Acta Biomaterialia 8 (2012) 2243–2253

Contents lists available at SciVerse ScienceDirect

Acta Biomaterialia

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

In vivo biostability of polyurethane–organosilicate nanocomposites

Katie E. Styan a,⇑,1, Darren J. Martin b, Anne Simmons a, Laura A. Poole-Warren a

a Graduate School of Biomedical Engineering, University of New South Wales, NSW 2052, Australiab Division of Chemical Engineering, School of Engineering, University of Queensland, QLD 4072, Australia

a r t i c l e i n f o a b s t r a c t

Article history:Received 4 October 2011Received in revised form 1 February 2012Accepted 1 March 2012Available online 8 March 2012

Keywords:NanocompositePolyurethaneSilicateOrganosilicateBiostability

1742-7061/$ - see front matter � 2012 Acta Materialhttp://dx.doi.org/10.1016/j.actbio.2012.03.004

⇑ Corresponding author. Tel.: +61 293853905; fax:E-mail address: katie.styan@gmail.com (K.E. Styan

1 Present address: Commonwealth Scientific and IndMaterials Science and Engineering, VIC 3069, Australia

Organically modified layered silicates were incorporated into a polyether soft-segment polyurethane toform composites of at least delaminated morphology. The primary organic modifier was a quaternaryammonium compound; however, one composite included an alternative amino undecanoic acid-modi-fied silicate. The composites’ biostability was assessed in an in vivo ovine model over a period of 6 weeks.Attenuated total reflectance–Fourier transform infrared analysis and semi-quantitative scanning electronmicroscopy image rating indicate a significant enhancement of the base polyurethane biostability withthe inclusion of silicate at 3 wt.%. The potential effect at 15 wt.% was confounded by probable leachingof the quaternary ammonium compound affecting the tissue response. The amino undecanoic acid com-posite compared favourably with the quaternary ammonium compound composite of equivalent silicateloading, and offers the promise of a more favourable tissue response.

� 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction

Polyurethane (PU) has, historically, been disadvantaged byhaving poor biostability in vivo. Such degradation is significant asit introduces a risk of loss of device function and/or compromisedbiocompatibility. Poly(ester)urethanes were among the earliestbiomedical PUs. The problem of the ester’s hydrolytic instabilitywas initially addressed by poly(ether)urethanes (PEUs), but thesehave since been shown to be susceptible to degradation via oxida-tion of the ether linkages [1,2], and are thus currently limited toshort-term applications in vivo. A PU originating from past workin our laboratory includes a polydimethylsiloxane (PDMS; –(CH3)3–Si–O–[Si–O–(CH3)2]n–Si–(CH3)3–) soft segment and hassignificantly increased biostability suitable for long-term in vivoapplications [3–5].

A current area of research in our laboratories is organosilicatenanocomposites. Nanocomposites are well represented in the liter-ature and have been the focus of much interest since researchersdiscovered mechanical property trends that were unique in thefield of composites [6,7]. Nanocomposites have been explored pre-dominantly for industrial application, with few studies aiming toexploit the enhancement of mechanical and barrier properties forapplication in biomaterials. However, increasingly of late, studieshave explored the potential of silicates to modulate drug releasefrom hydrogels (e.g. [8,9]).

ia Inc. Published by Elsevier Ltd. A

+61 296632108.).ustrial Research Organisation,.

In the present study, we explore the hypothesized potential oflayered silicates in a PEU matrix to mitigate in vivo degradation.Organosilicate nanocomposites have not been assessed in thepublished literature for in vivo biostability as far as we are aware.Our hypothesis was based on the fact that one layer of montmo-rillonite (MMT), a commonly used silicate, is 250 nm in twodimensions and 1 nm in the other, giving an aspect ratio of 250and total surface area of over 700 m2 g–1 [10]. Layers naturallystack in groups on the order of 1000 units, but these can be atleast partially delaminated by exploiting the cationic exchangecapacity (CEC) of the layered silicate to organically modify theinorganic silicate and increase its compatibility with the organicpolymer matrix. It is now well accepted that the dispersed highsurface area particles confer barrier effect properties to the poly-mer. Accordingly, we hypothesized that the included partiallydelaminated silicate layers would act as a barrier to attack and in-gress of degradation species in vivo, thus protecting ether linkagesand inhibiting degradation.

2. Materials and methods

2.1. Poly(ether)urethane

PEU with chemical structure illustrated in Fig. 1 was used in thisstudy. The polymer was supplied by Urethane Compounds (Mel-bourne, Australia) and contained�65 wt.% 1000 g mol–1 poly(tetra-methylene oxide) polyol ether soft segment, 4,40-diphenylmethanediisocyanate and 1,4-butanediol as the chain extender. The compo-nents were combined in the molar ratio 100:7.5:46.3, respectively,

ll rights reserved.

Fig. 1. Molecular structure of PEU.

Table 2Biostability-relevant infrared absorption wave numbers.

Wavenumber (cm�1) Description of proposed peak assignment

Literature Experimental a

1038 1038b Stretching of the Si–O–Si bonds in MMT1081 1075 ± 9.49 Aliphatic symmetric stretching of the hard

segment ether C–O–C1100 1099 ± 1.31 Aliphatic asymmetric stretching of the soft-

segment ether C–O–C1174 1176 ± 0.17 Bending of branched ether C–O–C

degradation product1208 1208b Twisting of CH2

1219 1220 ± 0.77 Stretching of C–N1251 1251 ± 0.24 Wagging of CH2

1365 1367 ± 0.16 Bending of CH2

1446 1446b Bending of CH2

1529 1530 ± 0.69 Stretching of C–N and bending of N–H1597 1597 ± 0.60 Stretching of C=C in aromatic ring1703 1699 ± 0.57 Stretching of hard segment C=O hydrogen

bonded to amine

2244 K.E. Styan et al. / Acta Biomaterialia 8 (2012) 2243–2253

with 0.003 dibutyltin dilaurate added as a catalyst. This is modelledon Pellethane� 2363-80A.

2.2. Layered silicates

A commercially available organically modified MMT, Cloisite�

30B (QACMMT; Southern Clay Products), was used. QACMMT hadthe quaternary ammonium compounds (QAC) methyl tallowbis-2-hydroxyethyl ammonium chloride as an organic modifier.This QAC is derived from animal adipose tissue, with the tallowindicating an alkyl chain of varying carbon atoms in length: 5% C14,30% C16 and 65% C18. As a control, MMT (Na0.33[(Al1.67Mg0.33)-Si4O10(OH)2]�H2O; Southern Clay Products), of CEC 92.6 m Eq/100 g, was organically modified with amino undecanoic acid(AUA; Aldrich) by a method described previously [11]. The organi-cally modified MMT thus prepared was referred to as AUAMMT.

2.3. Composite preparation

Composites were prepared from PEU and either (i) QACMMT or(ii) AUAMMT by a solvent casting method. Composites with load-ings ranging from �1 to 15 g nanoparticle/100 g PEU (1–15 wt.%)were prepared as detailed in Table 1. For details of the preparationmethods and characterization of the resultant material, the readeris referred to our previous publication [11].

2.4. Composite biostability evaluation

2.4.1. In vivo ovine modelMethods were based on those published previously [3–5].

Briefly, dumb-bells were punch-cut from composite materialsand strained to 150% over polymethylmethacrylate (PMMA) hold-ers. The holders were engraved and tracked by a random code thatwas not broken until completion of the analysis. Strained sampleswere washed for 30 min in 2% Decon90 with manual agitation,rinsed at least six times in Milli-Q water, subjected to EtO steriliza-tion and then degassed at room temperature for at least 7 days. Allsamples were taken from a single material batch, with the numberof samples tested detailed in the following sections.

Ethics approval was granted by the Animal Care and EthicsCommittee at The University of New South Wales (ACE #03/81).Samples were implanted subcutaneously in the dorsal region ofsheep (healthy crossbred males, weighing 40–60 kg and �2 years

Table 1Material loadings for PEU, PEU-QACMMT and PEU-AUAMMT materials.

Material Loading (g/100 g PEU)

Nanoparticle Silicate OM

PEU 0 0 0PEU-QACMMT-1 1.0 0.8 0.2PEU-QACMMT-3 3.0 2.4 0.6PEU-QACMMT-15 15.0 12.1 2.9PEU-AUAMMT-1 0.9 0.8 0.1

of age) for a time period of 6 weeks. Replicate samples were im-planted into each of four sheep. One sample was assigned for his-tological analysis and another randomly selected post-assay forFourier transform infrared (FTIR) analysis, while four samples werereserved for scanning electron microscopy (SEM) imaging and rat-ing. Unimplanted control samples were also prepared (three permaterial) for FTIR and SEM analysis. Explanted samples werewashed in 0.1 M NaOH for 5 days to remove attached biological tis-sue and proteins, rinsed in Milli-Q water, washed in 2% Decon-90for 3 days to remove fatty deposits, rinsed again with Milli-Q waterfor 4 h and finally dried in a laminar flow hood. Unimplanted con-trol samples were washed by the same process for consistency.Samples remained strained on the PMMA holders during both FTIRand SEM analysis.

2.4.2. Scanning electron microscopySamples were coated with a �20–30 lm thick layer of conduc-

tive gold using a Dynavac SC150 sputter coater and analysed usinga Cambridge Stereoscan 360 SEM operated at an accelerating volt-age (extrahigh tension) of 20 kV, with a typical working distance of�20–27 mm. The system for objectively rating degradation em-ployed was developed by Martin et al. [4]. SEM images from fiveset locations along the neck region of each sample were collectedat �50, �150, and �500 magnification per location, for each offour replicate samples, taken one per sheep. One image at �10

1730 1728 ± 0.43 Stretching of non-hydrogen bonded hardsegment C=O

2797 2795 ± 0.41 Symmetric stretching of C–H in –CH2O–2852 2852 ± 0.36 Asymmetric stretching of CH2CH2922 2919b Dichroic parallel stretching of CH2

2938 2938 ± 0.88 Asymmetric stretching of CH2

3305 3322 ± 0.79 Stretching of N–H hydrogen bonded tocarbonyl

a Average of all materials (unimplanted and implanted), not correcting for the Si–O–Si peak.

b Experimental peak not well defined, so the theoretical peak location was taken.

K.E. Styan et al. / Acta Biomaterialia 8 (2012) 2243–2253 2245

magnification was also collected at the centre location. Two exam-iners, blinded to the sample identity, individually rated all imagesaccording to a rating scheme, whereby degradation at �10, �50,�150 or �500 resulted in a score of 10, 5, 2 or 1, respectively. Amaterial rating ranging from 0 (no surface feature or degradationevident in any image) to 50 (surface features or degradation evi-dent in all images or sample broken) was determined, and the rat-ings from the examiners were averaged for plotting. Error bars are1 standard deviation of all data points. Statistical methods suitablefor non-normal samples were applied. Briefly, where a Kruskal–Wallisone-way analysis of variance indicated a statistically signif-icant difference, multiple comparisons of test samples with controlsamples were performed using Dunnett’s method. For normallydistributed data, Student’s t-test may also have been performed.

2.4.3. Attenuated total reflectance–Fourier transform infrared(ATR-FTIR) spectroscopy

ATR-FTIR spectroscopy was conducted on a single sample from4000 to 500 cm�1, at a resolution of 2 cm�1 for a total of 64 scans,on a Bruker IFS 55/S spectrometer. A diamond crystal ATR attach-ment was used. Absorbance wave numbers associated withpoly(ether)urethane and silicates were collected from varioussources [2,3,12–14] and are listed in Table 2. The ATR-FTIR trace

Fig. 2. SEM images of (a) PEU, (b) PEU-QACMMT-1, (c) PEU-AUAMMT-1, (d) PEU-QMagnification is �10 (the scale bar is 2 mm). The sample chosen is the worst case.

was baselined at numerous locations across the spectral rangeusing the supplied software (Opus), and the data normalized tothe C–C bond of the PEU aromatic ring at 1414 cm�1[12] usingMicrosoft Excel. Also shown in Table 2 are the wave numberassignments used in the data processing. Data are presented astraces of absorbance intensity vs. wave number, or as a comparisonof the intensities at important peak wave numbers. In the latter, fora particular wave number of interest the deviation of the im-planted test material absorbance intensity from the unimplantedPEU control absorbance was calculated and then taken as a ratioto the unimplanted PEU control absorbance, as shown in Eq. (1),to enable simultaneous plotting of data for numerous absorptionwave numbers.

deviation ¼ Atest;implanted � APEU;unimplanted

APEU;unimplanted� 100% ð1Þ

2.5. In vivo inflammatory response

2.5.1. HistologySamples for histology were explanted with fibrous capsules in-

tact and placed directly into 10% neutral buffered formalin. The

ACMMT-3 and(e) PEU-QACMMT-15 materials implanted in sheep for 6 weeks.

2246 K.E. Styan et al. / Acta Biomaterialia 8 (2012) 2243–2253

sample side of the holder within the capsule was later determinedand marked with Indian ink. An annulus of the capsule was thencut and histological sectioning and staining conducted. Briefly, sec-tions were cut transverse to the direction of sample strain and atabout the point halfway longitudinally along the dumb-bell. Sec-tions were treated for 12 h in a Miles Tissue Tek Vacuum Infiltra-tion Processor to exchange aqueous regions with paraffin wax,embedded in further paraffin wax, sectioned at 5 lm, mountedon glass microscope slides, and stained with haematoxylin and eo-sin Y (H&E). Note that the PMMA holder was removed prior to sec-tioning, and that the sample may have been lost with the PMMAholder and/or during the wash process.

3. Results

3.1. Biostability

SEM images of explanted materials displaying the qualitativelyworst degradation observable at �10 magnification are shown inFig. 2. A higher-magnification image of the far left of each of theFig. 2 samples is shown in Fig. 3. Severe cracks and degradation

Fig. 3. SEM images of (a) PEU, (b) PEU-QACMMT-1, (c) PEU-AUAMMT-1, (d) PEU-QMagnification is �150 (the scale bar is 200 lm). The sample is the same as in Fig. 3.

are evident on implanted PEU control materials. Much less severecracks are seen on 1 wt.% composites.

The average rating for each of four sample replicates per mate-rial (one per animal) is displayed in Fig. 4a for PEU-QACMMT andFig. 4b for PEU-AUAMMT. Unimplanted materials received ratingsof below 10. That some of these unimplanted silicate-containingmaterials received a non-zero score was due to surface discolor-ation rather than degradation, which had been observed previouslyin other studies in our laboratory. In some cases, this surface dis-coloration would have affected the implanted scores too, but, ifso, it would have artificially elevated the degradation score so thatthe degradation would have appeared worse than in actuality, andthus would not have compromised the assay. The implanted PEUcontrol material received an average rating approaching 50, withsome sample replicates achieving a score of 50, which indicatesthat degradation was observed in most regions and for most mag-nifications. Implanted PEU-QACMMT-1 had an average rating notstatistically significantly greater than PEU due to more congruentsample replicate ratings and surface discoloration; however, asthe silicate loading was increased to 3and 15 wt.%, a statisticallysignificant reduction in rating was seen. The average rating of

ACMMT-3 and (e) PEU-QACMMT-15 materials implanted in sheep for 6 weeks.

Fig. 4. Degradation ratings of SEM images of unimplanted and implanted PEUcontrol and (a) PEU-QACMMT and(b) PEU-QACMMT-1 compared to PEU-AUAMMT-1. A rating of 50 or 0 indicates visual evidence of degradation at all or nomagnifications, respectively. ⁄⁄Statistically significant difference (p < 0.05). Datapoints are the average of one rating from each of two raters, for four replicatesamples of each material. Values are mean ± one standard deviation.

K.E. Styan et al. / Acta Biomaterialia 8 (2012) 2243–2253 2247

PEU-QACMMT-15 is almost zero. The SEM image ratings shown inFig. 4b seem slightly better for PEU-AUAMMT-1 than for PEU-QACMMT-1, although not statistically significant.

The ATR-FTIR traces for unimplanted PEU, PEU-QACMMT andPEU-AUAMMT are shown in Fig. 5a (1800–800 cm�1) and Fig. 5b(3400–2600 cm�1). Unimplanted PEU-QACMMT-1, PEU-AUAMMT-1 and PEU-QACMMT-3 show similar infrared absor-bance to unimplanted PEU at all wave numbers, validating thedirect comparison of implanted composites with unimplantedPEU as a control. PEU-QACMMT-15 had an obvious divergence at�1100–900 cm�1 and more subtle divergences at 2922 and2938 cm�1.

The ATR-FTIR traces for unimplanted PEU control, implantedPEU control, and implanted PEU-QACMMT and implanted PEU-AUAMMT composites are presented in Fig. 6a (1340–800 cm�1),6b (1640–1340 cm�1), 6c (1800–1640 cm�1) and 6d (3400–2600 cm�1). Figs. 7 and 8 collate the most significant informationcontained in Fig. 6 for PEU-QACMMT and PEU-AUAMMT, respec-tively, by presenting the absorbance of each composite at a partic-ular wave number as a quantitative deviation from the absorbanceof unimplanted PEU control at that same wave number. Thus, bythe definition used in Eq. (1), a ‘‘negative deviation’’ occurs whenthe test material absorbance is lower than the unimplanted PEUcontrol absorbance (e.g. for 1081 cm�1 in Fig. 7a) and a ‘‘positivedeviation’’ occurs when the test material absorbance is higher thanthe unimplanted PEU control absorbance (e.g. 1174 cm�1 inFig. 7b). A ‘‘deviation’’ equal to zero means that the absorbancesof the test material and of the unimplanted PEU control are identi-cal: the desired case of no degradation measurable by ATR-FTIR.Thus, in Fig. 7 each line represents a wave number and followinga line from left to right illustrates the deviation from unimplantedPEU control as the QACMMT loading is increased from 0 to 15 wt.%.For almost all wave numbers of interest, the largest absorbancedeviation from control was seen for implanted PEU control. Also,for almost all wave numbers, when QACMMT was added at1 wt.%, the deviation from control was decreased relative to thatof implanted PEU control. For all wave numbers, as the loadingof QACMMT was increased from 1 to 3 wt.%, the deviation was fur-ther decreased, and at a QACMMT loading of 15 wt.% the deviationfrom unimplanted control approached zero. In Fig. 8, the deviationfrom PEU for each wave number is presented for both PEU-AUAMMT-1 and PEU-QACMMT-1. At most wave numbers theabsorbance of implanted PEU-AUAMMT-1 is closer to unimplantedPEU control than to implanted PEU-QACMMT-1, which in turn iscloser than implanted PEU control.

3.2. Inflammatory response

A single representative histological image of the sample-sideof the capsule is contained in Fig. 9a–e. Images of the PMMAside are not shown. Qualitatively, the capsules of all test materi-als excepting PEU-QACMMT-15 appeared to be composed of fi-brous tissue. The body’s response seems to have increased inseverity with silicate loading, since, with the exceptions dis-cussed below, the observed qualitative capsular thickness wasgreater for samples of higher QACMMT loading. That said, thecapsular thickness of the 1 wt.% materials was not obviouslythicker than that of PEU control. Comparing the PEU-AUAMMT-1 and PEU-QACMMT-1 materials, the cellularity and fibrous mor-phology visually appeared similar. PEU-QACMMT-3 had a muchthicker capsule wall on both the PMMA side and the sample sideof the capsule; however, the cellularity and fibrous morphologyappeared normal. The sample side capsule of PEU-QACMMT-15was virtually non-existent and was severely lacking in fibrousstructure, while the PMMA side was present but smaller thanfor implanted PEU control.

4. Discussion

4.1. Silicate dispersion

Styan et al. [11] contains a morphological study of the materi-als used in the present study. There, PEU-QACMMT and PEU-AUAMMT composites were considered to be nanocomposites ofpartially exfoliated morphology, with some layers completely del-aminated, and of intercalated morphology, with layer spacing ex-panded but the stacked nature retained, respectively. Moststudies aim for complete exfoliation, believing it to provide thegreatest property enhancement, but many, if not most, of thepublished studies on solution-cast PU organosilicate nanocom-posites achieve at best only partially exfoliated morphology[15–19]. PEU-QACMMT-15 displayed significant silicate align-ment, which had increased with silicate loading. Overall, the dis-persion of QACMMT appears greater than that of AUAMMT.Favourable thermodynamics resulting from the hydroxyl func-tionality of both the QAC and AUA were likely the primary drivingforce for silicate dispersion in PEU, specifically hydrogen bondingbetween the QAC/AUA and the hard segment urethane group Nand the soft-segment C–O–C [20,21].

Fig. 5. ATR-FTIR scans of unimplanted PEU control, PEU-QACMMT and PEU-AUAMMT at wave numbers from (a) 1800–800 cm�1 and (b) 3400–2600 cm�1.

2248 K.E. Styan et al. / Acta Biomaterialia 8 (2012) 2243–2253

4.2. Validation of assay for assessment of in vivo biostability

Two factors needed validation before the implanted silicate-containing samples could be analysed. The first was to confirm thatdegradation in PEU control is typical of environmental stress crack-ing (ESC), as seen in previous studies in this laboratory [3–5]. TheSEM images for implanted PEU control (Figs. 2 and 3) shows thedegradation to be manifested as microscopic surface cracks whenin the presence of mechanical stress such as that applied by thePMMA holders, which is typical. Then, comparing ATR-FTIR tracesfor unimplanted PEU control to those for implanted PEU control inFig. 6, it is further evident that implanted PEU control underwentclassic ESC. Most characteristic is the oxidative degradation of bothhard-and soft-segment ether, as evidenced by reduced absorbanceat 1081 and 1100 cm�1, respectively, and by introduction ofbranched ether degradation products at 1174 cm�1. Hydrolyticdegradation of the ‘‘inter-phase region’’, the urethane junctionbetween the hard and soft segments, is also indicated by loss ofnon-hydrogen-bonded carbonyl at 1730 cm�1, carbon–nitrogenstretching at 1219 cm�1 and the ether mentioned previously at1081 cm�1 [2,3]. That such severity of ESC was observed on thePEU material suggests that an implantation period of 6 weeks isof a sufficient time length to allow evaluation of any biostabilityconferred by the silicate.

The second factor requiring validation is that inclusion of the sil-icate does not result in a change in the ATR-FTIR traces of the unim-planted samples, as such a change would invalidate the comparisonof implanted silicate-containing samples with implanted PEU con-trol. Unimplanted composites of 1and 3 wt.% had ATR-FTIR tracescomparable to that of unimplanted PEU control, as shown inFig. 5a and b. Unfortunately, for PEU-QACMMT-15 the unimplantedATR-FTIR trace varied significantly from that of unimplanted PEUcontrol, particularly at 2922 and 2938 cm�1 and in the region of�1100–900 cm�1. It is suggested that the difference in the unim-planted PEU-QACMMT-15 trace arises as a consequence of the sili-cate’s presence, and that the effect is significant at 15 wt.% and notthe lower loadings due to ATR-FTIR sensitivity. The absorbanceincreases at 2922 and 2938 cm�1 are related to the vibration of theC–H bond in methylene groups (CH2) in the alkyl chain in the QAC(data not shown). The absorbance differences at �1100–900 cm�1

are more interesting, and are suggested to be a result of (i) absor-bance at �1038 cm�1 due to stretching of the Si–O–Si bonds in thelayered silicate and (b) silicate–ether interaction lowering the ether1081 cm�1 vibration. The outcome is that the affected �1100–900 cm�1 region of the PEU-QACMMT-15 trace was not consideredin the analysis of the degree of biostability of PEU-QACMMT-15. Thisexcluded the important ether degradation at 1081 cm�1, but wasinconsequential as other peaks, particularly the generation of

Fig. 6. ATR-FTIR scans of unimplanted PEU control and implanted PEU control, PEU-QACMMT and PEU-AUAMMT at wave numbers from (a) 1340–880 cm�1, (b) 1640–1340 cm�1, (c) 1800–1640 cm�1 and (d) 3400–2600 cm�1.

K.E. Styan et al. / Acta Biomaterialia 8 (2012) 2243–2253 2249

Fig. 8. Comparison of the deviation of biostability-relevant PEU infraredabsorptionsfrom control levels between PEU-QACMMT-1 and PEU-AUAMMT-1 composites.Also shown, as a reference, is implanted PEU data.

Fig. 7. The effect of increasing QACMMT loading on the maintenance, to the level ofunimplanted PEU control, of biostability relevant PEU infrared absorptions. Shownare wave numbers whose test material absorbance was (a) lower than that ofunimplanted PEU control and (b) higher than that of unimplanted PEU control.

2250 K.E. Styan et al. / Acta Biomaterialia 8 (2012) 2243–2253

branched ether degradation products at 1174 cm�1, could be reliedupon.

4.3. Biostability and inflammatory response of composites in vivo

The data strongly supports the hypothesis of enhanced biostabil-ity for PEU organosilicate composites when compared to neat PEUcontrol. SEM images clearly show a reduction in visible surface deg-radation with increased QACMMT loading, as in Figs. 2–4. The SEMimage ratings show that the 3 and 15 wt.% samples are significantlyless degraded, and it is thought that the stringency of the rating scalewas not sensitive enough to display small differences such as arethought to exist for the 1 wt.% materials. This qualitative assessmentwas strengthened by the ATR-FTIR data, as evident in the traces inFig. 6, and most readily understood by consideration of the pro-cessed data in Fig. 7. Of most importance is that absorbance, irre-spective of wave number and direction of deviation from control,tends to approach that of unimplanted PEU control as the silicateloading is increased. PEU-QACMMT-3 offers�50% improved biosta-bility over PEU control, and PEU- QACMMT-1 results in some minorenhancement of biostability. The implanted PEU-QACMMT-15absorption trace is nearly identical to that of unimplanted PEU con-trol at wave numbers 3400–2600 and 1800–1640 cm�1, suggestinga near complete avoidance of in vivo degradation at the urethanelinkages and much reduced degradation in other regions of thePEU chain. However, there is some concern as to the origin of this ef-fect, as discussed in the following paragraphs.

Previous in vitro studies have observed leaching of, and subse-quent cellular response to, organic modifiers from within PU–sili-cate composites [11,22]. Further, independent studies haveobserved Sepiolite inhalation to produce macrophage agglomer-ates and/or multinucleate giant cells that are persistent for up to3 months [23]. Sharma et al. [24] also recently observed oxidativestress and genotoxicity in liver cells when exposed to zinc oxidenanoparticles. Thus, it is plausible that the QAC and/or silicatecould be leaching from PEU-QACMMT and inciting surroundingcells to (i) produce a more corrosive environment than typicaland/or (ii) die, resulting in a less corrosive environment than typ-ical. This is supported by preliminary histology conducted onexplanted samples, as in Fig. 9. That is, the QAC is interfering inan assessment of the affect provided by the silicate layers per se,since we no longer have control of all variables or consistent treat-ment among samples. However, an argument for protection by thesilicate can still be made.

For PEU-QACMMT-1, the capsular thickness appeared similar tothat around PEU and cells appeared healthy. There was likely aslight protective benefit to the QACMMT inclusion, as seen in someSEM ratings and in the ATR-FTIR data. For PEU-QACMMT-3, thecapsular thickness was much thicker than that around PEU, butthe cells appeared healthy, suggesting a much stronger inflamma-tory response. Despite this, there was a significant protective ben-efit to the QACMMT inclusion, as seen in significantly different SEMratings and in the ATR-FTIR data. In this case, it seems fair to con-clude that the biostability enhancement observed was due to thesilicate inclusion and may even be an understatement of that pos-sible from an otherwise equivalent composite including a less irri-tating organic modifier. For PEU-QACMMT-15, the material clearlycaused severe tissue degradation as the capsule was almost non-existent. Thus, it is not possible to attribute the extreme biostabil-ity enhancement seen in SEM ratings and ATR-FTIR data entirely toa silicate protective effect. However, it is likely that at least some ofthe 15 wt.% effect is due to the silicate presence, based on the po-sitive 1 and 3 wt.% data.

Since this issue was pre-empted, PEU-AUAMMT-1 was includedin this study as a partial control. Unfortunately, animal constraintsmeant that 3 and 15 wt.% controls could not be included in this

study. A loading of 1 wt.% was chosen over the higher loadings asit was thought that if enhanced biostability was seen at the lowestAUAMMT loading then this could be extrapolated to higher loadings

Fig. 9. Light microscopy images of histological sections of the sample side of fibrous capsules of (a) PEU, (b) PEU-QACMMT-1, (c) PEU-QACMMT-3, (d) PEU-QACMMT-15 and(e) PEU-AUAMMT-1. Magnification in � 10 (the scale bar (red) is 200 lm). Staining is with H&E.

K.E. Styan et al. / Acta Biomaterialia 8 (2012) 2243–2253 2251

more confidently than a back-extrapolation from, say, 15 wt.% to1 wt.%. Pleasingly, PEU-AUAMMT-1 seems to have provided a smallprotective effect on the order of that seen for PEU-QACMMT-1, asseen in the SEM images/ratings and ATR-FTIR traces in Figs. 2–6.Also, when the ATR-FTIR data is summarized as in Fig. 8, PEU-AUAMMT-1 is seen to consistently possess slightly less degradedchemical bonds than PEU-QACMMT-1. Since the silicate dispersionof PEU-AUAMMT-1 is poorer than that of PEU-QACMMT-1, perhapsthe biostability enhancement observed for PEU-AUAMMT-1 is anunder-estimation of that expected if dispersion could be increased.

4.4. Potential mechanisms of biostability enhancement

The mechanisms responsible for the observed reduction indegradation are likely to stem from alterations in the barrier prop-erties and mechanical properties that occur in nanocomposites.However, it is difficult to separate the individual contributions ofthese factors. The modulation of barrier properties characteristicof nanocomposites has been well established in several polymer-based nanocomposite systems (e.g. [25,26]). Diffusion of gasessuch as oxygen can be significantly retarded in the presence ofdispersed silicates [27], and water vapour permeability has also

been shown to decrease significantly [28]. Given that oxidativemechanisms of degradation combined with applied mechanicalstress are primarily responsible for ESC in polyurethanes, it is likelythat perturbation of the diffusion of reactive oxygen species (ROS)impacts on the degradation rate in nanocomposites.

ROS, especially the hydroxyl radical, react with PEU chains byextracting hydrogen ions from methylene groups, leaving a freeelectron pair (�CH) in a process known as hydrogen abstraction. Ifoxygen is present, a peroxyl radical (COO�) forms on the polymerchain that is also capable of hydrogen abstraction, leading to fur-ther degradation and cross-linking. A hydrogen ion (H+) is also pro-duced, which may increase the degradation of the PEU etherpresent in the soft segment and the ester linking the soft and hardsegments. The density of the impermeable silicate layers at or verynear the surface likely inhibits the access of the ROS to the polymersurface, thereby preventing degradation. Internally, these layersmay also inhibit migration of the ROS through the polymer bulkby the so-called ‘‘tortuous path’’ effect. In preliminary studies donein our laboratories, and as seen throughout the literature, silicateinclusion reduces permeability to both water and oxygen. The bar-rier effect of the silicate layers appears to be akin to the action ofthe interconnected hard segments of Pellethane� 2363-55DA,

2252 K.E. Styan et al. / Acta Biomaterialia 8 (2012) 2243–2253

which has much greater biostability than Pellethane� 2363-80A[5].

Another consideration is that the PEU hard segments are phys-ically larger than the interlayer silicate spacing measured by XRDand are thus unlikely to exist between the silicate layers [29]. Bydeduction, it is probable that it is the soft, rather than the hard,segments which are associated with the silicate and are thus of-fered greatest protection. As the soft segments are more suscepti-ble to degradation, their association with the silicate layers wouldprovide greater biostability enhancement than if it were the hardsegment that was most protected. In support of this, the loweringof the PEU ether peak at 1081 cm�1 in the PEU-QACMMT-15 com-posite (Fig. 5a) is suggestive of silicate–ether interaction. Also, anycrack beginning through chain cleavage in the soft segment mightbe hindered upon reaching a silicate layer.

Mechanical stress has been shown to increase the severity andrate of degradation in many cases (e.g. [30]), so the in vivo biosta-bility assay used applied stress by straining samples over PMMAholders. A potential contributor that has not been resolved is thechanges to the internal structure caused by the applied strain. Inprevious studies on PU without included silicate [4], straining to�150% over the PMMA holders was done in order to acceleratedegradation. With the included silicate layers, however, the effectof the strain may have been different. The PEU of this studycombined with a similar organosilicate has been studied by Finne-gan et al. [29,31]. Small-angle X-ray scattering(SAXS) resultsshowed that silicate alignment in the direction of strain is detect-able, with orientation beginning immediately upon application ofstrain and significant rotation occurring by 150% strain. This hasalso been seen by other groups [32]. Conceptually, the path lengthto migrating species will be shortest if the silicate is aligned broad-side to the sample surface and longest if aligned thin side to thesample surface. The application of strain theoretically causes align-ment in the direction of the applied strain; thus, pulling the sampleover the PMMA holder may have also pulled silicate particles toshow a greater broadside surface area parallel to the samplesurface.

No studies have looked at the effect of silicate orientation on themagnitude of the barrier effect of nanocomposites, due presumablyto the difficulty of preparing a series of nanocomposites with var-ious degrees of silicate alignment. However, there is support bystudies such as that by Schubert et al. [33], where reduced degra-dation of a poly(etherurethane urea) elastomer was observed andattributed to the increased orientation and compaction of poly-ether segments following stretching, which also resulted in low-ered oxygen permeability. In the present case, in addition tosilicate alignment, it is possible that the soft segment has been ori-ented and compacted between silicate layers. If these orientationeffects were present, the observed biostability enhancement wouldbe overstated compared to the actual situation for an unstrainedsample. Promisingly, though, increased soft-segment crystallinitywas not detected by ATR-FTIR, as evidenced by the absence of apeak at �996 cm�1 and/or one at 1370 cm�1 [33], and no changein the soft-segment morphology of very similar composites wasdetected by SAXS/small-angle neutron scattering[29,31].

5. Conclusions

Much effort has been directed towards the generation of morebiostable PUs. Success has been achieved through replacement ofthe soft segment with more chemically resistant polymers, mostnotably polycarbonate and PDMS. The current study is the first toshow that biostability can also be improved by the addition of lay-ered silicates. The addition of dispersed QACMMT at loadings lowerthan 3 wt.% resulted in significantly enhanced biostability, likely

due primarily to decreased material permeability. The addition of15 wt.% QACMMT superficially seemed to result in a stronger pro-tective effect; however, the impact of a relatively larger immuneresponse on the reduced PEU degradation observed is unknown.A more biocompatible PEU-AUAMMT-1 material of poorer silicatedispersion and likely reduced host response displayed propertiescomparable toPEU-QACMMT-1. The significance of these resultsis that: (i) although the QAC had a detrimental effect on surround-ing tissue, a less irritating organosilicate may be found (such asAUAMMT) that provides significant biostability enhancements;and (ii) although this study employed a poly(ether)urethanespecifically because of its lack of biostability, it is reasonable toextrapolate that a similar effect could be seen for other PUs.

Acknowledgements

The assistance of Veronika Tatorinoff (Surgery) and Kate Noble(Surgery, SEM), from the University of New South Wales, and KarenBarnes (histological image preparation), from the University ofSydney, is gratefully acknowledged. The authors would also liketo acknowledge funding support in the form of both a Faculty ofEngineering Research Grant from The University of New SouthWales and an Australian Research Council Discovery Grant(DP0558561).

Appendix A. Figures with essential colour discrimination

Certain figures in this article, particularly Fig. 9, is difficult tointerpret in black and white. The full colour images can be foundin the on-line version, at http://dx.doi.org.10.1016/j.actbio.2012.03.004.

References

[1] Pinchuk L. A review of the biostability and carcinogenicity of polyurethanes inmedicine and the new generation of biostable polyurethanes. J Biomater SciPolym Ed 1994;6(3):225–67.

[2] Gunatillake PA et al. Designing biostable polyurethane elastomers forbiomedical implants. Aust J Chem 2003;56:545–57.

[3] Simmons A et al. Long-term in vivo biostability of poly(dimethylsiloxane)/poly(hexamethylene oxide) mixed macrodiol-based polyurethane elastomers.Biomaterials 2004;25(20):4887–900.

[4] Martin DJ et al. New methods for the assessment of in vitro and in vivo stresscracking in biomedical polyurethanes. Biomaterials 2001;22:973–8.

[5] Martin DJ et al. Polydimethylsiloxane/polyether-mixed macrodiol-basedpolyurethane elastomers: biostability. Biomaterials 2000;21(10):1021–9.

[6] Usuki A et al. Synthesis of nylon 6-clay hybrids. J Mater Res1993;8(5):1179–84.

[7] Kojima Y et al. Mechanical properties of nylon 6–clay hybrids. J Mater Res1993;8(5):1185–9.

[8] Thatiparti TR, Tammishetti S, Nivasu MV. UV Curable polyester polyol acrylate/bentonite nanocomposites: synthesis, characterization, and drug release. JBiomed Mater Res Part B–Appl Biomat 2010;92B(1):111–9.

[9] Cypes SH, Saltzman WM, Giannelis EP. Organosilicate–polymer drug deliverysystems: controlled release and enhanced mechanical properties. J ControlRelease 2003;90(2):163–9.

[10] Beall, G.W., New conceptual model for interpreting nanocompositebehaviour.In: G.W. Beall and T.J. Pinnavaia (eds.), Polymer–ClayNanocomposites. Brisbane: John Wiley & Sons, Ltd., 2000, p. 267–279.

[11] Styan KE, Martin DJ, Poole-Warren LA. In vitro fibroblast response topolyurethane organosilicate nanocomposites. J Biomed Mater Res2008;86A:571–82.

[12] Dillon JG. Infrared spectroscopic atlas of polyurethanes. Lancaster,PA: Tachnomic Publishing Company, Inc.; 1989.

[13] McCarthy SJ et al. In-vivo degradation of polyurethanes transmission–FTIRmicroscopic characterization of polyurethanes sectioned by cryomicrotomy.Biomaterials 1997;18(21):1387–409.

[14] Loo LS, Gleason KK. Fourier transform infrared investigation of thedeformation behaviour of montmorillonite in nylon-6/nanoclaynanocomposite. Macromolecules 2003;36:2587–90.

[15] Chen T-K, Tien Y-I, Wei K-H. Synthesis and characterization of novelsegmented polyurethane/clay nanocomposites. Polymer 2000;41:1345–53.

[16] Tien YI, Wei KH. High-tensile-property layered silicates/polyurethanenanocomposites by using reactive silicates as pseudo chain extenders.Macromolecules 2001;34(26):9045–52.

K.E. Styan et al. / Acta Biomaterialia 8 (2012) 2243–2253 2253

[17] Chang J-H, An YU. Nanocomposites of polyurethane with various organoclays:thermomechanical properties, morphology, and gas permeability. J Polym SciB: Polym Phys 2002;40:670–7.

[18] Finnigan B et al. Morphology and properties of thermoplastic polyurethanenanocomposites incorporating hydrophilic layered silicates. Polymer2004;45:2249–60.

[19] Han B et al. Effect of organophilic montmorillonite on polyurethane/montmorillonite nanocomposites. J Appl Polym Sci 2004;91(4):2536–42.

[20] Balazs AC, Singh C, Zhulina E. Modelling the interactions between polymersand clay surfaces through self-consistent field theory. Macromolecules1998;31(23):8370–81.

[21] Vaia RA, Giannelis EP. Polymer melt intercalation in organically-modifiedlayered silicates: model predictions and experiment. Macromolecules1997;30(25):8000–9.

[22] Poole-Warren LA et al. Controlling cell–material interactions with polymernanocomposites by use of surface modifying additives. Appl Surf Sci2008;255(2):519–22.

[23] Warheit DB et al. Pulmonary exposures to Sepiolite nanoclay particulates inrats: resolution following multinucleate giant cell formation. Toxicol Lett2010;192(3):286–93.

[24] Sharma V, Anderson D, Dhawan A. Zinc oxide nanoparticles induce oxidativestress and genotoxicity in human liver cells (HepG2). J Biomed Nanotechnol2011;7(1):98–9.

[25] Arora A, Padua GW. Review: nanocomposites in food packaging. J Food Sci2010;75(1):R43–9.

[26] Sanchez-Garcia MD, Lagaron JM. Novel clay-based nanobiocomposites ofbiopolyesters with synergistic barrier to UV light, gas, and vapour. J ApplPolym Sci 2010;118(1):188–99.

[27] Martino VP et al. Correlation between composition, structure and properties ofpoly(lactic acid)/polyadipate-based nano-biocomposites. Macromol Mater Eng2010;295(6):551–8.

[28] Xu R et al. Low permeability biomedical polyurethane nanocomposites. JBiomed Mater Res Part A 2003;64(1):114–9.

[29] Finnigan B et al. Segmented polyurethane nanocomposites: impact ofcontrolled particle size nanofillers on the morphological response to uniaxialdeformation. Macromolecules 2005;38(17):7386–96.

[30] Stokes K, McVenes R, Anderson JM. Polyurethane elastomer biostability. JBiomater Appl 1995;9(4):321–54.

[31] Finnigan B et al. Effect of the average soft-segment length on the morphologyand properties of segmented polyurethane nanocomposites. J Appl Polym Sci2006;102(1):128–39.

[32] Gaharwar AK et al. Highly extensible bio-nanocomposite films with direction-dependent properties. Adv Funct Mater 2010;20(3):429–36.

[33] Schubert MA et al. The effect of strain state on the biostability of apoly(etherurethane urea) elastomer. J Biomed Mater Res 1997;35(3):319–28.