Poly(vinylidene fluoride)-graft-Poly(N-vinyl-2-pyrrolidone) Copolymers Prepared via a RAFT-Mediated Process and their Use in Antifouling and Antibacterial Membranes

Full Paper

Poly(vinylidene fluoride)-graft-Poly(N-vinyl-2-pyrrolidone) Copolymers Preparedvia a RAFT-Mediated Process and their Use inAntifouling and Antibacterial Membranes

Qiang Peng,* Shiqiang Lu, Dezhi Chen, Xiaoqin Wu, Panfeng Fan,Rong Zhong, Yongwen Xu

PNVP-g-PVDF copolymers were synthesized and used to produce microfiltration membranes.The pore size and distribution varied with the grafting concentration and the density of graftpoints. A significant decrease of the amounts of adsorbed BSA indicated improved antifoulingproperties of PVDF. The MF membranes were further func-tionalized via surface-initiated block copolymerizationwith DMAEMA to obtain (PVDF-g-PNVP)-b-PDMAEMA MFmembranes. Quaternization of the tertiary amine groupsof the PDMAEMA brushes gave rise to a high concentrationof quaternary ammonium salt on the membrane surfaces.The bactericidal effect of the QAS-functionalized mem-branes on E. coli was also demonstrated and discussed.

Introduction

Because of its excellent chemical resistance, well-con-

trolled porosity, and good thermal properties,[1] poly(vinyl-

idene fluoride) (PVDF) has attracted considerable attention

Q. Peng, D. Chen, X. Wu, P. Fan, R. Zhong, Y. XuDepartment of Environmental and Chemical Engineering, Nan-chang Institute of Aeronautical Technology, Nanchang 330034,ChinaE-mail: [email protected]. LuDepartment of Materials and Engineering, Nanchang Institute ofAeronautical Technology, Nanchang 330034, China

Macromol. Biosci. 2007, 7, 1149–1159

� 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

for applications in ultrafiltration (UF) and microfiltration

(MF) membranes,[2] protein adsorption, immobilization

and separation,[3–5] and stimuli- responsive controlled

release applications.[6–9] However, the application of PVDF

membranes is limited by the hydrophobic nature of their

surfaces, especially the surfaces of the pores. Thus, protein

fouling occurs both on the membrane surface and within

the pores when the membranes are exposed to protein-

containing solutions during filtration.[10] Various ap-

proaches, such as coating[11,12] and grafting techni-

ques,[13–17] have been developed to endow the conventional

hydrophobic membranes with hydrophilic properties. For

the coating method, the coated surface layers may be easily

removed, especially by varying the pH of the solution. On

DOI: 10.1002/mabi.200700068 1149

Q. Peng et al.

1150

the other hand, surface modification of existing membranes

by grafting or graft copolymerization is likely to reduce their

permeability because of the change in membrane pore size

and its distribution. To overcome these shortcomings,

bulk-modified or chain-modified PVDF may be used as the

membrane-fabrication material.[18]

Living free-radical polymerizations, such as nitroxide-

mediated polymerization (NMP),[19] atom transfer poly-

merization (ATRP),[20] and reversible addition fragmenta-

tion chain transfer (RAFT)[21] polymerization, open up an

promising way of designing and controlling polymer

architectures under mild reaction condition.[22–25] The

chemical versatility by controlling the agents in the RAFT

processmakes RAFT-based procedures highly attractive for

the preparation of well-defined polymers with specific

polymer architectures.[26–28] Graft copolymerization by

RAFT-mediated process can be expected to prepare well-

defined side chains.[29,30] Thus, the uniformity of the

pore-size distribution can be improved through the

interaction of well-defined side chains with the casting

medium by phase-inversion film-formation. This molec-

ular design technique also provides an alternative approach

to prepare antifouling membranes with a well-defined

surface structure and controlled pore size and distribution.

On the other hand, antimicrobial agents in liquid form

have the inherent problem of residual toxicity. This

problem may be resolved if the antimicrobial agent could

be immobilized on a substrate surface. Methods of

immobilizing antimicrobial agents on various substrate

surfaces have been widely studied. A large number of

quaternary ammonium salts (QAS) exhibit good bacter-

icidal properties.[31–34] Quaternized cationic polymers can

exhibit a higher antimicrobial activity than the corre-

sponding low-molecular-weight model compounds.[35]

However, the activity evidently decreases after usage,

because a great deal of bacteria cadavers will be absorbed

on the solid surface. The above problems inspire us to

prepare the biocompatible membranes with permanent

antimicrobial activity.

In this work, we report on the synthesis of PVDF

with grafted living poly(N-vinyl-2-pyrrolidone) (PNVP)

side chains using a RAFT-mediated process. PNVP was

chosen to modify PVDF because of its has excellent

biocompatibility with living tissues and a relatively-low

cytotoxicity.[36–39] The copolymers can be readily made

into microfiltration (MF) membranes by phase inversion

with enriched, living PNVP graft chains on the pore

surfaces. Protein-adsorption experiments indicated that

these membranes exhibited good antifouling properties.

The living membranes were further functionalized via

surface-initiated block copolymerizationwith (2-dimethyl-

amino)ethyl methacrylate (DMAEMA), as well as the

quaternization of the PDMAEMA brushes, to obtain (PVDF-

g-PNVP)-b-PDEMAEMA MF membranes. The resulting

Macromol. Biosci. 2007, 7, 1149–1159


membranes exhibited an excellent and permanent anti-

microbial activity because of their hydrophilic surfaces.

Experimental Part

Materials

Poly(vinylidene fluoride) powders with a molecular weight of

441 000 g �mol�1 were obtained from Elf Atochem of North

America Inc. The monomers, N-vinyl-2-pyrrolidone and

2-(dimethylamino)ethyl methacrylate, were purchased from

Aldrich Chemical Co. and were used after removal of the inhibitor

in an Al2O3 column. The chain transfer agent (CTA), 1-phenylethyl

dithiobenzoate (PDB), was prepared according to published

procedures.[40] The protein, bovine serum albumin (BSA), was

obtained from the Sigma Chemical Co. of St. Louis. Dulbecco’s

phosphate-buffered saline (PBS, pH¼7.4), used for the protein

adsorption experiments, was freshly prepared. Filter papers of

No. 1 Grade were obtained fromWhatman Co. Escherichia coliwas

obtained from American Type Culture Collection (ATCC, # 14948).

RAFT-Mediated Graft Copolymerization of

PNVP with PVDF

The PVDF powders were dissolved in N,N-dimethylformamide

(DMF) to a concentration of 75 g � L�1. A continuous stream of O3/O2

mixture was bubbled through 25 mL of the solution at 25 8C.The O3/O2 mixture was generated from an Azcozon RMU 16-04EM

ozone generator. The gas flow rate was adjusted to 300 L �h�1 to

give rise to an ozone concentration of about 0.027 g � L�1 in the

gaseous mixture. A typical treatment time of about 15 min was

used. This pre-treatment time gave rise to a peroxide content of

about 10�4 mol � g�1 in the polymer.[41] The dependence of

ozone-treatment time and peroxide concentration has been

reported previously.[41] The ozone-pretreated PVDF solution

(containing about 1 g of PVDF) and a predetermined volume

(0.5, 1, or 2 mL) of the DMF solution of PDB (4.2� 10�1 mol � L�1)

were transferred to an ampoule. About 3 g of N-vinyl-2-

pyrrolidone monomer was then introduced into the ampoule.

The final volume of the reaction mixture was adjusted to 20 mL.

The ampoules were degassed with three freeze-evacuate-thaw

cycles. They were then sealed and heated at 60 8C for a desired

reaction time period. At the end of the reaction, the ampouleswere

cooled in an ice bath. The PVDF-g-PNVP was precipitated in excess

ethanol (a good solvent for the NVP homopolymer and PDB

residues). The copolymerswere purified three times by redissolving

in DMF and reprecipitating in ethanol. The graft copolymers were

further purified by extraction with an ethanol/water (50:50)

mixture for 48 h, in order to remove the residual PNVP. Finally, the

copolymers were dried by pumping under reduced pressure

overnight at room temperature.

Preparation of the PVDF-g-PNVP MF Membranes with

Living Surfaces

Each microfiltration (MF) membrane was prepared by phase

inversion in an aqueous medium from a 12 wt.-% DMF solution of

DOI: 10.1002/mabi.200700068

Poly(vinylidene fluoride)-graft-Poly(N-vinyl-2-pyrrolidone) Copolymers Prepared via a RAFT-Mediated Process. . .

Scheme 1. Schematic illustration of the process of PNVP graft copolymerization with an ozone-preactivated PVDF backbone, byRAFT-mediated polymerization and the preparation of the PVDF-g-PNVP MF membrane by phase inversion.

Figure 1. The 300 MHz 1H NMR spectra of the PVDF-g-PNVPcopolymer with a ([–PNVP–]/[–CH2CF2–])bulk ratio of 0.082.

the copolymer. The copolymer solutionwas cast onto a glass plate,

whichwas then immersed in a bath of doubly-distilledwater after

the polymer solution had been subjected to a brief period of

evaporation in air. The temperature of the water in the casting

bath was controlled to 25 8C. Each membrane was left in water for

about 20 min after separation from the glass plate. It was then

extracted in a second bath of doubly-distilled water at 70 8C for

15 min. Such a heat-treatment step is commonly performed

during the fabrication of commercialmembranes in order to refine

the pore-size distribution.[42] The purified membranes were dried

by pumping under reduced pressure for subsequent characteriza-

tion. The processes for the RAFT-mediated synthesis of the

PVDF-g-PNVP copolymer and the preparation of the MF mem-

brane are described in Scheme 1.

Preparation of the (PVDF-g-PNVP)-b-PDMAEMA

Antibacterial Membranes

The living PNVP side chains on the surface of the PVDF-g-PNVP

membranes were used as macro-CATs to further functionalize the

membrane in another round of surface-initiated graft copolymer-

ization. 30 mg of the PVDF-g-PNVP membrane, DMAEMA

monomer (2 mL), and 2,20-azoisobutyronitrile (AIBN) (0.4 mg)

were introduced in 40 mL of 2-propanol. The solution was

saturated with purified argon for 30 min under stirring. The

reactor flask was then sealed and heated to 70 8C. After the desiredreaction time of 12 h, the flaskwas cooled to stop the reaction. The

membrane was washed with tetrahydrofuran (THF), water and

ethanol. After themembranewas dried in vacuum, it was exposed

in 5 mL of 1-bromohexane for the quaternization reaction. The

reactionwas kept at 70 8C for 48 h, and themembranewaswashed

with THF, water and ethanol in turn, and then purified by

extraction with ethanol for 24 h, and dried under vacuum.

Macromol. Biosci. 2007, 7, 1149–1159


Characterization of the Graft Copolymers

Fourier-transform infrared (FT-IR) spectra of the graft copolymers

were recorded on a Bio-Rad FTS 135 FT-IR spectrophotometer, with

the copolymer samples dispersed in KBr pellets. 1H NMR spectros-

copy was performed on a Bruker ARX 300 instrument with

deuterated DMF as the solvent. The bulk C and N contents were

determined on a Perkin-Elmer 2400 elemental analyzer. The F

content was determined by the Schoniger combustion method.[43]

X-ray photoelectron spectroscopy (XPS)measurementsweremade

on a Kratos AXIS HSi spectrometer with a monochromatized Al Ka

www.mbs-journal.de 1151

Q. Peng et al.

1152

X-ray source (1 486.7 eV photons) at a constant dwelling time of

100 ms and a pass energy of 40 eV. The core-level signals were

obtained at the photoelectron takeoff angle (R, with respect to

the sample surface) of 90 8. Surface elemental stoichiometries

were determined from peak-area ratios, after correcting with the

experimentally-determined sensitivity factors, and were reliable

to �5%. The surface morphology of the MF membranes was

studied by scanning electron microscopy (SEM), using a JEOL 6320

electron microscope. The pore sizes of the MF membranes were

measured from SEM images, and the mean pore size was

calculated using the relationship Dm¼ (Pn

i¼ 1 Di)/n based on

200 pores (n¼ 200). Static water contact angles of the graft

copolymer films, cast by spin-coating, were measured at 25 8Cand 60% relative humidity on a telescopic goniometer (Rame-

Hart, model 10000-230). The telescope, with a magnification

power of 23�, was equipped with a protractor of 18 graduation.For each angle reported, at least five sample readings from

different surface locations were averaged. The angles reported

were reliable to �38.

Figure 2. (a) Effect of graft-copolymerization time on the bulk andsurface graft concentration of the PVDF-g-PNVP copolymersobtained from RAFT-mediated graft copolymerization ([PDB]¼0.5� 10�2 M). (b) The effect of the PBD concentration on the bulkand surface graft concentration of the PVDF-g-PNVP copolymersobtained from RAFT-mediated graft copolymerization (reactiontime¼ 12 h).

Protein Fouling Measurements

To investigate the antifouling properties of the PVDF-g-PNVP

membranes, the BSA adsorption experiment was conducted

at 25 8C. The membranes were hydrated initially by immersion

in methanol, followed by immersion in distilled water. The

membranes were subsequently washed with the phosphate

buffer saline (0.01 M PBS, pH¼7.4) for 1 h before being incubated

in PBS containing 2mg �mL�1 of BSA for 24 h at room temperature.

After removal from the protein solution, the membranes were

washed for 5 min in three batches of PBS, followed by three

batches of distilledwater. Finally, sampleswere dried in a vacuum

oven at room temperature. The surface coverage of BSA was

quantified by XPS using the nitrogen signal associated with BSA.

Permeation experiments were performed at room temperature

(25 8C) and 8 kPa transmembrane pressure using a stirred

microfiltration cell (Toyo Roshi UHP-25, Tokyo, Japan). The

concentration of the protein (BSA) solution was 1 mg �mL�1.

The effective membrane area was 3.14 cm2. The flux was

calculated from the weight of the solution permeated per unit

time and per unit area of the membrane surface.

Determination of Antibacterial Characteristics

E. coli was cultivated in 50 mL of a 3.1% yeast-dextrose broth[44] at

37 8C. The bacterial cell concentrationwasmeasured by the optical

density at 540 nm and the cell number was calculated based on a

standard calibration with the assumption that the optical density

of 1.0 at 540 nm is equivalent to approximately 109 cells permL.[45]

For thewaterborne antibacterial assay, the E. coli-containing broth

was centrifuged at 2 700 rpm for 10 min, and after the removal of

the supernatant, the cells were washed twice with PBS (pH¼7.4)

and resuspended in PBS at a concentration of 107 cells per mL. The

substrates were immersed in this suspension in a sterile

Erlenmeyer flask. The flask was then shaken at 200 rpm at

37 8C for 2 h. The substrates were removed from the above flask

andwashed three timeswith sterile PBS and placed in Petri dishes.

This was followed by the immediate addition of solid growth agar

Macromol. Biosci. 2007, 7, 1149–1159


(1.5% agar in yeast-dextrose broth, autoclaved, poured into a Petri

dish, and dried under reduced pressure at room temperature

overnight). The Petri dishes were then sealed and incubated at

37 8C for 24 h. The substrates, after the waterborne antibacterial

assays, were characterized by optical microscopy (using an

Olympus BX60).

Results and Discussion

Chemical Structure of the Graft Copolymers

The chemical structure of the graft copolymer PVDF-g-

PNVP was firstly characterized by 1H NMR spectroscopy.

The chemical shifts at 2.9 and 2.4 ppm are attributable

to the CH2 group in PVDF due to head-to-tail (ht) and

head-to-head (hh) bonding arrangements, respectively.[46]

Grafting of the PNVP polymer to PVDF resulted in

the appearance of peaks with chemical shifts in the

regions 2.32–2.51, 2.95–3.21, and at 2.79 ppm attributed

to the CH2CO and CH2 protons, the CH2 protons of

the grafted PNVP backbone hydrogen species, and the

CHN and CH2N protons, respectively. The dithiobenzoate

DOI: 10.1002/mabi.200700068


end group of the graft chain, which is associated with

the chain transfer agent, was discernible in the region

of 7.38–8.34 ppm in the spectrum. The presence of the

dithiobenzoate end group confirmed that the graft

copolymerization had proceeded via the RAFT-mediated

process. The 1H NMR spectrum of the PVDF-g-PNVP

copolymer with a ([–PNVP–]/[–CH2CF2–])bulk ratio of

0.082 is shown in Figure 1. The solvent peaks s2 and s3

were subtracted from the spectrum using their known

intensities relative to that of the solvent peak s1, obtained

from the analysis of pure deuterated DMF. The chemical

structure of PVDF-g-PNVP was also studied by FT-IR

spectroscopy. The FT-IR spectra of the PVDF-g-PNVP

copolymers all contained the characteristic band for C––O

stretching (n¼ 1657 cm�1) of the grafted PNVP side chains.

On the other hand, the adsorption band in the region of

1 120 to 1 280 cm�1, characteristic of the CF2 functional

groups of PVDF, was also present in all the copolymer

samples. The intensity for the characteristic peak of the

carbonyl group increased with the grafting degree.

Bulk Graft Concentrations of the PVDF-g-PNVPCopolymers: Elemental Analysis

The bulk graft concentration, defined as the number of

PNVP repeat units per PVDF repeat unit, or the ([–PNVP–]/

[–CH2CF2–])bulk molar ratio, can be obtained readily from

the elemental ([N]/[F])bulk molar ratio by taking into

account the nitrogen stoichiometries of the graft and

main chains, and the nitrogen-to-fluorine ratio of the PVDF

main chains. Thus, the ([–PNVP–]/[–CH2CF2–])bulk molar

ratio can be calculated from the following relationship:

([–PNVP–]/[–CH2CF2–])bulk¼ 2 � [N]/[F], where the factor

2 accounts for the fact that there are 2 fluorine atoms and

1 nitrogen atomper repeat unit of PVDF and PNVP polymer

chains, respectively. Figure 2(a) shows the effect of the

Table 1. Experimental parameters and compositions of the PVDF-g-P

Copolymer PVDF

[–CH2CF2–]

NVP

Monomer

PBD

(CAT)

mol � LS1 mol � LS1 10S2 mol � L

PVDF-g-PNVP (1) 0.4 2.0 0.5

PVDF-g-PNVP (2) 0.4 2.0 0.5

PVDF-g-PNVP (3) 0.4 2.0 0.5

PVDF-g-PNVP (4) 0.4 2.0 0.5

PVDF-g-PNVP (5) 0.4 2.0 1.0

PVDF-g-PNVP (6) 0.4 2.0 1.5

PVDF-g-PNVP (7) 0.4 2.0 2.0

a)Peroxide initiator is the peroxides and hydroperoxides generated f

peroxide initiator will generate free radicals.

Macromol. Biosci. 2007, 7, 1149–1159


polymerization time on the bulk graft concentration of the

PVDF-g-PNVP copolymers obtained from RAFT-mediated

graft copolymerization. The bulk graft concentration

increases almost linearly, which reflects the living charac-

teristics of the RAFT process. In a RAFT-mediated poly-

merization process, activation and deactivation occurs at

about the same rate or an active site migrates from one

chain to another. In the present work, the chains were

grown from PVDF carrying the peroxide initiating groups

that would probably decompose following the Arrhenius

rate. Nevertheless, the polymer side chains initiated from

the PVDF propagated very slowly because of the high

chain-transfer rate constant. As a result, well-defined

grafted chains of about equal sizes can still be achieved.

The slow and approximately linear increase in graft

concentration with the polymerization time in the presence

of the CTA is consistent with the presence of a controlled

polymerization process. Figure 2(b) shows the effect of

concentration of the chain-transfer agent on the bulk and

surface graft concentrations of the PVDF-g-PNVP copoly-

mers. At a constant reaction time, monomer concentration

and initiator concentration, the bulk graft concentration

increases with decreasing concentration of the chain-

transfer agent in the reaction system. The results indicate

that, as the concentration of the chain-transfer agent is

reduced, the normal events of radical polymerization gain

importance over those of the RAFT-mediated process.

Experimental parameters and composition of the

PVDF-g-PNVP copolymers including the bulk graft con-

centrations are shown in Table 1.

Surface Compositions of thePVDF-g-PNVP Membranes

The surface compositions of the PVDF- g-PNVPmembranes

were studied by XPS. Figure 3(a–e) show the respective C1s

NVP copolymers.

Peroxide

initiatora)Reaction

time

Bulk graft

concentration of

copolymer ([–PNVP–]/

[–CH2CF2–])bulkS1 10S3 mol � LS1 h

2.5 6 0.024

2.5 12 0.045

2.5 18 0.071

2.5 24 0.082

2.5 12 0.037

2.5 12 0.022

2.5 12 0.021

rom ozone treatment of PVDF; after heating decomposition, the


Q. Peng et al.

Figure 3. XPS C1s core-level, wide-scan and N1s core-level spectra of (a) the pristinePVDF membrane, and the PVDF-g-PNVP membranes (b) ([–PNVP–]/[–CH2CF2–])bulk¼0.024, reaction time¼6 h, [PDB]¼0.5� 10�2 M; (c) ([–PNVP–]/[–CH2CF2–])bulk¼0.045, reaction time¼ 12 h, [PDB]¼0.5� 10�2 M; (d) ([–PNVP–]/[–CH2CF2–])bulk¼0.071, reaction time¼ 18 h, [PDB]¼0.5� 10�2 M; and (e) ([–PNVP–]/[–CH2CF2–])bulk¼0.082, reaction time¼ 24 h, [PDB]¼0.5� 10�2 M).

1154

core-level, wide-scan and N1s core-level

spectra of the membranes prepared from

the pristine PVDF and the RAFT-mediated

PVDF-g-PNVP copolymers. The pristine PVDF

and the PVDF-g-PNVP membranes were cast

at 25 8C in an aqueous bath. In the case of the

pristine PVDF membrane (Figure 3a), the C1s

core-level spectrum can be curve-fitted with

two peak components of approximately

equal areas, with binding energy (BE) at

285.8 eV for the CH2 species and at 290.5 eV

for the CF2 species.[47] For the PVDF-g-PNVP

copolymer membrane, three new peak com-

ponents appeared in the C1s core-level spec-

trum. The component at the BE of 287.4 eV is

assigned to the N–C––O species of the grafted

PNVP polymer chains. The component with

BE at 284.6 eV, on the other hand, is attri-

buted to the hydrocarbon of the grafted PNVP

polymer chain. Finally, the remaining area of

the peak component with BE at about

285.8 eV is assigned to the CN species.[48]

As the CN and (CH2) (PVDF) peak components

have the same BE, they are combined and

shown as a single peak component in

Figure 3(b–e).

The surface graft concentration of the

PVDF-g-PNVP MF membrane can be deter-

mined from the XPS-derived nitrogen-

to-fluorine ratio and the same relationship

as that used to determined the bulk graft

concentration. The results are shown in

Table 2. Figure 2 compares the effect of the

polymerization time and the concentration

of the chain-transfer agent on the surface

graft concentration of the PVDF-g-PNVP

copolymers obtained from RAFT-mediated

graft copolymerization. The surface PNVP

polymer concentration for each copolymer

sample was always much higher than

the corresponding bulk concentration.

Obviously, significant surface aggregation

of the hydrophilic graft chains has occurred

during membrane fabrication by the phase

inversion in the aqueous medium, due to

the relatively-low interfacial energy be-

tween the PNVP graft chain and water.

To evaluate the wettability of PVDF-g-

PNVP, the static contact angles on the

membrane surfaces were measured. The

pristine PVDF had a water contact angle

of about 1148. By increasing the surface

graft concentration from 0 to 0.12, the

contact angle decreased slowly from 114

Macromol. Biosci. 2007, 7, 1149–1159

� 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mabi.200700068


Table 2. Characteristics of the PVDF-g-PNVP MF membranes.

MF Membrane Contact

angle

Surface

graft concentration

of membranes

([–PNVP–]/[–CH2CF2–])surface

Maximum

pore size

Minimum

pore size

Mean

pore size

Pore

density

- mm mm mm 109 cmS2

PVDF 114 0

PVDF-g-PNVP (1) 80 0.072 1.03 0.30 0.68 4.65

PVDF-g-PNVP (2) 76 0.094 1.44 0.39 0.75 3.06

PVDF-g-PNVP (3) 70 0.110 1.82 0.57 0.86 2.28

PVDF-g-PNVP (4) 65 0.120 2.26 0.61 1.04 1.92

PVDF-g-PNVP (5) 77 0.085 1.12 0.37 0.65 3.72

PVDF-g-PNVP (6) 82 0.065 0.91 0.28 0.61 3.41

to about 65 8. The phenomenon is attributed to the

hydrophilic nature of the grafted PNVP side chains and

their aggregation in the surface region. The water contact

angles for the PVDF-g-PNVP copolymers of various

graft concentrations are summarized in Table 2. Thus,

with the increase in the PNVP graft concentration, the

surface water contact angle of the copolymer film

decreased.

Figure 4. XPS C1s core-level, wide-scan and N1s core-level spectraof (a) the (PVDF-g-PNVP)-b-PDMAEMA membranes before qua-ternization and (b) the (PVDF-g-PNVP)-b-PDMAEMA membranesafter quaternization.

Surface Compositions of the(PVDF-g-PNVP)-b-PDMAEMA Membranes

The surface compositions of the (PVDF-g-PNVP)-b-PDMA-

EMA membranes were studied by XPS. Figure 4 shows the

respective C1s core-level, wide-scan and N1s core-level

spectra of the (PVDF-g-PNVP)-b-PDMAEMA membrane sur-

faces before and after quaternization with 1-bromohexane.

As shown in Figure 4a (before quaternization), the five peak

components in the C1s core-level spectrum at BEs of about

284.6, 285.8, 286.3, 287.4 and 288.5 eV are attributable

to the C–H(PNVP and PDMAEMA), C–N(PNVP)/C–H(PVDF),

C–N(PDMAEMA), C–O(PDMAEMA)/N–C––O(PNVP) and

O––C–O(PDMAEMA) species, respectively. The N1s core-level

spectrum shows only one peak at the BE of about 399.1 eV

attributed to the C–N species. After quaternization with

1-bromohexane, as shown in Figure 4b, the peak at 286.3 eV,

attributed to C–N(PDMAEMA) species, is not detectable. A

new peak component due to C–Nþ(PDMAEMA) appeared at

286.8 eV. The results indicate that the grafted PDMAEMA

polymer chains was quaternized by 1- bromohexane. Other

peak components in the C1s core-level spectrum at Es of

about 284.6, 285.8, 287.4 and 288.5 eV are attributable to

the C–H (PNVP and PDMAEMA), C–N (PNVP)/C–H (PVDF),

C–O (PDMAEMA)/N–C––O (PNVP) and O––C–O (PDMAEMA)

species, respectively, as described above. On the other

hand, in theN1s core-level spectrum, the C–Npeak at 399.1

Macromol. Biosci. 2007, 7, 1149–1159


eV was displaced by C–Nþ at 401.6 eV generated from the

quaternization reaction.

Morphology and Pore Size Distribution of thePVDF-g-PNVP Membranes

The morphology of the PVDF-g-PNVP membranes, pre-

pared from the PVDF-g- PNVP copolymers synthesized via

the RAFT-mediated process, was studied by SEM. The pore

sizes of the MF membranes were measured from SEM

images, and themean pore size was calculated using the

relationships Dm¼ (Pn

i¼ 1 Di)/n based on 200 pores (n¼


Q. Peng et al.

1156

200). The SEM images of themembranes prepared from (a)

the pristine PVDF, and the corresponding PVDF-g-PNVP

copolymers with different graft-concentration ([–PNVP–]/

[–CH2CF2–] surface ratios of (b) 0.072, (c) 0.094, (d) 0.11, (e)

0.12 and (f) 0.08 are shown in Figure 5(a–f). For the pristine

PVDF membrane, there are almost no obvious pores

present. A much more uniform pore size distribution is

observed for themembranes from copolymers prepared by

the RAFT-mediated process. Both the mean pore size and

porosity increase with the increase in graft-copolymeri-

zation time. The higher the graft concentration of the PNVP

chains in the copolymer, the larger the pore size and the

less uniform the pore-size distribution. During the RAFT-

mediated graft copolymerization, the growth of the PNVP

side chains was well controlled, and the graft concen-

trationwas lower. As a result, the distribution of PNVP side

chains in the RAFT-mediated copolymers is more uniform.

Thus, smaller, but more-uniform pores were obtained for

membranes cast from the RAFT-mediated copolymers. The

pore size and pore-size distribution of the various PVDF-

Figure 5. SEM micrographs of the MF membranes cast at 25 8C bycopolymers, with graft concentrations ([–PNVP–]/[–CH2CF2–] surfaceprepared by the RAFT-mediated process.

Macromol. Biosci. 2007, 7, 1149–1159


g-PNVPmembranes, as measured by SEM, are summarized

in Table 2.

Antifouling Properties of thePVDF-g-PNVP MF Membranes

The protein, BSA, was allowed to adsorb on the pristine

and the various PVDF-g-PNVP membrane surfaces. The

relative amount of protein adsorbed on each surface was

derived from the XPS measurements. The nitrogen signal

from the peptide bonds was used as an indicator or marker

of the relative amount of protein adsorbed on the surface.

The relative amount of BSA adsorbed onto the surface can

be simply expressed as the ([N] – [NPNVP])/[F] ratio. Here,

[NPNVP] and [N] are the peak areas of PVDF-g-PNVP mem-

branes before and after the protein-absorption experi-

ments. The dependence of the ([N] – [NPNVP])/[F] ratio on

the PNVP graft concentration of the PVDF-g-PNVP surface

is summarized in Figure 6. Thus, even with a low surface

coverage by the PNVP chains, a significant reduction in

phase inversion from 12 wt.-% DMF solutions of the PVDF-g-PNVPratios) of (a) 0, (b) 0.072, (c) 0.094, (d) 0.11, (e) 0.12 and (f) 0.085,

DOI: 10.1002/mabi.200700068


Figure 6. Decrease in the permeation rate of the BSA solution (1mg �mL�1) as a function of time through theMillipore hydrophilic(solid square) and hydrophobic (open square) PVDF membranes,and the PVDF-g-PNVP microporous membrane with ([–PNVP–]/[–CH2CF2–])bulk¼0.045 (open triangle) and 0.082 (open circle).Inset: the dependence of the extent of BSA adsorption (expressedas the ([N] – [N]PVDF)/[F] ratio) on the PNVP graft concentration inthe PVDF-g-PNVP membrane.

Figure 7. Optical micrographs of (a) the pristine (PVDF-g-PNVP)-b-PD(PVDF-g-PNVP)-b-PDMAEMA membrane after quaternization; (c) thetimes; and (d) the (PVDF-g-PNVP)-b-PDMAEMA membrane, after qua

Macromol. Biosci. 2007, 7, 1149–1159


BSA adsorption can be achieved. Generally, for ([N] –

[NPNVP])/[F] ratios greater than 0.045, the amount of BSA

adsorbed becomes negligible. It is well known that the

hydrophobic interaction between thematerial surface and

the protein plays a very important role for non-selective

adsorption of proteins onto biomaterials.[49] As a polymer

that is soluble in both water and organic solvents, PNVP

has been the focus of numerous applications with living

tissues and extremely-low cytotoxicity, including addi-

tives, cosmetics, coatings and biomedicines. The decrease

of the BSA adsorption could be mainly ascribed to the

improvement of the hydrophilicity by the introduced

PNVP chains on the PVDF surface.

Figure 6 shows the reduction in the flux of the BSA

solution through the pristine PVDF, PVDF-g-PNVP and

(PVDF-g-PNVP)-b-PDMAEMAmembranes of different graft

concentrations as a function of time. The concentration of

the protein solution was 1 mg �ml�1. The permeation

behavior of the commercial, Millipore, hydrophilic and

hydrophobicmembranes (Duropore, DVPP, pore size¼ 0.79

and 0.57 mm) is also shown in the figure for comparison

purposes. Previous studies have characterized the Milli-

pore hydrophilicmembrane as being ‘low-protein-binding’

with respect to proteins, such as bovine serum albumin

(BSA), porcine insulin, human chorionic gonadotropin

(HCG), and sheep immunoglobulin G (IgG).[50] In the

MAEMA membrane before quaternization; (b) the freshly-prepared(PVDF-g-PNVP)-b-PDMAEMA membrane, after quaternization twoternization five times.


Q. Peng et al.

1158

present study, all of the the PVDF-g-PNVP membranes

exhibit a better antifouling property in the dynamic

fouling process than that of the pristine PVDF membrane.

For the PVDF-g-PNVP membranes with a graft concentra-

tion higher than 0.045, the antifouling ability is compar-

able to that of the commercial, ‘low-protein-binding’,

Millipore, hydrophilic membranes. It is worth noting

that, although protein fouling occurs on all these mem-

branes in the permeation experiments, the protein adsorp-

tion on the PVDF-g-PNVP membranes is mostly reversible.

After thewashing or desorption process, close to 80% of the

initial flux was restored for the PVDF-g- PNVP membranes

and the Millipore hydrophilic membrane, while only 40%

of the initial flux was restored for the hydrophobic PVDF

membrane. The antifouling property of (PVDF-g-PNVP)-b-

PDMAEMA membrane in the dynamic fouling process is

decreased to some extent, but still much better than the

commercial, hydrophobic PVDF membrane, which is

attributed to the electro-attraction of the quaternary

ammonium salt and BSA protein.

Antibacterial Characteristics of Functionalized(PVDF-g-PNVP)-b-PDMAEMA AntibacterialMembranes

Waterborne tests were performed to simulate the natural

deposition of E. coli on a substrate. Two substrates were

chosen for these tests: PVDF-g-PNVP membrane and QAS-

functionalized (PVDF-g-PNVP)-b-PDMAEMA antibacterial

membrane. Figure 7 shows the optical-microscopy results

of the above two films after waterborne antibacterial tests.

In Figure 7a, numerous distinguishable bacteria colonies

can be observed on the PVDF-g-PNVP membrane surface.

The number of colonies on the QAS-functionalized (PVDF-

g-PNVP)-b-PDMAEMA surface, as shown in Figure 7b, is

significantly less. The permanence of the antimicrobial

activity of the QAS-functionalized (PVDF-g-PNVP)-b-

PDMAEMA surface was investigated by recovering the

original surface without washing, repeated several times.

The antimicrobial capability was almost similar to its

original level. Figure 7c and 7d show the optical-micros-

copy results of the antibacterial membranes after water-

borne antibacterial tests, two and five times, respectively.

It is likely that material from the dead cells cannot

accumulate on the surface of the membrane surface via

hydrophobic interactions, whichwould evidently decrease

the antimicrobial activity of the used films.

Conclusion

PVDF with living PNVP side chains (PVDF-g-PNVP) was

successfully synthesized through the RAFT-mediated

process. MF membranes prepared by phase inversion in

Macromol. Biosci. 2007, 7, 1149–1159


an aqueous medium from amphiphilic PVDF-g-PNVP

copolymers of different graft concentrations and of dif-

ferent graft chain density showed enrichment of the

hydrophilic PNVP polymer in the surface region. A uniform

pore-size distribution of the MFmembranes was obtained,

probably due to the presence of well-defined PNVP side

chains on the PVDF main chain. The present study has

shown that graft copolymerization using the RAFT-media-

ted technique is a promising approach to the preparation

of membranes with a uniform pore-size distribution, re-

fined pore size, and increased porosity. BSA adsorption

experiments were conducted to evaluate the antifouling

property of the PVDF-g-PNVP membranes. The significant

decrease of the statistical amounts of adsorbed BSA on the

membrane indicated that the antifouling property of PVDF

was improved. Permeation fluxes of BSA solutionwere also

measured to evaluate the antifouling property of the

PVDF-g-PNVP membranes. Most important of all, the MF

membranes with surface-tethered PNVP macro chain-

transfer agents could be further functionalized via

surface-initiated block copolymerization with DMAEMA

to obtain a (PVDF-g-PNVP)-b-PDMAEMA MF membrane.

Quaternization of the tertiary amine groups of the

PDMAEMA brushes gave rise to a high concentration of

quaternary ammonium salts on the membrane surfaces.

The resulting membranes exhibited an excellent and

permanent antimicrobial activity because of their hydro-

philic surfaces.

Received: February 27, 2007; Revised: May 28, 2007; Accepted:May 31, 2007; DOI: 10.1002/mabi.200700068

Keywords: antibacterial membranes; poly(N-vinyl-2-pyrroli-done); poly(vinylidene fluoride); protein antifouling; reversibleaddition fragmentation chain transfer (RAFT)

[1] K. Li, Chem. Eng. Technol. 2002, 25, 203.[2] M. L. Yeow, Y. T. Liu, K. Li, J. Appl. Polym. Sci. 2004, 92, 1782.[3] M. M. Vestling, C. Fenselau, Mass Spectrom. Rev. 1995, 14,

169.[4] J. Mueller, R. H. Davis, J. Membr. Sci. 1996, 116, 47.[5] C.W. Sutton, C. H. Wheeler, U. Sally, J. M. Corbett,M. J. Dunn,

Electrophoresis 1997, 18, 424.[6] S. Akerman, P. Viinikka, B. Svarfvar, K. Jarvinen, K. Kontturi,

J. Nasman, A. Urtti, P. Paronen, J. Controlled Release 1998, 50,153.

[7] S. Akerman, P. Viinikka, B. Svarfvar, K. Putkonen, K.Jarvinen, K. Kontturi, J. Nasman, A. Urtti, P. Paronen,Int. J. Pharm. 1998, 164, 29.

[8] S. Akerman, B. Svarfvar, K. Kontturi, J. Nasman, A. Urtti, P.Paronen, K. Jarvinen, Int. J. Pharm. 1999, 178, 67.

[9] T. Tarvainen, B. Svarfvar, S. Akerman, J. Savolainen,M. Karhu, P. Paronen, K. Jarvinen, Biomaterials 1999, 20,2177.

[10] P. Wang, K. L. Tan, E. T. Kang, K. G. Neoh, J. Membr. Sci. 2002,195, 103.

DOI: 10.1002/mabi.200700068


[11] K. J. Kim, A. G. Fane, C. J. D. Fell, Desalination 1998, 70,229.

[12] L. E. S. Brink, S. J. G. Elbers, T. Robbertsen, P. J. Both, J. Membr.Sci. 1993, 76, 281.

[13] Y. Wang, J. H. Kim, K. H. Choo, Y. S. Lee, C. H. Lee, J. Membr.Sci. 2000, 169, 269.

[14] M. Ulbricht, G. Belfort, J. Membr. Sci. 1996, 111, 193.[15] M. Ulbricht, K. Richau, H. Kamusewitz, Colloids Surf., A 1998,

138, 353.[16] H. Iwata, M. Oodate, Y. Uyama, H. Amemiya, Y. Ikada,

J. Membr. Sci. 1991, 55, 119.[17] H. Iwata, T. Matsuda, J. Membr. Sci. 1988, 38, 185.[18] J. F. Hester, P. Banerjee, Y.-Y. Won, A. Akthakul, M. H. Acar,

A. M. Mayes, Macromolecules 2002, 35, 7652.[19] C. J. Hawker, A. W. Bosman, E. Harth, Chem. Rev. 2001, 101,

3661.[20] J. S. Wang, K. Matyjaszewski, J. Am. Chem. Soc. 1995, 117,

5614.[21] R. T. A. Mayadunne, E. Rizzardo, J. Chiefari, Y. K. Chong, G.

Moad, S. H. Thang, Macromolecules 1999, 32, 6977.[22] H. Y. Huang, T. Kowalewski, K. L. Wooley, J. Polym. Sci., Part

A: Polym. Chem. 2003, 41, 1659.[23] E. Harth, H. B. Van, V. Y. Lee, D. S. Germack, C. P. Gonzales,

R. D. Miller, C. J. Hawker, J. Am. Chem. Soc. 2002, 124, 8653.[24] K. C. Barner, H. Dalton, T. P. Davis, M. H. Stenzel, Angew.

Chem. Int. Ed. Engl. 2003, 42, 3664.[25] D. Holzinger, G. Kickelbick, J. Polym. Sci., Part A.: Polym.

Chem. 2002, 40, 3858.[26] M. Pitsikalis, S. Pispas, M. Pitsikalis, C. Vlahos, H. Iatrou,

Adv. Polym. Sci. 1998, 135, 1.[27] H. Colfen, Macromol. Rapid Commun. 2001, 22, 219.[28] R. T. A. Mayadunne, J. Jeffery, G. Moad, E. Rizzardo,

Macromolecules 2003, 36, 1505.[29] S. J. Hou, E. L. Chaikof, D. Taton, Y. Gnanou,Macromolecules.

2003, 36, 3874.[30] M. H. Stenzel, T. P. Davis, A. G. Fane, J. Mater. Chem. 2003, 13,

2090.

Macromol. Biosci. 2007, 7, 1149–1159


[31] A. Kanazawa, T. Ikeda, T. Endo, J. Polym. Sci., Part A: Polym.Chem. 1993, 31, 335.

[32] C. Z. Chen, N. C. Beck-Tan, P. Dhurjati, T. K. v. Dyk, R. A.LaRossa, S. L. Cooper, Biomacromolecules 2000, 1, 473.

[33] C. Z. Chen, S. L. Cooper, Biomaterials 2002, 23, 3359.[34] J. Lin, J. C. Tiller, S. B. Lee, K. Lewis, A. M. Klibanov,

Biotechnol. Lett. 2002, 24, 801.[35] T. Ikeda, S. Tazuke,Makromol. Chem., Rapid Commun. 1983,

4, 459.[36] S. H. Park, Y. N. Xia, Adv. Mater. 1998, 10, 1045.[37] J. J. Kim, K. Park, J Controlled Release. 2001, 77, 39.[38] G. M. R. Wetzels, L. H. Koole, Biomaterials 1999, 20,

1879.[39] M. T. Razzak, D. Darwisb, S. Zainuddinb, Radiat. Phys. Chem.

2001, 62, 107.[40] J. Chiefari, Y. K. Chong, F. Ercole, J. Krstina, J. Jeffrey, P. T. Le,

R. T. A. Mayadunne, G. F. Meijs, C. L. Moad, G. Moad, E.Rizzardo, S. H. Thang, Macromolecules 1998, 31, 5559.

[41] P. Wang, K. L. Tan, E. T. Kang, K. G. Neoh, J. Mater. Chem.2001, 11, 783.

[42] H. Strathmann, K. Kock, Desalination. 1977, 21, 241.[43] H. F. Walton, ‘‘Principles &Methods of Chemical Analysis’’, 2nd

edition, Prentice Hall, Englewood Cliffs 1964, p. 175.[44] D. Cunliffe, C. A. Smart, S. Alexander, E. N. Vulfson, Appl.

Environ. Microbiol. 1999, 65, 4995.[45] A. H. Hogt, J. Dankert, J. Feijen, J. Biomed. Mater. Res. 1986,

20, 533.[46] Q. T. Pham, R. Petiaud, M. F. Llauro, H. Waton, ‘‘Proton and

Carbon NMR Spectra of Polymers’’, John Wiley & Sons, Chi-chester 1984, Vol. 3.

[47] D. Briggs, ‘‘Surface Analysis of Polymers by XPS and StaticSIMS’’, Cambridge University Press, New York 1998, p. 65.

[48] L. Ying, E. T. Kang, K. G. Neoh, Langmuir 2002, 18, 6416.[49] G. B. Sigal, M. Mrksich, G. M. Whitesides, J. Am. Chem. Soc.

1998, 120, 3464.[50] W. R. Bowen, Q. Gan, J. Colloid Interface Sci. 1991, 144,

254.


Poly(vinylidene fluoride)-graft-Poly(N-vinyl-2-pyrrolidone) Copolymers Prepared via a RAFT-Mediated Process and their Use in Antifouling and Antibacterial Membranes

Documents