Mechanical properties and interfacial adhesion of compositemembranes probed by in-situ nano-indentation/scratch technique

Yingting Hang, Gongping Liu n, Kang Huang, Wanqin Jin n

State Key Laboratory of Materials-Oriented Chemical Engineering, College of Chemistry and Chemical Engineering, Nanjing Tech University (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 28 February 2015Received in revised form9 May 2015Accepted 13 May 2015Available online 19 June 2015

Keywords:Mechanical propertyInterfacial adhesionComposite membranesNano-indentationNano-scratch

a b s t r a c t

The mechanical strength and interfacial behavior have been of great significance for compositemembranes. In this work, an in-situ nano-indentation/scratch technique was employed to probemechanical properties and interfacial adhesion of ceramic-supported polydimethylsiloxane (PDMS)composite membranes. The nanostructures of active, transition and support layers were systematicallycorrelated with measured mechanical and adhesive properties, as well as interfacial morphologies andseparation performance. It was found that the PDMS layer's mechanical strength was reinforced by theceramic substrate and its elastic recovery could be promoted with the existence of PDMS-ceramictransition layer. The interfacial adhesion between PDMS layer and ceramic support can be remarkablyenhanced with a critical load from 10 mN to 450 mN, by increasing the PDMS thickness within 3–14 μm. During long-term test in continuous practical separation process, the optimized PDMS compositemembrane exhibited excellent and stable performance and interfacial adhesion: total flux of 1244 g/m2 h, butanol separation factor of 29.9 and critical load of 37 mN.

& 2015 Elsevier B.V. All rights reserved.

1. Introduction

As an advanced separation technology, membrane separationhas been widely used in various applications in the past decades[1,2]. Membrane is regarded as the core of membrane technology,which can be generally classified into symmetric and asymmetricmembrane according to its cross-sectional structure. For practicalapplication, asymmetric membranes are always employed owing toits high permeate flux meanwhile maintaining sufficient mechan-ical strength. Until now, the majority of asymmetric membranes arecomposite membranes that consist of an active (separation) layerdeposited on top of porous support layer. An ideal membranedemands for high permeate flux and selectivity, as well as goodstability. Therefore, on the one hand, the selective separation layeris as thin as possible to achieve higher flux. On the other hand, eachlayers and interfacial properties are optimized to obtain a robustcomposite membrane with long-term stability.

Mechanical strength and interfacial adhesion properties playimportant roles in engineering of composite membranes, becausetheir reliability and stability would be limited by wear and staticfriction forces that occurred in practical application [3]. In addition

to minimized wear in contact, it is also important to take intoconsideration the swelling of active layer in membrane processes(e.g., pervaporation [4], nanofiltration [5]). Great efforts have beenmade to improve the structural stability of composite membraneby means of controlling intrinsic physico-chemical properties ofactive layer [6] and support layer [7], as well as interfacialadhesion between these two layers [8].Our previous work [9–11]demonstrated that rigid porous ceramic support offers greatpotential to improve mechanical property of composite mem-branes. Moreover, a variety of methods have been reported toimprove adhesion between separation layer and support, includ-ing surface modification [12–17] and introduction of a transitionlayer [8,18]. In addition, studies on mechanical strength of com-posite films suggested that the toughness and adhesion of activelayer is closely related to its thickness [19–22].

Although the importance of structural stability of compositemembranes has beenwell recognized, it is still a challenge to directlymeasure the mechanical properties and interfacial adhesion force ofactive layer on support. In the field of membrane technologies, themechanical properties of composite membranes were commonlystudied by tensile test [22–24]. This technique requires that the sizeof sample is standard enough to be clamped rigidly by the sampleholder without sliding. The most frequently-used method to estimatethe membrane interfacial adhesion is peeling test [25], which needsrelatively tedious sample preparation procedure and is difficult to be

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/memsci

Journal of Membrane Science

http://dx.doi.org/10.1016/j.memsci.2015.05.0760376-7388/& 2015 Elsevier B.V. All rights reserved.

n Corresponding authors. Tel.: þ86 25 83172266; fax: þ86 25 83172292.E-mail addresses: gpliu@njtech.edu.cn (G. Liu), wqjin@njtech.edu.cn (W. Jin).

Journal of Membrane Science 494 (2015) 205–215

applicable for membranes with different geometrical configurations(e.g., tubular, hollow fiber) except flat one. These drawbacks havesignificantly limited its applicability and effectiveness. Alternatively,nano-indentation test uses electromagnetic force application andcapacitive depth measurement to measure elastic and plastic proper-ties of materials on nano-scale [26,27]. And as an accessory of nano-indentation, nano-scratch test moves sample perpendicular to thescratch probe to determine interfacial adhesion strength throughfilm delamination caused by the accumulation of interfacial shearstress [28–29]. Compared with the tensile test and peeling test, thenano-indentation/scratch technique provides a simple, accuracy,versatile and rapid means to assess the mechanical strength andinterfacial adhesion of small-volume samples [30–33]. On the basis ofnano-indentation/scratch technique, we have developed an efficientcharacterization method to measure interfacial adhesion of variouscomposite membranes [9–12] used for pervaporation and gasseparation.

As benchmark organophilic membranes, polydimethylsiloxane(PDMS)-based membranes have been intensively investigated formolecular separations, including organic compounds removal [34],organic mixtures separation [35,36] and solvent resistant nanofil-tration [37]. Our group proposed a type of PDMS compositemembranes using macroporous ceramic as support layer. SuchPDMS/ceramic composite membranes exhibited high flux andgood selectivity for bio-fuels recovery [38], offering an excitingopportunity for industrial application. Recent studies revealed thatinevitable swelling effect induced by solvent is a great challengefor structural stability of organophilic membranes applied inseparation process [4,5]. So probing and controlling the structuralstability is indispensable for implementation of PDMS compositemembranes.

In this work, the mechanical and interfacial adhesion proper-ties of composite membranes were probed by in-situ nano-indentation/scratch technique. We demonstrated the probingapproach and structural optimization of composite membraneswith the example of PDMS active layer and curved tubular andhollow fiber ceramic support layer. Nanostructures of the active,transition and support layers were finely tuned to achieve thecomposite membranes with simultaneously high structural stabi-lity and separation performance. Moreover, long-term stability ofthe prepared PDMS composite membranes was evaluated bycontinuous operation in bio-butanol recovery process.

2. Experiment

2.1. Membrane preparation

PDMS (α, ω-dihydroxypolydimethylsiloxane, Mw¼60,000) waspurchased from Shanghai Resin Factory Co., Ltd., China. Tetraethy-lorthosilicate (TEOS), n-heptane, dibutyltin dilaurate were pur-chased from Sinopharm Chemical Reagent Co., Ltd, China withoutfurther purification. The tubular asymmetric ceramic substrateswith average pore size of 200 nm were supplied by MembraneScience and Technology Research Center, Nanjing Tech University,China. The ceramic tube consists of a 20 μm-thick ZrO2 top layerand an Al2O3 supporting layer, with the porosity of 30%. Theirlength, outer diameter and inner diameter were 68 mm, 12 mmand 8 mm, respectively. The ceramic hollow fiber substrates werefabricated in our lab [39] with an average pore size of 200 nm, andlength, outer diameter and inner diameter of 68 mm, 2 mm and1 mm, respectively. The Al2O3 hollow fiber exhibits an asymmetricstructure consisting of a sponge pore layer on top of figure porelayer, with the porosity of 45%. The PDMS polymer was dissolvedin n-heptane uniformly in a Teflon bottle, and then TEOS anddibutyltin dilaurate were added into the solution. n-heptane, TEOSand dibutyltin dilaurate were used as solvent, cross-linking agentand catalyst, respectively. The preparation of PDMS/ceramic com-posite membranes followed with the dip-coating method intro-duced in our previous work [39,40]. The outer surface of theceramic support was immersed in polymer solution for 60 s. Afterdrying at room temperature for 24 h, the composite membraneswere heat treated at 120 1C for 12 h.

2.2. Membrane characterization

2.2.1. Nano-indentation/scratch testThe Nano-Test (NanoTestTM, Micro Materials, United Kingdom)

system used in this work is capable of high resolution mechanicalproperty analysis in the Nano and Micro meter regimes. TheSchematic diagram of a Nano-Test experiment is shown in Fig. 1.The basic Nano-Test Vantage platform consists of the instrumentframe with a low and/or high load head, the electronic controlunit, the motor control box and a PC loaded with the Nano-TestPlatform software.

Hardness and elastic modulus were measured using a nano-indenter as shown in Fig. 2a. The indentation test was carried outby using continuous stiffness measurement (CSM) techniquewhich monitors and records the dynamic load and displacementof three-sided pyramidal diamond (Berkovich). The nano-indentation experiment was carried in the load-controlled mode,with load of the indentation controlled at around 0.5 mN for allcases. Other experimental conditions were: preset initial load0.03 mN, loading and unloading in 20 s, 20 s holding at peak load,60 s holding period at 90% unload for thermal drift correction. 30points were applied on each composite membrane.

The schematic representation of the nano-indentation techni-que is demonstrated in Fig. 2b. Oliver and Pharr method [41,42]was to be used to determine elastic modulus (E) and hardness (H)values from an analysis of indentation load–displacement data.The elastic modulus (E) and the hardness (H) are calculated fromseveral parameters, including maximum loading (P), displacement(h) and contact stiffness (S), which can be measured from theindentation load–displacement curve and unloading curves. Hard-ness, which is typically defined as the mean pressure under thenano-indenter, can be derived from the following equation:

H¼ Pmax

ACð1Þ

Fig. 1. Schematic diagram of a nano-test experiment. (a) a PC loaded with theNanoTest Platform software; (b) NanoTest NTX controller; (c) sample stagecontroller; (d) low load head; (e) low load magnet; (f) Low Load pendulum;(g) an indenter; (h) balance mass rod; (i) damping plate; (j) membrane sample;(k) sample mount; (l) directional control stage; (m) base board.

Y. Hang et al. / Journal of Membrane Science 494 (2015) 205–215206

where Pmax (mN) is the maximum applied force and obtaineddirectly from the load–displacement curve, and Ac (nm2) is theprojected contact area of the indenter tip with the material. Theinitial slope of the unloading curve (dp/dh) can be used for theindentation modulus (Mr) of the material, the following expres-sions are formulated:

S¼ dPdh

¼ β2ffiffiffiπ

p Mr

ffiffiffiffiffiffiAC

pð2Þ

where S is the contact stiffness, dP/dh is the initial slope of theunloading curve, P is the applied force to the indenter, β is acorrection factor that depends on the geometry of the nano-indenter, and Mr (GPa) is the reduced modulus of the nano-indenter. Therefore, the Es of the material can be determined by

1Mr

¼ 1�ν2sEs

þ1�ν2iEi

ð3Þ

where Es (GPa) and νs are Young's modulus and Poisson's ratio forthe specimen, respectively, and Ei (1141 GPa) and νi (0.07) are thesame parameters for the indenter. As the indenter is a rigidmaterial, whose Young's modulus goes to infinity, the Es of thematerial is very close to Mr, the reduced modulus then can beexpressed by

1Mr

¼ 1�ν2sEs

ð4Þ

Nano-scratch tests monitors and records the scratch distancerelationships with the dynamic load, penetration depth, frictionforce and corresponding friction coefficient. Scratch tests must becarried out by moving the sample in Z negative to a friction data. A

conical indenter was used for scratch with 601 angle, 5 μm radiusand 20 μm polished depth. The scratch data were obtained bytaking as multi-pass tests: (i) pre-scratch topographic scan, (ii) adistance followed by a ramped scratch and (iii) a post-scratch lowload scan. In the topographic scan, the applied load was constantat 0.05 mN (no wear occurs at this load). Surface roughness wasmeasured from this scan. In the second (scratch) scan, the appliedload was constant at 0.05 mN between 0 and 20 mm and then wasramped to the maximum load 70 mN at the end of the scan(500 mm) with the ramping rate about 2.5 mN/s. In the final scanthe resultant topography was observed by using a low applied loadof 0.05 mN. Three repeat tests were performed on each compositemembrane. During analysis, the data shown has been corrected fortopography, sample slope and instrument compliance in theinstrument software, allowing the real depth data to be displayedunder the scratch load [19].

In the experiment, an initial surface profile of the membranetested before scratching was detected by pre-scanning the samplesurface The surface roughness of membrane can be obtained fromthis step. During scratching, the surface profile could be sensedand recorded by the depth-sensing system. After ramped scratch-ing, the surface profile of the membrane was once again deter-mined by post-scratching. Then the elastic recovery afterscratching could be obtained. Therefore, in the scratch depth–displacement profile, the on-load depth and residual depth aredescribed for the ramped scratch of test (ii) and post-scratch oftest (iii), respectively. The on-load depth can be considered as thetotal amount of deformation while the residual depth in post-scratch can be regarded as the amount of plastic deformation, andthe difference between them can be regarded as the amount of

Fig. 2. (a) Detailed schematic diagram of a nano-indenter on a composite membrane; (b) typical profiles of nano-indentation test: force vs. depth curve, where hf is the finalpenetration depth; hmax is the maximum depth and Pmax is the maximum force applied to indenter; (c–d) typical profiles of nano-scratch test: (c) scratch depth–displacement curve and (d) scratch load–displacement curve.

Y. Hang et al. / Journal of Membrane Science 494 (2015) 205–215 207

elastic deformation [43]. The adhesion strength of active layer tosupport can be evaluated by the critical load at failure, which isobtained by noting the obvious discontinuity in the scratch depth–displacement curve (Fig. 2c–d) when a film peels from a support.Furthermore, the interfacial failure was confirmed by characteriz-ing the scratch morphology of sample. The surface and cross-section morphology of composite membranes were characterizedby field emission scanning electron microscopy (FESEM, Hitachi-4800, Japan). The sample was coated with gold powder undervacuum before SEM observation was performed.

2.2.2. Separation experimentPervaporation separation process was conducted on a home-

made apparatus [40]. The feed solution was maintained at 40 1Cand vigorous stirring and high flow rate on membrane surfacewere applied in order to minimize the concentration and tem-perature polarization. During experiments, solvents were added tomaintain a constant feed concentration. The permeate vapor wascollected in liquid nitrogen trap. The pressure at permeate sidewas below 400 Pa during collections. The concentrations of n-butanol, acetone and ethanol in feed and penetration side weremeasured by a gas chromatography (GC-2014, SHIMADZU, Japan)with a thermal conductivity detector (FID).

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

J ¼MAt

ð5Þ

α¼ Yi 1�Yið ÞXi 1�Xið Þ ð6Þ

whereM is the weight of the permeate (g), A is the effective area ofthe membrane (m2), and t is the permeation time interval for thepervaporation (h); Yi and Xi are the weight fractions of compo-nents in the permeate and feed side, respectively.

3. Results and discussion

It is well-known that integrity and transport resistance of theseparation layer related closely with it thickness. It has shown thatan appropriate thickness can also result in high mechanical

properties and well adhesion strength [19,20]. Thus, wesystematically studied the mechanical, interfacial and separationproperties of the PDMS composite membranes with various activelayer thicknesses. Moreover, porous ceramic substrate wasemployed as the support layer, because its excellent stiffnesswould be beneficial to the mechanical properties of the PDMSmembranes. Tubular and hollow fiber ceramic supports were usedto prepare the composite membranes in order to meet therequirement of practical application.

3.1. Morphologies of PDMS composite membranes

PDMS layer with various thicknesses were obtained by con-trolling the coating viscosity of PDMS solution [39]. The solutionviscosity was controlled by cross-linking time. With time, theviscosity of solution will increase due to the cross-linking reaction.Generally, higher coating viscosity leads to thicker active layerwhile thinner transition layer. In our case, the PDMS layer thick-ness of the composite membranes supported on ceramic tubeswas varied from 370.5 mm, 670.8 mm, 970.6 mm, 1171.1 mm, to1471.3 mm, and that of the composite membranes coated onceramic hollow fibers ranged from 3.570.2 mm, 570.7 mm,771 mm, 970.5 mm, to 1271.7 mm. Fig. 3 shows typical cross-sectional SEM images of PDMS composite membranes withdifferent active layer thickness. It can be observed that the PDMSlayers were coated uniformly on the surface of porous ceramicsupports. With the increase of PDMS thickness, the penetration ofPDMS into the support pores is restrained, owing to the effect ofcoating viscosity.

3.2. Mechanical properties of PDMS layers

The mechanical properties of the PDMS membrane supportedon the ceramic supports were measured by depth sensing nano-indentation test, and the load–depth curves are shown in Fig. 4aand b. It can be observed that the indentation depth in PDMS layerincreased with its thickness. According to Oliver and Pharr methodand Eqs. 1–4, the hardness and elastic modulus were calculated foreach sample. As displayed in Fig. 4c and d, both the hardness andreduced modulus of the PDMS/ceramic composite membranesreduced with increasing PDMS layer thickness. Theoretically, the

Fig. 3. Typical SEM images of the cross section of PDMS composite membranes supported on (a–c) ceramic tube and (d–f) hollow fiber with different PDMS layer thickness.

Y. Hang et al. / Journal of Membrane Science 494 (2015) 205–215208

hardness and elastic modulus belong to an intrinsic property ofmaterials. The unusual thickness-dependent mechanical proper-ties found in our composite membranes can be attributed to theeffect of support layer. The rigid ceramic layer could significantlyenhance the mechanical strength of PDMS layer deposited on itssurface. Moreover, the reinforced function of ceramic layer wouldbe progressively weakened as PDMS layer becomes thicker. Inaddition, the hardness and reduced modulus of thin PDMS layersupported on ceramic tube were slight higher than ceramic hollowfiber. This might be due to the higher mechanical strength ofceramic tube than that of ceramic hollow fiber. Likewise, as thePDMS layer thickness is more than 9 μm, the effect of differentsubstrates is negligible.

Besides targeting specific unique sites, it is also useful to mapthe distribution of hardness and modulus across a large membranearea, so as to check the uniformity of PDMS coating. The distribu-tions of hardness and reduced modules of PDMS/ceramic compo-site membranes are shown in Fig. 5. It can be found that themechanical data along radial and axial directions of the mem-branes distribute uniformly. Standard deviations of the hardnessand reduced modules for tubular and hollow fiber PDMS compo-site membranes are 0.34 MPa, 0.60 MPa, 1.59 MPa, and 0.97 MPa,respectively. The results suggest the formation of integrated andcontinuous PDMS layer on the surface of ceramic support.

3.3. Interfacial adhesion of PDMS composite membranes

The interfacial adhesion of PDMS layer onto the ceramicsupport was measured by the nano-scratch test. Fig. 6 gives a

typical scratch result of PDMS/ceramic composite membraneto demonstrate the determining of critical load (Lc) at thePDMS-ceramic interface. It is possible to observe two cleartransition points in scratch profile (Fig. 6a) and friction profile(Fig. 6b):

(i) At 121 mm scan displacement for the onset of edge cracking(Lc1¼15.9 mN).

(ii) At 337 mm scan displacement for the total film failure(Lc2¼50.5 mN).

These transitions can be confirmed by SEM characterization ofthe scratch morphology. In the range before 121 mm, PDMS layerwas fully recovered under elastic contact and no obvious scratchcan be found. When the displacement was over 121 mm, fluctua-tions in the on-load depth and residual depth are found, implyingthe beginning of cracking formation. The PDMS layer continued tocrack with the increase of applied load until it peeled from theceramic support indicated by the onset of larger fluctuations in thescratch depth at displacement of 337 mm. It can be seen in the SEMimages (Fig. 6c) that porous ceramic support has exposed withinthe scratch track after Lc2 on the PDMS composite membrane.Thus, such critical load (Lc2) at the failure onset of PDMS layer isconsidered as the adhesive strength of the PDMS layer onto theceramic support layer, namely, the interfacial adhesion force of thePDMS/ceramic composite membrane [19]. The total film removalafter Lc2 was accompanied with broad scratch trace and thescratch damage was remarkable at the end of trace.

Fig. 4. Load–distance curves of PDMS membrane supported on (a) tubular or (b) hollow fiber ceramic substrates. and Effect of active layer thickness on the (c) hardness and(d) reduced modulus.

Y. Hang et al. / Journal of Membrane Science 494 (2015) 205–215 209

Scratch profiles of PDMS/ceramic membranes with variousPDMS layer thicknesses are shown in Fig. 7. The displacementcorresponding to the onset of film failure is found to be increasedwith the thickness of PDMS layer of the composite membranes. Itmeans that thicker PDMS layer exhibits higher critical load sincethe applied load is linearly increased with the displacement(Fig. 2c). Furthermore, there are discordances between on-loaddepth residual depth in some scratch profiles (Fig. 7a and d–f),indicating the existence of elastic recovery after scratch test inPDMS layer. The phenomena occur to the PDMS membranes withthin PDMS layer (Fig. 7a and d) and PDMS membranes supportedon hollow fiber (Fig. 7d–f). This may be owing to the influence oftransition layer between PDMS and ceramic support [44]. As theload was applied into the PDMS surface, the deformation of PDMSlayer forced more polymer chain to penetrate into the rigidceramic pores. Once the applied load was repealed from thescratch test, such penetrated polymers would provide a restoringforce to push PDMS layer to recover itself deformation. Thus, theformation of interfacial layer in composite membrane wouldbenefit for the elastic recovery. From the interfacial morphologiesdisplayed in Fig. 3, thinner PDMS layer resulting from low coatingviscosity leaded to more PDMS penetration into support pores. Sothe elastic recovery is more significant for the composite mem-branes with thinner PDMS layer.

The interfacial morphologies of PDMS membranes supportedby tubular and hollow fiber ceramic substrates were further

compared by using EDX analysis. Fig. 8 gives the Silicone (Si)element distributions across the cross-section of PDMS/ceramiccomposite membranes with PDMS layer thickness of 6 mm. EDXanalysis shows the tubular ceramic and ceramic hollow fibersupported PDMS composite membrane consists of two parts,active layer rand transition layer (as also shown in Fig. 2a). Thereis nearly no Si element in the support, indicating little PDMSpenetration in the tubular ceramic-supported 6-mm thick PDMSmembrane. In contrast, for the hollow fiber ceramic-supportedmembrane, it is clearly observed an interfacial layer betweenPDMS and ceramic, according to the Si element distribution inceramic substrate. Although the ceramic tubes and hollow fibersused here have the same average pore size, the differences inchemical groups and nanostructures on the ceramic surface (asdescribed in Section 2.1) could lead to different PDMS penetra-tions. This may account for the discordance of elastic recoverybehaviors in tubular and hollow fiber ceramic-supported PDMScomposite membranes.

Furthermore, the effect of PDMS layer thickness on the criticalloads of the PDMS composite membranes is shown in Fig. 9. It isfound that the interfacial adhesion is proportional to the activelayer thickness for the PDMS composite membranes supported onceramic substrates. There is a linear advance in the critical loadfrom 10 mN to over 50 mN, as the PDMS layer thickness increasesfrom 3 mm to 14 mm. It has been proposed that the critical load forfilm failure is a marked function of indenter radius and film

Fig. 5. Distribution of (a, c) hardness and (b, d) reduced modules of PDMS membrane supported on (a, b) tubular (indent array: 40 μm�200 μm, PDMS layer thickness:9 μm); or (c, d) hollow fiber ceramic substrates (indent array: 10 μm�200 μm, PDMS layer thickness: 12 μm).

Y. Hang et al. / Journal of Membrane Science 494 (2015) 205–215210

thickness [19]. In this work, a same spherical indenter was used forall membranes. Thus, we attribute the result found here to that theintrinsic stress of the PDMS layer is increased with its thickness,leading to demand larger critical load to break into the membrane.It should be noted that the layer thickness does not always play apositive effect on the interfacial adhesion of composite mem-branes. If active layer is too thick, it would lead to a highstress formation in the layer. Similar phenomenon was alsofound by Sheeja and coworkers [20]. As a result, thethick layer may be not able to deflect with the substrate andhence the driving force for delamination is highly increased. Toverify this, we prepared PDMS composite membranes with thickactive layer and then measured the interfacial adhesions. Asdisplayed in Table 1, the critical loads of the thick PDMS compositemembranes were far lower than that of the thin compositemembranes in Fig. 9. It is referred that high stress is existed inthe thick PDMS layer that results in a very low critical breakingstress. The thick PDMS composite membranes are expected toperform low flux which is also not desirable for practicalapplication.

3.4. Practical application of PDMS composite membranes

One of the promising applications of PDMS composite mem-branes is bio-fuels recovery from aqueous mixtures [34]. Thefermentation productivity could be effectively improved with theremoval of solvent from the broth to eliminating the productexhibition on microbial growth. The prepared PDMS compositemembranes were applied for pervaporation recovery of bio-butanol from butanol–water mixtures and acetone–butanol–etha-

nol (ABE)–water mixtures, which are the model systems of bio-butanol fermentation broth [38].

Fig. 10 displays the effect of PDMS layer thickness on the fluxand separation factor of PDMS composite membranes supportedon ceramic tubes or hollow fibers. Generally, there is a trade-offbetween total flux and separation factor with varying the thick-ness of active layer. Namely, the separation factor decreasessharply and total flux rises greatly with reducing the membranethickness. The inverse proportion between membrane flux andthickness is usually attributed to the effect of transport resistance.Membrane selectivity is the intrinsic property of membranematerial. So the selectivity of a perfect PDMS membrane istheoretically neither dependent on its thickness nor the transportresistance of support. In our work, it is interesting to find that theseparation factor of PDMS composite membrane increases gradu-ally with its thickness. Such result reveals that there are severalnon-selective defects in the PDMS separation layer that wasprogressively eliminated by increasing the PDMS thickness. Similarphenomena were also reported in literatures [35,45]. Comparedwith ceramic tubes, the hollow fiber could allow a thicker PDMSlayer to obtain higher selectivity, meanwhile maintain high fluxowing to the low transport resistance of itself. Besides of the lowtransport resistance hollow fiber also shows advantages of high-packing density and low-preparing cost. Thus, the hollow fiberceramic-supported PDMS composite membranes show greaterpotential in the practical application.

In the practical application of organophilic membranes forsolvent recovery, the solvent-induced swelling would also influ-ence the interfacial adhesion strength and stability of the compo-site membranes [4]. Thus, long-term stability of the 9 μm-thick

Fig. 6. Typical results of nano-scratch test of PDMS/ceramic composite membranes (PDMS layer thickness: 12 mm): (a) scratch profile, (b) friction profile and (c) SEM imagesof scratch morphology.

Y. Hang et al. / Journal of Membrane Science 494 (2015) 205–215 211

Fig. 7. Scratch profiles of PDMS membranes supported by (a–c) tubular or (d–f) hollow fiber ceramic substrates with various PDMS layer thicknesses: (a) 3 mm; (b) 6 mm;(c) 14 mm; (d) 3.5 mm; (e) 5 mm; (f) 12 mm.

Fig. 8. EDX line scans of Si element for the cross-section of PDMS membrane supported by (a) tubular or (b) hollow fiber ceramic substrate; PDMS layer thickness: 6 mm.

Y. Hang et al. / Journal of Membrane Science 494 (2015) 205–215212

PDMS membrane supported on ceramic hollow fiber was evalu-ated by the continuous pervaporation process in model ABE-watersolution. As shown in Fig. 11, the separation performance keepsstable over a period of 200 h operation, exhibiting a high andstable total flux of 1244 g/m2 h and separation factor of 29.9 for n-butanol, 30.2 for acetone and 6 for ethanol. Most PV processeswere generally operated at high vacuum, which might distort andeven destroy the structure of soft PDMS membranes. However, therigid ceramic support can provide excellent mechanical stability tocomposite membrane at high negative pressure.

Moreover, the interfacial adhesions of the composite mem-brane before and after the long-term process were measured usingNano-scratch test. The obtained critical loads are listed in Table 2.It can be found that the interfacial adhesion of the PDMScomposite membrane keeps stable after over a 200 h continuousoperation in real separation process. This phenomenon indicatesceramic hollow fiber can effectively suppress the excessive swel-ling during the pervaporation process. In addition, appropriatepenetration of PDMS into ceramic support pores forms a favorable

interface between support and active layer. The mechanical inter-locking of PDMS polymer chain with ceramic pores benefits for thestructural stability of PDMS composite membrane in the practicalapplication.

4. Conclusion

The mechanical and interfacial adhesion properties of ceramic-supported PDMS composite membranes were probed by in-situnano-indentation/scratch technique. The results indicated that theceramic substrates reinforced the mechanical strength and thePDMS-ceramic transition layer promoted the elastic recovery ofPDMS layer in the composite membranes. Meanwhile, by increas-ing the PDMS layer thickness within a suitable range of 3–14 μm,the adhesive force of the composite membranes can be signifi-cantly improved from 10 mN to 450 mN. By optimizing thenanostructures of active, transition and support layers, the ceramichollow fiber-supported PDMS composite membrane with PDMS

Fig. 9. Effect of active layer thickness on the critical load of PDMS compositemembranes supported on tubular or hollow fiber ceramic substrates.

Table 1Interfacial adhesion of PDMS composite membranes with thick active layer.

PDMS composite membranes PDMS layer thickness (mm) Critical load (mN)

Tubular ceramic support 20 20.5Hollow fiber ceramic support 18 17.7

Fig. 10. Effect of PDMS layer thickness on the PV performance of PDMS composite membranes supported on (a) tubular or (b) hollow fiber ceramic substrates (1 wt% n-butanol–water, 40 1C).

Fig. 11. Long-term stability of PDMS composite membrane supported on ceramichollow fiber during the continuous pervaporation process in model ABE-watersolution (0.6 wt% acetone, 1.2 wt% n-butanol, 0.2 wt% ethanol, 98 wt% water).

Table 2Interfacial adhesion of PDMS composite mem-branes before and after stability test.

PDMS compositemembranes

Critical load(mN)

Before stability test 37.6After stability test 36.2

Y. Hang et al. / Journal of Membrane Science 494 (2015) 205–215 213

thickness of 9 μm exhibited excellent separation performance(total flux: 1244 g/m2 h; butanol separation factor: 29.9) andmeanwhile highly stable interfacial adhesion (critical loadE37 mN), during 200 h continuous bio-butanol recovery process.It can be expected that the proposed nano-indentation/scratchtechniques could be a universal method for characterizing andoptimizing the mechanical and interfacial adhesion of compositemembranes.

Acknowledgments

This work was supported by the National Natural ScienceFoundation of China (Nos. 21406107, 21490585, 21476107), Inno-vative Research Team Program by the Ministry of Education ofChina (No. IRT13070), Natural Science Foundation of JiangsuProvince (No. BK20140930) and the Project of Priority AcademicProgram Development of Jiangsu Higher Education Institutions(PAPD).

Nomenclature

J mass flux (g/m2 h)M weight of the permeate (g)A effective area of hollow fiber membrane (m2)t operation time (h)Yi weight fractions of component i in permeateXi weight fractions of component i in feedH hardness (MPa)Pmax maximum applied force (mN)AC contact area of the indenter tip (nm2)S contact stiffness (N/m)dP/dH initial slope of the unloading curveMr reduced modulus of the nanoindenter (GPa)Es Young's modulus for the specimen (GPa)γs Poisson's ratio for the specimenEi Young's modulus for the indenter (1141 GPa)γi Poisson's ratio for the indenter (0.07)

Greek letters

αij selectivityβ correction factor for the indenter

References

[1] K.K. Sirkar, Membranes, phase interfaces, and separations: novel techniquesand membranes – an overview, Ind. Eng. Chem. Res. 247 (2008) 5250–5266.

[2] P. Shao, R.Y.M. Huang, Polymeric membrane pervaporation, J. Membr. Sci. 287(2007) 162–179.

[3] J.A. Williams, H.R. Le, Tribology and MEMS, J. Phys. D: Appl. Phys. 39 (2006)201–214.

[4] P. Shao, R.Y.M. Huang, X. Feng, W. Anderson, R Pal, Burns C.M. Burns,Composite membranes with an integrated skin layer: preparation, structuralcharacteristics and pervaporation performance, J. Membr. Sci. 254 (2005)1–11.

[5] E.S. Tarleton, J.P. Robinson, M. Salman, Solvent-induced swelling of mem-branes -Measurements and influence in nanofiltration, J. Membr. Sci. 280(2006) 442–451.

[6] J. Wei, X.G. Jian, C.R. Wu, S.H. Zhang, C. Yan, Influence of polymer structure onthermal stability of composite membranes, J. Membr. Sci. 256 (2005) 116–121.

[7] S.M. Liang, G.Y. Xu, Y. Jin, Z.C. Wu, Z.Q. Cai, N. Zhao, Z.L. Wu, Annealing ofsupporting layer to develop nanofiltration membrane with high thermalstability and ion selectivity, J. Membr. Sci. 476 (2015) 475482.

[8] Y.F. Li, Y.L. Su, J.Y. Li, X.T. Zhao, R.N. Zhang, X.C. Fan, J.N. Zhu, Y.Y. Ma, Y. Liu, Z.Y. Jiang, Preparation of thin film composite nanofiltration membrane with

improved structural stability through the mediation of polydopamine, J.Membr. Sci. 475 (2015) 10–19.

[9] W. Wei, S.S. Xia, G.P. Liu, X.H. Gu, W.Q. Jin, N.P. Xu, Interfacial adhesionbetween polymer separation layer and ceramic support for composite mem-brane, AIChE J. 56 (2010) 1584–1592.

[10] J. Shen, G.P. Liu, K. Huang, W.Q. Jin, K.R. Lee, N.P Xu, Membranes with fast andselective gas-Transport channels of laminar graphene oxide for efficient CO2

capture, Angew. Chem. Int. Ed. 54 (2015) 578–582.[11] K. Huang, G.P. Liu, Y.Y. Lou, Z.Y. Dong, J. Shen, W.Q. Jin, A graphene oxide

membrane with highly selective molecular separation of aqueous organicsolution, Angew. Chem. Int. Ed. 53 (2014) 6929–6932.

[12] Y.Y Lou, G.P. Liu, S.N. Liu, J. Shen, W.Q. Jin., A facile way to prepare ceramic-supported graphene oxide composite membrane via silane-graft modification,Appl. Surf. Sci. 307 (2014) 631–637.

[13] R.Y.M. Huang, R. Pal, G.Y. Moon, Pervaporation dehydration of aqueous ethanoland isopropanol mixtures through alginate/chitosan two ply compositemembranes supported by poly(vinylidene fluoride) porous membrane, J.Membr. Sci. 167 (2000) 275–289.

[14] S. Belfer, R. Fainchtain, Y. Purimson, O. Kedem, Surface characterization byFTIR-ATR spectroscopy of polyethersulfone membranes-unmodified, modifiedand protein fouled, J. Membr. Sci. 172 (2000) 113–124.

[15] W.C. Chao, S.H. Huang, Q.F. An, D.J. Liaw, Y.C. Huang, K.R. Lee, J.Y. Lai, Novelinterfacially-polymerized polyamide thin-film composite membranes: studieson characterization, pervaporation, and positron annihilation spectroscopy,Polymer 52 (2011) 2414–2421.

[16] Y.L. Liu, C.H. Yu, K.R. Lee, J.Y. Lai, Chitosan/poly(tetrafluoroethylene) compositemembranes using in pervaporation dehydration processes, J. Membr. Sci. 287(2007) 230–236.

[17] Y.L. Liu, C.H. Yu, L.C. Ma, G.C. Lin, H.A. Tsai, J.Y. Lai, The effects of surfacemodifications on preparation and pervaporation dehydration performance ofchitosan/polysulfone composite hollow-fiber membranes, J. Membr. Sci. 311(2008) 243–250.

[18] S.-Y Li, R. Srivastava, R.S. Parnas, Separation of 1-butanol by pervaporation using anovel tri-layer PDMS composite membrane, J. Membr. Sci. 363 (2010) 287–294.

[19] B.D. Beake, S.R. Goodes, B. Shi, Nanomechanical and nanotribological testing ofultra-thin carbon-based and MoST films for increased MEMS durability, J.Phys. D: Appl. Phys. 42 (2009) 065301–065308.

[20] D. Sheeja, B.K. Tay, K.W. Leong, C.H. Lee, Effect of film thickness on the stressand adhesion of diamond-like carbon coatings, Diam. Relat. Mater. 11 (2002)1643–1647.

[21] B.D. Beake, M.I. Davies, T.W. Liskiewicz, V.M. Vishnyakov, S.R. Goodes, nano-scrtach, nano-indetation and fretting tests of 5–80 nm ta-c films on Si(100),Wear 301 (2013) 575–582.

[22] M. Liu, J.R. Sun, Y. Sun, C. Bock, Q.F. Chen, Thickness-dependent mechanicalproperties of polydimethylsiloxane membranes, J. Micromech. Microeng. 19(2009) 035028–035031.

[23] P.S. Rachipudi, M.Y. Kariduraganavar, Kariduraganavar, A.A. Kittur, Kittur, A.M. Sajjan, Sajjan, Synthesis and characterization of sulfonated-poly(vinylalcohol) membranes for the pervaporation dehydration of isopropanol, J.Membr. Sci. 383 (2011) 224–234.

[24] C.Y. Lai, A. Groth, S. Gray, M. Duke, Enhanced abrasion resistant PVDF/nanoclayhollow fibre composite membranes for water treatment,, J. Membr. Sci. 449(2013) 146–157.

[25] C.H. Zhao, H. Wu, X.S. Li, F.S. Pan, Y.F. Li, J. Zhao, Z.Y Jiang, P. Zhang, X.Z. Cao, B.Y. Wang, High performance composite membranes with a polycarbophilcalcium transition layer for pervaporation dehydration of ethanol, J. Membr.Sci. 429 (2013) 409–417.

[26] H. Nili, K. Kalantar-zadeh, M. Bhaskaran, S. Sriram, In situ nanoindentation:probing nanoscale multifunctionality, Prog. Mater. Sci. 58 (2013) 1–29.

[27] L.R. Meza, S. Das, J.R. Greer, Strong, lightweight, and recoverable three-dimensional ceramic nanolattices, Science 345 (2014) 1322–1326.

[28] AC. Fischer-Cripps, Critical review of analysis and interpretation of nanoin-dentation test data, Surf. Coat. Technol. 200 (2006) 4153–4165.

[29] AC. Fischer-Cripps, Nanoindentation, Springer-Verlag, New York, 2004.[30] P.R. Chalker, S.J. Bull, D.S. Rickerby, A review of the methods for the evaluation

of coating substrate adhesion, Mater. Sci. Eng. 140 (1991) 583–592.[31] W.E. Fu, Y.Q. Chang, C.W. Chang, C.K. Yao, J.D. Liao, Mechanical properties of

ultra-thin HfO2 films studied by nano scratches tests, Thin Solid Films 529(2013) 402–406.

[32] A. Hodzic, Z.H. Stachurskia, J.K. Kim, Nano-indentation of polymer–glassinterfaces Part I. Experimental and mechanical analysis, Polymer 41 (2000)6895–6905.

[33] N. Wang, L. Wang, R. Zhang, J. Li, C. Zhao, T. Wu, S. Ji, Highly stable “pore-filling”tubular composite membrane by self-crosslinkable hyperbranched polymers fortoluene/n-heptane separation, J. Membr. Sci. 474 (2015) 263–272.

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

[35] H.L. Zhou, L. Lv, G.P. Liu, W.Q. Jin, W.H Xing, PDMS/PVDF compositepervaporation membrane for the separation of dimethyl carbonate from amethanol solution, J. Membr. Sci. 471 (2014) 47–55.

[36] R. Xu, G.P. Liu, X.L. Dong, W.Q. Jin, Pervaporation separation of n-octane/thiophene mixtures using polydimethylsiloxane/ceramic composite mem-branes, Desalination 258 (2010) 106–111.

[37] A.F.M. Pinheiro, D. Hoogendoorn, A. Nijmeijer, L. Winnubst, Development of aPDMS-grafted alumina membrane and its evaluation as solvent resistantnanofiltration membrane, J. Membr. Sci. 463 (2014) 24–32.

Y. Hang et al. / Journal of Membrane Science 494 (2015) 205–215214

[38] G.P. Liu, W. Wei, W.Q. Jin, Pervaporation membranes for biobutanol produc-tion, ACS Sustain. Chem. Eng. 2 (2014) 546–560.

[39] Z.Y. Dong, G.P. Liu, S.N. Liu, Z.k. Liu, W.Q. Jin, High performance ceramic hollowfiber supported PDMS composite pervaporation membrane for bio-butanolrecovery, J. Membr. Sci. 450 (2014) 38–47.

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

[41] W.C. Oliver, G.M. Pharr, Improved technique for determining hardness andelastic modulus using load and displacement sensing indentation experi-ments, J. Mater. Res. 7 (1992) 1564–1580.

[42] W.C. Oliver, G.M. Pharr, Measurement of hardness and elastic modulus byinstrumented indentation: advances in understanding and refinements tomethodology, J. Mater. Res. 19 (2004) 3–19.

[43] W. Tang, X.L. Weng, L.J. Deng, K.W. Xu, J. Lu, Nano-scratch experiments of Au/NiCr multi-layered films for microwave integrated circuits, Surf. Coat. Technol.201 (2007) 5664–5666.

[44] W. Wei, S.S. Xia, G.P. Liu, X.L. Dong, W.Q. Jin, N.P. Xu, Effects of polydimethyl-siloxane (PDMS) molecular weight on performance of PDMS/ceramic compo-site membranes, J. Membr. Sci. 375 (2011) 334–344.

[45] A. Raisi, A. Aroujalian, Aroma compound recovery by hydrophobic pervapora-tion: the effect of membrane thickness and coupling phenomena, Sep. Purif.Technol. 82 (2011) 53–62.

Y. Hang et al. / Journal of Membrane Science 494 (2015) 205–215 215