Analysis of pH-dependent Structure and Mass Transfer Characteristics of Polydopamine Membranes by Molecular Dynamics Simulation☆

Fusheng Pan,Ruisi Xing,Zhongyi Jiang,*

Separation Science and Engineering

Analysis of pH-dependent Structure and Mass Transfer Characteristics of Polydopamine Membranes by Molecular Dynamics Simulation☆

Fusheng Pan1,2,Ruisi Xing1,2,Zhongyi Jiang1,2,*

1Key Laboratory for Green Chemical Technology of Ministry of Education,School of Chemical Engineering and Technology,Tianjin University,Tianjin 300072,China
2Collaborative Innovation Center of Chemical Science and Engineering(Tianjin),Tianjin 300072,China

A R T I C L EI N F O

Article history:

Membranes

Polydopamine

Molecular dynamics simulation

Free volume

Diffusion

Detailedatomistic structuresare constructed for polydopaminemembranescontainingdifferent amountsofcatechol and quinone groups to investigate the effect ofpH value inthe membrane casting solution on sorption and diffusion of small gas molecules(water and propylene)in the membranes.Interactions between dopamine oligomers are calculated,and it is found that the interactions decrease from−2356.52 kJ·mol−1in DOP-1 to−1586.69 kJ·mol−1in DOP-3 when all of the catechol groups are converted to quinone groups.The mobility of polymer segments and free volume properties of polydopamine membranes are analyzed.The sorption quantities of water and propylene in the membrane are calculated using Grand Canonical Monte Carlo method.The sorption results show that water adsorbed in DOP-1,DOP-2 and DOP-3 are 17.3,18.6 and 20.0 mg water per gram polymer,respectively,and no propylene molecule can be adsorbed.The diffusion behavior of water molecules in the membrane is investigated by molecular dynamics simulation.The diffusion coeff i cients of water molecules in DOP-1,DOP-2 and DOP-3 membranes are(1.80±0.52)×10−11,(3.40±0.64)×10−11and(4.50±0.92)×10−11m2·s−1,respectively.The predicted sorption quantities and diffusion coeff i cients of water and propylene in the membrane present the same trends as those from experimental results.

©2014TheChemicalIndustry andEngineeringSocietyofChina,andChemicalIndustryPress.Allrightsreserved.

1.Introduction

Inspired by the strong adhesion ability of mussel onto various surfaces,dopamine,a commercially available chemical containingcatechol and amine groups(origins of the extraordinarily strong adhesion[1,2]), hasbeenwidelyusedassurfacecoatingagentandimmobilizedinitiator to graft functionalgroupsforsurface modif i cation[3-6].Since it can adhere f i rmly onto the substrate and self-polymerize rapidly to form a dense,ultrathin f i lm,dopamine presents intrinsic advantage to be utilized as the dense active layer in composite membranes for separation. Composite membranes with polydopamine active layer(about 14 nm) on porous polysulfone(PS)hollow f i ber were fabricated and exhibited superior separation performance in the dehumidif i cation of propylene gas[7].Composite membranes with a thin polydopamine layer (＜100 nm)on the porous PS substrate were utilized for pervaporative desulfurization and exhibited satisfactory separation performance[8]. Apolydopamine layerabout50nmwascoatedontoNaf i onmembranes as the methanol barrier in direct methanol fuel cell[9].The methanol crossover of the modif i ed membranes was dramatically suppressed by 79%from 3.14×10−6to about 0.65×10−6cm2·s−1,while the proton conductivity was decreased slightly.As a dense membrane,the separation performance relies heavily on the inter-chain interaction and the microstructure.

Catechol groups play an important role in the redox reactions in the self-polymerization of dopamine[1,3,10].They are easily oxidized to quinone and can react with each other in an oxidative process under basic conditions.And quinone groups exhibit lower adhesion interaction than parent catechols[11].In the polymerization of dopamine,pH regulation is usually employed to manipulate its structure and performance of polydopamine[12],since pH of the dopamine solution can alter the equilibrium between catechol and quinone groups.At higher pH,more catechol groups of dopamine are deprotonated and oxidized to quinone groups[1].The intermolecular interaction,microstructure and separation performance of polydopamine are subsequently changed.However,inf l uences of pH on the structure and performance of polydopamine are poorly understood.

It is ratherdiff i cult tocharacterizethestructuraland transport propertiesofpolydopaminemembranesbyexperimentalmethodbecauseof its ultra-thin thickness.Generally accepted alternative methods to characterize the homogenous membranes do not work either since polydopamine cannot form a free-standing f i lm.Molecular simulationhas been proved as a useful tool for addressing membrane material and transport phenomena in separation processes at molecular level [13-17].Herein,molecular simulation is used to describe how pH inf l uences the structure and performance of polydopamine.The models for polydopamine containing different amounts of catechol and quinone groups are constructed to simulate the membranes fabricated at different pH values.The interaction between dopamine oligomers within the polydopamine and its inf l uences on the segment mobility and free volume properties are examined.The sorption positions and amount of water/propylene are calculated by Grand Canonical Monte Carlo (GCMC)method.The diffusion of water molecules in the membranes is investigated by molecular dynamics(MD)simulation.The simulation results are compared with experimental results to check the validity of the simulation approach.

2.Details of the Simulation

Fig.1.Atomisticmodelsofpolydopaminewithdifferentratiosofcatecholandquinoneforms(a)oligomerofDOP-1,withfunctionalgroupsofcatecholgroups;(b)oligomerofDOP-2,with functional groupsof 50%catechol groupsand 50%quinonegroups;(c)oligomerof DOP-3,with functional groups of quinone groups;(d)DOP-1,amorphouscells of(a);(e)DOP-2,amorphous cells of(b);(f)DOP-3,amorphous cells of(c).

2.1.Model construction

Duetothecomplexityofthepolymerizationprocess,thestructureof polydopamineissubjecttodebateatpresent.However,itiswellaccepted that 5,6-dihydroxylindole is one of the main intermediate component in dopamine polymerization[10,18-20].In this study,based on Stark et al.'s study[21],dopamine oligomer is simplif i ed as a homopolymer comprised of 5,6-dihydroxyindole and its oxidation form 5,6-indole quinone in the redox polymerization.The oligomer,as the basic structural units,consists of 5 dopamine molecules and exhibits a platelet with lateral extents of 1.5-2.0 nm.

Three models of dopamine oligomer with different amounts of catechol groups and quinone groups are developed as shown in Fig.1(a,b,c).Oxygen-containing functional groups on DOP-1 and DOP-3 are catechol groups and quinone groups,respectively,while those on DOP-2 are composed of 50%catechol groups and 50%quinone groups.It should be pointed out that the models only ref l ect the inf l uence tendency of pH on the structure of polydopamine,since the content of catechol and quinone groups at different pH values is unclear.

A packing model,with an initial density of 1.68 g·cm−3[22]and containing 20 oligomers,is constructed by amorphous cell module using the combination of an algorithm developed by Theodorou and Suter[23]and thescanningmethod of Meirovitch[24].Periodic boundary conditions are applied to the cubic simulation box.

2.2.Structure optimization and MD simulation procedure

MD simulations are carried out using the Discover and Amorphous Cell module of Materials Studio,developed by Accelrys Software Inc. The COMPASS(condensed-phase optimization molecular potentials for atomistic simulation studies)force f i eld is used in this study.Nonbond cutoff distance of 0.950 nm(with a spline width of 0.100 nm and a buffer width of 0.050 nm)is employed to evaluate the nonbond interaction.Long-tail corrections to the energy due to the nonbond cutoff are employed in the dynamics simulation.The temperature and pressure are both controlled by the Berendsen method[25]with decay constant of 0.1 ps.The equations of motion are integrated with a time step of 1 fs for all dynamic runs.

An optimization procedure is applied to the initially constructed atomistic structure as our previous work[26,27].A 2000-step energy minimization is performed f i rst to eliminate the undesirable contact (overlappingorclosecontact).Theannealingmethodisappliedtoovercome local energy minima and yield equilibrated conf i gurations with a globalenergyminimum.Inthismethod,themodels areheatedupstepwisefrom300Kto600Katintervalsof50K,andthencooledto300Kat intervals of 20 K.At each step a 250-ps NPT dynamics is applied to the cell.Afterwards,a 200 ps NPT(T=300 K,P=1.01×105Pa)dynamics is performed to obtain the equilibrium density.An additional 500 ps NVT(T=300 K)dynamics is performed on the endpoint of the NPTdynamics to obtain equilibrium molecular structures,and the atomic trajectory is recorded every picosecond for the subsequent analysis.

Fig.2.Mean square displacement of polymer segments in polydopamine membranes.

3.Results and Discussion

3.1.Morphology of the optimized polydopamine models

The morphologies of polydopamine after optimization are shown in Fig.1(d,e,f).The basic structural unit of polydopamine is a fragment of dopamine oligomers(like a fragment of a graphite sheet),or several fragments stacked together with～0.34 nm spacing.These basic structural units range between several nanometers and exhibit a multilayer structure.Polydopamine is based on a random packing of these multilayer structures.The optimized structure of polydopamine is consistent with experimental results[28].

3.2.Interaction between dopamine oligomers

The interaction energy(ΔE)between dopamine oligomers is calculated by the degree of overall oxidation that affords an optimal mixture of catechol and quinone groups.

where Ebulkis the potential energy of polydopamine bulk system,n is thenumberof dopamine oligomers,and Eoligomeris thepotentialenergy of single oligomer.

ΔEvaluesbetweendopamineoligomersinDOP-1,DOP-2andDOP-3 are−2356.52,−2301.76,and−1586.69 kJ·mol−1,respectively.High amount of polar groups(catechol groups,quinone groups,amine groups)on dopamine oligomers could form hydrogen bonds and polar group interaction with other oligomers,while π-electron conjugated structure in the dopamine oligomer will form strong π-π interaction among oligomers.Such strong π-π interactions drive oligomers to stack together and exhibit a multilayer structure as graphite.

The interaction energies in polydopamine follow the order of DOP-1＞DOP-2＞DOP-3.The decrease of interaction energy is resulted from the increase content of quinone groups.The simulated results show that quinone-quinone interaction(−57.74 kJ·mol−1)is weaker than catechol-quinone interaction(−76.36 kJ·mol−1)and catecholcatechol interaction(−99.23 kJ·mol−1).The result is consistent with the single-molecule measurements by atomic force microscopy[11].

Oxidation degree of catechol residues to corresponding quinones determines the bulk cohesiveness and the relative amount of catechol and quinone groups.At higher pH value,more quinone groups by oxidized catechols endow polydopamine with higher cohesiveness.Meanwhile,the weaker interaction energy of quinones than parent catechols lowers the interfacial adhesion ability[29].A balance between interfacial adhesion and bulk cohesiveness should be maintained by f i nding

3.3.Segment mobility of polydopamine

Segment mobility of the polymer controls the dynamic properties (generation and disappearance)of free volume voids within the membranes,and constitutes a key issue for transport properties[30,31].Segment mobility is investigated by examining the mean-square displacement(MSD)of polymer segment as a function of time.

where ri(t)and ri(0)are the position of atom i at time t and 0,respectively,andthebracket denotestheensembleaverage,whichis obtained from averaging all atoms and all time origins t=0.

Fig.2 shows the MSDs of polydopamine in the membranes.Larger slope of MSD curve ref l ects higher segmentmobility.In thestudied system,the change of segment mobility depends on the interaction between dopamine oligomers.In DOP-1,the high interaction among catecholgroupsreducesthesegmentmobilityofsurroundingpolymers. With the increasing content of quinone groups,the segment mobility is enhanced since theweakerattractive interactiondecreases theconf i nement effects among these oligomers.

Fig.3.Fractional free volume(a)and cavity size distribution(b)of the polydopamine membranes.

3.4.Free volume properties of the polydopamine membranes

Free volume plays a crucial role in diffusion behavior of penetrant molecules in membranes.Positron annihilation spectroscopy(PAS)is commonlyemployed as a directapproach to probe free volume properties.In the experimental study,PAS analysis only indicates that thepolydopamine membranes possess a looser structure when fabricated athigherpHvalues.ToinvestigatehowpHvalue affects themicrostructure,the fractional free volume(FFV),size distribution of free volume voids and free volume morphology are analyzed.Free volume properties of polydopamine are analyzed by the Connolly surface method [26,27].Fig.3(a)showstheFFVofpolydopaminemembranesusingmolecular probes,with diameter ranging from 0.02 to 0.8 nm.The FFV probed by water and propylene,modeled by spheres with diameter 0.26 and 0.47 nm,respectively,are shown in Table 1.The FFV of the membranes follows the order of DOP-3＞DOP-2＞DOP-1,which indicates that higher pH induces a looser structure.The simulated results present the same tendency as the experimental PAS results[7].

The FFV analysis indicates the overall free space in the membranes, but it cannot ref l ect the actual free volume voids through which the penetrant molecules pass.To solve such problem,size distribution of free volume voids is calculated.The volume(FV)of free volume voids between r−Δr and r+Δr is obtained by

Table 1The FFV of polydopamine membranes obtained from geometrical analysis by water and propylene

where FV(d−Δd)and FV(d+Δd)are the free volumes that are accessible for probes with diameter d−Δd and d+Δd,respectively.The interval of probe diameter,Δd,is set to 0.005 nm in this study.

The size distribution of free volume voids in polydopamine membranes is shown in Fig.3(b).The diameter of free volume voids mostly is around 0.13 nm.The size distributions of Dop-2 and Dop-3 shift toward the right-handside in comparison tothat of Dop-1.Withtheincrease of pH in the fabrication process,smaller void region reduces and larger void region increases.The variation indicates that the polydopamine changes from tight to loose,which is more benef i cial for transport of penetrant molecules.From Fig.3(b),it can be observed that the intensity of void size larger than 0.45 nm approaches zero in DOP-1,while DOP-2 and DOP-3 provide a moderate intensity.

Fig.4 shows the morphology maps of free volume voids accessible to water and propylene.Free volume voids are mainly created by ineff i cient packing or transient gaps among dopamine oligomers. The increase of FFV with pH is ascribed to the increasing amount and larger size of free volume voids by the weaker interactions among dopamine oligomers.The dispersion of water-accessible free volume voids in polydopamine is heterogeneous.The voids are much denser in the gaps among the multilayer structures than those in the multilayer structures.In the dense dispersion area,the free volume voids interconnect with neighboring ones and form water channels,especially in DOP-2 and DOP-3.Such water channels would provide a convenient pathway for diffusion of water molecules.On the contrary,much less propylene-accessible free volume voids are present in the polydopamine,always with longer distance.

Fig.4.Morphology of free volume voids(bright region)accessible to water(a)and propylene(b)in polydopamine membranes.

3.5.Sorption behavior of water/propylene in the membranes

Sorption of water and propylene in polydopamine membranes is investigated by the GCMC method implemented in the Sorption module.The method of Metropolis et al.[32]is employed for accepting or rejecting conf i gurational moves(rotation and translation of sorbate molecules)as well as for sorbate insertion and deletion,in which the trial conf i gurations are generated without bias and the adsorbate structure is treated as rigid.A total of 10000000 steps are used.

The GCMC simulation is conducted at a f i xed pressure of 350 kPa.The partial pressure of water and propylene is set as the experimental value (water mass content 0.5%).In the f i xed pressure simulation,the conf i gurations are sampled from a grand canonical ensemble,in which thefugacitiesofallcomponents,aswellasthetemperature,are fi xedasifthe framework is in open contact with an in fi nite sorbate reservoir with a fi xed temperature.The GCMC calculations are therefore carried out over the equilibrated con fi gurational snapshot of the membrane.

The amountofwateradsorbedinDOP-1,DOP-2 and DOP-3 are17.3, 18.6 and 20.0 mg water per gram polymer,respectively.According to the simulation results,with the increase content of quinone groups in the membrane,the water uptake increases slightly.The sorption of water in the polymeric membranes is mainly governed by the polymer-penetrant interaction and adsorbed positions within the membranes.With the increasing content of quinone groups,the polydopamine-water interaction is lower,since quinone-water interaction(−58.16 kJ·mol−1)is lower than catechol-water interaction (−76.93 kJ·mol−1).Meanwhile,the larger FFV originated from the increasing content of quinone groups induces more adsorbed positions (as shown in Fig.5).Such two factors with opposite effects change the water sorption slightly.The calculated sorption amount of propylene in polydopamine is zero for all models of polydopamine membranes. Low polydopamine-propylene interaction and big size of propylene make it dif fi cult to be adsorbed.The calculated results prove the ultra high sorption selectivity of polydopamine membranes towards water vapor over propylene as in the experiment[7].

3.6.Diffusion behavior of water/propylene in the membranes

The diffusion coef fi cient D of water in equilibrated models of membranes is calculated from the slope of the MSD for long time by Einstein relation:

Fig.5.Adsorbed position of water molecules in polydopamine membranes(atoms of polydopamine not shown).

In the simulation,a certain amount penetrant molecules(according to the GCMC simulation)are inserted into the membrane models. The models are equilibrated using the same procedure as mentioned in Section 2.2.The diffusion runs are performed under the NVE conditions for 5 ns.The diffusion coeff i cient is an averaged value from all penetrant molecules.Due to the limitation of short calculation time, it is rather diff i cult to obtain exact value of diffusion coeff i cient. Therefore,this study simply offers the changing trends of diffusivity in different polydopamine membranes.

The simulated diffusion coeff i cients of water molecules in DOP-1,DOP-2 and DOP-3 polydopamine membranes are(1.80±0.52)× 10−11,(3.40±0.64)×10−11,and(4.50±0.92)×10−11m2·s-1, respectively.The diffusion coeff i cient of water increases with the pH value in the fabrication process.The main factors determining the diffusion of water in the membranes are free volume properties,polymer segment mobility and polymer-water interaction. With the increase of pH,free volume voids are enlarged,the polymer segment mobility increases,and the polymer-water interaction decreases.All these factors enhance the diffusion of water.

The spatial distribution of water in polydopamine membranes (shown in Fig.6)indicates that water molecules are not uniformly distributed within the membranes.Most water molecules are gathered in waterchannels(gapsamongmultilayerstructures)andquitelesslocate between dopamine oligomers in the multilayer structure.Diffusion coeff i cients of these two kinds of water molecules in the last 100 ps of thediffusion runare calculated.The results display that diffusion coeff icient of water molecules in the water channel(5.60×10−11m2·s-1)is muchlarger than that between oligomers(1.67×10−12m2·s-1).Plenty of water channels with high water permeability endow polydopamine membranes with high water vapor permeance.

Fig.6.Snapshots of the spatial distribution of water in polydopamine membranes(atoms of polydopamine not shown).

Although the size distribution of free volume in the membranes indicates that some free volume voids in DOP-2 and DOP-3 are larger than that of propylene,the diffusion behavior of propylene molecules cannot be achieved until a new free volume void generated near the free volume voids is occupied by propylene molecules,since the propylene-accessible free volume voids are always with large distance.Thegenerationofthefreevolumevoidsaccessibletopropyleneisrather diff i cult because of the low segment mobility of polydopamine.Hence, these polydopamine membranes fabricated at different pH values possess high diffusion selectivity.Thehigh diffusion selectivity and adsorption selectivity endow polydopamine membranes with high separation factors for water vapor over propylene.

4.Conclusions

A combination of GCMC and MD simulation methods is employed to investigate the structure properties and sorption-diffusion behavior of water/propylene in polydopamine membranes fabricated at different pH values in membrane casting solution.Higher content of quinone groups athigherpHvalue induces weakerinteraction amongdopamine oligomers,subsequently leading to higher polymer segment mobility and larger FFV.The morphology and distribution of free volume voids accessible to water and propylene are shown explicitly.Water channels are formed by interconnected free volume voids,especially in the gaps among the multilayer structures stacked by the dopamine oligomers. Such channels endow thepolydopaminewith high water vapor permeability.Moreover,appropriatesizedistributionoffreevolumevoidsand low polymersegmentmobility inhibit thepenetrationofpropyleneand ensurehighseparationfactors.Thecalculatedsorptionanddiffusioncoeff i cients of water molecules in the polydopamine membranes exhibit thesametrendsasthosefromtheexperimentalinvestigations.Itproves thatmolecularsimulationisausefultooltoprobethestructureatatomistic level and dynamic properties and transportation behavior in picosecond level.The visible structure of membranes and transportation process of penetrant molecules in the simulation are helpful to understand the transportation behavior of penetrant molecules in the membranes.

Acknowledgments

Gratitude is also expressed to R&D Center for Petrochemical Technology,and Advanced Instrumental Detecting&Analytical Center, School of Chemical Engineering and Technology,Tianjin University,for providing access to the Material Studio molecular modeling software.

[1]H.Lee,S.M.Dellatore,W.M.Miller,P.B.Messersmith,Mussel-inspired surface chemistry for multifunctional coatings,Science 318(5849)(2007)426-430.

[2]H.Lee,B.P.Lee,P.B.Messersmith,A reversible wet/dry adhesive inspired by mussels and geckos,Nature 448(7151)(2007)338-342.

[3]J.Sedó,J.Saiz-Poseu,F.Busqué,D.Ruiz-Molina,Catechol-based biomimetic functional materials,Adv.Mater.25(5)(2013)653-701.

[4]L.Q.Xu,W.J.Yang,K.G.Neoh,E.T.Kang,G.D.Fu,Dopamine-induced reduction and functionalization of graphene oxide nanosheets,Macromolecules 43(20)(2010) 8336-8339.

[5]S.H.Yang,S.M.Kang,K.B.Lee,T.D.Chung,H.Lee,I.S.Choi,Mussel-inspired encapsulation and functionalization of individual yeast cells,J.Am.Chem.Soc.133(9)(2011) 2795-2797.

[6]C.Y.Li,W.C.Wang,F.J.Xu,L.Q.Zhang,W.T.Yang,Preparation of pH-sensitive membranes via dopamine-initiated atom transfer radical polymerization,J.Membr.Sci. 367(1-2)(2011)7-13.

[7]F.S.Pan,H.P.Jia,S.Z.Qiao,Z.Y.Jiang,J.T.Wang,B.Y.Wang,Y.R.Zhong,Bioinspired fabrication of high performance composite membranes with ultrathin defect-free skin layer,J.Membr.Sci.341(1-2)(2009)279-285.

[8]B.Li,W.P.Liu,Z.Y.Jiang,X.Dong,B.Y.Wang,Y.R.Zhong,Ultrathin and stable active layer of dense composite membrane enabled by poly(dopamine),Langmuir 25(13) (2009)7368-7374.

[9]J.T.Wang,L.L.Xiao,Y.N.Zhao,H.Wu,Z.Y.Jiang,W.Q.Hou,A facile surface modif i cation of Naf i on membrane by the formation of self-polymerized dopamine nanolayer to enhance the methanol barrier property,J.Power Sources 192(2)(2009) 336-342.

[10]D.R.Dreyer,D.J.Miller,B.D.Freeman,D.R.Paul,C.W.Bielawski,Elucidating the structure of poly(dopamine),Langmuir 28(15)(2012)6428-6435.

[11]H.Lee,N.F.Scherer,P.B.Messersmith,Single-molecule mechanics of mussel adhesion,Proc.Natl.Acad.Sci.103(35)(2006)12999-13003.

[12]D.G.Barrett,D.E.Fullenkamp,L.H.He,N.Holten-Andersen,K.Y.C.Lee,P.B. Messersmith,pH-based regulation of hydrogel mechanical properties through mussel-inspired chemistry and processing,Adv.Funct.Mater.23(9)(2013) 1111-1119.

[13]J.Zhao,J.Ma,J.Chen,F.S.Pan,Z.Y.Jiang,Experimental and molecular simulation investigations on interfacial characteristics of gelatin/polyacrylonitrile composite pervaporation membrane,Chem.Eng.J.178(2011)1-7.

[14]Y.Jing,L.Wei,Y.D.Wang,Y.X.Yua,Molecular simulation of MCM-41:structural properties and adsorption of CO2,N2and f l ue gas,Chem.Eng.J.220(2013)264-275. [15]D.Hofmann,M.Heuchel,Y.Yampolskii,V.Khotimskii,V.Shantarovich,Free volume distributions in ultrahigh and lower free volume polymers:comparison between molecular modeling and positron lifetime studies,Macromolecules 35(6)(2002) 2129-2140.

[16]D.L.Sun,J.Zhou,Molecular simulation of oxygen sorption and diffusion in the poly (lactic acid),Chin.J.Chem.Eng.21(3)(2013)301-309.

[17]Q.Y.Yang,Q.Xu,B.Liu,C.L.Zhong,B.Smit,Molecular simulation of CO2/H2mixture separation inmetal-organic frameworks:effectof catenation and electrostatic interactions,Chin.J.Chem.Eng.17(5)(2009)781-790.

[18]K.L.Double,L.Zecca,P.Costi,M.Mauer,C.Griesinger,S.Ito,D.Ben-Shachar,G. Bringmann,R.G.Fariello,P.Riederer,M.Gerlach,Structural characteristics of human substantia nigra neuromelanin and synthetic dopamine melanins,J. Neurochem.75(6)(2000)2583-2589.

[19]A.Pezzella,L.Panzella,A.Natangelo,M.Arzillo,A.Napolitano,M.d'Ischia,5,6-Dihydroxyindole tetramers with“Anomalous”interunit bonding patterns by oxidative coupling of 5,5,6,6-tetrahydroxy-2,7-biindolyl:emerging complexities on the way toward an improved model of eumelanin buildup,J.Org.Chem.72(24) (2007)9225-9230.

[20]J.H.Waite,Mussel power,Nat.Mater.7(1)(2008)8-9.

[21]K.B.Stark,J.M.Gallas,G.W.Zajac,J.T.Golab,S.Gidanian,T.P.McIntire,J.Farmer, Effect of stacking and redox state on optical absorption spectra of melaninscomparison of theoretical and experimental results,J.Phys.Chem.B 109(5) (2005)1970-1977.

[22]R.L.Davidson,Handbook of Water-soluble Gums and Resins,Mac.Graw-Hill,New York,1980.

[23]T.D.N.Heodorou,U.W.Suter,Detailed molecular structure of a vinyl polymer, Macromolecules 18(1985)1467-1478.

[24]H.Meirovitch,Computer simulation of self-avoiding walks:testing the scanning,J. Chem.Phys.79(1)(1983)502-508.

[25]H.J.C.Berendsen,J.P.M.Postma,W.F.Funsteren,Molecular dynamics with coupling to an external bath,J.Chem.Phys.81(8)(1984)3684-3690.

[26]F.S.Pan,F.B.Peng,Z.Y.Jiang,Diffusion behavior of benzene/cyclohexane molecules in poly(vinyl alcohol)-graphite hybrid membranes by molecular dynamics simulation,Chem.Eng.Sci.62(3)(2007)703-710.

[27]F.S.Pan,J.Ma,L.Cui,Z.Y.Jiang,M.H.Zhang,Water vapor/propylene sorption and diffusion behavior in PVA-P(AA-AMPS)blend membranes by GCMC and MD simulation,Chem.Eng.Sci.64(24)(2009)5192-5197.

[28]J.M.Gallas,K.C.Littrell,S.Seifert,G.W.Zajac,P.Thiyagarajan,Solution structure of copper ion-induced molecular aggregates of tyrosine melanin,Biophys.J.77(2) (1999)1135-1142.

[29]M.Yu,T.J.Deming,Synthetic polypeptide mimics of marine adhesives,Macromolecules 31(15)(1998)4739-4745.

[30]N.C.Karayiannis,V.G.Mavrantzas,D.N.Theodorou,Molecular simulation of permeation of small penetrants through membranes.1.Diffusion coeff i cients,Macromolecules 27(16)(1994)4498-4508.

[31]C.Nagel,E.Schmidtke,K.Gunther-Schade,D.Hofmann,D.Fritsch,T.Strunskus,Free volume distributions in glassy polymer membranes:comparison between molecular modeling and experiments,Macromolecules 33(6)(2000)2242-2248.

[32]N.Metropolis,A.W.Rosenbluth,M.N.Rosenbluth,A.H.Teller,Equation of state calculations by fast computing machines,J.Chem.Phys.21(1953)1087-1092.

14 December 2013

☆Supported by the National Science Fund for Distinguished Young Scholars (21125627),the National Natural Science Foundation of China(21306131),Specialized Research Fund for the Doctoral Program of Higher Education(20120032120009),Seed Foundation of Tianjin University,and the Programme of Introducing Talents of Discipline to Universities(B06006).

*Corresponding author.

E-mail address:zhyjiang@tju.edu.cn(Z.Jiang).

http://dx.doi.org/10.1016/j.cjche.2014.09.014

1004-9541/©2014 The Chemical Industry and Engineering Society of China,and Chemical Industry Press.All rights reserved.

Received in revised form 2 April 2014

Accepted 9 April 2014

Available online 16 September 2014