Chemical Engineering and Processing 86 (2014) 162–172

PDMS/ceramic composite membrane for pervaporation separation ofacetone–butanol–ethanol (ABE) aqueous solutions and its applicationin intensification of ABE fermentation process

Gongping Liu, Lin Gan, Sainan Liu, Haoli Zhou, Wang Wei, Wanqin Jin *State Key Laboratory of Materials-Oriented Chemical Engineering, College of Chemistry and Chemical Engineering, Nanjing Tech University (the formerNanjing University of Technology), 5 Xinmofan Road, Nanjing 210009, PR China

A R T I C L E I N F O

Article history:Received 8 March 2014Received in revised form 1 June 2014Accepted 19 June 2014Available online 7 September 2014

Keywords:PDMS/ceramic composite membranePervaporationAcetone–butanol–ethanol (ABE)FermentationIntensification

A B S T R A C T

Pervaporation (PV) has attracted increasing attention because of its potential application in bio-butanolrecovery from fermentation process. In this work, PDMS/ceramic composite membrane was employedfor PV separation of acetone–butanol–ethanol (ABE) aqueous solutions. The influence of coupling effecton the molecular transport during the PV process was systematically investigated. The separationperformance and transport phenomena of ABE molecules were discussed based on the analysis andcalculation of physicochemical properties such as solubility parameter, polarity parameter, interactionparameter, activity coefficient. The results suggested that the ABE separation factor was mainlydetermined by the intrinsic solubility parameter and driving force. Coupling effect in the ABEmulticomponent system was closely related to the interaction parameters between componentsthemselves and between component and membrane. Also, the PDMS membrane was integrated with ABEfermentation to construct an efficient intensification process. It was found that the rate matching offermentation and in situ removal could improve the ABE productivity by 2 times.

ã 2014 Elsevier B.V. All rights reserved.

Contents lists available at ScienceDirect

Chemical Engineering and Processing:Process Intensification

journal home page : www.elsevier .com/ loca te /cep

1. Introduction

Pervaporation (PV) is a membrane technology that could realizemolecular-level separation [1], which has been widely used invarious applications, such as biofuels production [2–5], VOCsremoval [6,7], solvent dehydration [8–11], organic/organic mix-tures separation [12,13] and so on. In the past decades, byintegrating with chemical reaction or fermentation, PV techniquewas also employed for process intensification [14,15]. Our previouswork found that the integration of PV with ethyl acetate reactivedistillation [16] or ethyl lactate hydrolysis [17] could improve thereaction conversion and product purity.

Recently, because of the shortage of fossil energy and climaticchange, biofuels, as one of the important renewable energies, hasreceived increasing attention [18–20]. As an advanced biofuel,bio-butanol has the advantages of less volatile and flammable,higher energy content, water insensitivity and less hazardous tohandle compared with bio-ethanol [19]. In the bio-butanolproduction process, a critical issue is the end-product inhibition

* Corresponding author. Tel.: +86 25 83172266; fax: +86 25 83172292.E-mail address: wqjin@njtech.edu.cn (W. Jin).

http://dx.doi.org/10.1016/j.cep.2014.06.0130255-2701/ã 2014 Elsevier B.V. All rights reserved.

effect during the acetone–butanol–ethanol (ABE) fermentation[20]. The butanol would restrain the microbial growth, resultingin a low solvent content in the fermentation broth. Generally, theconcentration of butanol is lower than 13 g/L [19], leading to a lowproductivity and high energy cost for obtaining bio-butanol. Thus,separation technologies were used to integrate with thefermentation to continuously remove butanol from the fermen-tation broth as it is produced, in order to relief the inhibitoryeffect and implement continuous fermentation. Among thesetechniques for butanol in situ removal (e.g., adsorption [21],extraction [22], gas stripping [23]), PV is considered to be thegreatest potential candidate owing to its energy-saving andefficiency, as well as harmless to microorganisms [19].

The key point of applying PV for butanol removal fromfermentation broth is the PV membranes with high selectivepermeation of butanol. Membranes for recovering butanol frommodel solutions or fermentation broths are typically organophilicPV membranes [24], including polydimethylsiloxane (PDMS)membranes [25–27], poly[1-(trimethylsilyl)-1-propyne] (PTMSP)membranes [28], poly(ether block amide) (PEBA) membranes[29,30], polypropylene (PP) membranes [31], polytetrafluoro-ethylene (PTFE) membranes [22], liquid membranes [32], zeolitemembranes [33,34] and mixed matrix membranes (MMMs)

G. Liu et al. / Chemical Engineering and Processing 86 (2014) 162–172 163

[29,35,36]. Until now, it is believed that the commonly used PDMS-based membranes with good selectivity and stability are regardedas the most potential membranes for butanol recovery [37]. Ourprevious work has also reported the fabrication of a novel PDMS/ceramic composite membrane and its application for recoveringbutanol from aqueous solution or fermentation process, whichexhibited high flux and good selectivity [5,25–42]. More detailsabout the PV membranes for bio-butanol production can bereferred in our recent review [3].

Although a lot of progress has been made in the developing PVmembranes and integrating PV with fermentation process, littleattention was paid on the influence factors that would affect thetransport of butanol molecules during the PV process [13]. First,besides butanol, the fermentation broth also mainly containsacetone and ethanol. In the multicomponent mixtures (e.g.,acetone/butanol/ethanol/water), the butanol transport would bedifferent from that in the binary butanol/water mixture [30].Coupling effects usually occur because of the mutual interactionsbetween the permeate components in the membrane, as well as byinteractions between the components and the membrane material[43–45]. As a result, the separation performance of butanol in theABE aqueous solution cannot be predicted by only using the valuesin the binary mixtures based on the solution-diffusion theory.Furthermore, there are many differences between real fermenta-tion broth and aqueous solution, not only the different density, pHand viscosity, but also with or without the inorganic salts, glucose,microbial cells and several other metabolic compounds. Some ofthem would have positive or negative effects on the membraneperformance in the fermentation process [26]. Therefore, when thepervaporation membrane is used in the fermentation process, theseparation performance would be determined not only by thebutanol concentration, but also by the different compositions ofethanol and acetone, as well as the salts, glucose, microbes in thebroth. It is necessary to systematically study the effects of differentABE model solutions on membrane performance.

In this work, therefore, a systematical investigation on butanolrecovery from multicomponent aqueous solutions (binary, ternaryand quaternary systems) was carried out to study the influence ofcoupling effect on the molecular transport during the PV process.Based on the analysis and calculation of physicochemical proper-ties such as solubility parameter, polarity parameter, interactionparameter, activity coefficient, the PV performance and transportphenomena of ABE molecules were thoroughly discussed. Themembrane used here was PDMS/ceramic composite membrane,which has showed high performance in the PV-integratedfermentation process in our previous work [26,42]. Also, thePDMS membrane was brought into efficient process intensificationby integrating PV with ABE fed-batch fermentation, in order toimprove the butanol productivity via in situ product removal.

2. Experimental

2.1. PV and PV-integrated fermentation experiment

The PV experiment was conducted on a homemade apparatus[26], in which the PDMS/ceramic composite membrane (thepreparation was followed by our previous work [5]) was sealed in anylon PV cell, with an effective membrane region of 48.9 cm2. Thefeed tank was maintained at 37 �C and the flow rate was fixed at15 L/h during the PV experiment. The permeate vapor wascollected in liquid nitrogen trap. Permeate pressure was below400 Pa during collections. After a steady state was obtained, thecold trap was exchanged every one hour with a consecutivepermeate collection. The concentrations acetone, butanol andethanol were determined by gas chromatography (GC-2014,Shimadzu, Japan) equipped with a thermal conductivity detector

(TCD) using a Porapak Q packed column and helium (He) as thecarrier gas. Iso-butanol was used as an internal standard. If thepermeate separated into two phases, the permeate sample wasdiluted with deionized water to one phase prior to injection. Inorder to make sure the reproducibility, all the experiment resultswere repeated at least three times, and the errors were <10%.

Before the ABE fermentation-PV coupled process, the PVmembrane and any part of the PV system that would contactwith the fermentation broth were sterilized at 110 �C for 15 min bycirculating sterile deionized water. When the n-butanol concen-tration in the fermentation broth reached to ca. 4.5 g/L, the sterilePV system was coupled to fermentor and conducted continuouslywith the fermentation.

The PV performance of a membrane is usually expressed interms of the permeation flux J and separation factor b:

J ¼ WAt

(1)

b ¼ y=ð1 � yÞx=ð1 � xÞ (2)

where W is the weight of the permeate, A is the effective area of themembrane, and t is the permeation time interval for thepervaporation; y and x are the weight fractions of componentsin the permeate and feed, respectively.

Since the permeant flux is dependent on the operatingconditions, normalizing the permeant flux with respect to thedriving force for permeation will be useful to further understandthe permeant-specific intrinsic membrane properties. This intrin-sic membrane property is the permeability P and selectivity a,which are defined based on the solution-diffusion mechanism:

Pi ¼Ji � l

Pvapori;feed � ni;permeatePpermeate

(3)

a ¼ Pi

Pj(4)

where Ji is the partial permeate flux of component i,l themembrane thickness, pvapori;feed the equilibrium partial vapor pressureof i in the feed, ni, permeate the mole fraction of i in the permeate andppermeate the permeate pressure; Pi and Pj are the permeability ofsolvent and water, respectively.

2.2. Fermentation

2.2.1. Culture and inoculum preparationClostridium acetobutylicum XY16, (China Center for Type Culture

Collection, CCTCC No.: M 2010011) was used in all experiments.Cells were grown in 50 mL sealed anaerobic bottles containing25 mL seed medium with the following composition: solublestarch 10 g/L, yeast extract 3 g/L, peptone 5 g/L, ammonium acetate2 g/L, sodium chloride 2 g/L, KH2PO41 g/L, K2HPO41 g/L, MgSO4 3 g/L and FeSO4�7H2O 0.01 g/L. The medium was autoclaved at 121 �Cfor 15 min and cooled to 37 �C. Anaerobic bottles were inoculatedwith 1 mL of a �70 �C glycerol stock culture and incubated at 37 �Cfor 20–24 h as the primary seed culture. Before inoculation,nitrogen was bubbled through the medium for 1 min to removeoxygen. 7.5 mL of the primary seed culture was inoculated into250 mL sealed anaerobic bottles containing 150 mL of seedmedium and incubated at 37 �C for 12 h as the secondary seedculture. Before inoculation, nitrogen was bubbled through themedium for 3 min to remove oxygen.

Table 1Basic physical properties of ABE and water [46].

Physical properties Acetone Butanol Ethanol Water

Molecular formula CH3COCH3 C4H9OH C2H5OH H2OMolecular weight/g/mol 58.08 74.12 46.07 18.01Density/g/cm3 (20 �C) 0.793 0.810 0.789 0.998Boiling point/�C 57 117 79 100Solubility in water/g/100 mL (20 �C) Miscible 7.7 Miscible –

Saturated vapor pressure/kPa (37 �C) 50.082 1.934 15.299 6.256

164 G. Liu et al. / Chemical Engineering and Processing 86 (2014) 162–172

2.2.2. FermentationBatch fermentation was conducted in a 3.5 L serum bottle

(Boeco, Germany). The fermentation medium contained: glucose60 g/L, ammonium acetate 2.2 g/L, corn steep liquor (CSL) 1 g/L,sodium chloride 0.01 g/L, KH2PO4 0.5 g/L, K2HPO4 0.5 g/L, MgSO4

0.2 g/L, MnSO4�7H2O 3 g/L and FeSO4�7H2O 0.01 g/L. The pH was setat 6.8 using 5 M NaOH. The fermentor containing 1.5 L fermenta-tion medium was autoclaved at 121 �C for 15 min and cooled underoxygen-free nitrogen gas atmosphere to 37 �C for 20 min in cleanbench area, then 150 mL of the secondary seed culture wasinoculated into fermentor. Batch fermentation mode was precededfor 32 h and then integrated with PV process. When the unutilizedglucose decreased to 20 g/L, 130 mL feed medium was added to thefermenter at regular intervals (pH of the feed medium was set at6.8 using 5 M NaOH, glucose 200 g/L, other components wereidentical to the fermentation medium). Glucose concentration wasanalyzed by an SBA-40C biosensor analyzer (Shandong ProvinceAcademy of Sciences, China). The rate of glucose utilization wasdefined as grams of glucose utilized over a given time interval inthe working volume of the fermenter (L) and is expressed as g/(L/h). The productivity of solvent was defined as ABE (g) produced fora given time interval in the working volume of fermenter (L) and isexpressed as g/(L/h). The yield was calculated as the total amountof solvent produced divided by the total amount of glucose utilizedand is expressed as g/g.

3. Results and discussion

3.1. Physicochemical properties of feed components and membrane

In order to study the transport of different molecules in themembrane, it is necessary to know the physicochemical propertiesof the permeate components and membrane material. The basicphysical properties of acetone, butanol, ethanol and water arelisted in Table 1. It is found that butanol has the highest molecularweight and boiling point, while lowest solubility and saturatedvapor pressure, which are much different from acetone andethanol. These characters would have a certain influence on themolecular diffusion and driving force in the PV process.

Generally, the separation mechanism of polymeric pervapora-tion membranes is described by solution-diffusion model. Thenature of the solution of feed components in the polymericmembrane is breaking up the polymer chain packing and thenswelling the membrane. Thus, the solution process is determined

Table 2Interaction parameters between permeate components and membrane [47,49].

Component Solubility parameter/(MPa)1/2

PDMS 14.9

Acetone 19.7

Butanol 23.32

Ethanol 26.2

Water 47.9

a Reported by Ref. [47].b Reported by Ref. [50].

by the mutual interactions of component–component andcomponent-membrane. The molecular interaction between con-densed matters can be evaluated by the Hildebrand solubilityparameter d, which is defined as the square root of the cohesiveenergy density. Solubility parameter can be a good indication ofaffinity between polymeric membrane and solvent, and thisaffinity could be estimated by Flory–Huggins equation [47,48]:

x ¼ Vi

RT

� �ðd2 � d1Þ2 (5)

where x is the interaction parameter, d1 and d2 are the solubilityparameter of polymeric membrane and solvent, respectively; Vi, Rand T are the molar volume, gas constant and kelvin temperature,respectively. The smaller interaction parameter, the strongeraffinity, and the higher solubility of solvent in polymericmembrane.

Table 2 lists the solubility parameters of ABE, water and PDMSmembrane, and the calculated interaction parameters betweenfeed components and PDMS membrane. The result indicates thatthe affinity between ABE and PDMS membrane follows the ordersof acetone > butanol > ethanol. It can be expected that there wouldbe much difference in the solution process of ABE molecules in thePDMS membrane.

The membrane performance would also be affected by thevariation of ABE concentrations during the fermentation process.According to Eq. (3), the main reason should be the difference ofdriving force resulted from variation of activity coefficient g . NRTLequation was used to calculate the activity coefficient of ABE andwater by using Aspen Plus program. The effects of feedconcentration in binary mixtures (i.e., acetone/water, butanol/water, ethanol/water) on the activity coefficients of ABE and waterwere studied under the feed temperature (37 �C) and pressure(1 atm). As shown in Fig. 1, with the increase of ABE concentration,the activity coefficients of ABE gradually decreased while that ofthe water increased slightly. With the same concentration, thevalues and variations of ABE activity coefficients follow the ordersof butanol > acetone > ethanol.

3.2. Comparison of individual behaviors of ABE in binary systems

The PV performance of PDMS/ceramic composite membrane inthe binary mixtures (butanol/water, acetone/water and ethanol/water) was separately tested under different feed concentrations.The concentration ranges of butanol, acetone and ethanol were

Polarity parameter (P) Interaction parameter (x)

– –

0.695 1.5a (0.75b)0.096 7.2a (2.39b)0.268 8.8a (3.05b)0.819 30.3a (7.91b)

Fig. 2. Effect of feed concentration on total flux and separation factor (a) butanol,(b) acetone, and (c) ethanol.

Fig. 1. Effect of feed concentration on activity coefficients of ABE and water.

G. Liu et al. / Chemical Engineering and Processing 86 (2014) 162–172 165

fixed at 0.47–2.3 wt%, 0.22–1.27 wt%, 0.05–0.28 wt%, respectively,all of which were within the limits of product contents of ABEfermentation.

Fig. 2a–c shows the variation of total flux and separation factorwith the feed concentration. It is found that, the PV performance ofABE exhibited similar tendency with increasing feed concentra-tion, i.e., the total flux increased linearly while the separationfactor progressively declined. Comparing the total flux andseparation factor at the same feed concentration, one can findthat the separation factor of PDMS/ceramic composite membranefor ABE follows the orders of acetone > butanol > ethanol. It waswell agreed with the analysis result of solubility parameters inSection 3.1. However, the conclusion “stronger affinity leads tohigher selectivity” still cannot be drawn from this phenomenon,and the reason will be discussed later.

To further investigate the effects of feed concentration on theABE molecular transports, the individual fluxes of butanol, acetoneand ethanol were plotted vs. feed concentration, as displayed inFig. 3a–c. The individual fluxes also linearly increased with feedconcentration, and the corresponding linear fittings were given inthe figures. From their slopes, it is reasonable to speculate thatwith the driving force of feed concentration, the order of transportresistance of ABE in the PDMS membrane was ethanol > butanol >acetone. This is relevant with the affinity with membrane and alsothe size and configuration of the molecular structure. According tothe solubility parameters of ABE and PDMS, the affinity betweenABE and PDMS membrane follows the orders of acetone > butanol >ethanol, which is the main reason for the differences of transportresistance of ABE in the PDMS. Moreover, the molecular structureof acetone is more favorable for its diffusion in the membranecompared with butanol. Although the molecular size of butanol isbigger than ethanol, the stronger affinity with PDMS membraneleads to the lower transport resistance than that of ethanolmolecule.

Moreover, the permeance and selectivity were calculated withthe aim of distinguishing the effects operating conditions(component’s activity variation resulted from concentration’schange) from the intrinsic properties (variation of transportresistance) on the PV performance. As shown in Fig. 4, withincreasing feed concentration, the entire ABE permeance exhibiteda remarkable decline, while the water permeance almost keptconstant, leading to the decrease of membrane selectivity.

It is interesting to note that the individual permeance wasdifferent from the flux at the same feed concentration: butanolowned the highest permeance, and the next was acetone and

ethanol. Although acetone flux was a little higher than butanol, thesaturated vapor pressure of acetone (50 kPa) is much higher thanbutanol (1.93 kPa), leading to higher driving (Pf,i� Pp,i) force ofacetone and thus lower permeance according to Eq. (3). Therefore,the conclusion “membrane separation factor for ABE followed theorder of acetone > butanol > ethanol” (Fig. 2) could be for the

Fig. 3. Effect of feed concentration on individual flux (a) butanol, (b) acetone, and(c) ethanol.

166 G. Liu et al. / Chemical Engineering and Processing 86 (2014) 162–172

whole PV process, which was affected by the differences of drivingforce. The intrinsic selectivity of PDMS membrane for ABE shouldbe the order of butanol > ethanol > acetone. In spite of the strongestaffinity with PDMS membrane, acetone also has the highest

polarity parameter (as listed in Table 2), resulting in the lowestpermeance and selectivity. As for butanol, its polarity parameter isthe lowest, making fastest permeance in the PDMS membrane [51].Additionally, the A/B/E permeance decline is mainly attributed tothe coupling effect, in which the A/B/E–water clusters were formedwith the aid of hydrogen bonds. The clusters have larger kineticdiameters and reduced diffusivity in the membrane, leading to thedecrease of A/B/E permeance [9,52].

3.3. Coupling effect in the multicomponent systems

Based on the study of individual behaviors of butanol, acetoneand ethanol in binary systems, the coupling effect on moleculartransport was investigated in the multicomponent systems step-by-step, from butanol-containing ternary mixtures (A/B/W, E/B/W) to quaternary mixtures (A/B/E/W). By keeping the butanolconstant meanwhile varying the acetone or ethanol concentra-tion in the multicomponent systems, the separation performanceof butanol, acetone and ethanol in the binary, ternary andquaternary systems were compared in details. The concentrationranges of acetone (0.30–1.49 wt%) and ethanol (0.12–0.35 wt%)were well controlled in accordance with their binary systems, andthe butanol concentration was fixed at 1.2 wt% (typical concen-tration in the ABE fermentation broth). As for the ternary andquaternary systems with varying acetone concentration (Figs. 5a,6a, 7a, 8a, and 9a ), the butanol concentrations were both kept at1.2 wt%, and the ethanol concentration in the quaternary systemwere kept at 0.2 wt%. As for the ternary and quaternary systemswith varying ethanol concentration (Figs. 5b, 6b, 7b, 8b, and 9b),the butanol concentrations were still both kept at 1.2 wt%, and theacetone concentration in the quaternary system were kept at0.6 wt%.

The effect of acetone or ethanol concentration in themulticomponent systems on the flux and permeance of acetoneor ethanol, water flux and separation factor are shown in Figs. 5–7,respectively. As shown in Fig. 5, as similar as in the binary systems,with increasing the feed concentration the acetone or ethanol fluxin the ternary and quaternary systems increased linearly while theacetone or ethanol permeance in the ternary and quaternarysystems decreased linearly. The flux and permeance decreased alittle when compared with that in the binary system at the sameconcentration. The least square method was also applied for fittingthe flux of acetone and ethanol with their concentration variation,getting similar results in Fig. 3a–c all of them are listed in Table 3. ais defined as the systemic error of PV test, because when the feedconcentration c is 0, the flux J should be 0; k represents the extentof the flux’s change resulted from the variation of feedconcentration. It is suggested that the slope k in the multicompo-nent systems were lower than k in the binary system, indicatinglower extent of the flux’s change resulted from the variation of feedconcentration. One could speculate that the transport resistancesof acetone or ethanol were higher in their multicomponentsystems. It is supposed that the existence of butanol maybe havecompetitive adsorption and diffusion in the membrane, resultingin the increase of transport resistance and flux decline of acetoneor ethanol.

According to the polarity parameters listed in Table 2, it couldbe inferred that the order of polarity: water > acetone > ethanol >butanol and the molecular interaction: acetone–water >acetone–ethanol > ethanol–water > butanol–water. Thus, in thequaternary system, the interaction between acetone and ethanolwould impede the ethanol transport, resulting in higher transportresistance of ethanol, as shown in Fig. 5b. Since the molecularinteraction of acetone–ethanol is smaller than acetone–water, thereplacement of water with ethanol could promote the acetonetransport, thus as shown in Fig. 5a, the transport resistance of

Fig. 4. Effect of feed concentration on permeability and selectivity.

G. Liu et al. / Chemical Engineering and Processing 86 (2014) 162–172 167

acetone in the quaternary system (A/B/E/W) could be a little lowerthan that in the ternary system (A/B/W).

As displayed in Fig. 6, the water flux in the binary, ternary andquaternary systems had little difference, and nearly kept stablewith the increase of acetone or ethanol concentration. The mainreason is that the low ABE content in the feed was not enough tohave impact on the water transport. Fig. 7 shows the effect of feedconcentration on the separation factor of acetone and ethanol indifferent systems. Overall, the separation factors decline with thefeed concentration in all the systems. It is mainly attributed to thereduction of molecular sieving selectivity resulted from the higherswelling degree under higher feed concentration. Because the fluxof acetone or ethanol in the multicomponent systems was lowerthan that in the binary ones, the separation factor of acetone orethanol showed a similar decline like the variation of flux when thefeed system was shifted from binary to ternary and quaternaryones (as shown in Fig. 7).

In addition, the influence of acetone and ethanol on the butanoltransport was also studied by analyzing the butanol permeationdata in the binary and multicomponent systems. The results aregiven in Figs. 8 and 9. Firstly, with the concentration increase ofacetone or ethanol, the butanol flux, permeance and separationfactor declined. Secondly, compared with the PV performance in B/W system, the presence of acetone or ethanol had an overallpositive effect on butanol flux and permeance but negative effecton butanol separation factor.

The polarity parameters listed in Table 2 reveals that, therewere strong molecular interactions of acetone–water and ethanol–water. It would lead to more hydrogen bonds formation betweenacetone and water or ethanol and water. The formed acetone–water and ethanol–water clusters possess larger size that wouldhinder the transport of acetone and ethanol molecules, which canbe observed in Fig. 5 (flux in ternary and quaternary systems is

lower than that in binary system). As a result, the competitivesorption and diffusion effect from the acetone and ethanol wasweakened, facilitating the transport of butanol molecules. Becauseacetone’s polarity is larger than ethanol’s, it would form strongerinteraction with water, thus its positive effect on the butanol fluximprovement was more significant. However, with increasing theacetone or ethanol concentration, more and more acetone–waterand ethanol–water clusters in the feed would also reduce thesorption sites and diffusion cavities for the butanol molecules, andthus decrease the butanol flux. It is interesting to find an oppositetrend of butanol flux variation with the feed acetone concentrationin the quaternary system. A possible explanation may be that withthe strong hydrogen bonding between acetone and water as well asbetween acetone and ethanol, more acetone in the feed couldpromote the transport of butanol molecules. Meanwhile, acetoneand ethanol is more permeable than water in the organophilicPDMS membrane, bringing about lower butanol concentration inthe permeate. Thus, as shown in Fig. 9, the separation factor ofbutanol in the multicomponent system was lower than that in thebinary mixture. Moreover, the selectivity of acetone is higher thanthat of ethanol, so there was more remarkable decrease of butanolseparation factor in the acetone-containing systems.

In summary, the selective permeation of ABE binary systemsthrough PDMS membrane was mainly determined by the intrinsicsolubility parameter and driving force of ABE molecules. Never-theless, there would be a certain degree of coupling effect in themulticomponent systems, depending on the form and strength ofmolecular interaction between the components themselves andbetween the component and the membrane. For instance, thepresent of butanol molecules and acetone–ethanol interactioncould hinder the transport of acetone and ethanol; but theacetone–water or ethanol–water interaction could attract watermolecules meanwhile liberate butanol molecules, promoting the

Fig. 5. Effect of feed concentration on the individual flux and permeance (a and c) acetone and (b and d) ethanol.

168 G. Liu et al. / Chemical Engineering and Processing 86 (2014) 162–172

butanol transport. Sometimes, the influence of competitivesorption and diffusion should be also considered in the moleculartransport.

3.4. Intensification of fermentation process by PV integration

The PV performance of PDMS/ceramic composite membrane inaqueous solution and PV-integrated fermentation process arecompared in Table 4. It is found that both the total flux and

Fig. 6. Effect of feed concentration on the w

separation factor decreased dramatically in the fermentationprocess. Compared with our previous work, the membraneperformance had no obvious improvement by reducing the ABEconcentration in the broth. This is mainly due to the membranebiofouling resulted from the adsorption of microbes on themembrane surface that would hinder the transport and decreasethe surface hydrophobicity of PDMS membrane [26]. Effectiveapproaches to prevent the biofouling include development of anti-biofouling PV membranes and pretreatment of the broth before the

ater flux (a) acetone and (b) ethanol.

Fig. 7. Effect of feed concentration on the separation factor (a) acetone and (b) ethanol.

G. Liu et al. / Chemical Engineering and Processing 86 (2014) 162–172 169

PV process. Table 4 also lists the reported performance of PVmembranes for bio-butanol recovery from ABE model solution andfermentation process. It is found that the PDMS/ceramic compositemembrane exhibited higher flux and separation factor both in theaqueous solution and fermentation-PV integrated process.

In the initial work, the PDMS/ceramic composite membranewas integrated with ABE fermentation process for in situ solventremoval [26,42]. As shown in Table 5, by using PV integrationtechnique, the fermentation time has been extended almost3 times (from 72 h of batch to 202 h of fed-batch fermentation).However, because the PV membrane with low capacity could

Fig. 8. Effect of feed concentration on the butanol flux and

hardly reduce the product inhibition effect, the glucose consump-tion rate and ABE productivity had little improvement. It is neededto match the solvent’s producing rate with removing rate. ForPV-integrated fermentation process, the most effective way isadjusting the membrane area to optimize the intensificationefficiency. Thus, in this work, the membrane area was enlargedmore than 3 times to immediately remove the produced ABEsolvent from the fermentation broth. The result suggests that theABE concentration in the fermentation broth was effectivelyreduced from 14.6 g/L to 5.3 g/L, which is harmless for microbialgrowth (i.e., the product inhibition effect was relieved).

permeance (a and c) acetone and (b and d) ethanol.

Fig. 9. Effect of feed concentration on the butanol separation factor (a) acetone and (b) ethanol.

Table 3Linear fitting results of individual flux of acetone and ethanol.

J = a + k � c Component Binary Ternary Quaternary

a Acetone 0.0502 0.0365 0.0464k 0.2746 0.2481 0.2592

a Ethanol 0.0047 0.0007 0.0049k 0.0822 0.0673 0.0472

Table 4PV performance comparison in the model solution and fermentation process.

Feed system Membrane Composition (g/L) Total flux (g/m2h) Separation factor Reference

A B E A B E

Aqueous solution PDMS/ceramic 6.0 11.6 1.8 1211 34.6 20.0 6.5 This workPERVAP1060 10 10 – 340 14.5 18.9 – [53]PEBA 6.4 19 6.7 34 5.1 12.4 3.5 [30]

Fermentation-PV integrated process PDMS/ceramic 1.5 3.5 0.3 661 14.9 13.8 5.6 This workPDMS/ceramic 3.9 9.1 1.9 524 13.9 10.8 4.3 [42]PDMS/CA 3.1 7.0 0.7 566 – 7.0 2.8 [54]

170 G. Liu et al. / Chemical Engineering and Processing 86 (2014) 162–172

Consequently, the glucose consumption rate was accelerated from0.69 g/L/h to 1.40 g/L/h, leading to a 2 times higher ABE productivity(0.410 g/L/h). It is reasonable to expect that the fermentationefficiency can be further improved by prolonging the fermentationtime or even carry out the continuous fermentation process.

4. Conclusion

PDMS/ceramic composite membrane was used for pervapo-ration separation of ABE aqueous solutions with binary, ternary

Table 5Performance of fermentation process with and without PV integration.

Fermentation mode Batch

Membrane area (cm2) 0

Fermentation time (h) 72

ABE in fermentation (g/L) 14.6

Glucose consumption rate (g/L/h) 0.69

ABE productivity (g/L/h) 0.20

ABE yield (g/g) 0.29

Reference [42]

and quaternary systems. It was revealed that ABE separationfactor was closely related to the intrinsic solubility parameterand driving force. Coupling effect in the ABE multicomponentsystem was dependent on the molecular interactions betweencomponents themselves and between component andmembrane. Membrane biofouling was the main cause for PVperformance decline in the fermentation process. The ABEproductivity can be improved by 2 times via matching the rateof fermentation and in situ removal. Our work demonstratedthat the PV-integrated ABE fermentation intensification process

PV-integrated fed-batch

49 163202 12014.9 5.30.76 1.400.21 0.4100.28 0.292[42] This work

G. Liu et al. / Chemical Engineering and Processing 86 (2014) 162–172 171

can be expected to be a promising technology for the bio-butanol production.

Acknowledgments

This work was supported by the National Natural ScienceFoundation of China (Nos. 21406107, 21206071), InnovativeResearch Team Program by the Ministry of Education of China(No. IRT13070), Nature Science Foundation of Jiangsu Province(Nos. BK2014044011, BK2012423) and the Project of PriorityAcademic Program Development of Jiangsu Higher EducationInstitutions (PAPD).

References

[1] X.S. Feng, R.Y.M. Huang, Liquid separation by membrane pervaporation: areview, Ind. Eng. Chem. Res. 36 (1997) 1048–1066.

[2] L.M. Vane, A review of pervaporation for product recovery from biomassfermentation processes, J. Chem. Technol. Biotechnol. 80 (2005) 603–629.

[3] G. Liu, W. Wei, W. Jin, Pervaporation membranes for biobutanol production,ACS Sustain. Chem. Eng. 2 (2014) 546–560.

[4] G. Liu, F. Xiangli, W. Wei, S. Liu, W. Jin, Improved performance of PDMS/ceramiccomposite pervaporation membranes by ZSM-5 homogeneously dispersed inPDMS via a surface graft/coating approach, Chem. Eng. J. 174 (2011) 495–503.

[5] F. Xiangli, Y. Chen, W. Jin, N. Xu, Polydimethylsiloxane (PDMS)/ceramiccomposite membrane with high flux for pervaporation of ethanol–watermixtures, Ind. Eng. Chem. Res. 46 (2007) 2224–2230.

[6] M. Peng, L.M. Vane, S.X. Liu, Recent advances in VOCs removal from water bypervaporation, J. Hazard. Mater. 98 (2003) 69–90.

[7] A. Hasano�glu, Y. Salt, S. Keleşer, S. Özkan, S. Dinçer, Pervaporation separation oforganics from multicomponent aqueous mixtures, Chem. Eng. Process.:Process Intensif. 46 (2007) 300–306.

[8] B. Bolto, M. Hoang, Z. Xie, A review of membrane selection for the dehydrationof aqueous ethanol by pervaporation, Chem. Eng. Process.: Process Intensif. 50(2011) 227–235.

[9] W.F. Guo, T.-S. Chung, T. Matsuura, Pervaporation study on the dehydration ofaqueous butanol solutions: a comparison of flux vs. permeance separationfactor vs. selectivity, J. Membr. Sci. 245 (2004) 199–210.

[10] Y.K. Ong, H. Wang, T.-S. Chung, A prospective study on the application ofthermally rearranged acetate-containing polyimide membranes in dehydra-tion of biofuels via pervaporation, Chem. Eng. Sci. 79 (2012) 41–53.

[11] G.M. Shi, T. Yang, T.S. Chung, Polybenzimidazole (PBI)/zeolitic imidazolateframeworks (ZIF-8) mixed matrix membranes for pervaporation dehydrationof alcohols, J. Membr. Sci. 415–416 (2012) 577–586.

[12] M. Peivasti, A. Madandar, T. Mohammadi, Effect of operating conditions onpervaporation of methanol/methyl tert-butyl ether mixtures, Chem. Eng.Process.: Process Intensif. 47 (2008) 1069–1074.

[13] K.W. Böddeker, G. Bengtson, H. Pingel, Pervaporation of isomeric butanols, J.Membr. Sci. 54 (1990) 1–12.

[14] A. Hasanoglu, Y. Salt, S. Keleser, S. Ozkan, S. Dincer, Pervaporation separation oforganics from multicomponent aqueous mixtures, Chem. Eng. Process. 46(2007) 300–306.

[15] S. Steinigeweg, J. Gmehling, Transesterification processes by combination ofreactive distillation and pervaporation, Chem. Eng. Process.: Process Intensif.43 (2004) 447–456.

[16] B.D. Lv, G.P. Liu, X.L. Dong, W. Wei, W.Q. Jin, Novel reactive distillation–pervaporation coupled process for ethyl acetate production with waterremoval from reboiler and acetic acid recycle, Ind. Eng. Chem. Res. 51 (2012)8079–8086.

[17] W.X. Li, X.J. Zhang, W.H. Xing, W.Q. Jin, N.P. Xu, Hydrolysis of ethyl lactatecoupled by vapor permeation using polydimethylsiloxane/ceramic compositemembrane, Ind. Eng. Chem. Res. 49 (2010) 11244–11249.

[18] G. Stephanopoulos, Challenges in engineering microbes for biofuels produc-tion, Science 315 (2007) 801–804.

[19] N. Qureshi, T.C. Ezeji, Butanol, ‘a superior biofuel’ production from agriculturalresidues (renewable biomass): recent progress in technology, BiofuelsBioprod. Bior. 2 (2008) 319–330.

[20] V. Garcia, J. Pakkila, H. Ojamo, E. Muurinen, R.L. Keiski, Challenges inbiobutanol production: how to improve the efficiency? Renew. Sustain. EnergyRev. 15 (2011) 964–980.

[21] B.M. Ennis, N. Qureshi, I.S. Maddox, In-line toxic product removal duringsolvent production by continuous fermentation using immobilized Clostridiumacetobutylicum, Enzyme Microb. Technol. 9 (1987) 672–675.

[22] N.G. Grobben, G. Eggink, F. Petrus Cuperus, H.J. Huizing, Production ofacetone, butanol and ethanol (ABE) from potato wastes: fermentation withintegrated membrane extraction, Appl. Microbiol. Biotechnol. 39 (1993)494–498.

[23] T.C. Ezeji, N. Qureshi, H.P. Blaschek, Acetone butanol ethanol (ABE) productionfrom concentrated substrate: reduction in substrate inhibition by fed-batchtechnique and product inhibition by gas stripping, Appl. Microbiol. Biotechnol.63 (2004) 653–658.

[24] F. Lipnizki, S. Hausmanns, P.-K. Ten, R.W. Field, G. Laufenberg, Organophilicpervaporation: prospects and performance, Chem. Eng. J. 73 (1999) 113–129.

[25] G.Liu,D.Hou,W.Wei,F.Xiangli,W.Jin,Pervaporationseparationofbutanol–watermixtures using polydimethylsiloxane/ceramic composite membrane, Chin. J.Chem. Eng. 19 (2011) 40–44.

[26] G. Liu, W. Wei, H. Wu, X. Dong, M. Jiang, W. Jin, Pervaporation performance ofPDMS/ceramic composite membrane in acetone–butanol–ethanol (ABE)fermentation-PV coupled process, J. Membr. Sci. 373 (2011) 121–129.

[27] Z. Dong, G. Liu, S. Liu, Z. Liu, W. Jin, High performance ceramic hollow fibersupported PDMS composite pervaporation membrane for bio-butanolrecovery, J. Membr. Sci. 450 (2014) 38–47.

[28] S. Claes, P. Vandezande, S. Mullens, K. De Sitter, R. Peeters, M.K. Van Bael,Preparation and benchmarking of thin film supported PTMSP-silica pervapo-ration membranes, J. Membr. Sci. 389 (2012) 265–271.

[29] S. Liu, G. Liu, X. Zhao, W. Jin, Hydrophobic-ZIF-71 filled PEBA mixed matrixmembranes for recovery of biobutanol via pervaporation, J. Membr. Sci. 446(2013) 181–188.

[30] F.F. Liu, L. Liu, X.S. Feng, Separation of acetone–butanol–ethanol (ABE) fromdilute aqueous solutions by pervaporation, Sep. Purif. Technol. 42 (2005) 273–282.

[31] A. Friedl, N. Qureshi, I.S. Maddox, Continuous acetone–butanol–ethanol (ABE)fermentation using immobilized cells of Clostridium acetobutylicum in apacked bed reactor and integration with product removal by pervaporation,Biotechnol. Bioeng. 38 (1991) 518–527.

[32] H.R. Cascon, S.K. Choudhari, 1-Butanol pervaporation performance andintrinsic stability of phosphonium and ammonium ionic liquid-basedsupported liquid membranes, J. Membr. Sci. 429 (2013) 214–224.

[33] J.A. Stoeger, J. Choi, M. Tsapatsis, Rapid thermal processing and separationperformance of columnar MFI membranes on porous stainless steel tubes,Energy Environ. Sci. 4 (2011) 3479–3486.

[34] X.L. Liu, Y.S. Li, Y. Liu, G.Q. Zhu, J. Liu, W.S. Yang, Capillary supported ultrathinhomogeneous silicalite-poly(dimethylsiloxane) nanocomposite membranefor bio-butanol recovery, J. Membr. Sci. 369 (2011) 228–232.

[35] N. Qureshi, M.M. Meagher, J. Huang, R.W. Hutkins, Acetone–butanol–ethanol(ABE) recovery by pervaporation using silicalite–silicone composite mem-brane from fed-batch reactor of Clostridium acetobutylicum, J. Membr. Sci. 187(2001) 93–102.

[36] H.L. Zhou, Y. Su, X.R. Chen, S.L. Yi, Y.H. Wan, Modification of silicalite-1 byvinyltrimethoxysilane (VTMS) and preparation of silicalite-1 filled polydime-thylsiloxane (PDMS) hybrid pervaporation membranes, Sep. Purif. Technol. 75(2010) 286–294.

[37] N. Qureshi, H.P. Blaschek, Butanol recovery from model solution/fermentationbroth by pervaporation: evaluation of membrane performance, BiomassBioenergy 17 (1999) 175–184.

[38] F. Xiangli, W. Wei, Y. Chen, W. Jin, N. Xu, Optimization of preparationconditions for polydimethylsiloxane (PDMS)/ceramic composite pervapora-tion membranes using response surface methodology, J. Membr. Sci. 311(2008) 23–33.

[39] W. Wei, S. Xia, G. Liu, X. Gu, W. Jin, N. Xu, Interfacial adhesion between polymerseparation layer and ceramic support for composite membrane, AIChE J. 56(2010) 1584–1592.

[40] W. Wei, S. Xia, G. Liu, X. Dong, W. Jin, N. Xu, Effects of polydimethylsiloxane(PDMS) molecular weight on performance of PDMS/ceramic compositemembranes, J. Membr. Sci. 375 (2011) 334–344.

[41] G. Liu, W. Wei, W. Jin, N. Xu, Polymer/ceramic composite membranes and theirapplication in pervaporation process, Chin. J. Chem. Eng. 20 (2012) 62–70.

[42] H. Wu, X.P. Chen, G.P. Liu, M. Jiang, T. Guo, W.Q. Jin, P. Wei, D.W. Zhu, Acetone–butanol–ethanol (ABE) fermentation using Clostridium acetobutylicumXY16 and in situ recovery by PDMS/ceramic composite membrane, Bioproc.Biosyst. Eng. 35 (2012) 1057–1065.

[43] E. Drioli, S. Zhang, A. Basile, On the coupling effect in pervaporation, J. Membr.Sci. 81 (1993) 43–55.

[44] S. Tan, L. Li, Z. Xiao, Y. Wu, Z. Zhang, Pervaporation of alcoholic beverages—thecoupling effects between ethanol and aroma compounds, J. Membr. Sci. 264(2005) 129–136.

[45] S. Ray, S.K. Ray, Dehydration of tetrahydrofuran (THF) by pervaporation usingcrosslinked copolymer membranes, Chem. Eng. Process.: Process Intensif. 47(2008) 1620–1630.

[46] R.H. Perry, Chemical Engineer’s Handbook, McGraw Hill, New York, 2001.[47] P.J. Flory, Principles of Polymer Chemistry, Cornell University Press, New York,

1953.[48] J.P.G. Villaluenga, M. Khayet, P. Godino, B. Seoane, J.I. Mengual, Pervaporation

of toluene/alcohol mixtures through a coextruded linear low-densitypolyethylene membrane, Ind. Eng. Chem. Res. 42 (2002) 386–391.

[49] S. Mandal, V.G. Pangarkar, Separation of methanol–benzene and methanol–toluene mixtures by pervaporation: effects of thermodynamics and structuralphenomenon, J. Membr. Sci. 201 (2002) 175–190.

[50] M. García, M.T. Sanz, S. Beltrán, Separation by pervaporation of ethanol fromaqueous solutions and effect of other components present in fermentationbroths, J. Chem. Technol. Biotechnol. 84 (2009) 1873–1882.

[51] W.-d. Zhang, W. Sun, J. Yang, Z.-q. Ren, The study on pervaporation behaviors ofdilute organic solution through PDMS/PTFE composite membrane, Appl.Biochem. Biotechnol. 160 (2010) 156–167.

[52] N.L. Le, Y. Wang, T.-S. Chung, Pebax/POSS mixed matrix membranes for ethanolrecovery from aqueous solutions via pervaporation, J. Membr. Sci. 379 (2011)174–183.

172 G. Liu et al. / Chemical Engineering and Processing 86 (2014) 162–172

[53] A. Jonquières, A. Fane, Filled and unfilled composite GFT PDMS membranes forthe recovery of butanols from dilute aqueous solutions: influence of alcoholpolarity, J. Membr. Sci. 125 (1997) 245–255.

[54] C. Chen, Z. Xiao, X. Tang, H. Cui, J. Zhang, W. Li, C. Ying, Acetone–butanol–ethanolfermentation in a continuous and closed-circulating fermentation system withPDMS membrane bioreactor, Bioresour. Technol. 128 (2013) 246–251.