 20%SAPO-34,are shown in Fig.7.The results demonstrated that SAPO-34 nanocrystals tend to increase the surface roughness,with average roughness values of bare PU and one with 20 wt%SAPO-34 loaded of 2.92 nm and 24.94 nm,respectively.Also,it can be concluded that the bare polyurethane membrane surface was smoother.After adding up to 20 wt%of SAPO-34,the average roughness value has increased,and this could also indicate the presence of SAPO-34 nanocrystals at the membrane surface.The heterogeneity contributes to extended membrane surface area,improving the gas permeation.Therefore,it can be said that by increasing the surface roughness,the membrane surface can be contacted with more gases and it can dissolve more gases to the polymer surface.Also,absorbing the gases on their active pores of SAPO-34 nanocrystals on the membrane surface can help further dissolve the gases.

3.6.Gas permeation results

The permeation properties of CO2,CH4,and N2gases within these membranes were investigated at various pressures(0.6-1.2 MPa).The effects of pressure on the gas permeation properties of neat and polyurethane nanocomposite membranes are shown in Fig.8.According to Fig.8,with increasing pressure,the permeability of CO2gases increased while the permeability of CH4and N2gases almost is constant.In overall it can be said that by increasing pressure feed flow rate increased and penetration of gas molecules becomes more[67].

According to Fig.8,the permeability rate of CO2is significantly higher than CH4and N2gases in pure and nanocomposite membranes.High CO2permeation rate compared to that of CH4and N2is due to low kinetic diameter,high condensability,and more interaction of CO2,as a polar gas,with polar groups on polymer[45].In rubbery polymers such as PU and Pebax higher permeability of CO2is quite obvious[67].Due to higher condensability of CH4compared to N2in polyurethane membranes,the permeability of CH4is higher[42].In other words,the higher permeability of CH4in comparison with the smaller gas molecule such as N2shows that the solution mechanism dominates,while in glassy polymers permeability of N2is higher than CH4and this phenomenon indicated that diffusion mechanism is predominant[68].Therefore,it can be said that the PU and PU-SAPO-34 membranes have a rubber property.

The rate of increase of gas permeability of CO2was higher than N2and CH4,whichmeansenhancement of CO2/CH4and CO2/N2selectivity.The effects of pressure on the selectivity of CO2/CH4and CO2/N2gases for neat PU and PU20 are shown in Fig.9.As shown in Fig.9,by increasing SAPO-34 content,the selectivity of CO2/CH4and CO2/N2gases improved.

Fig.7.AFM analysis of PU-based membranes:neat PU membrane(a),and PU membrane with 20 wt%SAPO-34 loaded(b).

Fig.8.Effect of pressure on permeability of CO2(a)and permeability of CH4and N2gases(b).(1 Barrer=10?10cm3·cm·cm-2·s-1·(cm Hg)-1).

Fig.10 shows the effect of SAPO-34 content on permeability(a)and selectivity of CO2,CH4and N2gases(b)on studied polyurethane membranes at pressure 1.2 MPa.

According to permeation results,CO2,CH4,and N2gases had their permeabilities reduced by increasing the SAPO-34 content.Hence,one can suggest that,in PU membranes,the soft segment should be formed via a microphase separation,to be permeable concerning molecules,whereas the hard segment should act as an impermeable barrier[69].Although based on latter sections,SAPO-34 nanocrystals seem to exist in both soft and hard phases of the polyurethane.According to the entropic view,it seems that SAPO-34 nanocrystals prefer to distribute in the hard segment of polyurethane[70].

FTIR results confirmed that the SAPO-34 particles are probably distributed within the soft phase.Reduction of CO2,CH4and N2permeations through polyurethane membranes after the addition of SAPO-34 nanocrystals serves as a good reason for the existence of nanocrystals in the soft phase.Another reason for the reduction of permeation in the presence of SAPO-34 particles is the phase interposition between soft and hard phases of the polyurethane[42].Therefore,one can argue that most SAPO-34 particles are distributed in the soft phases(permeable region).The soft andflexible(polyol)parts of the polyurethane alone provide the necessaryflexibility and movement to create space for the movement of gas molecules.The presence of SAPO-34 particles in the soft section,can be an obstacle to the passage of gas molecules and reduce free volume[44].

According to the permeability results at 1.2 MPa,permeability was observed to have declined for PU(5 wt%SAPO-34),PU(10 wt%SAPO-34),and PU(20 wt%SAPO-34),as compared to the neat PU,as follows:

Fig.9.Effect of pressure on the selectivity of CO2/CH4and CO2/N2.

By reducing the free volume of the polyurethane due to the presence of SAPO-34 particles,only smaller molecules can cross through the membrane rather than larger molecules.So,their permeability will be reduced further[45].According to the obtained results,a greater reduction in the permeability of N2compared to CO2due to its large molecular size and presence of SAPO-34 nanocrystals can be noted.Although the molecular size of CH4is larger than N2,less reduction in permeability of CH4than N2is observed as a result of more condensation in the membranes.The kinetic diameter of CH4molecules are equal to the pore size of SAPO-34 particles(0.38 nm).Also by incorporation of SAPO-34 particles in polymer matrix,free volume size of polyurethane reduced.Given that the kinetic diameter of CO2is 0.32 nm,it can be concluded that permeation rate of CH4is more limited than CO2.Therefore,a more reduction in the permeability of CH4in comparison with CO2is reasonable.

Based on the permeation results,it is observed that the permeation rate of the CO2gas is higher than that of N2gas,which indicates improved CO2/N2selectivity.Fig.9 shows that at a gas pressure of 1.2 MPa,CO2/CH4and CO2/N2selectivity increases from 21.932 to 25.633 and 36.64 to 58.59 for PU-SAPO-34(20 wt%)compared to neat PU.Also,the increase in selectivity of CO2/CH4and CO2/N2was measured at 14.43%and 37.46%,respectively.

Membrane selectivity depends on permeability and solubility[71,72].Molecular sieve mechanism and selective surface affect the performance separation of CO2,CH4and N2gases by PU-SAPO-34 nanocomposite membranes.High permeability and selectivity serve as two important factors contributing to the industrialization of membranes for gas separation.The obtained results from the prepared PU-SAPO-34 membranes were compared with Robeson's upper bound line[73],as shown in Fig.11.According to the figure,capabilities of polyurethane membranes in the presence of SAPO-34 nanocrystals were compared to similar studies,regarding CO2separation from N2.The closer the data points to Robson's line,the better separation capability the membrane possesses.It is clear that increasing the dose of SAPO-34 particles improves the sorption capability of the membrane.

3.7.Membrane permeability prediction using polynomial regression model

Nonlinear PR is a general method,which can be employed to estimate model parameters,and it can be applied even if the probable model cannot be linearized.This function fits a PR model to powers of an individual predictor by linear least squares method.If a polynomial model is acceptable,one can use the following function:

where Y caret is the predicted result for the polynomial model with regression coefficients b1to k for every single degree and Y-intercept b0.The model is a linear regression with k predictors elevated to the power of i where i=1 to k.More details about this method are available in the literature[74,75].

The accuracy of the PR model was investigated by calculating error and correlation between measured and predicted gas fluxes.Therefore,the mean-squared-error(MSE)and the coefficient of determination(R2)can be written as:

Fig.10.Effect of SAPO-34 content in the gas permeation properties of polyurethane-SAPO-34 membranes on the permeability of CO2,CH4and N2gases in 1.2MPa pressure(a)and selectivity of CO2/CH4and CO2/N2gases in 1.2 MPa pressure(b).1 Barrer=10-10cm3·cm·cm-2·s-1·(cm Hg)-1

In order to predict the values of flux utilizing the PR model,70%of the data were employed for training purpose.The remainders were used for testing.

The mass percent of filler in the polymer(L),feed pressure(P)and kinetic diameter of the permeating species(D)was considered as the input variable of flux function.The result of polynomial regression modeling can be expressed by the following formula:

Fig.11.Comparison of the permeation results with Robeson's upper bound.1 Barrer=10-10cm3·cm·cm-2·s-1·(cm Hg)-1

The accuracy of developed PR model is evaluated with R2and MSE and are tabulated in Table 2 which are 0.9969 and 0.4943,for all data,respectively.This indicates that PR can predict effects of operating conditions on permeability with high accuracy.An important disadvantage of polynomial curve fitting is that a low order polynomial might not fit well,whereas a high-order polynomial might fit the data precisely by having an extreme oscillator shape that is irrelevant to the main function[76],but in this work,polynomial regression could predict permeability without the oscillatory shape of the curve.

In order to investigate the effect of feed pressure and SAPO-34 content on selectivity and gas flux through the composite membrane,PR was used.Generalization performance of polynomial regression,effects of SAPO-34 loading and pressure on permeability of(a)CO2,(b)CH4,(c)N2and CO2/CH4(a),CO2/N2(b)selectivity can be seen in Figs.12 and 13.

Table 2 Comparison between the developed model's performance

According to Figs.12 and 13,permeability was enhanced with feedpressurewhichisconsistentwiththeliterature[77,78].Inaddition,permeability was reduced by increasing SAPO-34 content,and selectivity was improved.Thus,the PR model proposed in this study can be applied to simulate and optimize the gas separation membrane process.

4.Conclusions

In this study,the permeability properties of polyurethanebased MMMs were investigated as a function of pressure and SAPO-34 nanocrystal content.The synthesis and characterization of PU-SAPO-34 MMMs were undertaken with the aim to separate CO2,CH4and N2gases.The effect of the incorporation of SAPO-34 nanocrystals into the bare PU membrane was investigated.The prepared membranes were characterized by TGA,AFM,FTIR,and SEM.FTIR and the SEM results verified the presence of SAPO-34 nanocrystals in the prepared PU membrane.The results of permeabilities concerning CO2,CH4,and N2gases showed that the CO2gas permeation increased by raising the pressure,while the CH4and N2gas permeation remained almost unchanged.Moreover,the presence of SAPO-34 particles in the polymer substrate reduced the CO2and CH4gas permeation,with a much lower reduction for the polar molecules of CO2.The reduced permeation of gas molecules through the membrane could be attributed to the presence of SAPO-34 particles in the soft phase of PU.At a pressure of 1.2 MPa,an increase in SAPO-34 loading to up to 20 wt%caused the permeation of CO2,CH4,and N2gases to decline to 4.55%,18.24%,and 40.24%,respectively.Also,enhancement of selectivity of CO2/CH4and CO2/N2compared to neat PU was 14.4%and 37.3%.

Fig.12.Generalization performance of PR,effects of SAPO-34 loading and pressure on permeability of CO2(a),CH4(b)and N2(c)gases.1 Barrer=10-10cm3·cm·cm-2·s-1·(cm Hg)-1

Besides,based on the gas permeation data,adding SAPO-34,as an inorganic filler to the polymer matrix,CO2/CH4,and CO2/N2selectivity was enhanced.However,our findings showed that PU-SAPO-34 MMMs provide promising potentials for gas separation purposes.According to Robeson's upper bound line,increased SAPO-34 content may contribute to improved separation capability of such membranes.PR was able to predict the membrane permeability with appropriate accuracy.This work indicated that PR could be used as an accurate tool in predicting gas separation processes.

Fig.13.Generalization performance of PR,effects of SAPO-34 loading and pressure on the selectivity of CO2/CH4(a)and CO2/N2(b).