In vitro fibroblast response to polyurethane organosilicate nanocomposites

In vitro fibroblast response to polyurethaneorganosilicate nanocomposites

K. E. Styan,1 D. J. Martin,2 L. A. Poole-Warren1

1Graduate School of Biomedical Engineering, University of New South Wales, Sydney, NSW 2052 Australia2Division of Chemical Engineering, School of Engineering, University of Queensland, QLD 4072 Australia

Received 19 January 2007; revised 15 June 2007; accepted 28 June 2007Published online 9 November 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.31667

Abstract: The term nanocomposite refers to organic:inor-ganic composites where one phase, typically the inorganicphase, has dimensions on the nanoscale. Several authorshave noted the potential benefit of biomedical applicationof nanocomposite technology, and have suggested usingquaternary ammonium compounds (QAC) as an organicmodification to enhance dispersion of nanoparticles withinpolymer matrices. This study aimed to examine fibroblastresponses in vitro to a range of nanocomposites using dif-ferent organic modifiers. Composite materials were pre-pared from a polyether urethane (PEU) and various un-modified and organically modified montmorillonite (MMT)nanoparticles. QAC and amino undecanoic acid (AUA)modified-MMT were added to PEU at loadings rangingfrom �1 to 15 wt %. Composites with organically modifiedQAC and AUA particles displayed partially exfoliated and

intercalated silicate morphology, respectively. Nanocompo-sites showed increases in ultimate tensile properties formaterials with lower QACMMT loadings. However QACwas shown to significantly inhibit cell growth followingrelease from PEU-QACMMT under extraction conditionsmimicking those of the physiological environment. Materi-als containing silicate modified using AUA were cytocom-patible. The results of this study suggest that QAC may beunsuitable as organic modifiers for nanoparticles destinedfor biomedical use. Alternative modifiers based on AUAconfer equivalent dispersion and are of low toxicity.� 2007 Wiley Periodicals, Inc. J Biomed Mater Res 86A:571–582, 2008

Key words: nanocomposite; polyurethane; silicate; biocom-patibility; organic modification

INTRODUCTION

Nanocomposites are well represented in the litera-ture and have been the focus of much interest sinceresearchers discovered mechanical property trendsthat were unique in the field of composites. Specifi-cally, Usuki et al.1 synthesized a Nylon-6 compositeby polymerizing e-caprolactam within the layers of anatural silicate, montmorillonite (MMT), and showedusing X-ray diffraction (XRD) and transmission elec-tron microscopy (TEM) that the silicate was dis-persed on the nanoscale. In a subsequent study,Kojima et al.2 reported that for a composite contain-ing 4.7 wt % MMT the tensile strength, elastic modu-lus, and the heat distortion temperature increased to140, 168, and 234% (respectively) of Nylon-6 alone.

Achieving optimal dispersion of nanoparticles incomposites is of prime importance since materialswith poor nanoparticle dispersion are unlikely tohave improvements in their physical properties. Atypical MMT layer is 250 nm in two dimensions and1 nm in the other, giving an aspect ratio of 250 andtotal surface area of over 700 m2/g).3 Layers naturallystack in groups on the order of 1000 units that needto be delaminated for generation of nanocomposites.Usually, the cationic exchange capacity (CEC) of thelayered silicate is exploited to organically modify theinorganic silicate and increase its compatibility withthe organic polymer/prepolymer/monomer. Quater-nary ammonium compounds (QAC) are the mostfrequently employed organic modification (OM), how-ever theoretically any positively charged species is ca-pable of cationic exchange onto the MMT.

The current study focuses on biomedical applica-tion of polyurethane (PU) nanocomposites, where theproposed enhancement of mechanical and barrierproperties may have significant advantages over con-ventional PU. In these material systems, another criti-cal design specification is the appropriate biologicalperformance, or biocompatibility. Previous studies

Correspondence to:L.A. Poole-Warren; e-mail: [email protected] grant sponsor: The University of New SouthWalesContract grant sponsor: Australian Research Council Dis-

covery Grant ; contract grant number: DP0558561

� 2007 Wiley Periodicals, Inc.

have shown intercalated silicate dispersion, mainte-nance or improvement of mechanical properties andenhanced barrier properties in solution cast interca-lated nanocomposites fabricated using the biomedicalPU BiospanTM and Cloisite1 15A (OM: dimethylditallow ammonium chloride) or Nanomer 1.30TC(OM: octadecyl ammonium chloride).4,5 The samestudies proposed that the reported barrier propertyenhancement could be advantageous to thin-walledblood-contacting devices such as the blood sacs inventricular assist devices, where high materialstrength and low permeability are required, but didnot support this proposal with biological evaluationstudies. Other research groups have recognized thepotential for silicate particles to influence drug releasekinetics from hydrogels and have conducted studiesusing silicates organically modified using QAC tosupport their hypotheses.6–10 These hydrogel-silicatenanocomposites showed promise in the proposedapplications, however nanocomposite biocompatibilitywas not evaluated.

The components of the composite in the currentstudy were a polyether urethane (PEU) with similarchemistry to polymers typically used in medicalapplications, and a layered silicate MMT, both ofwhich are not expected to possess inherent cytotoxic-ity. Layered silicates are used in topical and oralpharmaceutical applications as fillers, and in foodproduction processes as filtering and/or clarifyingagents,11 and as long as they are not released in vivoshould have acceptable biological performance.However, the OM may have a large influence onbiocompatibility depending on its chemical proper-ties and whether it is released or exposed in vivo.QACs are the most frequent class of OM used in thenanocomposite literature. They are used commer-cially in clinical, domestic, and food manufacturingenvironments due to their antibacterial activity, how-ever, they have been associated with some cytotoxic-ity and lysis of mammalian cells.12–14 Although im-plantation of benzalkonium chloride-containing in-travenous catheters into human subjects showed noadverse recipient effects,15 the biological perform-ance must be evaluated for any composite materialcontaining OM.

Alternative OM that might not introduce biocom-patibility issues include a range of alkyl chains withprotonated terminal amine groups. Amino undeca-noic acid (AUA) was selected in the current study asit possesses amine and carboxyl functionality locatedat opposite ends of a C10 alkyl chain. When acidifiedto a pH below the isoelectric point (pI), AUA has astructure similar to that of a QAC in that it containsboth a cationic head group and a long alkyl chain.AUA and similar molecules have been used as anOM with some success in the literature,16,17 althoughthe resulting AUA-containing materials have not

been tested for cytotoxicity. The similarity of AUAwith biological macromolecules such as the alkylchain amino acids (for example amino oleic acid)suggests lowered toxicity relative to QAC.

The objective of this study was to compare QACand AUA OM methods for their impact on both thedispersion of the nanoparticles and the mammaliancell response to the material. Specifically, the studyaimed to determine the importance of OM for MMTdispersion in a PU matrix and the comparative abilityof QAC and AUA-modified MMT to disperse withinthis polymer matrix. Also studied were the in vitrocytotoxicity of leachables from the QAC and AUA-containing materials in order to determine the poten-tial for both inhibition of cell growth and cytotoxicity.

MATERIALS AND METHODS

Poly(ether)urethane

The base PEU used contained 1000 g/mol poly(tetra-methylene oxide) polyol (PTMO), 4,40 diphenylmethanediisocyanate (MDI), and 1,4-butanediol (BDO) as the chainextender and was supplied by Urethane Compounds (Mel-bourne, Australia). The components were combined in theratio 100:7.5:46.3, respectively, with 0.003 dibutyltin dilau-rate added as catalyst,18 thus the PEU contained �65 wt %PTMO (soft segment).

Nanoparticles

MMT of chemical formula Na0.33[(Al1.67Mg0.33)-Si4O10(OH)2]�H2O sourced from Southern Clay Products(Texas, USA) and CEC of 92.6 mEq/100 g19was used asthe base silicate. An organically modified MMT, Cloisite30B (referred to as QACMMT) with QAC methyl tallowbis-2-hydroxyethyl ammonium chloride as an OM wasalso sourced from Southern Clay Products [see Fig. 1(a)].The alkyl chain in this OM had a range of lengths includ-ing 5% C14, 30% C16, and 65% C18, designated QAC-C14,QAC-C16, and QAC-C18, respectively. A second organosili-cate was prepared by modifying the MMT with AUA asan OM as shown in Figure 1(b) (referred to as AUAMMT:Aldrich). Figure 1(c) shows the AUA as it would occur atphysiological pH. AUAMMT was prepared by stirringMMT in distilled, deionised Milli-Q water at 1 wt % for 24 hfollowed by addition of AUA at 0.2 g/100 g MMT (110%CEC). The solution was then acidified using 10M HCl to apH just below pH 4 (the pI of the similar NH2CH2COOHis 4.89), and left to stir vigorously for a further 24 h at�608C. Product was isolated by 10 min of centrifugation at30,100g and 158C, and then dried at 608C overnight. Driedorganosilicate was crushed in a glass mortar and pestle,sieved with a 325-mesh (45 lm) sieve, and then redried at608C for several h before use. Nanocomposites were thenprepared from the PEU and either MMT alone (no OM),QACMMT, or AUAMMT.

572 STYAN, MARTIN, AND POOLE-WARREN

Journal of Biomedical Materials Research Part A

Nanocomposite preparation

A 5 wt % solution of PEU in dimethylacetamide(DMAc, Sigma Aldrich) was prepared at 608C for 7 days.A 5 wt % suspension of QACMMT in toluene (SigmaAldrich) was also prepared. As a polar activator, methanol(Univar) and distilled deionised water were then added at30 wt % and 6 wt % of the QACMMT mass, respectively.The suspension was vigorously stirred for 30 min, andplaced in an ultrasonic water bath for 10 min. MMT andAUAMMT were added as dry nanoparticle powderdirectly to the PEU/DMAc solution.

The PEU/DMAc solution was combined with the nano-particle in appropriate proportions then stirred on a mag-netic stirrer for at least 5 h at �508C to result in compositeswith loadings ranging from �1 to 15 g nanoparticle/100 gPEU. Following stirring, mixtures were poured into glasspour plates 220 3 70 mm2. DMAc was removed at 608Cunder a dry air atmosphere at partial vacuum of 400–500mbar in a laboratory vacuum oven (Binder VD). After 24 h,the material films were peeled from the glass and subjectedto a further 24 h at the same conditions while resting on aTeflon1 sheet. Materials were designated as polymer-nano-particle-nanoparticle loading, for example PEU-QACMMT-3was a material containing 3 g QACMMT/100 g PEU. Mate-rials were prepared with nanoparticle loadings from 1 to 15wt % as detailed in Table I.

Thermal gravimetric analysis

PEU-QACMMT and PEU-AUAMMT composites wereanalyzed by thermal gravimetric analysis (TGA) to deter-mine percent organic component present. A Perkin ElmerPyris 1 TGA operated with an air atmosphere and a20 mL/min N2 purge was used to determine percent or-ganic component (the loading of OM on the MMT, that is,the OM/MMT) with OM weight loss occurring from �100to 8008C. Scans were collected from 50–8008C at 308C/min.Nanoparticle loadings were calculated as both g silicate/100 g PEU and g OM/100 g PEU using the measuredOM/MMT and Eqs. (1), (2), and (3) as shown in Table I.

OM

nanoparticle¼

OM

MMTOM

MMTþ 1

� � ð1Þ

MMT

PEU¼ nanoparticle

PEU3

OM

nanoparticle3MMT

OMð2Þ

OM

PEU¼ MMT

PEU3

OM

MMTð3Þ

Nanocomposite morphological characterization

X-ray diffraction

Diffractograms were taken on samples from cast sheetusing a Siemens D5000 wide angle XRD generating CuKa

radiation of wavelength 0.15406 nm from 1.88 to �6–108 2hat a rate of 0.088/min, a 0.018 step size, and with sourceconditions of 40 kV, 35 mA. Parameters used for the PEU-MMT materials were 18/min and 0.028 step size, sincepeaks were of higher intensity. A 2 mm anti-scatter slitand 0.05 mm receiving slit were used to increase the signalto noise ratio at the lower angles. The raw data was proc-essed using the XRD software’s fitting and deconvolutionfunction, applying the pseudo-Voigt function to manuallyselected peaks.

This data was converted to 2h data using Bragg’s Law[Eq. (4)] where k is the wavelength of the incident X-raybeam (nm), d is the distance (nm) between the silicatelayers (from the top of one layer to the top of the nextlayer), and h is the incident beam angle (radians). Dis-placement correction (shift) was conducted using Eq. (5)where s is the sample thickness (mm), R is the goniometerradius (mm, 200 mm for the Siemens D5000), and h is theincident X-ray beam angle (radians) and the shift wasadded using Microsoft Excel. Data was then converted to d(nm) [using a rearrangement of Eq. (4)]. Results areexpressed as plots of the shifted pseudo-Voigt fitted curve

Figure 1. Molecular structure of (a) QAC on Cloisite1

30B, (b) AUA at pH < 4, and (c) AUA at pH � 4 (e.g.in vivo pH 7.4).

TABLE IMaterial Loadings for PEU-MMT, PEU-QACMMT,

and PEU-AUAMMT Materials

Material

Loading

g nanoparticle/100 g PEU

g silicate/100 g PEU

g OM/100 g PEU

PEU-MMT-1 1 1 0PEU-MMT-3 3 3 0PEU-MMT-7 7 7 0PEU-MMT-15 15 15 0PEU-QACMMT-1 1 0.8 0.2PEU-QACMMT-2 2 1.6 0.4PEU-QACMMT-3 3 2.4 0.6PEU-QACMMT-7 7 5.7 1.3PEU-QACMMT-15 15 12.1 2.9PEU-AUAMMT-1 1.0 0.8 0.2PEU-AUAMMT-3 3.0 2.4 0.6PEU-AUAMMT-7 6.4 5.5 0.9

IN VITRO FIBROBLAST RESPONSE TO PU ORGANOSILICATE NANOCOMPOSITES 573

Journal of Biomedical Materials Research Part A

of XRD intensity versus 2h, and as shifted silicate spacing,d (nm). The MMT silicate spacing is referred to herein asthe natural spacing, while for QACMMT and AUAMMT,spacing is referred to as organosilicate basal spacing.When multiple peaks were detected in composite materi-als, peaks were referred to as first peak and second peak,and are presented from right to left on the graphs.

k ¼ 2d sin u ð4Þ

D2u ¼ 2s cos u

Rð5Þ

Transmission electron microscopy

A sample of �1 3 �4 mm2 dimension was inserted intoa holder and frozen in a sucrose solution using liquidnitrogen. TEM sections were then prepared using a Reich-ert Ultracut E cryo-ultramicrotome equipped with a dia-mond knife and operated at 2808C. Sections were placedon 400-mesh copper grids and analyzed in a Hitachi H-7000 TEM operated with a beam current of 100 keV. Sev-eral sections from the one sample were prepared andimaged, with at least five images at each magnificationbeing collected per material.

Tensile testing

Tensile tests were conducted on strip-shaped samples asrecommended by ASTM D 882.20 Strips of from 7 to12 mm width were cut from solution cast materials and33 mm or 45 mm gauge lengths were marked on samplesusing a soft-tipped marker. Thickness was measured usinga digital micrometer at the extreme and mid positions ofthe marked gauge length region and the three measureswere averaged. ASTM D 88220 suggests a variation inthickness of no more than 10% over the gauge length beused however this was difficult to achieve in this system.Actual thickness variations were 12% for PEU, 13% for

PEU-MMT, 19% for PEU-QACMMT and 13% for PEU-AUAMMT. An INSTRON 4302 was then used to obtainforce versus elongation data, from which ultimate strain(e), ultimate stress (s), and elastic modulus (E), were calcu-lated using Eqs. (6), (7), and (8). Strips were gripped usingthe combination of line contact and smooth rubber faces ata grip pressure of 500 kPa, and were tested at a strain rateof 100 mm/min until failure. A two-tailed student’s t-testwith the assumption of equal variances was conducted tocompare among materials. Results are expressed as mean61 standard deviation. The number of samples andbatches from which the mechanical property data wasobtained are shown in Table II.

e ¼elongation

��breakage

�length��initial

length��initial

3 100ð%Þ ð6Þ

s ¼forcejbreakageðNÞ

cross.sectional.areajinitialðm2Þ ðPaÞ ð7Þ

E ¼ stressðPaÞstrainð%Þ

100

�������linear.toe.region

ðPaÞ ð8Þ

Nanocomposite biological characterization

In vitro mammalian cell response

In accordance with AS ISO 10993,21 samples were punchcut from cast materials to give an area of 6 cm2, or weighedto give 0.1–0.2 g, per mL of extraction vehicle. Sampleswere washed with 2% Decon-90 for 24 h, rinsed with Milli-Q water for 2 days, sterilized with 100% ethylene oxide gasand allowed at least 7 days to degas. The samples wereextracted at 378C for 24 h in complete media (Eagle’s mini-mum essential media (EMEM, Sigma) þ 10 vol % fetal calf

TABLE IIMechanical Properties of all Materials

MaterialNumber of

Samples/BatchesUltimate Strain

(% 6 S.D.)Ultimate Stress(MPa 6 S.D.)

Elastic Modulus(MPa 6 S.D)

PEU 27/12 481 6 33 56 6 11 13 6 2PEU-MMT-1 4/2 469 6 5 47 6 3a 13 6 1PEU-MMT-3 5/1 439 6 21a 40 6 4a 16 6 1a

PEU-MMT-7 4/2 469 6 24 47 6 8 22 6 1a

PEU-MMT-15 4/2 431 6 11b 31 6 3a 35 6 5b

PEU-QACMMT-1 9/5 480 6 13 60 6 7 15 6 3b

PEU-QACMMT-2 4/3 503 6 17 62 6 8 19 þ 1b

PEU-QACMMT-3 5/3 476 6 10 45 6 5 23 6 1b

PEU-QACMMT-7 19/5 477 6 16 41 6 7b 30 6 7b

PEU-QACMMT-15 16/4 450 6 20b 28 6 5b 45 6 12b

PEU-AUAMMT-1 7/2 503 6 13c 57 6 6 13 þ 1PEU-AUAMMT-3 7/2 457 6 42 54 6 8 18 6 2a

PEU-AUAMMT-7 8/2 465 6 16 40 6 4b 30 6 3b

aSignificantly different to PEU (b) or corresponding (by closest loading) PEU-QACMMT (c) respectively (student’s t-test,95% confidence).

574 STYAN, MARTIN, AND POOLE-WARREN

Journal of Biomedical Materials Research Part A

serum (FCS, Gibco)) supplemented with 2% penicillin/streptomycin (P/S, CSL Biosciences). Controls of PEU, silas-tic (negative, Dow Corning), latex (positive, gloves), andduplicate extraction controls (glass vials) were included. Afibroblast monolayer (L929 mouse fibroblasts, ATCC CCL-1)was prepared by seeding at 105 cells/mL in complete mediaand incubating for 24 h at 378C and 5% CO2. After removalof media and washing of the cell layer, extraction fluid wasapplied to the cell monolayers and incubated for 48 h. Nullsamples were not exposed to extraction fluid, but werereplenished with fresh complete media and incubatedunder identical conditions.

After 48 h, monolayers were washed with Dulbecco’sphosphate buffered saline without calcium and magne-sium (DPBS, Sigma), trypsinised (0.12% trypsin, 0.02% eth-ylene diamine tetra-acetic acid (EDTA), 0.04% glucose, JRHBiosciences), and assessed by flow cytometry. Fluorescentbeads (Bangs Laboratories) were added at a ratio of �1:10with cells to allow quantitation. Fluorescent propidiumiodide (PI, Sigma-Aldrich) was added to assess cellularmembrane integrity at a working concentration of 1 lg/mL. The cell/bead mixture was analyzed on a BectonDickinson FACSort flow cytometer, with final data manip-ulation conducted using WinMDI 2.8 and Microsoft Excelsoftware. Measured events were plotted as PI fluorescenceversus forward scatter. Null samples were used to deter-mine the baseline PI fluorescence, with events fallingbelow this baseline considered PI negative (PI2) mem-brane intact cells (likely viable), and those above consid-ered PI positive (PIþ) membrane compromised cells (likelydead). The total cell numbers used for the majority of cal-culations are a sum of the PIþ and PI2 cell populations.

Where cell growth was inhibited after contact withextractables, materials were sequentially extracted over tenconsecutive days to determine whether cytotoxic leach-ables could be completely extracted. Briefly, extractionfluid was removed from n 5 3 material samples of eachtype after 24 h of extraction, and fresh extraction mediaadded. This procedure was repeated ten times and theextracts applied cells as described. Where toxicity wasobserved, modified nanoparticles alone were tested andthe results compared with unmodified MMT silicates.Briefly, nanoparticles were extracted in complete media for24 h at 0.007 g/mL, which is equivalent to extraction ofPEU-QACMMT-7 at 0.1 g/mL. After removal of the silicateby filtration with a 0.45 lm syringe filter, a standard in vitrocell growth assay as described above was conducted on theextraction fluid and dilutions of 1021 and 1022.

All assays were conducted in triplicate and wererepeated at least three times on different days for bothPEU-QACMMT and PEU-AUAMMT materials. Total cellcounts obtained for PEU from all assays conducted werepooled and averaged. Total cell counts for test materialswere then presented as a ratio with a control, which waseither the extraction controls described above or the aver-age PEU value. Plots show the test : control ratio on the y-axis, and the nanoparticle loading or material type on thex-axis. A test sample was considered bio-inhibitory or cy-totoxic if the test : control ratio was below 0.7. Membraneintegrity data is presented as the percent of total cell popu-lation that is PI2. A two-tailed student’s t-test with theassumption of equal variances was conducted to compare

the cell response to materials with equivalent nanoparticleloading.

Liquid chromatography mass spectroscopy

Where toxicity of extracts was observed, they were ana-lyzed using a reverse-phase liquid chromatography sepa-ration system (Thermofinnigan Surveyor) and an electro-spray ionization mass spectrometer (LCQ Deca XP plus)(LCMS). The eluting solvent used had initial composition97.9% water, 2% acetonitrile, and 0.1% formic acid, andwas gradually changed to 100% acetonitrile over the40 min run. The solvent flow rate was set to 200 lL/min,with 10 lL of sample being analyzed. The chromatographycolumn had a C8 stationary phase, internal dimensions100 3 2.1 mm2, and pore size 7 lm (Aquapore, AppliedBiosystems). In these studies, the QAC containing materialswere associated with toxicity and therefore chromatogra-phy traces were scanned for molecules of molecularweight (MW) corresponding to the QAC: that is 316.5(QAC-C14), 344.6 (QAC-C16), 372.6 (QAC-C18) 6 0.5 toallow for instrumental error. MMT composites were usedas controls and AUAMMT was not tested.

RESULTS

Silicate dispersion

Silicate dispersion was assessed using XRD andTEM. Raw XRD peaks were diffuse and tended toincrease in intensity and area with silicate loading asthe number of constructively interfering diffractionsincreased. It is likely that a range of silicate spacingsand/or orientations was present in all materials. Theshifted silicate spacings are shown in Table III and

TABLE IIIXRD Peaks of MMT, QACMMT, and AUAMMTNanoparticles, and PEU-MMT, PEU-QACMMT,

and PEU-AUAMMT Materials

Material

Silicate Spacing (nm)a

First peak Second peak

MMT 1.24PEU-MMT-1 1.29PEU-MMT-7 1.30PEU-MMT-15 1.25QACMMT 1.82PEU-QACMMT-1 1.74PEU-QACMMT-2 1.78 3.35PEU-QACMMT-3 1.84 3.66PEU-QACMMT-7 1.73 3.23PEU-QACMMT-15 1.77 3.32AUAMMT 1.39 n/aPEU-AUAMMT-1 1.70 2.91PEU-AUAMMT-3 1.67 2.79PEU-AUAMMT-7 1.68 2.84

aOne silicate layer thickness plus space between layers(�0.97 nm).

IN VITRO FIBROBLAST RESPONSE TO PU ORGANOSILICATE NANOCOMPOSITES 575

Journal of Biomedical Materials Research Part A

Figure 2(a,b). The natural spacing of MMT of 1.24 nmincreased marginally to �1.30 nm following inclusionin PEU at any loading. Contrarily, the MMT naturalspacing increased to 1.82 nm and 1.39 nm with theintercalation of the QAC and AUA, respectively, andmost PEU composites of QACMMT and AUAMMTcontained two peaks. PEU-QACMMT materials con-tained a first peak at the QACMMT organosilicate ba-sal spacing of (1.8 nm, and a second peak at anexpanded silicate spacing of �3.2 to �3.7 nm. Thissecond peak was not observed for the material withlowest organosilicate loading, PEU-QACMMT-1.PEU-AUAMMT materials contained a first peak thatwas significantly greater at �1.7 nm than theAUAMMT organosilicate basal spacing of 1.39 nm,and a further expanded second peak at �2.8 to

�2.9 nm. Data compares well with that reported bythe silicate manufacturer of 1.17 nm and 1.85 nm forMMT and QACMMT, respectively.19,22

Representative TEM images of PEU-MMT, PEU-QACMMT, and PEU-AUAMMT materials are pre-sented in Figure 3(a–f). TEM images of PEU controlmaterials contained no observable inclusions at anymagnification, and are thus not presented. MMT isevident as dark inclusions existing only as largeclumps on the order of �1000 to �5000 nm dis-persed throughout the sample. QACMMT is evidentas fine groupings of few silicate layers with thick-ness of approximately several nm, as well as smallclumps on the order of �100 to 500 nm. The thick-ness of the silicate groups and the number of clumpstends to increase with the QACMMT loading.AUAMMT is evident as both large and small orga-nosilicate clumps of �2000 nm and �200 to �600nm in dimension, respectively. Higher magnificationimages of PEU-QACMMT and PEU-AUAMMTmaterials are presented in Figure 4(a–d), where theoverall dispersion of QACMMT appears greater.This supports the XRD results which also suggestthat the layered silicate spacing was slightly greaterin QACMMT compared with the AUAMMT.

Mechanical properties

The ultimate mechanical properties and the elasticmoduli for all materials are presented in Table II. Incomparison to PEU control, ultimate strain wasreduced upon incorporation of MMT with someloadings resulting in significantly lowered valuesand ultimate stress was significantly lowered acrossall MMT loadings except PEU-MMT-7. Elastic modu-lus significantly increased for all MMT loadingsexcept 1 wt %. The inclusion of nanoparticles ofQACMMT resulted in maintenance of PEU ultimatestrain until the highest loading of 15 wt %. Thelower QACMMT loadings of 1 and 2 wt % causedan increase in the ultimate stress (not significant),however as the loading increased the ultimate stresswas significantly lowered compared to PEU. Itshould be noted that the host PEU employed in thiswork had an inherently high tensile strength withrespect to other published values for similar materi-als.16,23–25 Elastic modulus increased with QACMMTloading as for MMT, however the rate of increasewas significantly greater with QACMMT inclusion.Inclusion of AUAMMT caused similar changes inPEU mechanical properties as QACMMT, howeverthe increased stress at low loadings was notobserved and the elastic modulus of PEU-AUAMMT-3 was significantly lower than the PEU-QACMMT-3 modulus.

Figure 2. XRD traces for (a) MMT and QACMMT nano-particles, and PEU-QACMMT materials, and (b) MMT,QACMMT, and AUAMMT nanoparticles, and PEU-AUAMMT materials. The XRD intensity has been adjustedto increase presentation clarity. Pseudo-voigt deconvolu-tion and displacement correction have been applied.

576 STYAN, MARTIN, AND POOLE-WARREN

Journal of Biomedical Materials Research Part A

In vitro mammalian cell response

The cell response to silicone negative controls andlatex positive control materials was as expected inall assays confirming the validity of the assay. Thecell response to PEU-MMT, PEU-QACMMT, andPEU-AUAMMT materials is presented in Figure 5(a).PEU-MMT materials did not significantly inhibit cellgrowth at any MMT loading. PEU-AUAMMT mate-rials also caused minimal cell growth inhibition atall AUAMMT loadings tested, and materials withhigher AUAMMT loadings tended to cause greatercell growth reduction.

PEU-QACMMT materials caused a dramatic re-duction in the number of cells present following ex-

posure to the sample extraction fluid. The membraneintegrity of cells following exposure to PEU andPEU-QACMMT and PEU-AUAMMT materialextracts is presented in Figure 5(b). The majority ofthe cell population was not affected by exposure toPEU or PEU-AUAMMT extracts, however membranepermeability was affected by PEU-QACMMTextracts. Specifically, there appeared to be increasedmembrane damage with increased QACMMT load-ing, although this was not significant.

In vitro cell response assays were also used toassess extracts of the QACMMT nanoparticles com-pared with the unmodified MMT particles. MMTextraction fluid did not affect cell growth whileQACMMT extraction fluid caused significant reduc-

Figure 3. TEM images of (a) PEU-MMT-1 and (b) PEU-MMT-7 at 35003 showing the large clumps of silicate and a scalebar of 5000 nm. Higher magnification images were not shown as most were featureless; The remaining TEM images are at10,0003 magnification showing (c and d) PEU-QACMMT-1 and PEU-QACMMT-7 and (e and f) PEU-AUAMMT-1 andPEU-AUAMMT-7, with the scale bar representing 2000 nm, respectively.

IN VITRO FIBROBLAST RESPONSE TO PU ORGANOSILICATE NANOCOMPOSITES 577

Journal of Biomedical Materials Research Part A

tion of cell numbers. The entire cell population expe-rienced disruption to the membrane permeability toPI following exposure to QACMMT extraction fluid,except after a hundred-fold dilution of the extractwas conducted (data not shown). The results of therepeated extraction assay are presented in Figure 6.PEU-QACMMT materials showed reduced cellgrowth inhibition with successive extractions over 10days, however unacceptable levels of inhibition werestill present after the tenth successive extraction.

Liquid chromatography mass spectroscopy

Figure 7(a–f) show the partial Liquid Chromatog-raphy Mass Spectroscopy (LCMS) data for completemedia alone and complete media extracts of MMTand QACMMT nanoparticles, PEU, and PEU-QACMMT materials. There was no observable dif-ference between the data for complete media, andMMT extracts [see Fig. 7(a,b)]. The QACMMT extracthowever was markedly different, with a group ofspecies eluting between �19 and �21 min [see Fig.7(c)]. Further, peaks were detected when these elut-ing species’ mass spectrometric data was scannedfor presence of species with MW equivalent to thatof the QAC. The extraction of PEU controls alsoresulted in leachables that elute within the �19 to

�21 min residence time period [see Fig. 7(d)], how-ever these PEU leachate species did not have MWequivalent to that of QAC. PEU-QACMMT-1 andPEU-QACMMT-15 material extraction fluids con-tained species whose MW matched that of the QAC,with the amount appearing to increase with theQACMMT loading as shown in Figure 7(e,f).

DISCUSSION

Silicate dispersion

On the basis of XRD and TEM analysis PEU-QACMMT and PEU-AUAMMT materials were con-cluded to be nanocomposites of partially exfoliatedand intercalated morphology, respectively. Contra-rily, nanocomposites were not formed by incorpora-tion of MMT without OM into PEU. PEU-MMTmaterials can thus be described as unintercalatedmicrocomposites. Approximately half of the studiespresented in the literature on solution cast PU orga-nosilicate nanocomposites concluded partially exfoli-ated morphology.16,17,25–27 In TEM images of PEU-QACMMT both finely-dispersed groups (exfoliated)and small clumps or tactoids (intercalated) of silicatelayers were observed, while in images of PEU-

Figure 4. TEM images at 60,0003 magnification of (a) PEU-QACMMT-1 and (b) PEU-QACMMT-7; (c) PEU-AUAMMT-1and (d) PEU-AUAMMT-7, with the scale bar representing 200 nm. These high magnification images show the silicate tac-toids separating into individual layers.

578 STYAN, MARTIN, AND POOLE-WARREN

Journal of Biomedical Materials Research Part A

AUAMMT the finely-dispersed groups are less prev-alent and silicate-rich tactoid regions appear denser.

Favorable thermodynamics resulting from thehydroxyl functionality of both the QAC and AUAwere likely the primary driving force for silicate dis-persion in both materials, specifically hydrogenbonding between the nanoparticle OM and the hardsegment urethane group N and the soft segmentC��O��C.28,29 However, the ��CH2��CH2��OHfunctionality of the QAC likely provides a more en-ergetically favorable, hydrophilic molecular environ-ment near to the QACMMT surface, thus likelyencouraging strong intercalation by PEU and molec-ular interaction of PEU and QAC organic sequences.The position of the AUA polar ��OH functionalityon the molecule end that is farthest from the silicatesurface might actually encourage a ‘stratified’arrangement with overlaying of AUA by PEU chainsas opposed to molecular intermingling, since PEUchains are not expected to associate as readily withthis region closer to the silicate surface. The observa-

tion that the organosilicate basal spacing ofQACMMT at 1.82 nm was substantially greater thanthat of AUAMMT at 1.39 nm partly supports thispostulation. The significant difference in organosili-cate basal spacing is most likely due to the relativeamounts of OM exchanged (see Table I), the shorterlength of AUA compared to QAC, and possibly avariation in the interlayer conformation adopted bythe OM. That is, the two ��CH2��CH2��OH arms onthe QAC head group likely cause the OM to extendmore perpendicularly away from the silicate surface,while the AUA is likely to lie more parallel to thesilicate surface.

For PEU-AUAMMT materials, the organosilicatebasal spacing of AUAMMT at 1.39 nm was notobserved by XRD and instead all AUAMMT wasexpanded to at least �1.7 nm. This was not the samefor PEU-QACMMT materials where all materialshad some proportion of QACMMT nanoparticlesexisting in an undispersable state at �1.8 nm layerspacing. It is suggested that this could be due toinconsistent silicate size, shape, charge density, orOM surface coverage. All PEU-AUAMMT and PEU-QACMMT materials also displayed a secondexpanded layer spacing at �2.85 nm and �3.5 nm,respectively. The exception was PEU-QACMMT-1which displayed only the organosilicate basal spac-ing of �1.8 nm, and did not display a secondexpanded spacing. Supported by supplementaryTEM evidence, the absence of an expanded peak inPEU-QACMMT-1 indicates a degree of QACMMTdispersion beyond the limit of detection of the XRD.This proposition is also supported by the analysisconducted during this research being extremely rig-

Figure 5. The effect of 24 h, 378C complete media extractsof PEU-MMT, PEU-QACMMT, and PEU-AUAMMT materi-als on (a) the growth of mouse fibroblasts, and (b) mousefibroblast membrane permeability to PI. Mean 6 1 standarddeviation. Number of assay replicates shown below data.

Figure 6. The effect of 24 h, 378C complete media extractsof QACMMT on the growth of mouse fibroblasts. Thecomplete media was collected and then replaced withfresh complete media each day for 10 consecutive days.Also shown is data for control materials Silastic (S) andLatex (L). Mean 6 one standard deviation. Assay was notrepeated (n 5 1); standard deviation generated from intra-assay replicates.

IN VITRO FIBROBLAST RESPONSE TO PU ORGANOSILICATE NANOCOMPOSITES 579

Journal of Biomedical Materials Research Part A

orous in terms of scatter minimization and data col-lection time (0.088/min, 0.018 step size) when com-pared with others presented in the literature.16,30,31

That only PEU-QACMMT-1 has all silicate (exceptthe hypothesized undispersable silicate at �1.8 nm)dispersed beyond the XRD detection limit is sup-ported by the theoretical argument of Vaia et al.32 Inthis study, it was reported that in order for organosi-licate to have a chance of achieving complete exfolia-tion in a given host polymer, spatial restrictionsimpose that the volume fraction of silicates must beless than 1/aspect ratio. The natural MMT employedhere has an average aspect ratio in the order of100,33 meaning that true exfoliation will only occurat volume fractions of 1% or less where ‘‘over-crowding’’ of the silicate layers will not limit disper-sion. In the case of PEU-AUAMMT, the presence ofthe second expanded silicate spacing for all loadingssuggests that no material achieved complete exfolia-

tion, which was in agreement with silicate dispersionobserved in TEM images.

The majority of literature studies on PU nanocom-posites utilize a QAC organic modifier, howeverHan et al.16 recently reported on an AUA-containingnanocomposite comprising a 5 wt % composite of anAUAMMT with a soft grade MDI/polypropyleneoxide/BDO PEU. On the basis of the absence of anXRD peak above 1.58 2h and TEM imaging, an exfo-liated nanocomposite was concluded. The XRD oper-ating parameters were not detailed in the study andonly a high magnification TEM was included. As aresult it is difficult to compare between the currentstudy and the study by Han et al. However, thelikely cause of the seemingly better silicate disper-sion achieved by Han et al. compared with that ofthis study is the softer PEU grade and the use of ul-trasonic energy during nanocomposite processing.Softer polymers often display improved dispersion

Figure 7. LCMS data for (a) complete media, (b) MMT extract, (c) the extract from QACMMT alone, (d) PEU extract, and(e and f) PEU-QACMMT-1 and PEU-QACMMT-15 extracts. For each sample, two sets of data are shown; the upper dataset corresponds to all eluting species, while the lower data set is specific to the MW of the QAC. Only the region of inter-est is shown. X-axis is time (min) to elute through chromatography column. Y-axis is amount of substance eluting.

580 STYAN, MARTIN, AND POOLE-WARREN

Journal of Biomedical Materials Research Part A

and properties,4,16,30 and ultrasound has been seenin this laboratory to result in damage and scission ofthe PEU chain. Thus, the current study and that ofHan et al. have almost certainly been conducted onsignificantly different polymer systems and thus can-not be readily compared.

Mechanical properties

PEU-QACMMT and PEU-AUAMMT materials dis-played similar mechanical integrity except that PEU-QACMMT had greater mechanical strength at thelower loadings. Both materials displayed propertiesthat were greater than that of PEU-MMT. Reducedultimate properties for PEU-MMT materials, particu-larly ultimate stress, was likely due to materialdefects caused by chemical incompatibility at theinterface between PEU and MMT. It is understand-able given the proposed partially exfoliated silicatemorphology of PEU-QACMMT materials that thesematerials display the most improved mechanicalstrength. In general, properties began to declinebeyond 2–3 wt % nanoparticle. It is possible that asthe silicate loading increases above 2 wt %, the‘‘overcrowding’’ of silicate layers restricts thedynamic motion of polymer chains predicted toallow polymer stress shielding and stress transfer tothe silicate.32 Also, at the higher organosilicate load-ings, the poorly dispersed silicate tactoids becomemore frequent and larger, and it is likely that poorinterfacial connections between the PEU and theorganosilicate simply lead to a higher chance of voidformation and thus poorer tensile properties whenunder tensile stress. The observed trends upon addi-tion of QACMMT are in agreement with studies byFinnigan et al. on the same PEU nanocomposite sys-tem, although in this study the lowest QACMMTloading tested was 3 wt %.27

In vitro fibroblast response

Experimental results suggested that the QAC OMcaused inhibition of cell growth and loss of cellularmembrane integrity. QAC release was strongly sug-gested by LCMS analysis of extraction fluids, whereit was shown that species of equivalent MW to thatof QAC molecules were observed only in extractionsof materials containing QACMMT. QACs have beenshown to interact with cellular membranes renderingthe cell incapable of maintaining homeostasis andthus slowing cell growth and even causing celldeath.12,34 That QAC leached from the nanocompo-sites in sufficient quantities to cause cell growth dis-ruption is not surprising. QAC is associated with thesilicate via a relatively weak electrostatic interaction

as opposed to being covalently bound, and therelease of small MW substances from similar PU hasbeen studied previously.35 Further, Edwards et al.36

recently reported that excess QAC is present in Cloi-site organosilicates due to incomplete washing afterthe alkylammonium exchange process by the manu-facturer, and such an excess would render moreQAC available for leaching than expected.

Since the strength of AUA interaction with silicateis likely on the same order of magnitude as that forQAC with silicate, and since AUA is a smaller mole-cule than the QAC it is probable that a similar oreven more rapid release pattern occurred for AUAas for QAC. This suggests that AUA released is lesscytotoxic than QAC released. At pH above the pI theAUA amine group is neutral as opposed to posi-tively charged, and the carboxyl group is negativelycharged, as shown in Figure 1(c). Also, the anti-bac-terial activity of QAC is generally considered to bedue to the cationic ammonium head group facilitat-ing insertion of the hydrophobic alkyl chain into thecellular membrane, disrupting maintenance of cellhomeostasis, and leading to cell death.37 Thus, at pH7.4 as in cell culture media, the absence of cationicgroups on AUA may explain its reduced capacityfor insertion in the cellular membrane leading to celldeath. Further, the AUA molecule is C11 in lengthwhile the QACMMT QAC is C14 to C18 in length,and longer alkyl chain lengths have been shown tohave greater anti-bacterial efficacy.38

Although the in vitro cytotoxicity of PEU-QACMMT materials has been demonstrated by thisstudy, the potential for successful in vivo applicationis unknown. In the proposed applications of PUnanocomposites, the release of nanoparticles isunlikely. However, it is clear that the OM must becarefully evaluated for cytotoxicity given the highlikelihood of its release from the material in vivo.

CONCLUSIONS

OM of MMT nanoparticles is essential to achievenanoparticulate dispersion. Unmodified MMT didnot disperse on the nanoscale, while both QAC andAUA organically modified MMT displayed interca-lated or partially exfoliated dispersion. Compositesbased on silicates dispersed using QACs showed sig-nificant cell growth inhibition and disruption of cellmembranes. However, an intercalated dispersionwith no associated cell growth inhibition wasobserved for an AUA modified MMT. AUA modi-fied MMT dispersed in PU represents the first gener-ation of nanocomposite elastomers with improvedmechanical properties and with potential for in vivobiomedical usage.

IN VITRO FIBROBLAST RESPONSE TO PU ORGANOSILICATE NANOCOMPOSITES 581

Journal of Biomedical Materials Research Part A

The assistance of the following University of New SouthWales staff members is appreciated: Sigrid Fraser (Electronmicroscopy Unit, TEM), Mark Raftery (Bioanalytical MassSpectroscopy Facility, LCMS), and Yu Wang (Materials Sci-ence, XRD).

References

1. Usuki A, Kojima Y, Kawasumi M, Okada A, Fukushima Y,Kurauchi Y, Kamigaito O. Synthesis of nylon 6-clay hybrids.J Mater Res 1993;8:1179–1184.

2. Kojima Y, Usuki A, Kawasumi M, Okada A, Fukushima Y,Karauchi T, Kamigaito O. Mechanical properties of nylon 6-clay hybrids. J Mater Res 1993;8:1185–1189.

3. Beall GW. New conceptual model for interpreting nanocom-posite behaviour. In: Beall GW, Pinnavaia TJ, editors. Poly-mer-Clay Nanocomposites. Brisbane: Wiley; 2000. p 267–279.

4. Xu R, Manias E, Snyder AJ, Runt J. New biomedical poly(ur-ethane urea)-layered silicate nanocomposites. Macromole-cules 2001;34:337–339.

5. Xu R, Manias E, Snyder AJ, Runt J. Low permeability bio-medical polyurethane nanocomposites. J Biomed Mater ResA 2003;64:114–119.

6. Cypes SH, Saltzman WM, Giannelis EP. Organosilicate-poly-mer drug delivery systems: Controlled release and enhancedmechanical properties. J Controlled Release 2003;90:163–169.

7. Lee WF, Chen YC. Effect of bentonite on the physical proper-ties and drug-release behavior of poly(AA-co-PEGMEA)/ben-tonite nanocomposite hydrogels for mucoadhesive. J ApplPolym Sci 2004;91:2934–2941.

8. Lee WF, Chen YC. Effect of hydrotalcite on the physicalproperties and drug-release behavior of nanocompositehydrogels based on poly[acrylic acid-co-poly(ethylene gly-col) methyl ether acrylate] gels. J Appl Polym Sci 2004;94:692–699.

9. Lee WF, Fu YT. Effect of montmorillonite on the swellingbehavior and drug-release behavior of nanocomposite hydro-gels. J Appl Polym Sci 2003;89:3652–3660.

10. Lee W-F, Jou L-L. Effect of the intercalation agent content ofmontmorillonite on the swelling behaviour and drug releasebehaviour of nanocomposite hydrogels. J Appl Polym Sci2004;94:74–82.

11. http://www.ima-na.org/about_industrial_minerals/bentonite.asp, 3rd Feb 2006.

12. Fabreguette A, Hua SZ, Lasne F, Damour O. Cytotoxicityevaluation of antiseptics commonly used on cultured fibro-blasts and keratinocytes. Pathologie Biologie 1994;42:888–892.

13. Steinsvag SK, Bjerknes R, Berg OH. Effects of topical nasalsteroids on human respiratory mucosa and human granulo-cytes in vitro. Acta Oto-Laryngologica 1996;116:868–875.

14. Giovannelli D, Abballe F. Aliphatic long chain quaternaryammonium compounds analysis by ion-pair chromatographycoupled with suppressed conductivity and UV detection inlysing reagents for blood cell analysers. J Chromatogr A 2005;1085:86–90.

15. Moss HA, Tebbs SE, Faroqui MH, Herbst T, Isaac JL, BrownJ, Elliot TSJ. A central venous catheter coated with benzalko-nium chloride for the prevention of catheter-related microbialcolonization. Eur J Anaesthesiol 2000;17:680–687.

16. Han B, Cheng AM, Ji GD, Wu SS, Shen J. Effect of organo-philic montmorillonite on polyurethane/montmorillonite nano-composites. J Appl Polym Sci 2004;91:2536–2542.

17. Chen T-K, Tien Y-I, Wei K-H. Synthesis and characterizationof novel segmented polyurethane/clay nanocomposites. Poly-mer 2000;41:1345–1353.

18. Cartledge J. Urethane Compounds, Melbourne. 2002.19. http://www.scprod.com/product_bulletins/PB Cloisite NAþ.pdf,

8th Nov 2005.20. ASTM-D882-02. Standard test method for tensile properties

of thin plastic sheeting. 1997.21. AS ISO 10993-2002. Biological evaluation of medical devices.

SAI Global Limited.22. http://www.scprod.com/product_bulletins/PB Cloisite 30B.pdf,

10th May 2005.23. Tien YI, Wei KH. Hydrogen bonding and mechanical properties

in segmented montmorillonite/polyurethane nanocomposites ofdifferent hard segment ratios. Polymer 2001;42:3213–3221.

24. Mishra JK, Kim I, Ha CS. New millable polyurethane/orga-noclay nanocomposite: Preparation, characterization and pro-perties. Macromol Rapid Commun 2003;24:671–675.

25. Tien YI, Wei KH. High-tensile-property layered silicates/poly-urethane nanocomposites by using reactive silicates aspseudo chain extenders. Macromolecules 2001;34:9045–9052.

26. Chang J-H, An YU. Nanocomposites of polyurethane withvarious organoclays: Thermomechanical properties, morphol-ogy, and gas permeability. J Polym Sci B: Polym Phys2002;40:670–677.

27. Finnigan B, Martin DJ, Halley PJ, Truss RT, Campbell K.Morphology and properties of thermoplastic polyurethanenanocomposites incorporating hydrophilic layered silicates.Polymer 2004;45:2249–2260.

28. Balazs AC, Singh C, Zhulina E. Modelling the interactionsbetween polymers and clay surfaces through self-consistentfield theory. Macromolecules 1998;31:8370–8381.

29. Vaia RA, Giannelis EP. Polymer melt intercalation in organi-cally-modified layered silicates: Model predictions andexperiment. Macromolecules 1997;30:8000–8009.

30. Xu R, Manias E, Snyder AJ, Runt J. Low permeability bio-medical polyurethane nanocomposites. J Biomed Mater ResA 2003;64:114–119.

31. Chen TK, Tien YI, Wei KH. Synthesis and characterization ofnovel segmented polyurethane clay nanocomposite via poly(epsilon-caprolactone)/clay. J Polym Sci A: Polym Chem1999;37:2225–2233.

32. Vaia RA, Liu W, Koerner H. Analysis of small-angle scatter-ing of suspensions of organically modified montmorillonite:Implications to phase behaviour of polymer nanocomposites.J Polym Sci B: Polym Phys 2003;41:3214–3236.

33. Fornes TD, Paul DR. Modeling properties of nylon 6/claynanocomposites using composite theories. Polymer 2003;44:4993–5013.

34. Nagamune H, Maeda T, Ohkura K, Yamamoto K, NakajimaM, Kourai H. Evaluation of the cytotoxic effects of bis-quater-nary ammonium antimicrobial reagents on human cells. Toxi-col In Vitro 2000;14:139–147.

35. Ikeda Y, Kohjiya S, Takesako S, Yamashita S. Polyurethaneelastomer with PEO-PTMO-PEO soft segment for sustainedrelease of drugs. Biomaterials 1990;11:553–560.

36. Edwards G, Halley P, Kerven G, Martin D. Thermal stabilityanalysis of organo-silicates, using solid phase microextractiontechniques. Thermochimica Acta 2004;429:13–18.

37. Gilbert P, Moore LE. Cationic antiseptics: Diversity ofaction under a common epithet. J Appl Microbiol 2005;99:703–715.

38. Campanac C, Pineau L, Payard A, Baziard-Mouysset G, RoquesC. Interactions between biocide cationic agents and bacterialbiofilms. Antimicrobial Agents Chemother 2002;46:1469–1474.

582 STYAN, MARTIN, AND POOLE-WARREN

Journal of Biomedical Materials Research Part A