Preparation and characterization of a composite PDMS membrane on CA support

Research Article

568

Received: 25 November 2008, Revised: 21 April 2009, Accepted: 22 April 2009, Published online in Wiley InterScience: 21 June 2009

(www.interscience.wiley.com) DOI: 10.1002/pat.1467

Preparation and characterization of acomposite PDMS membrane on CA support

Mohtada Sadrzadeha, Ehsan Saljoughia, Kazem Shahidia

and Toraj Mohammadia*

Polydimethylsiloxane (PDMS) is the most commonl

Polym. Adv

y used membrane material for the separation of condensablevapors from lighter gases. In this study, a composite PDMSmembranewas prepared and its gas permeation propertieswere investigated at various upstream pressures. A microporous cellulose acetate (CA) support was initially preparedand characterized. Then, PDMS solution, containing crosslinker and catalyst, was cast over the support. Sorptionand permeation of C3H8, CO2, CH4, and H2 in the prepared composite membrane were measured. Using sorption andpermeation data of gases, diffusion coefficients were calculated based on solution-diffusion mechanism. Similar toother rubbery membranes, the prepared PDMSmembrane advantageously exhibited less resistance to permeation ofheavier gases, such as C3H8, compared to the lighter ones, such as CO2, CH4, and H2. This result was attributed to thevery high solubility of larger gas molecules in the hydrocarbon-based PDMS membrane in spite of their lowerdiffusion coefficients relative to smaller molecules. Increasing feed pressure increased permeability, solubility, anddiffusion coefficients of the heavier gases while decreased those of the lighter ones. At constant temperature (25-C),empirical linear relations were proposed for permeability, solubility, and diffusion coefficients as a function oftransmembrane pressure. C3H8/gas solubility, diffusivity, and overall selectivities were found to increase withincreasing feed pressure. Ideal selectivity values of 9, 30, and 82 for C3H8 over CO2, CH4, and H2, respectively, atan upstream pressure of 8 atm, confirmed the outstanding separation performance of the prepared membrane.Copyright � 2009 John Wiley & Sons, Ltd.

Keywords: composite polydimethylsiloxane (PDMS) membrane; gas sorption; permeability; solubility; diffusivity; idealselectivity

* Correspondence to: T. Mohammadi, Research Center for Membrane Separ-ation Processes, Department of Chemical Engineering, Iran University ofScience and Technology (IUST), Narmak, Tehran, Iran.E-mail: [email protected]

a M. Sadrzadeh, E. Saljoughi, K. Shahidi, T. Mohammadi

Research Center for Membrane Separation Processes, Department of

Chemical Engineering, Iran University of Science and Technology (IUST),

Narmak, Tehran, Iran

Contract/grant sponsors: Shiraz Oil Refinery Company (Iran, Shiraz); Iran

National Science Foundation (INSF).

INTRODUCTION

The separation of higher hydrocarbons, such as C3H8, from lowerones, like CH4, in a flue or flare gas in a refinery, is of great eco-nomic importance. As the rates of these gas streams are usuallymodest, applying membrane gas separation process seems to berational.The most important part of a membrane separation process is

the membrane itself, and polymeric membranes are the mostcommon ones used in membrane gas separation process. Gaspermeability through dense polymeric membranes is typicallycalculated by multiplying diffusion and solubility coefficients ofthe penetrants. In all polymer materials, diffusion coefficientdecreases with increasing molecular size. It is due to the fact thatlarge molecules interact with more segments of the polymerchains than the smaller molecules, thereby favoring the passageof small molecules, such as H2, over larger ones such as C3H8.However, solubility increases with increasing condensability andtherefore increases with increasing molecular size.In glassy, rigid polymers, such as polysulfone (PS), permeant

diffusion coefficient is more important than solubility coefficient.Therefore, these polymers preferentially permeate the smaller,noncondensable gases, H2 and CH4, over the larger condensablegases, C3H8 and CO2. On the other hand, in rubbery polymers,such as polydimethylsiloxane (PDMS), permeant solubility coeffi-cient is the most important. Thus, these polymers preferentiallypermeate the larger, more condensable gases over the smaller,

. Technol. 2010, 21 568–577 Copyright �

noncondensable gases.[1] Higher hydrocarbons are usually theminor components of flue or flare gas streams in a refinery.[2]

Hence, when rubbery membranes, such as PDMS, are used, asmall portion of feed has to permeate through the membrane toremove from the feed stream, requiring relatively smallmembrane areas.PDMS is the most commonly used rubbery membranematerial

for separation of higher hydrocarbons from permanent gases.Recently, many studies have been carried out on transportproperties of pure and binary gas mixtures of O2, N2, H2, CO, CO2,CH4, and C2–C4 olefins and paraffins using PDMSmembrane.[3–25]

PDMS membranes, which have been prepared heretofore andevaluated in gas separation applications, were in the form of asingle layer,[3–8] a composite with a microporous support,[9–20] amixedmatrix,[21–23] and a copolymer.[24,25] Most of the composite

2009 John Wiley & Sons, Ltd.

PREPARATION AND CHARACTERIZATION OF A COMPOSITE PDMS/CA MEMBRANE

PDMS membranes were prepared by casting the PDMS solutionon a hydrophilic microfiltration (MF) membrane, such aspolyacrylonitrile (PAN), polyamide (PA), and cellulose acetate(CA), to avoid intrusion of the organic casting solution into thepores of the support.[9,14–16]

Among the different polymeric materials that are used forpreparation of MF membranes to act as support, CA is verycommon. CA MF membranes have maximum uniformity (whichmakes casting easier) and optimum physical properties such asstrength and flexibility. Furthermore, CA polymers are veryconvincing, with characteristics such as good toughness,biocompatibility, high potential flux, and relatively low cost.[26–28]

In the present work, a microporous CA support was initiallyprepared and characterized. Then, a permselective PDMS layerwas formed on the support. Permeability and solubilitycoefficients of C3H8, CO2, CH4, and H2 in the prepared compositePDMS/CA membrane were measured at various transmembranepressures. Based on solution-diffusion mechanism, the concen-tration-averaged diffusion coefficients were also calculated. Idealselectivity values of C3H8 over CO2, CH4, and H2 were calculated toconfirm the ability of the prepared composite PDMS/CAmembrane for separation of higher gases from lower ones.

5

THEORY

Gas transport in polymeric membranes is widely modeled usingthe solution-diffusion mechanism and is expressed by apermeability coefficient, P, defined as follows:

P ¼ Jl

p2 � p1(1)

where J is the steady-state gas flux through the membrane ofthickness l due to a partial pressure difference (p2� p1) across themembrane, p1 is the permeate or downstream pressure, and p2 isthe feed or upstream pressure. In the simplest case, penetrantdiffusion is modeled using Fick’s law of diffusion:[29]

J ¼ � Dloc

ð1� vÞdC

dx

� �(2)

where Dloc is the local concentration-dependent diffusioncoefficient and v is the penetrant mass fraction in the polymerat concentration C. Combining eqns (1) and (2) and integratingacross the membrane thickness yields:

P ¼ 1

p2 � p1

ZC2C1

Deff dC (3)

where C2 and C1 are the penetrant concentrations in the polymerat the upstream and the downstream faces of the membrane,respectively, at a given temperature and Deff is the local, effectivediffusion coefficient, defined for convenience as Deff ¼Dloc=ð1� vÞ. If the diffusion coefficient is not a function ofconcentration

P ¼ C2 � C1p2 � p1

Deff (4)

If the diffusion coefficient is dependent on concentration, Deff

is replaced with D, the concentration-averaged effective

Polym. Adv. Technol. 2010, 21 568–577 Copyright � 2009 John

diffusivity. If the downstream pressure is negligible comparedto the upstream pressure, eqn (4) can be simplified:

P ¼ SD (5)

where, D is Deff or D and the solubility coefficient, S, is defined asfollows:

S ¼ C

p(6)

In eqn (6), S should be evaluated at the upstream conditions.Equation (5) is widely used to rationalize gas transport propertiesof polymer membranes.Expression of resistances in series can be used to describe the

gas transport properties of composite membranes:[30]

JA ¼ DplMS

PMSA

þ lSL

PSLA

� ��1

(7)

where JA is the steady-state flux of gas A; PA is the permeabilitycoefficient of permeant A; l is the thickness, and MS and SLsuperscripts denote microporous support and selective layer,respectively.Ideal selectivity, aA/B, of component A over B is a measure of

potential separation ability of the membrane material. It can bewritten as the ratio of the fluxes of the pure gases measuredunder equal pressure driving force:[30]

aA=B ¼JAJB

(8)

As a result, the selectivity of a composite membrane can thenbe written from eqns (7) and (8) as:

aA=B ¼lMS=PMS

B þ lSL=PSLBlMS=PMS

A þ lSL=PSLA(9)

It is apparent from this expression that the selectivity of acomposite membrane is determined by both layers of thecomposite structure. When the resistance of the compositemembrane to gas permeation lies within the permselective toplayer (as in the present work), that is, lSL=PSLA >> lMS=PMS

A andlSL=PSLB >> lMS=PMS

B , eqn (9) simplifies to:

aA=B ¼lSL=PSLBlSL=PSLA

¼ PSLAPSLB

(10)

When permeability is viewed as the product of solubility anddiffusivity (eqn 5), this expression may be rewritten as theproduct of two ratios in the selective PDMS layer:

aA=B ¼SASB

� �� DA

DB

� �(11)

where the first term is the solubility selectivity and the second isthe diffusivity or mobility selectivity. In addition to operatingconditions (i.e., temperature, pressure, and gas composition),penetrant solubility depends on condensability and polymer–penetrant interactions.[31] In the absence of specific interactions(e.g., hydrogen bonding), the first parameter is dominant and thesolubility increases as penetrant condensability, characterized bycritical temperature, normal boiling point or Lennard Jones forceconstant, increases.[31] Thus, solubility selectivity increases as the

Wiley & Sons, Ltd. www.interscience.wiley.com/journal/pat

69

M. SADRZADEH ET AL.

570

difference between condensabilities of two penetrants in amixture increases. Often, larger penetrants are more condensableand, therefore, more soluble than smaller penetrants. However,diffusion coefficient decreases as penetrant size increases and,therefore, diffusivity selectivity increases as the differencebetween molecular sizes of two penetrants increases, withthe smaller penetrant having the higher diffusivity.[31] Thus, atrade off often exists between solubility selectivity and diffusivityselectivity. It means that the overall selectivity depends on therelative magnitudes of these two selectivities.

EXPERIMENTAL

Support preparation

CAwith an average molecular weight of 52 000 g/mol (Fluka) wasused as a polymer material to form microporous support.1-Methyl-2-pyrrolidone (NMP) with analytical purity of 99.5%(Merck) and distilled water were used as solvent and nonsolventagents, respectively. Polyethylene glycol (PEG), with an averagemolecular weight of 400 g/mol (Lobe Chemie Ltd), was used as anadditive.Initially, a solution of 15.5wt% CA and 10.0wt% PEG in NMP

was prepared. The prepared solution was stirred so as to ensurethat its polymer content was completely dissolved. Then, it wasdegassed in an ultrasonic bath for about 2 h to remove any airbubbles. In order to slow down the aging process of the solution,it was kept away from direct sunlight. After that, the solution wascast on a glass plate using a casting knife with a thickness of180mm. The cast film was subsequently immersed in a distilledwater bath to complete the phase separation, where exchangebetween the solvent (NMP) and the nonsolvent (water) wasinduced. Finally, the polymer film was heat-treated in a deionizedwater bath at 508C for 10min to remove the excess NMP and PEG.The prepared microporous support was kept in a container ofdeionized water to be ready for characterization as well aspreparation of the composite membrane.Full description of the support formation mechanism was

presented in our previous studies.[32,33] CA and PEG concen-trations as well as coagulation bath temperature (CBT) wererealized to affect the support morphology. Levels of PEG and CAconcentrations (10 and 15.5wt%, respectively) and CBT (258C)were chosen so as to prepare a support with high porosity. Hence,it was assured that the prepared support induced negligibleresistance against gas transport through the membrane.In order to evaluate the porosity of the support, it was initially

impregnated with distilled water and then weighed after wipingsuperficial water with filter papers. Thereafter, the wet polymerwas placed in an air-circulating oven at 808C for 24 h to becompletely dried and finally, the dry polymer was weighed.The porosity of the support polymer was calculated using thefollowing equation:[34,35]

Pð%Þ ¼ ðQ0 � Q1ÞAh

� 1000 (12)

where P is the polymer porosity; Q0 and Q1 are the weights of wetand dry polymers (g), respectively; A is the polymer surface area(cm2), and h is the polymer thickness (mm). In order to minimizethe experimental errors, the polymer porosity was measuredthree times and the results were reported in average.

www.interscience.wiley.com/journal/pat Copyright � 2009

SEM images were also taken to confirm the formation of aporous support. The prepared support was dried and cut underliquid nitrogen. Then, it wasmounted on a brass plate and sputtercoated with a thin film of gold to be ready for taking SEMphotographs. The SEM images are presented in the next sections.

Composite membrane preparation

The thin film composite PDMSmembranes were made as follows:A mixture of 20wt% PDMS (Dehesive 944 silicone from WackerSilicones Corporation, Adrian, MI) in toluene was prepared. Thesolution was then stirred for 1 h to ensure that the silicone oil wascompletely dissolved. Before casting, the proprietary CrosslinkerV24/Catalyst OL system, provided by Wacker, was added to thepolymer solution. Finally, it was degassed in an ultrasonic bath forabout 1 h to remove any air bubbles.The prepared CA support was impregnated with distilled water

and fixed on a glass plate, with the smallest pores directed to thetop. The water was used to fill the pores to prevent intrusion ofthe polymer solution into the support layer. The excess of waterwas wiped off with a humid tissue. Before casting the solution onthe support, the glass plate was inclined (308) to allow the poly-mer solution to flow down over the support. After 5min, the glassplate was turned upside down and the coating procedure wasrepeated. Five coatings were done. Afterward, the membranewas put into an oven at 808C for 2 h to remove residual solventand to fully crosslink the polymer. Finally, the crosslinked filmswere cooled to room temperature and housed in the gasseparation membrane cell. The membranes had an effective areaof approximately 10 cm2 in the module.The thickness of permselective layer was determined with a

digital micrometer (Mitutoyo Model MDC-25SB) readable to�1mm by subtracting the thickness of composite membranefrom that of the support. As the PDMS solution casting on the CAsupport is controlled by inclining or turning down the glass plate(as described above), the thickness of the PDMS layer on thesupport is predicted to be incompletely uniform. However, bycontrolling sources of errors (such as setting the glass plateprecisely at 308, accurate adjusting the membrane preparationtime, using the same coating procedure at each five coatingsteps), the thickness of the permselective layer is expected to bein a reasonable range. Thicknesses of the CA support and thecomposite membrane were measured at 20 random points ofsurface area and the results were reported in average to be 145.8and 171.4mm, respectively, with standard deviation of less than1.2. SEM images were also taken to confirm the formation of anintegrally skinned layer and evaluate the membrane thicknessmore precisely.Cross-sectional images of the membranes were obtained using

CamScan scanning electron microscopy (SEM) Model MV2300microscope.

Permeability measurement

A crossflow membrane cell made from stainless steel (grade 316)was used to conduct the experiments (Fig. 1). The membrane washoused in the cell that consisted of two detachable parts. RubberO-rings were used to provide a pressure-tight seal between themembrane and the cell. The membrane had an effective area ofapproximately 10 cm2.H2, CO2, and CH4 gases with purity of 99.5% supplied by

Technical Gas Services, Inc., and C3H8 gas with purity of 99.9%

John Wiley & Sons, Ltd. Polym. Adv. Technol. 2010, 21 568–577

Figure 1. Schematic view of the gas permeation module.

Figure 2. Schematic view of the gas sorption module.


5

supplied by Air Products and Chemicals, Inc. were used as feedgases. The feed flow rate was controlled by Dwyer mass flowcontrollers (MFC), model GFC 2111 (0–15 000 normal mL/minrange). Constant transmembrane pressure was controlled by aback pressure regulator (BPR), model 26.60 SCFBXE262C086.Permeate flow rates were measured using a bubble flow meter(BFM). In most of the experiments, a digital mass flow meter(MFM) was used instead of the BFM for convenience. The MFMwas formerly calibrated by the BFM.At steady-state condition, gas permeability was calculated

using the following equation:

P ¼ 22 414

A

l

p2 � p1

p1RT

dV

dt(13)

where A is the membrane area (cm2), R is the universal gasconstant [6236.56 cm3�cmHg/(mol�K)], T is the absolute tempera-ture (K), dV/dt is the volumetric displacement rate of the soap filmin the BFM (cm3/s), and 22 414 is the number of cm3(STP) ofpenetrant per mole.

Sorption measurement

Sorption experiments were conducted with a dense, filler freePDMS film that was approximately 300mm thick. Crosslinking wasperformed employing the same components and conditions asthose used for preparation of the PDMS permselective layer in thecomposite PDMS/CA membranes. Hence, the PDMS permselec-tive layer is presumed to have a crosslink density similar to that ofthe thick, single layer membrane and consequently the same gassorption properties.[5,13,36,37] The crosslink density of the PDMSfilms used in the sorption and gas permeation experiments wasestimated to be 1.5� 10�4mol/cm3.[37]

Pure gas sorption measurements were carried out using apressure decaymodule as shown in Fig. 2. Gas sorption apparatusconsisted of a stainless steel (grade 316) module of knownvolume. The module was connected to vacuum pump and gascylinder by stainless steel valves. The gas pressure in this modulewas monitored using sensitive pressure transducers andrecorded automatically by a data acquisition system employingLabTech software. The vacuum pump was connected to thisapparatus to degas the module, whenever required.


Initially, a polymer film was placed in the module and exposedto vacuum to remove all sorbed gases from the polymer. Then, V1was closed and the gas was introduced into the module byopening then closing V2, impulsively.The number of moles of the gas introduced into the module

could be calculated from the module pressure (initial pressure ofthe module), gas temperature (298 K) and knownmodule volume(290 cm3). Gas sorption into the polymer started exactly afteropening V2 and the module pressure decreased. When thesystem reached equilibrium (sorption and desorption of gasmolecules became equal), the pressure reduction stopped.The difference between initial and final moles of gas existing inthe module was the moles of gas sorbed into the polymer at theinitial pressure.Concentration of the penetrant gas in the polymer at a given

temperature and pressure was calculated from the followingrelation:[38]

C ¼ 22 414

RTðpi � pfÞ

VmVp

(14)

where C is the uniform concentration of the dissolved penetrantat equilibrium state [cm3(STP)/cm3 polymer]; pi and pf designate,respectively, the initial and the final pressure in the module; Vmand Vp are the volumes of the module and the polymer sample(cm3), respectively, and 22 414 is the number of cm3(STP) ofpenetrant per mole.An additional penetrant was then introduced into the module

and the procedure was repeated. In this incremental manner, thepenetrant uptake could be determined as a function of pressure.


71

M. SADRZADEH ET AL.

572

RESULTS AND DISCUSSION

Morphology of the support and the composite membrane

The cross-sectional SEM image of the support is shown in Fig. 3.As can be observed, the support layer is microporous so the effectof the support layer on the gas transport across the compositemembranes is negligible in comparison with that of the densePDMS layer. Thickness of the support layer is in the range of145.8� 1.2mm.Figure 4 shows the cross-section structure of the prepared com-

posite membrane investigated by SEM. As indicated in this figure,thickness of the PDMS skin layer is in the range of 25.6� 2.4mm.

Gas permeability

The permeability values of H2, CH4, CO2, and C3H8 at 258C as afunction of the pressure difference across the composite PDMS

Figure 3. Cross-sectional SEM image of the CA support.

Figure 4. Cross-sectional SEM image o


membrane are represented in Fig. 5. H2, CH4, and CO2 exhibitconstant or slightly decreasing permeabilities at all upstreampressures tested while permeability of C3H8 increases withincreasing upstream pressure. This result is in accordance withwhat Merkel et al.,[13] Prabhakar et al.,[36] and Stern et al.[39]

reported. As shown in Fig. 5, permeability coefficients of thepenetrants increase in the following order:H2< CH4<CO2<C3H8

Solubility coefficients of theses penetrants also increase asabove with an exception of H2, which is the smallest consideredpenetrant in this study. Based on solution-diffusion mechanism(eqn 5), permeability is the product of diffusivity, D, and solubility,S, of the gas in the membrane material. Rubbery membraneshave weak molecular sieve ability due to their weak inter-molecular forces, resulting in broad distribution of inter-segmental gap sizes responsible for gas diffusion. Diffusioncoefficients of penetrants often change less than their solubilitycoefficients so that more soluble penetrants are more permeable.

f the PDMS composite membrane.

Figure 5. Permeability of the penetrants through the prepared composite

PDMS membranes as a function of transmembrane pressure difference.


Figure 6. Sorption isotherms for penetrants in the prepared composite

PDMS membrane at 258C.


Consequently, the relative permeability of penetrants throughthe composite PDMS membrane is mainly determined by theirrelative solubility.For all penetrants, the permeability is a linear function of Dp

and can be represented as follows:

P ¼ P0 þmDp (15)

where P0 is the permeability coefficient at Dp¼ 0, the slope, m,characterizes the pressure dependence of permeability with Dp.P0 and m values for four gases examined in this study arerecorded in Table 1. The value of m is determined principally byinterplay between three factors: plasticization, hydrostaticpressure, and penetrant solubility. Plasticization refers to anincrease in penetrant diffusivity resulting from increased polymerlocal segmental motions caused by the presence of penetrantmolecules in the polymer matrix.[4,5,13]

Therefore, with upstream pressure , penetrant concentration inthe polymer and tendency to plasticize the polymer matrixincrease, particularly for strongly sorbing penetrants. On theother hand, high upstream pressure acting on the polymer filmcan slightly compress the polymer matrix, thereby reducing theamount of free volume available for penetrant transport andreducing the penetrant diffusion coefficient. In addition to thesedual effects, which affect the penetrant diffusion coefficient, thepenetrant solubility in rubbery polymers frequently increaseswith pressure, especially for organic vapor penetrants, leading toa corresponding increase in permeability.[13,40]

Hence, the permeability coefficients of low-sorbing penetrants,such as H2, do not plasticize the composite PDMSmembrane and,as indicated before, have essentially pressure independentsolubility coefficients, which decrease very slightly with increas-ing pressure. This was confirmed by negative values of m aspresented in Table 1. In contrast, the permeability coefficients ofmore soluble penetrants, such as C3H8, induce significantplasticization and have solubility coefficients that also increasesignificantly with pressure, increase with increasing pressure. Forthese penetrants, m is positive.

Gas solubility

Sorption isotherms for H2, CH4, CO2, and C3H8 in the preparedsingle layer PDMS membrane at 258C are presented in Fig. 6. Theisotherms for all penetrants are linear (H2, CO2, and CH4) or nearlylinear (C3H8), which is consistent with previously reported gasand vapor sorption isotherms in PDMS.[5,12,13,36–38]

C3H8 sorption isotherm is convex to the pressure axis, which isconsistent with the behavior of highly sorbing penetrants inrubbery polymers.[5,12,13,36–38] The concentration of C3H8 in thePDMS is 191, 30, and 8 times higher than that of H2, CH4, and CO2

Table 1. Permeability (P0 and m), solubility (S1 and n), and diffusmembrane at 258C

Penetrant P0 (Barrer) m (Barrer/atm) S1 (cm3/cm3 atm)

H2 365 �4.66 0.11CO2 3577 �94.53 0.57CH4 1040 �24.16 2.15C3H8 6100 856.21 9.22


at 8 atm. The convex curvature of C3H8 sorption isotherms is dueto the high levels of sorbed penetrants at high pressures.From the sorption isotherms, the solubility of each penetrant

was calculated using eqn (6). The values are presented as afunction of pressure in Fig. 7. As shown, in contrast to C3H8 that itssolubility in PDMS increases largely with pressure, the solubilitycoefficients of other lighter gases are almost constant. However,solubility coefficients of all gases can be well represented by thefollowing linear equation:

S ¼ S1 þ np (16)

where S1 is the infinite dilution solubility which is defined asfollows:

S1 ¼ lim ðC=pÞp!0

(17)

In eqn (16), n characterizes the pressure dependence ofsolubility. S1 and n values for H2, CH4, CO2, and C3H8 arepresented in Table 1.For the relatively low-sorbing penetrants as H2 and CH4, the

pressure dependence of solubility, n, is almost zero. Hence, asrealized before, the solubility of these penetrants in PDMS isessentially independent of penetrant pressure and is welldescribed by Henry’s law, which is typical for the sorption oflight gases in many polymers:[13,39]

C ¼ kdp (18)

ivity (D0 and q) parameters in the prepared composite PDMS

n (cm3/cm3 atm2) D0 � 106 (cm2/s) q� 106 (cm2/s atm)

�0.0017 28.04 �0.0476�0.0028 13.06 �0.2940�0.0125 14.04 �0.23730.9265 5.34 1.1712


573

Figure 7. Solubility of the penetrants in the prepared composite PDMS

membrane as a function of pressure.

M. SADRZADEH ET AL.

574

where C [cm3(STP) of penetrant sorbed per cm3 of polymer] is theequilibrium penetrant concentration in the polymer at pressure p(atm) and kd [cm

3(STP)/(cm3 atm)] is the Henry’s law constant. Inthe present work, Henry’s law constants were estimated to be0.083, 0.532, and 2.009 for H2, CH4, and CO2, respectively.The more strongly sorbing penetrants, C3H8, exhibit larger

values of n, which confirm strong solubility dependence of thesepenetrants with pressure. This is also a typical behavior for manyorganic vapors in rubbery polymers.[13,27]

In the absence of specific interactions between the penetrantmolecules and the polymer matrix, gas solubility coefficients areusually scaled with measures of penetrant condensability such ascritical temperature, Tc.

[13,27,36] In order to compare solubilities ofpenetrants on a consistent basis, solubility coefficient in the limitof zero pressure, S1, was utilized. Figure 8 presents S1 (extractedfrom Fig. 7) as a function of critical temperature of the penetrants.According to the literature, the logarithm of gas solubility inpolymers often increases linearly with critical temperature of thepenetrants.[13] This trend is confirmed in Fig. 8.

Figure 8. Infinite dilution solubility as a function of critical temperature

of penetrants.


Gas diffusivity

Concentration-averaged diffusion coefficient, D, was estimatedfrom the permeability and sorption data using the rearrangedform of eqn (4). The results are reported as a function of pressuredifference across the PDMS membrane at 258C in Fig. 9. For allpenetrants, the pressure dependence of diffusion coefficient canbe well described by the following linear equation:

D ¼ D0 þ qDp (19)

where D0 is the diffusion coefficient at Dp¼ 0 and q is aparameter that characterizes the pressure dependence ofdiffusion coefficient. The values of D0, q for all penetrants arereported in Table 1. For the least soluble penetrants (H2, CO2, andCH4), diffusivity decreases slightly with increasing pressure due tohydrostatic compression effects and q is consequently negative.As discussed earlier, solubility of these gases is independent ofupstream pressure. Hence, permeability coefficient of light gasesdecreases slightly as pressure increases. In contrast, more solublepenetrants (C3H8) exhibit an increase in both solubility anddiffusivity coefficients with increasing penetrant pressure. Basedon the data presented in Table 1, both n and q are positive forthese gases. Therefore, permeability of heavy gases in thecomposite PDMS membrane increases with pressure.At low pressures, diffusion coefficient of the penetrants

increases in the following order:C3H8<CO2<CH4<H2

This is the order of decreasing size of the penetrants and is atypical behavior for both rubbery and glassy polymers. Figure 10compares D0 values of gases in the prepared rubbery PDMSmembrane, polyisoprene (PIP),[41] and poly(tetrafluoroethylene-co-perfluoromethyl vinyl ether) or TFE/PMVE 49[42] with those inglassy polyvinyl chloride (PVC),[13] polysulfone (PS),[43] and poly(2,2,4-trifluoro-5-trifluoromethoxy-1,3-dioxole-co-tetrafluoroethylene) com-mercially knownHyflon AD 80[43] as a function of critical volume ofthe gases. As can be seen, diffusion coefficient of the penetrantsin PDMS and PIP, like other rubbery polymers, is relatively a weak

Figure 9. Concentration-averaged diffusion coefficients of the pene-trants, D, in the prepared composite PDMS membrane as a function of

transmembrane pressure difference.


Figure 10. Comparison of the variation of infinite dilution diffusioncoefficients with penetrant critical volume, Vc, in the prepared PDMS

membrane with that in rubbery polymers PIP[41] and TFE/PMVE 49[42] and

in glassy polymers PVC,[13] PS,[43] and Hyflon AD 80.[43]

Table 2. C3H8/gas solubility, diffusivity, and overall idealselectivity values at Dp¼ 0

Penetrant SC3H8=Sgas DC3H8

=Dgas PC3H8=Pgas

H2 83.81 0.20 16.76CO2 4.28 0.41 1.75CH4 16.17 0.37 5.98

Figure 11. C3H8/gas solubility selectivity of the prepared composite

PDMS membrane as a function of upstream pressure.


5

function of their size. For the entire range of the penetrantsconsidered, diffusion coefficient varied by less than two orders ofmagnitude. Diffusion coefficient values in PDMS are larger thanthose in amorphous rubber, PIP, and amorphous randomcopolymer rubber TFE/PMVE 49. In fact, PDMS has the lowestdiffusivity selectivity compared with any other rubbery polymers.It has a very flexible polymer backbone as indicated by itsextremely low glass transition temperature (Tg��1208C). As aresult, PDMS has a very weak ability to sieve penetrants based ontheir size.Variation of diffusion coefficient with critical volume (as a

measure of penetrant size) is usually described by the followingequation:

D ¼ t

Vhc

(17)

where t and h are adjustable parameters. h provides a measure ofthe reduction rate of diffusion coefficient with increasingpenetrant size. The higher the value of h, the greater thediffusivity selectivity of the polymer. Polymers with larger valuesof h exhibit diffusion coefficients that depend more strongly onpenetrant size compared with polymers with smaller values of h.Based on the data in Fig. 10, the values of h are 10.5, 8.4, 5.7, 2.6,2.4, and 1.7 for PVC, PS, Hyflon AD 80, TFE/PMVE 49, PIP, and theprepared composite PDMSmembrane, respectively.[13,41–43] For avariety of low molar mass organic liquids, such as hexane andbenzene, the value of h is equal to 0.45.[13] This resultdemonstrates that the relative ability of PDMS to separatemolecules based on their sizes is closer to that of organic liquids,such as hexane and benzene, than that of glassy polymers such asPVC, PS, and Hyflon AD 80.

Ideal selectivity

The selectivity of the prepared composite PDMS membrane isexpressed as the product of solubility selectivity and diffusivityselectivity. Solubility selectivity, diffusivity selectivity, and overallselectivity of the PDMS membrane for C3H8, relative to other


gases at Dp¼ 0, are presented in Table 2. C3H8 is significantlymore condensable than other gases. Hence, C3H8/gas solubilityselectivity is larger than 1. As mentioned earlier, C3H8 sorptionisotherm is convex to the pressure axis due to the high levels ofsorbed penetrants at higher pressures. Hence, the increasing rateof C3H8 solubility is much greater than that of other gases andconsequently, C3H8/gas solubility selectivity increases asupstream pressure increases (Fig. 11).C3H8/gas diffusivity selectivity values are smaller than 1. Similar

to solubility selectivity, diffusivity selectivity also increases asupstream pressure increases (Fig. 12). It means that diffusioncoefficient of C3H8 is lower than those of other lighter gases.However, increasing rate of C3H8 diffusivity is greater than thoseof H2, CO2, and CH4.The overall selectivity values, aA/B, of the prepared composite

PDMS membrane for C3H8 over CO2, CH4, and H2 (calculatedusing eqn 10) as a function of upstream pressure are presented inFig. 13. The increasing rate of ideal selectivity with pressure isobserved to be completely dependent on the difference betweencondensabilities or critical volumes of C3H8 and other gases. Atp¼ 8 atm, the prepared composite PDMS membrane is approxi-mately 9, 38, and 82 times more permeable to C3H8 than CO2,CH4, and H2, respectively.Based on eqn (11), the membrane selectivity depends on the

relative diffusion coefficients (DA, DB) of the two gases (A and B) inthe polymer and on the relative solubilities (SA, SB) of the gases inthe polymer. As mentioned earlier, in all polymer materials, thediffusion coefficient decreases with increasing molecular size


75

Figure 12. C3H8/gas diffusivity selectivity of the prepared compositePDMS membrane as a function of upstream pressure.

Figure 13. C3H8/gas overall selectivity of the prepared composite PDMS

membrane as a function of upstream pressure.

M. SADRZADEH ET AL.

576

while the solubility coefficient increases. Thus, these polymerspreferentially permeate the larger, more condensable gases,C3H8, over the smaller, less condensable gases, H2, CH4, and CO2.As a result, it can be concluded that the prepared composite

PDMS membrane is an appropriate choice for the separation ofliquefied petroleum gases (LPG) from natural gas.

CONCLUSION

Sorption, diffusion, and permeation of C3H8, CO2, CH4, and H2 in aprepared composite PDMS/CA membrane were studied. Solubi-lity coefficient of C3H8 was significantly higher than those of othergases due to more condensable nature of C3H8 regarding itscritical properties. However, diffusion coefficient of C3H8 in the


PDMSmembrane was lower than those of other gases. It is due tothe fact that larger molecules interact with more segments of thepolymer chains than smaller molecules, thereby favoring thepassage of smaller molecules, such as H2, over larger ones such asC3H8. This is typical behavior for both rubbery and glassypolymers. In rubbery membranes, such as PDMS, diffusioncoefficients of penetrants often vary less than their solubilitycoefficients. Consequently, relative permeability of penetrantsthrough PDMS is mainly determined by their relative solubility.The dependency of pressure with permeability, solubility, anddiffusivity coefficients was well described by empirical linearequations. Solubility and diffusivity of C3H8 increased withincreasing upstream pressure whereas for other gases, thesevalues decreased slightly. Hence, both C3H8/gas solubility anddiffusivity selectivities increased as upstream pressure increased.High C3H8 permeability at upstream pressure of 8 atm (26 493Barrer) along with high C3H8/gas ideal selectivities (9, 30, and 82over CO2, CH4, and H2, respectively) confirmed reasonableperformance of the prepared composite PDMS membrane forseparation of organic vapors from supercritical gases. Applicationof this membrane for separation of LPG from natural gas is underinvestigation.

Acknowledgements

This study was partially supported by Shiraz Oil Refinery Com-pany (Iran, Shiraz) and Iran National Science Foundation (INSF).The authors would like to appreciate Dr. M. Bruetsch, WackerSilicones Corporation, for supplying silicone oil, crosslinker, andcatalyst.

REFERENCES

[1] I. Pinnau, Z. He, J. Membr. Sci. 2004, 244, 227.[2] R. W. Baker, J. G. Wijmans, J. H. Kaschemekat, J. Membr. Sci. 1998, 151,

55.[3] S. H. Choi, J. H. Kim, S. B. Lee, J. Membr. Sci. 2007, 209, 54.[4] R. D. Raharjo, B. D. Freeman, D. R. Paul, G. C. Sarti, E. S. Sanders, J.

Membr. Sci. 2007, 306, 75.[5] R. D. Raharjo, B. D. Freeman, E. S. Sanders, J. Membr. Sci. 2007, 292, 45.[6] C. K. Yeom, S. H. Lee, J. M. Lee, H. Y. Song, J. Membr. Sci. 2002, 204, 303.[7] C. K. Yeom, S. H. Lee, H. Y. Song, J. M. Lee, J. Membr. Sci. 2002, 205, 155.[8] C. K. Yeom, S. H. Lee, H. Y. Song, J. M. Lee, J. Membr. Sci. 2002, 198, 129.[9] F. Wu, L. Li, Z. Xu, S. Tan, Z. Zhang, Chem. Eng. J. 2006, 117, 51.[10] P. Jha, L. W. Mason, J. D. Way, J. Membr. Sci. 2006, 272, 125.[11] U. Beuscher, C. H. Gooding, J. Membr. Sci. 1999, 152, 99.[12] T. C. Merkel, R. P. Gupta, B. S. Turk, B. D. Freeman, J. Membr. Sci. 2001,

191, 85.[13] T. C. Merkel, V. I. Bondar, K. Nagai, B. D. Freeman, I. Pinnau, J. Polym. Sci,

Part B: Polym. Phys. 2000, 38, 415.[14] Y. Shi, C. M. Burns, X. Feng, J. Membr. Sci. 2006, 282, 115.[15] L. Gales, A. Mendes, C. Costa, J. Membr. Sci. 2002, 197, 211.[16] H. Byun, B. Hong, B. Lee, Desalination 2006, 193, 73.[17] X. Jiang, A. Kumar, J. Membr. Sci. 2005, 254, 179.[18] F. Peng, J. Liu, J. Li, J. Membr. Sci. 2003, 222, 225.[19] M. B. Hagg, J. Membr. Sci. 2000, 170, 173.[20] M. L. Yeow, R. W. Field, K. Li, W. K. Teo, J. Membr. Sci. 2002, 203, 137.[21] J. H. Kim, J. Won, Y. S. Kang, J. Membr. Sci. 2004, 241, 403.[22] L. Lebrun, S. Bruzaud, Y. Grohens, D. Langevin, Eur. Pol. J. 2006, 42,

1975.[23] G. Clarizia, C. Algieri, E. Drioli, Polymer 2004, 45, 5671.[24] H. B. Park, C. K. Kim, Y. M. Lee, J. Membr. Sci. 2002, 204, 257.[25] K. Madhavan, B. S. R. Reddy, J. Membr. Sci. 2006, 283, 357.[26] A. Idris, L. K. Yet, J. Membr. Sci. 2006, 280, 920.[27] M. Hayama, K. Yamamoto, F. Kohori, K. Sakai, J. Membr. Sci. 2004, 234,

41.



[28] E. Ferjani, R. H. Lajimi, A. Deratani, Desalination 2002; 146, 325.[29] R. S. Prabhakar, R. Raharjo, L. G. Toy, H. Lin, B. D. Freeman, Ind. Eng.

Chem. Res. 2005, 44, 1547.[30] I. Pinnau, J. G. Wijmans, I. Blume, T. Kuroda, K. V. Peinemann, J. Membr.

Sci. 1988, 37, 81.[31] K. Ghosal, B. D. Freeman, Polym. Adv. Technol. 1993, 5, 673.[32] E. Saljoughi, M. Amirilargani, T. Mohammadi, J. Appl. Polym. Sci. 2009,

111, 2357.[33] E. Saljoughi, M. Sadrzadeh, T. Mohammadi, J. Membr. Sci. 2009, 326,

627.[34] K. Scott, Handbook of Industrial Membranes. Elsevier Advanced

Technology, Kidlington, 1998.[35] Q. Z. Zheng, P. Wang, Y. N. Yang, D. J. Cui, J. Membr. Sci. 2006, 286, 7.


[36] R. S. Prabhakar, T. C. Merkel, B. D. Freeman, T. Imizu, A. Higuchi,Macromolecules 2005, 38, 1899.

[37] G. K. Fleming, W. J. Koros, Macromolecules 1986, 19, 2285.[38] V. M. Shah, B. J. Hardy, S. A. Stern, J. Polym. Sci., Part B: Polym. Phys.

1986, 24, 2033.[39] S. A. Stern, V. M. Shah, B. J. Hardy, J. Polym. Sci., Part B: Polym. Phys.

1987, 25, 1263.[40] A. Singh, B. D. Freeman, I. Pinnau, J. Polym. Sci., Part B: Polym. Phys.

1998, 36, 289.[41] G. J. Van Amerongen, Rubber Chem. Technol. 1964, 37, 1065.[42] R. S. Prabhakar, M. G. De Angelis, G. C. Sarti, B. D. Freeman, M. C.

Coughlin, Macromolecules 2005, 38, 7043.[43] R. S. Prabhakar, B. D. Freeman, I. Roman,Macromolecules2004, 37, 7688.


577

Preparation and characterization of a composite PDMS membrane on CA support

Documents