Ab

Ra

b

a

ARRAA

KSPSNB

1

gRbftmaasatlPms

me

0d

Journal of Membrane Science 348 (2010) 287–297

Contents lists available at ScienceDirect

Journal of Membrane Science

journa l homepage: www.e lsev ier .com/ locate /memsci

pplication of polymeric solvent resistant nanofiltration membranes foriodiesel production

ahimah Othmana, Abdul Wahab Mohammada,∗, Manal Ismail a, Jumat Salimonb

Department of Chemical and Process Engineering, Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia, 43600 UKM Bangi, Selangor, MalaysiaDepartment of Chemical Science and Food Technology, Faculty of Science and Technology, Universiti Kebangsaan Malaysia, 43600 UKM Bangi, Selangor, Malaysia

r t i c l e i n f o

rticle history:eceived 1 July 2009eceived in revised form 5 November 2009ccepted 9 November 2009vailable online 13 November 2009

eywords:RNF

a b s t r a c t

The potential of incorporating polymeric solvent resistant nanofiltration (SRNF) membranes for biodieselseparation processes was investigated. Eight types of commercial polymeric nanofiltration membranes(Solsep 030705, Solsep 030306F, Starmem 240, Starmem 120, Desal-DL, Desal-DK MPF-34 and MPF-44)were chosen and screened for their abilities to separate the methyl esters-rich effluent (biodiesel) fromthe mixture of the homogeneous catalyst, free glycerin and excess methanol after the transesterificationprocess at various separation pressures and constant temperature. Scanning Electron Microscope (SEM)and Fourier Transform Infrared Spectroscopy (FTIR) were used to examine any changes to all the mem-branes studied. In order to enhance the SRNF membrane performance, the transesterification product

ermeability

electivityanofiltrationiodiesel separation

properties was modified by reducing the alkalinity value. Results showed that for 3 membranes (Solsep030705, Solsep 030306F, Starmem 240), the permeability of transesterification product after the alkalin-ity modification increased linearly to the operation pressures. The other 5 membranes namely Starmem120, Desal-DL, Desal-DK, MPF-44, MPF-34 membranes gave dismal results. Out of the eight tested mem-branes, it was found that Solsep 030705 membrane gave the most promising result. Analysis of the used

or dif

membranes showed min

. Introduction

Membrane-based separations are well-established technolo-ies in water purification, protein separations and gas separations.ecently, polymeric solvent resistant nanofiltration (SRNF) mem-ranes have been shown to provide a potentially viable alternativeor separation and purification process in non-aqueous applica-ions [1]. The new developments in the area of solvent resistant

embranes should lead to many novel applications in the foodnd pharmaceutical industries for selective recovery of organicsnd recycle of solvent streams. Various applications have beentudied on the use of membranes (including SRNF) to treat non-queous fluids for separation [2,3]. The well investigated area ishe dry degumming process which is to remove the phospho-ipids in the hexane miscella by membrane ultrafiltration [4,5].roduction of biodiesel is another important application in whichembranes could be utilized in the separation and purification

tages [6].For commercial biodiesel production process, there are two

ain hindrances that may affect the biodiesel production cost andnergy consumption which must be overcome: (i) the removal

∗ Corresponding author. Tel.: +60 3 8921 6102; fax: +60 3 8925 2546.E-mail address: wahabm@eng.ukm.my (A.W. Mohammad).

376-7388/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.memsci.2009.11.012

ferences on functional groups after the application.© 2009 Elsevier B.V. All rights reserved.

of residual triglyceride, glycerol and unreacted alcohol from thebiodiesel products due to the homogeneous nature and theirphysical properties like boiling point that were close to oneanother, and (ii) the homogeneous catalyst removal by wash-ing will tend to produce waste water containing small amountof soap and free glycerol which should be carefully handledbefore discharge [6–9]. Both these problems can potentially beovercome through the use of SRNF membranes separation pro-cess.

Recently, Dube� et al. [6] studied the use of a membrane reac-tor for biodiesel production. The process has been shown to offermany advantages, such as improved product yield, elimination offeed loss, reduced effluent quantity and thereby lowering energyconsumption. The use of membrane in the reactor allowed forthe removal of unreacted oil from the FAME product, yieldingbiodiesel of high purity and shifting the reaction equilibrium tothe product side. The membrane acts a phase barrier which lim-its the presence of TG and non-reacting lipids in the product.Four different carbon membrane pore sizes were tested: 0.05 �m,0.2 �m, 0.5 �m, and 1.4 �m [6,7]. However, different results have

been obtained for the presence of bonded glycerides in the per-meate. For example, DG was observed in the permeate whenthe FAME concentration was larger than 25–35 wt%, but neitherTG nor MG were detected by Cao et al. in the permeate stream[7].

2 mbran

bbgshti[atonahfnthpoe

te[tadmrclboebTrpmtbtnltAb

tmsmrcb

1

2

88 R. Othman et al. / Journal of Me

Wang et al. [9] used ceramic membrane to refine the crudeiodiesel instead of using hot water washing step in the commercialiodiesel production system. Most of the reported works used inor-anic membranes compared to organic membranes in the biodieseleparation and purification system where inorganic membranesave the advantages in the industrial operations in terms of high-emperature durability, sufficient mechanical strength, chemicalnertness, organic solvent resistance, unique surface characteristics3] and less likelihood of bacteria contamination [10]. While therere so many advantages for inorganic membranes, some disadvan-ages are very critical and serious enough to hamper the progressf industrial application of inorganic membranes such as brittle-ess (easy to crack), difficult sealing problems at high-temperaturepplications, low surface-to-volume ratio, few membranes withigh selectivity and expensive capital and repair costs [11]. There-

ore, Bournay and Baudot [12] recommended the use of polymericanofiltration and reverse osmosis (RO) membranes of differentype of polymer such as polyethersulphone, silicone, polyvinyl alco-ol (PVA), polyamide (PI), polyimide (PI) to separate at least oneortion of an alkyl esters-rich effluent from an alcohol-rich effluent,r to separate of an alcohol-rich effluent from an alkyl esters-richffluent.

Biodiesel or also known as methyl ester is usually produced byransesterification of vegetable oils, largely consisting of triglyc-rides, with methanol in the presence of strong alkali or acid13,14]. The transesterification process consists of a sequence ofhree consecutive reversible reactions where monoglycerides (MG)nd diglycerides (DG) will be formed as the intermediate productsuring the reaction process of biodiesel production [14]. The excessethanol is generally added prior to drive the reaction equilib-

ium towards the products side. However, reaching the completeonversion of the TG is a challenge in light of the chemical equi-ibrium of the reaction. Typical single-step batch processes yieldiodiesel with the present of TG, DG, and MG, which implies a lossf reactants as well as a failure to meet the bound glycerol lev-ls required by the ASTM standard. Specifically, in the Europeaniodiesel standard, DIN EN 14214, the contents of MG, DG andG are limited to be no more than 0.8 wt%, 0.2 wt%, and 0.2 wt%,espectively. It is also regulated in the ASTM standard for 100%ure biodiesel (B100), the total glycerin content is limited to noore than 0.24 wt%. For instance, free and bonded glycerin con-

ent reflects the quality of biodiesel that must be removed from theiodiesel before it undergoes further refinement process. This iso prevent from injector fouling, formation of deposits at injectionozzles, pistons, and valves, and severity of engine durability prob-

ems. Furthermore, the combustion of these chemicals can lead tohe formation of acrolein, a photochemical smog ingredient [15].s a result, the issue of separating this minor amount of glyceridesecomes very important.

With the above observations in view, this study is aimed athe selection and experimental verification of the polymeric SRNF

embranes performance in terms of permeation properties andelectivities for biodiesel separation and purification process. Theethyl ester-rich effluent is desired as a permeant, whilst alcohol-

ich effluent specifically methanol is expected as a retentate whichan be considered to be recycled into the reaction medium of theiodiesel production. Particularly, the objectives are:

. To understand the pure solvent permeation behavior of polar andnon-polar organic solvents, which are the reactants and productsof the biodiesel production reaction through the use of different

polarity of polymeric SRNF membranes.

. To investigate the possibilities of using the polymeric SRNFmembranes for biodiesel products separation and purificationprocess at certain process conditions by monitoring the selec-tivity of the tested membranes.

e Science 348 (2010) 287–297

3. To observe the differences on the morphological structures andfunctional groups of the tested polymeric SRNF membranes afterbeing applied for the separations.

2. Materials and methods

2.1. Chemicals

The following chemicals were used: anhydrous methanol andn-heptane were of HPLC grade, anhydrous glycerol (ChemAR®),and sodium hydroxide (pellets; R&M Chemicals). RBD olein waskindly supplied by Intercontinential Speciality Fats Berhad (ISF),Malaysia and BIO-N palm methyl esters mixture (crude biodiesel)was provided by Golden Hope Biodiesel Sdn.Bhd., Malaysia. Glyc-erin, monolein, diolein, triolein, butanetriol, tricaprin and methyloleate were supplied by Sigma–Aldrich Co.

2.2. Membranes

Eight polymeric solvent resistant nanofiltration membranesfrom different manufactures were examined and their propertiesare shown in Table 1. Starmem (Starmem 120 and Starmem 240),Solsep (Solsep 030705 and Solsep 030306F) and Desal (Desal-DLand Desal-DK) membranes have been designed for aqueous andnonequeous systems and were applied in a “dry” form. MPF (MPF-34 and MPF-44) membranes were supplied in a “wet” form in apreserving solution of 0.7% Roccal. All membranes are specified assolvent resistant nanofiltration membranes.

2.3. Synthesis of biodiesel

Biodiesel or also known as methyl ester was produced by trans-esterification process in which, RBD palm olein was reacted withanhydrous methanol in the presence of strong base catalyst ofsodium hydroxide. The transesterification process is one of thereversible reactions and proceeds essentially by mixing the reac-tants [6–8,13,14]. Based on the stoichiometry, the reaction needsa 3:1 molar alcohol-to-oil ratio, but the excess methanol is usu-ally added to drive the equilibrium towards the products side[16]. In this study, the synthesis of biodiesel was carried out ina 1 L three-neck round-bottomed flask fitted with thermometer,reflux condenser, and glass stopper on a magnetic stirrer [17]. Fil-tered RBD palm olein, methanol (10:1, methanol:oil mole ratio)with dissolving of base catalyst and sodium hydroxide (1 wt%of oil) were then transesterified into the round bottom flask.The reaction was performed at the agitation speed of 600 rpmand temperature of 60 ◦C for 2 h. At the normal pressure, themethanol vapor was condensed by the cooling water. After 2 hreaction, the transesterification product from the reaction wastransferred to a SEPA ST Stirred Cell (Osmonic USA) for permeationexperiment. The pH value and temperature of the transesteri-fication products were recorded before the separation processbegin.

Transesterification products modification by neutralization ispotentially the most sustainable solution of improving the capa-bilities of the polymeric SRNF membranes for biodiesel separationprocess. The idea is to reduce the alkalinity of the transesterificationproducts by using phosphoric acid, H3PO4 (100% purity), so that theseparation medium will be suited enough for the separation processwithout exceeding the specific range of working pH value for eachtested polymeric SRNF membranes, as listed in Table 1. The neu-

tralization was carried out by adding 0.1 mL of H3PO4 to 200 mL oftransesterification product and any changes on the pH value wererecorded. Potassium phosphate, Na3PO4 was produced after theneutralization can be used as valuable byproduct for instance, asa fertilizer [14,18]. The pH value of the transesterification product

R. Othman et al. / Journal of Membrane Science 348 (2010) 287–297 289

Table 1Solvent resistant nanofiltration membranes and their properties by the manufacturers.

Manufacturer Membrane type MWCO Membrane property Polymer type Tmax (◦C) pH

Osmonicsa Desal-DL 150–300 Hydrophilic Polyamide 90 2–11Desal-DK 150–300 Hydrophilic Polyamide 90 2–11

Kochb MPF-34 200 Hydrophilic PDMSc 40 0–14MPF-44 250 Hydrophilic PDMSc 40 3–10

METd STARMEM 120 200 Hydrophobic Polyimide 50 –e

STARMEM 240 400 Hydrophobic Polyimide 50 –e

Solsepf Solsep 030306F –e Hydrophobic –g 120 2–10.5Solsep 030705 –e Hydrophobic –g 120 2–10.5

a GE Osmonics, Vista, CA, USA.b Koch Membrane Systems, Wilmington, MA, USA.c Polydimethylsiloxane.

wptv

2

UwlatTsts0(

ppswEiPsdthreH

aaw

J

wuois

d Membrane Extraction Technology, London, UK.e Not specified.f SolSep BV, Apeldoorn, The Netherlands.g Covered by secrecy and non-analysis agreement.

as modified from 12.43 (after reaction) to 8.68 whereby this finalH value was found suitable for the polymeric membranes separa-ion process and it is approximately equal to the pure methanol pHalue.

.4. Permeation experimental set-up

Experiment was done by using a SEPA ST Stirred Cell (OsmonicSA). A circular membrane disk with a diameter of 0.049 m (49 mm)as placed at the bottom of the filtration module, the active top

ayer towards the feed solution. The membrane was sealed betweenTeflon O-ring and a stainless steel porous support. In this way,

he membrane surface has an active area of 0.00152 m2 (15.2 cm2).he feed solution was magnetically stirred with a (Teflon coated)tirrer bar. The maximum volume capacity of the cell is 0.3 L andhe maximum operating pressure is 3100 kPa (SI unit). The stirringpeed was fixed at 400 rpm and range of operating pressures of–3000 kPa. Pressure at the feed side was supplied by an inert gasN2).

Preliminary permeation experiment was conducted for pureolar solvents (water, anhydrous methanol and glycerol) and non-olar solvents (RBD olein and methyl esters) at various ranges ofeparation pressure (600–3000 kPa) and temperature (28–60 ◦C)hich were the reactants and products of the biodiesel reaction.

xperiments at different temperatures were carried out by immers-ng the apparatus in the thermostatically controlled water bath.olymeric SRNF membranes screening experiment for biodieseleparation then was continued for transesterification productirectly after the reaction and after the alkalinity modification ofransesterification product. Transesterification product is definedere as product mixture solution obtained after transesterificationeaction experiment is completed. Crude biodiesel refers to metylsters mixture obtained from a local-based biodiesel plant Goldenope Biodiesel Sdn.Bhd., Malaysia.

The performance of polymeric SRNF membranes was examinedfter the transesterification product separation prior to identifyingny changes on the tested membranes. The volume flux of permeateas expressed as (m3 m−2 s−1) [19]:

= �

A�t(1)

here J (m3 m−2 s−1) is the solvent flux, v (m3) is the permeate vol-me and t (s) is the separation time. The experiments were carriedut at fixed separation temperature of 40 ◦C and the range of operat-ng pressures of 600–3000 kPa. The percentage observed rejectionolutes i for each separation permeate collected was calculated

according to the following equation [19]:

R =(

1 − Ci,p

Ci,f

)× 100% (2)

where Ci,p is the permeate concentration and Ci,f is the feed con-centration.

2.5. Chemical and membrane analysis

To investigate the solute rejection in the nanofiltration pro-cess, the composition of biodiesel in the feed and separationpermeate samples was analyzed using Agilent 6890N gas chro-matography (GC) with a flame ionization detector (FID) employinga DB-5HT capillary column (J&W Scientific) of 30 m length,0.32 mm ID, 0.1 �m film thickness with 5% phenyl and 95% methylaccording to ASTM-D6584 [20]. A deactivated fused silica tubing(3 m × 0.32 mm) coated with syanophenylmethyl was used as theprecolumn. Data collection and analysis were performed using Agi-lent Chemstation software. 1,2,4-Butanetriol and tricaprin wereused as the internal standard 1 and 2, respectively. A stock solutionof n-heptane was added to the vial after the sample was silylatedwith N-methyl-N-trimethylsilyltrifluoracetamide (MSTFA). Sam-ples were prepared for analysis by adding 1 mg of oil to 10 ml septavials. One microliter of the sample was injected into the column.The oven temperature program starts at 50 ◦C (hold 1 min) andramps at 15 ◦C/min to 180 ◦C, 7 ◦C/min to 230 ◦C, and 30 ◦C/minto 380 ◦C (hold 30 min). Methyl ester of oleic acid, 1-monoolein,1, 3-diolein and triolein were used as standard samples to deter-mine the retention time of FAME, MG, DG and TG. These standardsamples retention time would be applied as a method indica-tor in the chromatogram. The methanol content of the feed andseparation permeate samples was detected using Agilent 7890 Agas chromatography (GC) with the flame ionization detector (FID)employing a DB-WAX capillary column of 30 m length, 0.25 mmID and 0.25 �m film capillary column thickness. The temperatureof the GC injector was fixed at 200 ◦C with the column detec-tor of 300 ◦C and the oven temperature was set isothermally at35 ◦C. The calibration curves of the standard solutions showed goodlinearity. The retention times of the all standards are shown inTable 2.

The differences on the surface and cross-sectional morphol-

ogy of the membranes samples were observed for before andafter transesterification product separation process by employingscanning electron microscopy (SEM), using an Oxford Instru-ment Scanning Electron Microscopy Model 7353 (England), underaccelerating voltage of 20 kV. Fourier transform infrared (FTIR)

290 R. Othman et al. / Journal of Membrane Science 348 (2010) 287–297

Table 2The retention times of standards for GC analyses.

Standard Retention time (min)

Triolein (TG) 47.566Diolein (DG) 32.143Monoolein (MG) 24.460Methyl oleate (FAME) 21.46Butanetriol (1st internal standard) 10.52Tricaprin (2nd internal standard) 27.36Glycerin 9.71Methanol 3.30

F(

si4too[

3

3

aaofhwmta

TT

Fig. 2. Non-linear effect of pressure on methanol flux at ambient temperature(∼28 ◦C).

byproduct of transesterification process [14]. From the permeationexperiment, Fig. 1 shows the water permeation flux where waterflux was increased linearly with respect to the applied pressure. Thedata obtained shows a linear profile which conforms to the Hagen

ig. 1. Water flux versus different separation pressure at ambient temperature∼28 ◦C).

pectroscopy (Perkin-Elmer Spectrum One, USA) was used to exam-ne the changes in functional groups at wave numbers between000 cm−1 and 400 cm−1. The wavenumbers of the FTIR func-ional groups are indicated in Table 3. The ester frequency rangef the functional group (C O) that would point to the presencef methyl ester is between 1830–1800 cm−1 and 1780–1640 cm−1

21].

. Results and discussion

.1. Permeability experiments with pure solvents

Preliminary permeation experiment for biodiesel separationnd purification was carried out with pure polar solvents (water,nhydrous methanol and glycerol) and non-polar solvents (RBDlein and methyl esters) which are the reactants and productsrom the transesterification process for biodiesel production. Theydrophobic and hydrophilic membranes that were listed in Table 1

ere tested in this experiment. The behavior of the polymeric SRNFembranes was identified at different separation pressure and

emperature in order to study the potential and capabilities of thepplied membranes for biodiesel separation.

able 3he FTIR functional group types.

Wavenumbers (cm−1) Group Class of compound

3200–3640 O–H Polymeric O–H and water impurities3000–2800 C–H Alkanes1830–18001780–1640 C O Ester1650–1580 C C Alkenes1475–1350 C–H Alkanes950–1300 C–O Primary, secondary and tertiary alcohol

O–H900–650 O–H Aromatic compounds

Fig. 3. Crude biodiesel flux versus different separation pressure at ambient temper-ature (∼28 ◦C).

3.1.1. Effect of applied pressureThe effect of operating pressure on pure solvent flux was exam-

ined in the range of 0–3000 kPa at room temperature (∼28 ◦C)with different types of polymeric SRNF membranes polarity. Purewater permeation flux was investigated since water is the minor

Fig. 4. Effect of temperature on pure methanol flux at constant separation pressureof 600 kPa.

R. Othman et al. / Journal of Membrane Science 348 (2010) 287–297 291

Fp

Ptan(dv

l

ig. 5. Transesterification product flux versus separation time at separation tem-erature of 40 ◦C, pressure of 600 kPa and separation feed pH value of 12.43.

oiseuille equation. Based on the figure, Desal-DL membrane gavehe highest permeate water flux compared to other membranesnd it was followed by Desal-DK, MPF-34 and MPF-44. However,o water flux was obtained through the other tested membranes

Starmem 120, Stramem 240, Solsep 030705 and Solsep 030306F)ue to hydrophobicity effects of the membranes on the polar sol-ent (water).

Pure methanol permeability experiment represented the non-inear behavior or specifically known as asymptotic behavior on

Fig. 7. The SEM images of Solsep 030705 mem

Fig. 6. Modified transesterification product flux versus applied pressure at constantseparation temperature of 40 ◦C and separation feed pH value of 8.68.

the polymeric SRNF membranes with the increasing of separa-tion pressure at constant operation temperature (∼28 ◦C). Fig. 2shows the non-linear increment effect of pressure on methanol

flux with respect to the applied pressure and constant separationtemperature. Desal-DL and Desal-DK exhibit a higher perme-ate flux compared to the other membranes. Desal-DL showedthe highest permeate flux on anhydrous methanol, while Solsep030306F indicated the lowest methanol permeate flux which

brane (a) surface and (b) cross-section.

2 mbran

iThomau[

pepmcstatepbaebmd

92 R. Othman et al. / Journal of Me

ndicates that Solsep 030306F was the less hydrophilic membranes.he behavior of hydrophilic membranes could be explained byydrogen-bonding theory [19]. However, no methanol flux wasbserved for Starmem 120 and Starmem 240. Starmem seriesembranes are hydrophobic membranes and immediately curled

fter being immersed in methanol solution. The permeability val-es obtained here compares favorably with other reported works19,22–23].

Machado et al. [24] described the non-linear behavior ofermeate flux in the case of all tested alcohols (methanol,thanol, iso-propanol, n-butanol and n-penthanol). The flux-ressure curves exhibited a falling rate behavior, which becomesore accentuated with increasing molecular weight. This is an indi-

ation of the asymptotic behavior. Dijkstra et al. [25] also found theame result of non-linear behavior for alcohol solvent permeationhrough laboratory made SRNF membranes. They found some devi-tion from the pore-flow model and a contribution of diffusion inransport of alcohols through a dense PDMS membrane. Bhanushalit al. [1,26] showed the contribution of diffusive and viscous trans-ort of the solvents in rejection of different solutes by use of alack-box model. The same approach was followed by Geens et

l. [27] for transport of six reference components in methanol andthanol for different SRNF membrane. They calculated the contri-ution of diffusion and convection to solute transport through theembranes and concluded that the balance between viscous and

iffusive transport is mainly directed toward convective transport.

Fig. 8. The SEM images of Solsep 030306F mem

e Science 348 (2010) 287–297

Darvishmanesh et al. [28] proposed a model equation for the per-meation of organic solvents from different chemical classes throughpolymeric and ceramic SRNF membranes by considering the trans-port phenomena in SRNF membranes, like diffusive and viscoustransport (viscosity), entrance resistance effects (surface tension),and the polarity effects between the membrane and the solvent(dielectric constant).

Crude biodiesel flux or crude methyl ester flux was obtainedthrough Solsep 030306F and Solsep 030705 membranes, whichincreased linearly to the operating pressure as represented in Fig. 3.Solsep 030705 obtained the highest permeability flux comparedto Solsep 030306F. This observation pointed to the suitability ofSolsep 030705 for crude biodiesel product separation. Accordingto Bhanushali et al. [1], the chemical composition of the mem-branes has considerable impact on organic or inorganic solventtransport. Non-polar solvents show greater flow rates throughhydrophobic membranes, notably polymeric ones, where the sep-aration mechanisms involve polymer–solvent, solvent–solute andsolute–polymer interactions. However, there was no permeate fluxobtained for pure glycerol (polar solvent) and RBD olein (non-polarsolvent) through all the tested membranes. The higher viscos-

ity values of these two pure solvents was the major hidrance forthis permeation test whereby at room temperature and atmo-spheric pressure, the viscosity values of pure glycerol and RBDolein were 0.64 N s m−2 and 0.016 N s m−2, respectively, comparedto the viscosity of water was 0.0008 N s m−2 and methanol was

brane (a) surface and (b) cross-section.

mbran

0es

3

2wbcpit0htwTsed[pai

R. Othman et al. / Journal of Me

.0005 N s m−2. Machado et al. [24] emphasized that the fluxes ofither pure or mixed solvents were mainly affected by surface ten-ion and viscosity of the solvents.

.1.2. Effects of temperatureFig. 4 exhibits the effect of temperature in the range of

8–60 ◦C on the pure anhydrous methanol. Each data pointas measured with different samples of the membranes mem-

rane. It is seen that the different temperature value had aonsiderable effect on the permeate flux value. The increase ofermeate flux was observed to be linearly proportional to the

ncrease in temperature. The order of increase (as measured fromhe slopes) follows: Solsep 030306F < MPF-34 < MPF-44 < Solsep30705 < Desal-DK < Desal-DL. Desal-DL and Desal-DK exhibitigher permeate flux with the increasing separation tempera-ure compared to the other tested membranes. Solsep 030306Fas only minimally affected by the increase in temperature.

he flux-temperature is usually well correlated by the hypothe-is that a rise in temperature increases permeation flux throughither a reduction in solvent viscosity or an increase in solvent

iffusion coefficient or by an increase in polymer chain mobile24]. According to the free volume theory, an increase in tem-erature can increase the thermal motion of the polymer chainnd generate more free volume in the polymer matrix to facil-tate sorption and diffusion of permeant in the membrane. The

Fig. 9. The SEM images of Starmem 240 mem

e Science 348 (2010) 287–297 293

increase in the free volume makes the membrane more perme-able but less discriminative to the permeation of the permeants.The temperature effect on flux of aqueous solutions is well cor-related by an Arrhenius plot, given by the following equation[24,29,30]:

J = AJ exp

(− EJ

RT

)(3)

where AJ is a frequency factor and EJ is a permeation activationenergy.

3.2. Permeability experiments with the unmodifiedtransesterification product

The permeation experiments were carried out to exam-ine the permeability flux of the transesterification productsthrough the all the membranes listed in Table 1 at con-stant separation temperature of 40 ◦C and range of operatingpressures from 600 kPa to 3000 kPa. The feed weight percent-age of transesterification products were as follows: triglyc-

erides (TG) 0.0009 wt%, monoglycerides (MG) 0.4654 wt%, diglyc-erides (DG) 0.0357 wt%, free glycerin 9.1893 wt%, methyl esters90.3087 wt% and unreacted excess methanol was 0.0917 wt%.The pH value of the transesterification product was recorded at12.43.

brane (a) surface and (b) cross-section.

294 R. Othman et al. / Journal of Membran

laSafl0onittnDh(iboaa

3t

aloo(e9pr(ptb5tc0l

3

i

standard reference to examine the differences on the FTIR spec-trums for the tested membranes. Membrane Solsep 030705, Solsep030306F and Starmem 240 were used in this analysis, thus these3 membranes showed considerable permeate flux of the alkalinity

Fig. 10. The FTIR spectrums of pure methanol and crude biodiesel.

The permeability of transesterification products improved withonger separation time (of up to 60 min) at the lowest range of oper-tion pressure of 600 kPa and constant temperature of 313.15 K fortarmem 120, Starmem 240, Solsep 030705 and Solsep 030306Fs presented in Fig. 5. Starmem 240 gave the highest permeationux compared to the other membranes. It was followed by Solsep30705, Starmem 120 and Solsep 030306F. However, at the endf the separation process, the membranes were found to be sig-ificantly damaged and the permeability flux for each membrane

ncreased rapidly. Consequently, the permeability experiments ofransesterification product could not be carried out further forhe higher range of the separation pressure. On the other hand,o transesterification product permeate flux was observed foresal-DL, Desal-DK, MPF-34 and MPF-44. This was because of theydrophilicity effects of the membranes on the non-polar solventbiodiesel), where a hydrophilic molecule or portion of a molecules one that is typically charge-polarized and capable of hydrogenonding, enabling it to dissolve more readily in water than in oil orther hydrophobic solvents. These appear when a hydrogen atomttached to an electronegative atom such as oxygen (hydroxyl), isttracted by an electronegative group in another chain [31].

.3. Permeability experiment with the modified alkalinity ofransesterification product

The permeation experiments for transesterification productfter alkalinity modification were then tested for all the membranesisted at constant separation temperature of 40 ◦C and range ofperating pressures of 600–2200 kPa. The feed weight percentagef transesterification product after neutralization was triglyceridesTG), 1.0011 wt%, monoglycerides (MG), 0.0036 wt%, diglyc-rides (DG), 1.4179 wt%, free glycerin, 5.0440 wt%, methyl esters,2.3963 wt% and unreacted excess methanol was 0.1370 wt%. TheH value of the transesterification product after neutralization wasecorded at 8.68. Fig. 6 shows that at the moderate levels of pressure≤2200 kPa) the flux behavior for the modified transesterificationroduct increased linearly with respect to the applied pressurehrough Solsep 030705, Solsep 030306 F and Starmem 240 mem-ranes. However, no permeate flux was obtained through the otherlisted membranes. Out of all tested membranes, it was found

hat Solsep 030705 membrane gave the highest permeate fluxompared to the other membranes and it was followed by Solsep30306 F and Starmem 240 membrane. The data obtained show a

inear profile which conforms to the Hagen Poiseuille equation.

.4. Scanning electron microscope (SEM) analysis

The SEM images of surface and cross-section were visualizedn Figs. 7–9 for three selected membranes namely Solsep 030705,

e Science 348 (2010) 287–297

Solsep 030306F and Starmem 240 membranes. For each membrane,the imaging was carried out for clean membrane (Figs. 7–9a(i)and b(i)), membrane after permeation of transesterification prod-uct without modification (Figs. 7–9a(ii) and b(ii)), membraneafter permeation of transesterification product with modification(Figs. 7–9a(iii) and b(iii)). The SEM images of each clean membranewere employed as the standard indicator to examine any changeson the tested membranes after permeation with the unmodifiedand alkalinity modified transesterification product separation pro-cess. As can be seen, the morphological structures of the testedmembranes were physically affected after permeation reflectingthe permeation behavior of the unmodified transesterificationproduct as discussed previously. However, there were much lesseffects on the morphological structure of the membranes after thepermeation of alkalinity modified transesterification product. Thisagain supported the permeation behavior of the alkalinity modifiedtransesterification product separation process as shown in Fig. 6.

3.5. Fourier transform infrared spectroscopy (FTIR) analysis

The wavenumbers of the FTIR functional groups were indicatedin Table 3. The ester frequency range of the functional group (C O)that would point to the presence of methyl ester was between1830–1800 cm−1 and 1780–1640 cm−1 [21]. Fig. 10 shows theFTIR spectrums for crude methyl esters and pure methanol as a

Fig. 11. The FTIR spectrums of Solsep 030705 membrane (a) after unmodified trans-esterification product separation process and (b) after modified transesterificationproduct separation process, based on the original membrane spectrum as standardindication.

R. Othman et al. / Journal of Membrane Science 348 (2010) 287–297 295

Ftfis

ma

atFsfimiiipc

3

3

irtmcolcp

St

order to evaluate the capabilities of the membrane to separatethe alkalinity modified transesterification product. The perme-ation experiment was carried out for 30 min at constant separationtemperature of 40 ◦C and pressure 1000 kPa. Fig. 16 shows theselectivity of the tested membrane which exhibited the highest

ig. 12. The FTIR spectrums of Solsep 030306F membrane (a) after unmodifiedransesterification product separation process and (b) after modified transesteri-cation product separation process, based on the original membrane spectrum astandard indication.

odified transesterification product with respect to the increasingpplied pressure.

The FTIR spectrums of each tested membranes before andfter the permeation of the unmodified and modified transes-erification product were presented in Figs. 11–13. As seen inigs. 11a and 13a, the functional groups of the membranes showedome differences before and after the permeation of the unmodi-ed transesterification product. The differences were however veryinimal for the modified transesterification product case as shown

n Figs. 11b and 13b. These results supported the previous find-ngs from the permeation experiments and the SEM analysis. It alsondicated that the integrity of the membranes remained intact afterermeation of the modified transesterification within the operatingonditions studied here.

.6. Performance of polymeric SRNF membrane

.6.1. Flux reduction after permeationThe performance of Solsep 030705 and Solsep 030306F were

nvestigated to evaluate the efficiency of the membrane to sepa-ate the alkalinity modified transesterification product. Accordingo Wang et al. [9], methanol could be used as a solvent to clean the

embrane after being used in the biodiesel microfiltration pro-ess. Thus methanol permeation flux was used here as an indicatorf the cleanability of the membrane after permeation of the alka-inity modified transesterification product. The experiments were

◦

arried out at constant temperature of 40 C and range of operatingressures from 600 kPa to 3000 kPa.

Figs. 14 and 15 represent the flux permeation of methanol forolsep 030705 and Solsep 030306F before and after permeation ofhe modified transesterification product. The flux reduction was

Fig. 13. The FTIR spectrums of Starmem 240 membrane (a) after unmodified trans-esterification product separation process and (b) after modified transesterificationproduct separation process, based on the original membrane spectrum as standardindication.

determined to be about 74.97% and 78.19% for Solsep 030705 andSolsep 030306F, respectively. Thus even though the membraneshave the potential to be used for biodiesel applications, furtherwork on the extent and causes of fouling especially for long-termapplication should be carried out.

3.7. Selectivity of the Solsep 030705 membrane

The selectivity of Solsep 030705 membrane was determined in

Fig. 14. Methanol flux versus different pressure at separation temperature of 40 ◦Cthrough Solsep 030705 membrane.

296 R. Othman et al. / Journal of Membran

Fig. 15. Methanol flux versus different pressure at separation temperature of 40 ◦Cthrough Solsep 030306F membrane.

Fta

roomtteetdtb

4

wdtgeatmmgi0

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

ig. 16. Rejection of Solsep 030705 membrane for the alkalinity modified transes-erification product separation process at constant separation pressure of 1000 kPand temperature of 40 ◦C and separation feed pH value of 8.68.

ejection for TG of about 99.80% followed by DG of about 97.16%, MGf 40.65%, free glycerin of 75.24% and ester of 25.37%. The rejectionf methyl ester was further reduced after 30 min, which means thatore methyl ester permeated across the membrane. Meanwhile,

he highest percentage of methanol rejection was of 74.98% rejec-ion. The results indicated a possibility of separating the methylsters from the mixture of excess (unreacted) methanol and glyc-rides groups. The methanol can be considered to be recycled intohe reaction medium after physical separation from the glyceridesue to the viscosity difference. Future work will be focused inhis direction including the long-term performance of the mem-rane.

. Conclusion

In this study eight polymeric solvent resistant NF membranesere tested for the potential to be used within the biodiesel pro-uction process. The membranes were challenged to permeation ofransesterification products, which consist of triglycerides, mono-lycerides, diglycerides, free glycerin, methyl esters and unreactedxcess methanol, at pH of 12.43 and 40 ◦C and it was found thatt the end of the permeation process, the membranes were foundo be significantly damaged and the permeability flux for each

embrane increased rapidly. SEM and FTIR analyses confirmed theajor changes occurred on the membrane surface and functional

roups. The pH of the transesterification products was then mod-fied to 8.68 and improved performance was observed for Solsep30705, Solsep 030306F and Starmem 240. Study on the selectiv-

[

[

e Science 348 (2010) 287–297

ity of the Solsep 030705 membrane during permeation with themodified transesterification product showed the possibility of sep-arating the methyl esters from methanol and glycerides group.Future work will be focused in this direction including the long-term performance of the membrane.

Acknowledgements

The authors gratefully acknowledge the financial support givenfor this work through the grants 03-01-02-SF0054 and UKM-KK-02-FRGS0002-2006.

References

[1] D. Bhanushali, S. Kloos, C. Kurth, D. Bhattacharryya, Performance of solventresistant membranes for non-aqueous systems: solvent permeation results andmodelling, J. Membr. Sci. 189 (2001) 1–21.

[2] S. Koile, R. Subramanian, H. Nabetani, M. Nakajima, Separation of oil con-stituents in organic solvents using polymeric membranes, J. Am. Oil Chem. Soc.79 (2002) 937–940.

[3] A. Lobo, A. Cambiella, J.M. Benito, C. Pazos, J. Coca, Ultrafiltration of oil-in-wateremulsions with ceramic membranes: influence of pH and crossflow velocity, J.Membr. Sci. 278 (2006) 328–334.

[4] C. Pagliero, N. Ochoa, J. Marchese, M. Mattea, Degumming of crude soybean oilby ultrafiltration using polymeric membranes, J. Am. Oil Chem. Soc. 78 (2001)793–796.

[5] N. Ochoa, C. Pagliero, J. Marchese, M. Mattea, Ultrafiltration of vegetableoils degumming by polymeric membranes, Sep. Purif. Technol. 22–23 (2001)417–422.

[6] M.A. Dube�, A.Y. Tremblay, J. Liu, Biodiesel production using a membrane reac-tor, Bioresource Technol. 98 (2007) 639–647.

[7] P. Cao, A.Y. Tremblay, M.A. Dube�, K. Morse, Effect of membrane pore size on theperformance of a membrane reactor for biodiesel production, Ind. Eng. Chem.Res. 46 (2007) 52–58.

[8] P. Cao, M.A. Dube�, A.Y. Tremblay, Methanol recycling in the production ofbiodiesel in a membrane reactor, Fuel 87 (2008) 825–833.

[9] Y. Wang, X. Wang, Y. Liu, S. Ou, Y. Tan, S. Tang, Refining of biodiesel by ceramicmembrane separation, Fuel Process. Technol. 90 (2008) 422–427.

10] H. Zhou, H.F. Lu, B. Liang, Solubility of multicomponent systems in the biodieselproduction by transesterification of Jatropha curcas L. oil with methanol, J.Chem. Eng. Data 51 (2006) 1130–1135.

11] Z.H. Sun, Inorganic membranes and membrane reactors, Korean J. Chem. Eng.18 (2001) 775–787.

12] L. Bournay, A. Baudot, Processing for Producing Fatty Acid Alkyl Esters andGlycerol of High Purity, US Patent 7,138,536 (2006).

13] L.C. Meher, D. Vidya Sagar, S.N. Naik, Technical aspects of biodiesel productionby transesterification—a review, Renew. Sust. Energy Rev. 10 (2006) 248–268.

14] J.V. Gerpen, Biodiesel processing and production, Fuel Process. Technol. 86(2005) 1097–1107.

15] L.H. Cheng, Y.F. Cheng, S.Y. Yen, Ultrafiltration of triglycerides from biodieselusing the phase diagram of Oil–FAME–MeOH, J. Membr. Sci. 330 (2009)156–165.

16] H.W. Alex, D. Posarac, N. Ellis, Assessment of four biodiesel production pro-cesses using HYSYS. Plant, Bioresource Technol. 99 (2008) 6587–6601.

17] S.M. Akers, J.L. Conkle, S.N. Thomas, K.B. Bider, Determination of the heat ofcombustion of biodiesel using bomb calorimetry, J. Chem. Educ. 83 (2006)260–262.

18] Y. Zhang, M.A. Dube�, D.D. McLean, K. Kates, Biodiesel production from wastecooking oil. 1. Process design and technological assessment, Bioresource Tech-nol. 89 (2003) 1–16.

19] X.J. Yang, A.G. Livingston, L.F. Santos, Experimental observations of nanofiltra-tion with organic solvents, J. Membr. Sci. 190 (2001) 45–55.

20] ASTM-D 6584, Test method for determination of free and total glycerine inB-100 biodiesel methyl esters by gas chromatography, ASTM USA, 2000.

21] D.L. Pavia, G.M. Lampman, G.S. Kriz, Introduction to Spectroscopy: A Guide forStudents of Organic Chemistry, ThomsanLearning, Australia, 2000.

22] J.A. Whu, B.C. Baltzis, K.K. Sirkar, Nanofiltration studies of larger organicmicrosolutes in methanol solutions, J. Membr. Sci. 170 (2000) 159–172.

23] Y. Zhao, Q. Yuan, Effect of membrane pretreatment on performance of solventresistant nanofiltration membranes in methanol solutions, J. Membr. Sci. 280(2006) 195–201.

24] D.R. Machado, D. Hasson, R. Semiat, Effect of solvent properties on perme-ate flow through nanofiltration membrane. Part I. Investigation of parametersaffecting solvent flux, J. Membr. 163 (1999) 93–102.

25] M.F.J. Dijkstra, S. Bach, K. Ebert, A transport model for organophilic nanofiltra-

tion, J. Membr. Sci. 286 (2006) 60–68.

26] D. Bhanushali, D. Bhattacharyya, Advances in solvent-resistant nanofiltrationmembranes, Adv. Membr. Technol. 984 (2003) 159–177.

27] J. Geens, B.V. Bruggen, C. Vandecasteele, Transport model for solvent per-meation through nanofiltration membranes, Sep. Purif. Technol. 48 (2006)255–263.

mbran

[

[

[

R. Othman et al. / Journal of Me

28] S. Darvishmanesh, A. Buekenhoudt, J. Degrève, B.V. Bruggen, General

model for prediction of solvent permeation through organic and inor-ganic solvent resistant nanofiltration membranes, J. Membr. Sci. 334 (2009)43–49.

29] P. Sampranpiboon, R. Jiraratananon, D. Uttapap, X. Feng, R.Y.M. Huang, Per-vaporation separation of ethyl butyrate and isopropanol with polyether blockamide (PEBA) membranes, J. Membr. Sci. 173 (2000) 53–59.

[

e Science 348 (2010) 287–297 297

30] J.J. Shieh, R.Y.M. Huang, Pervaporation with chitosan membranes. II. Blend

membranes of chitosan and polyacrylic acid and comparison of homogeneousand composite membrane based on polyelectrolyte complexes of chitosan andpolyacrylic acid for the separation of ethanol–water mixtures, J. Membr. Sci.127 (1997) 185–202.

31] M. Mulder, Basic Principles of Membrane Technology, Kluwer Academic Pub-lisher, Netherlands, 1996.