Nanoreinforcement of Poly(propylene fumarate) - American Chemical

Biomacromolecules 2004, 5, 1990-1998

1990

Nanoreinforcement of Poly(propylene fumarate)-Based Networks with Surface Modified Alumoxane Nanoparticles for Bone Tissue Engineering R. Adam Horch,†,§ Naureen Shahid,‡,§ Amit S. Mistry,†,§ Mark D. Timmer,†,# Antonios G. Mikos,*,†,| and Andrew R. Barron‡,| Department of Bioengineering, Rice University, MS-142, Houston, Texas 77251-1892, Department of Chemistry, Rice University, MS-60, Houston, Texas 77251-1892, and Center for Nanoscale Science and Technology, Rice University, MS-100, Houston, Texas 77251-1892 Received April 17, 2004; Revised Manuscript Received June 12, 2004

A novel composite material has been fabricated for bone tissue engineering scaffolds utilizing the biodegradable polymer poly(propylene fumarate)/poly(propylene fumarate)-diacrylate (PPF/PPF-DA) and surface-modified carboxylate alumoxane nanoparticles. Various surface-modified nanoparticles were added to the polymer including a surfactant alumoxane, an activated alumoxane, a mixed alumoxane containing both activated and surfactant groups, and a hybrid alumoxane containing both groups within the same substituent. These nanocomposites, as well as polymer resin and unmodified boehmite composites, underwent flexural and compressive mechanical testing and were examined using electron microscopy. Hybrid alumoxane nanoparticles dispersed in PPF/PPF-DA exhibited over a 3-fold increase in flexural modulus at 1 wt % loading compared to polymer resin alone. No significant loss of flexural or compressive strength was observed with increased loading of hybrid alumoxane nanoparticles. These dramatic improvements in flexural properties may be attributed to the fine dispersion of nanoparticles into the polymer and increased covalent interaction between polymer chains and surface modifications of nanoparticles. Introduction Each year, approximately 1 million fractures requiring hospitalization occur in the U.S., more than half of which affect the load-bearing bones of the lower extremities.1 Current treatments for severe bone injuries replace a nonunion defect with a permanent biomaterial that may corrode, wear, and cause infection.2,3 Additionally, these materials often possess mechanical strength far greater than bone, resulting in stress shielding and eventual bone resorption around the implant.4 Thus, there is a significant need for a bone tissue engineering scaffold with mechanical properties similar to bone that will gradually degrade, facilitate bone growth into the defect, and ultimately be replaced by natural bone tissue. The primary challenge toward this goal is mimicking the unique mechanical properties of bone tissue with a degradable, synthetic material. Human cortical bone has a compressive modulus of 17-20 GPa and a compressive strength of 106-133 MPa.5 Reported values of flexural modulus and flexural strength of bone are 15.5 GPa and 180 MPa, * To whom correspondence should be addressed. (A.G.M.) Address: Department of Bioengineering, Rice University, P.O. Box 1892, MS-142, Houston, TX 77251-1892. E-mail: [email protected]. Phone: (713) 3485355. Fax: (713) 348-4244. † Department of Bioengineering. ‡ Department of Chemistry. § These authors contributed equally to this work. # Current Address: Center for Biomaterials and Advanced Technologies, Ethicon, P.O. Box 151, Somerville, NJ 08876-0151. | Center for Nanoscale Science and Technology.

respectively.6 These unique tissue properties are partially derived from the nanoscale interactions between inorganic and organic components of bone. Hydroxyapatite (HA) nanocrystals, providing compressive strength to bone, precipitate onto bundles of collagen fibers that impart tensile strength to bone.7 Henceforth, a degradable biomaterial composed of inorganic and organic components with mechanical properties in the range of natural bone tissue may be suitable for load-bearing bone tissue engineering applications. Poly(propylene fumarate) (PPF) cross-linked with poly(propylene fumarate)-diacrylate (PPF-DA) is an injectable, biodegradable biomaterial developed in our laboratory (Figure 1). PPF degrades via ester hydrolysis, provides a conducive substrate for bone tissue growth, and yields only a mild inflammatory response in rats and rabbits.8,9 PPF shows great potential as a material for trabecular bone tissue engineering; however, its mechanical properties are far below those required for human cortical bone applications. A promising strategy to improve the mechanical properties of a polymer is the incorporation of inorganic filler particles into the organic matrix of the polymer. In our laboratory, we have seen significant increases in compressive mechanical properties of PPF upon the incorporation of β-tricalcium phosphate.10 The extent to which the fillers modify polymer properties is closely associated with the size, shape, and dispersion uniformity of the filler as well as the degree of interaction between the inorganic filler and the organic matrix. Therefore, an ideal performance is achieved with

10.1021/bm049768s CCC: $27.50 © 2004 American Chemical Society Published on Web 07/22/2004

Nanoreinforcement of a PPF-Based Polymer

Figure 1. Chemical structure of the PPF/PPF-DA network.

inorganic fillers consisting of small particles that are uniformly dispersed throughout the polymer and interact strongly with the organic matrix.11 A major limitation to the use of ceramic nanomaterials in polymer composites is the problem of dispersion of hydrophilic nanoparticles in a hydrophobic polymer. Unmodified nanoparticles often aggregate, losing their nanoscale size and corresponding properties;12 thus, some form of surface modification is required to provide miscibility. The performance of the polymer-filler composite is also strongly dependent on the strength of the interaction between polymer and filler, independent of size. The weakest to strongest interactions between the polymer and the inorganic filler are van der Waals and dipole-dipole interactions, followed by hydrogen bonding, and then finally covalent bonding. The polymer-filler covalent bonding interactions are on the order of 30-100 times greater than those that can be obtained by hydrogen bonding.13 Covalent bonding of the particles to the polymer lattice allows better transfer of mechanical loads to the particles and also promotes toughening of the polymer composite.14 It is therefore desirable to identify small size (i.e., nanometer scale) chemically functionalized inorganic fillers that can readily be incorporated into polymer matrixes through covalent bonds. We have previously demonstrated that chemically functionalized alumina nanoparticles (carboxylate-alumoxanes) may be used as the inorganic component of a new class of inorganic-organic composite materials for which a significant increase in thermal stability and tensile strength is observed.15 This work was originally aimed at tensile testing of epoxide resins (diglycidyl ether of bisphenol A). Bone tissue is commonly fractured in bending and loaded in compression in its biological environment. Thus, flexural and compressive mechanical properties are more relevant than tensile properties for testing of novel biomaterials for bone tissue engineering applications. Other studies of inorganic nanofillers dispersed in organic polymer matrixes have also suggested that dispersion and bonding are the most important factors for nanocomposite reinforcement.16-20 In general, the ideal reinforcing filler for polymers is a stable nanomaterial that easily disperses into polymer and covalently binds to its matrix.15 Layered silicates are nanosized platelets that are widely used in a variety of engineering applications for mechanical reinforcement of base resins.16,21 Lee et al. incorporated these materials into poly(L-lactic acid) (PLLA) for hard tissue treatment strategies and observed a 40% increase in tensile modulus with a 5.79 vol. % loading of montmorillonite (MMT) nanoplatelets.22 Similarly, Yoo et al. integrated silica nanoparticles into the

Biomacromolecules, Vol. 5, No. 5, 2004 1991

biodegradable polymer poly(-caprolactone) yielding favorable mechanical properties.20 Nanosized hydroxyapatite particles of various shapes are popular nanofillers for bone tissue engineering applications as they are similar to the inorganic components of bone. Liu et al. incorporated surface-modified HA nanocrystals into poly(ethylene glycol)/poly(butylene terephthalate) block copolymer and observed increases in tensile modulus.17 Further work with this material involved the use of chemical linkages between nanoparticles and polymer and yielded more significant increases in tensile modulus as well as tensile strength.23 This study was designed to determine the effects of incorporating functionalized alumoxane nanoparticles into the biopolymer PPF/PPF-DA. Specifically, four general types of functionalized alumoxanes were examined: activated alumoxane with reactive double bonds for enhanced polymer interaction, surfactant alumoxane with an organophilic carbon chain, mixed alumoxane modified with both activating and surfactant groups as separate ligands, and hybrid alumoxane with activating and surfactant groups within the same substituent. The questions addressed by this study are as follows: (1) What effect do these surface modifications of alumoxane nanoparticles have on dispersion within PPF/PPFDA? (2) What effect does each surface modification of nanoparticles have on the flexural and compressive mechanical properties of the composite material? (3) How does the loading concentration of each type of functionalized alumoxane nanoparticle affect mechanical reinforcement? Materials and Methods Materials. Research grade pseudo-boehmite (100%) was graciously provided for this work by Sasol North America, Inc. (Houston, TX). 11-Undecanoic acid, acryloyl chloride, diethyl fumarate, fumaric acid, hydroquinone, L-lysine, stearic acid, and triethylamine (NEt3) were purchased from Sigma-Aldrich (St. Louis, MO) and used as received. Calcium acetylacetonate hydrate (Ca(acac)2.(H2O)x) was purchased from Strem Chemicals (Newburyport, MA) and used as received. Hydrochloric acid (HCl), propylene glycol, propylene oxide, pyridine, sodium hydroxide (NaOH), sodium sulfate, zinc chloride, and all other organic solvents were purchased from Fisher-Acros (Fair Lawn, NJ). Bis(2,4,6-trimethylbenzoyl) phenylphosphine oxide (BAPO) was provided by Ciba Specialty Chemicals (Tarrytown, NY). PPF and PPF-DA were synthesized using previously described methods.24,25 Nanoparticle Synthesis. Three types of carboxylatealumoxane were prepared for comprehensive studies: diacryloyl lysine-alumoxane (“activated”), stearic acid-alumoxane (“surfactant”), and acryloyl undecanoic amino acidalumoxane (“hybrid”) (Figure 2a-c, respectively). Two additional alumoxanes were prepared for comparison. These were a “mixed” alumoxane with both diacryloyl lysine and stearic acid substituents and a Ca-doped diacryloyl lysinealumoxane (“Ca-activated”). In each case the base alumoxanes were prepared using one of two methods. The first method, hereafter referred to as “original synthesis”, was

1992

Biomacromolecules, Vol. 5, No. 5, 2004

similar to previously published methods26 producing dry alumoxane nanoparticles. The second method (“modified synthesis”) was aimed at removal of any carboxylic acid impurities and allowed nanoparticles to remain in organic solvent until incorporation into polymer. The modified synthesis is described in detail below. Lysine-alumoxane was prepared by previously published methods26 and by the modified synthesis as follows. Boehmite mineral was combined with an excess of lysine in a large volume of 1 N HCl with stirring at 115 °C. The reactor vessel was covered in foil to prevent exposure to light and kept open through a reflux condenser to prevent evaporation. After reacting for a minimum of 12 h, the reactor solution was diluted 1:1 with 1 N HCl. Nanoparticles were then brought to their isoelectric point by dropwise addition of supersaturated NaOH. Precipitated particles were centrifuged, redissolved in acid, and reprecipitated twice more to remove unreacted amino acids. Reactive acrylate groups were added to the amino termini of the lysine substituents by SchottenBaumann acylation.27 Lysine-alumoxane (from either synthesis) dissolved in double-distilled water was mixed with an equal volume of hexanes and chilled to 10 °C in a large three-neck reaction flask with stirring. An excess of acryloyl chloride was gradually added to the reaction mixture. The interphase reaction was carefully monitored, and the concentrated NaOH solution was periodically added to maintain a slightly basic pH. The reaction mixture was then removed, brought to a slightly acidic pH, and centrifuged. Particles were washed with water, chilled, and washed with acetone in the absence of light. Particles were then dissolved in methylene chloride (CH2Cl2), and acetone was removed by rotary evaporation. Similarly, stearic acid-alumoxane (surfactant) was prepared by two methods. By the original synthesis method, a mixture of stearic acid (173 g) and boehmite (3 g) was refluxed in toluene (150 mL). The mixture was refluxed for 4 days to yield a clear viscous solution that solidified to an immobile gel upon cooling to room temperature. All volatiles were removed under vacuum. The resulting solid was washed with diethyl ether (Et2O) resulting in a white powder. Stearic acidalumoxane was also prepared by a method similar to the modified synthesis described for the activated alumoxane except that boehmite was reacted with stearic acid instead of lysine. The mixed alumoxane was prepared by the reaction of L-lysine (20 g) and stearic acid (20 g) with boehmite (20 g) in refluxing toluene (500 mL). The product was slowly poured while stirring in a large beaker of Et2O. White flakes dropped to the bottom of the beaker, which were then filtered and washed many times with Et2O. The white product was dried in a vacuum overnight. The hybrid alumoxane was prepared by reaction of boehmite with 11-amino undecanoic acid as described in the modified synthesis. Both mixed and hybrid alumoxanes were further modified by the SchottenBaumann acylation as described above. Ca-activated alumoxane was prepared by the reaction of lysine alumoxane (19.88 g) and Ca(acac)2 in deionized water (150 mL). The yellow solution was stirred overnight. Water was removed using a rotary evaporator, and the resulting

Horch et al.

bright yellow product was dissolved in ethanol. This mixture was added dropwise to Et2O resulting in the precipitation of a white powder. The solvent was removed by decanting, followed by repeated washing with Et2O. The resulting Caactivated alumoxane was converted to the diacryloyl derivative by the reaction of a toluene solution with acryloyl chloride in the presence of NEt3. The mixture was constantly stirred, and the pH of the solution was monitored until it reached approximately 10. The reaction was assumed to have been complete once the reaction stopped giving off heat. The product was isolated by removal of the volatiles and washing with water. Nanocomposite Fabrication. PPF and PPF-DA were mixed in CH2Cl2 in a 1:2 mass ratio. Alumoxane nanoparticles in CH2Cl2 solution from the modified synthesis were then mixed into the polymer blend at loading concentrations of 0.5, 1, 2.5, and 5 wt %. Dried alumoxane nanoparticles from the original synthesis were mixed into the polymer solution at a loading concentration of 2 wt %. Solvent was removed by rotary evaporation and high vacuum-drying. The cross-linking photoinitiator, BAPO, was prepared in a 0.1 g/mL CH2Cl2 solution and added to the composite mixture at 0.5 wt %. Samples for flexural testing were prepared by injecting the nanocomposite mixture into cylindrical silicon molds and cross-linking in an Ultralum UV light box (Paramount, CA). Samples were placed 20 cm below four bulbs producing light at 365 nm and 2 mW/cm2 intensity (measured at sample distance). Samples were removed from silicon molds after 5 min and exposed to UV light for another 30 min. Flexural testing samples were typically 50 mm long and 2 mm in diameter. Samples for compressive testing were prepared by pouring the nanocomposite mixture into cylindrical glass vials (6.5 mm diameter, 40 mm length). Samples were subjected to vacuum to remove air bubbles within the polymer and then placed in the UV light box. After 5 min, samples were released from the vials by breaking the glass, and then returned to the light box for an additional 30 min. Samples were cut using a diamond saw into compression testing bars of approximately 6.5 mm diameter and 13 mm height. Nanoparticle Characterization. Functionalization of nanoparticles was verified by Fourier transform infraredattenuated reflectance spectroscopy (FTIR-ATR). IR spectra (4000-400 cm-1) were obtained using a Nicolet 760 FT-IR infrared spectrometer (Thermo Nicolet Corporation, Waltham, MA). Particle size measurements were made on a Coulter N4 Plus Submicron Particle Sizer (Beckman Coulter, Fullerton, CA) at scattering angles of 30° and 60°, using deionized water and a concentration of approximately 2 g/L. Mechanical Testing. Mechanical properties of solid nanocomposite samples were determined by an 858 Material Testing System mechanical testing machine (MTS System Corporation, Eden Prairie, MN) with a sample size of five for each group (except for comparison studies with mixed and Ca-activated alumoxane which were conducted with a sample size of three). Flexural testing was conducted in accordance with ASTM D790M-92. Flexural testing samples were placed on a three-point bending apparatus with a



Figure 2. Chemical structures of modified alumoxanes: (a) diacryloyl lysine-alumoxane (activated), (b) stearic acid-alumoxane (surfactant), and (c) acryloyl undecanoic amino acid-alumoxane (hybrid).

Figure 3. Representative stress-strain curve for mechanical testing data. Modulus was calculated from initial linear slope (/). Fracture strength was derived from maximum stress prior to failure (#).

support span of 40 mm. The cross-head was lowered at a rate of 10 mm/min to the center of each specimen until failure. Force and displacement measurements were recorded and converted to stress and strain based on sample dimensions. Flexural modulus was calculated as the slope of the initial linear region of the stress-strain curve (Figure 3). Flexural fracture strength was calculated as the maximum stress applied prior to failure. Compressive mechanical testing was conducted similarly in accordance with ASTM D695-95. Cylindrical samples were placed between two plates as the cross-head lowered onto the sample at a constant rate of 1 mm/min until failure. Compressive modulus and fracture strength were calculated in the same manner as flexural properties. Electron Microscopy. Fracture plane samples from threepoint bending testing of 1 wt % particle loading were sputter coated with approximately 20 nm of chromium and examined using either a FEI XL-30 environmental scanning electron microscope (FEI Company, Hillsboro, OR) or a JEOL 6500F scanning electron microscope (SEM; JEOL U.S.A., Peabody, MA). Energy-dispersive X-ray analysis (EDX) was carried out on a JEOL 5300 SEM.

Statistical Analysis. Mechanical data are presented as mean ( standard deviation for each experimental group (n ) 5, except for original synthesis data in Table 1 for which n ) 3). Statistical significance within a data set was determined by single factor analysis of variance (ANOVA). Sample measurements at each nanoparticle loading were compared to pure polymer resin and statistically significant differences were assessed utilizing Dunnett’s method for comparing treatment means with control.28 Statistically significant differences within loading concentration or material groups were determined using Tukey’s honestly significantly different (HSD) multiple comparison test. Additional comparisons between two populations (i.e., original vs modified synthesis) were conducted using an unpaired Student’s t-test. All tests were conducted with 95% confidence intervals (p < 0.05). Results Nanoparticle Characterization. The base alumoxane nanoparticles were characterized by IR spectroscopy and particle size measurements, and were shown to be equivalent to previously reported materials.29 FTIR-ATR spectroscopy was used to verify the appropriate functionalization of alumoxane nanoparticles. Figure 4 depicts the FTIR-ATR spectra of boehmite (1), acryloyl undecanoic amino acidalumoxane (2), and undecanoic amino acid (3). Bands observed at 1596-1586 and 1473-1466 cm-1 in the undecanoic amino acid-alumoxane spectrum are consistent with a bridging mode of coordination of the carboxylate to the boehmite core. Additionally, amino acid-associated peaks at 1620, 1570, and 1410 cm-1 indicate alumoxane modification. The presence of an amide peak (1580 cm-1) in the acryloyl derivative verifies the acrylation reaction. Similar FTIR-ATR spectral changes were observed between the lysine and diacryloyl lysine-alumoxane. Mechanical Testing. Flexural moduli of the four modifications of alumoxanes are shown at four loading concentrations with polymer resin in Figure 5a. The flexural modulus of PPF/PPF-DA alone was 1560 ( 110 MPa. Composites

1994


Horch et al.

Table 1. Mechanical Properties of Additional Test Materials for Comparisona material

flexural modulus (MPa)

flexural fracture strength (MPa)

PPF/PPF-DA alone diacryloyl lysine-alumoxaneb Ca-doped diacryloyl lysine-alumoxane mixed diacryloyl and stearic acid-alumoxane hybrid acryloyl undecanoic amino acid-alumoxane

1740 ( 150 2300 ( 90 2070 ( 320 3300 ( 270 4230 ( 340

105 ( 9 46 ( 6 77 ( 3 83 ( 7 58 ( 4

a All nanocomposite samples were prepared by the original synthesis at 2 wt % loading except where indicated. Data represent means ( standard deviation for n ) 3, except where indicated. b Prepared by modified synthesis at 2.5 wt % loading and tested at n ) 5.

Figure 4. FTIR-ATR spectra of (1) boehmite, (2) acryloyl undecanoic amino acid-alumoxane, and (3) undecanoic amino acid-alumoxane (without acrylate group). Successful alumoxane modification is shown by amino acid-associated peaks at 1620, 1570, 1470, and 1410 cm-1 (indicated by “/”) in the undecanoic amino acid-alumoxane spectrum (3) that are not present in the boehmite spectrum (1). Successful acrylation of the modified alumoxane is shown by the evolution of an amide peak at 1580 cm-1 (indicated by “#”) in the acryloyl undecanoic amino acid-alumoxane spectrum (2).

prepared using unmodified boehmite showed no significant difference in flexural modulus compared to polymer resin at all loading concentrations. For all nanocomposite materials, the flexural modulus increased to a maximum and then slowly decreased with increased nanoparticle loading. Nanocomposite samples activated with diacryloyl lysine showed significant improvements of flexural modulus compared to polymer alone at 2.5 and 5 wt % loadings. Surfactant alumoxane/polymer composites (stearic acid functionalization) showed significant increases in flexural modulus at all loading concentrations, with a maximum modulus of 2470 ( 130 MPa at 2.5 wt % loading. The greatest improvement in flexural modulus was observed with the hybrid, acryloyl undecanoic amino acid-alumoxane/polymer composite. With a loading of 1 wt %, the hybrid nanocomposite exhibited a flexural modulus of 5410 ( 460 MPa, which is more than a 3-fold increase compared to PPF/PPF-DA polymer alone or unmodified boehmite composite and is significantly greater than all other test samples. Flexural fracture strength showed little variation between material groups; however, a noticeable trend was observed with respect to loading concentration (Figure 5b). Significant decreases in flexural fracture strength compared to blank polymer were observed only at higher loading concentrations for unmodified, activated and surfactant alumoxanes. For these three materials, a significant decrease was also observed

Figure 5. Flexural modulus (a) and flexural fracture strength (b) of the different nanocomposites tested as a function of nanoparticle loading weight percentage. Error bars represent mean ( standard deviation for n ) 5. The symbol “/” indicates a statistically significant difference compared to the pure polymer resin (p < 0.05).

between the 0.5 and 5 wt % loadings. No significant difference was observed between hybrid alumoxane nanocomposites and blank polymer resin or with an increase in hybrid nanoparticle loading. Compressive moduli and fracture strength varied little with differing material groups and loading concentrations (Figure 6a,b). Unmodified boehmite/polymer composites showed a significantly lower modulus and fracture strength compared to blank polymer resin only at 5 wt % loading. No significant differences were observed for modified alumoxanes compared to polymer alone. Table 1 shows the flexural properties of nanocomposites from additional testing done for comparison purposes. Diacryloyl lysine-alumoxane was prepared by the modified synthesis at 2.5 wt % loading while all other presented


Nanoreinforcement of a PPF-Based Polymer Table 2. Comparison of Original and Modified Synthesis Methods at 1 wt % Loadinga flexural modulus (MPa) material unmodified boehmite diacryloyl lysine-alumoxane stearic acid-alumoxane acryloyl undecanoic amino acid-alumoxane a

modified synthesis

original synthesis

1720 ( 120 1650 ( 130 2360 ( 120 5410 ( 460

1800 ( 130 1850 ( 50 2080 ( 110 3890 ( 180

Data represent means ( standard deviation for n ) 5.

nanocomposites were prepared by the original synthesis with 2 wt % loading. The Ca-doped diacryloyl lysine-alumoxane showed no significant difference from regular diacryloyl lysine-alumoxane. Though both the mixed (diacryloyl and stearic acid-alumoxane) and hybrid (undecanoic amino acidalumoxane) nanocomposites exhibited significant increases in flexural modulus compared to polymer alone, the hybrid alumoxane nanocomposite showed a significantly higher flexural modulus compared to the mixed alumoxane nanocomposite. Modified and original synthesis methods as described above were also compared (Table 2). Significant differences were observed for the diacryloyl lysine, stearic acid, and acryloyl undecanoic amino acid alumoxane, though no consistent trend was observed. Electron Microscopy. SEM images of unmodified boehmite particles in the polymer show large aggregated clusters that are several microns in diameter (Figure 7a). Diacryloyl lysine-alumoxane nanoparticles (activated) were dispersed better in the polymer matrix but still aggregated into submicron-sized clusters (Figure 7b). Alumoxane nanoparticles modified with stearic acid (surfactant) achieved the best dispersion compared to other modifications, as demonstrated by the SEM image of a smooth fracture surface (Figure 7c). Aggregates of stearic acid-alumoxane particles in the polymer composite were roughly estimated to be no more than 50 nm in diameter based on these images. Similarly, the acryloyl undecanoic amino acid-alumoxane (hybrid) also demonstrated a fine dispersion of particles with clusters approximately 100 nm in diameter (Figure 7d). The enhanced dispersion of the stearic acid-alumoxane (surfactant) as compared to the diacryloyl lysine-alumoxane (activated) was also confirmed by the aluminum EDX maps of the composites. As seen from Figure 8, there was significant agglomeration of diacryloyl lysine-alumoxane particles within the polymer matrix. The aggregation of the diacryloyl lysine-alumoxane particles in the polymer composite was in the range of 1-10 µm, significantly larger than for the stearic acid-alumoxane, which showed no particles larger than 50 nm (data not shown). Discussion The objective of this study was to examine the effect of functionalized alumoxane nanoparticle incorporation into PPF/PPF-DA polymer networks on dispersion and mechan-

Figure 6. Compressive modulus (a) and compressive fracture strength (b) of the different nanocomposites tested as a function of nanoparticle loading weight percentage. Error bars represent mean ( standard deviation for n ) 5. The symbol “/” indicates a statistically significant difference compared to the pure polymer resin (p < 0.05).

Figure 7. SEM images of fracture planes of nanocomposite samples after flexural testing (1 wt % loading): (a) unmodified boehmite crystals in polymer, bar is 1 µm; (b) diacryloyl lysine-alumoxane nanocomposite (activated), bar is 10 µm; (c) stearic acid-alumoxane nanocomposite (surfactant), bar is 100 nm; (d) acryloyl undecanoic amino acid-alumoxane nanocomposite (hybrid), bar is 1 µm.

ical properties of the composite materials. The functional groups investigated in this study had different effects on

1996


Horch et al.

Figure 8. Aluminum elemental maps (EDX) of a diacryloyl lysinealumoxane (activated alumoxane) composite showing the aggregation of alumoxane particles.

nanoparticle dispersion within polymer matrixes, which thereby affected the mechanical properties of composite materials. Before discussing mechanical reinforcement, it is important to verify the chemical identities of alumoxane nanoparticles. Observed FTIR-ATR bands of alumoxanes (Figure 4) are consistent with a bridging mode of coordination of the carboxylate to the boehmite core30 and are within the ranges observed previously for carboxylate-alumoxanes.26,31 Furthermore, spectral changes during reactions verify proper chemical modification of activated, surfactant, mixed, and hybrid alumoxanes (data not shown). Unmodified boehmite particles impart little, if any, mechanical reinforcement to PPF/PPF-DA polymer in flexion (Figure 5), likely due to aggregate formation within the polymer matrix (Figure 7a). Applied loads are concentrated at these large aggregates, which are several microns in diameter and provide sites for crack propagation. Further evidence of this weakening of mechanical properties was seen at the 5 wt % loading of unmodified boehmite into polymer, which demonstrated a significant decrease in compressive modulus and compressive fracture strength compared to blank polymer resin. The flexural modulus for all nanocomposite material groups increased to a maximum and then gradually decreased with increasing nanoparticle loading. This rise and fall trend has been observed by others for tensile modulus in relation to nanoparticle loading.17 This may be due to higher occurrence of aggregation at higher loading concentrations, which may lead to lower mechanical properties. Diacryloyl lysine-alumoxane/polymer composites showed slightly improved dispersion in polymer matrixes as seen by SEM (Figure 7b) and EDX (Figure 8), though they still existed in micron-sized clusters which may serve as sites of crack propagation. Increased flexural moduli observed at loadings greater than 1 wt % may be attributed to the two double bonds on the alumoxane functionality available for cross-linking with the polymer matrix (Figure 5a). However, interactions between nanoparticles and polymer are probably hindered by aggregate formation, thus demonstrating only a modest improvement in modulus over polymer alone. The rationale for the application of an activated alumoxane is to provide covalent bonding to the polymer matrix and thus to allow better transfer of mechanical loads in the composite.14 We have previously shown that for epoxy systems, the presence of functional groups on the alumoxane significantly enhances the structural strengths of polymer composites as compared to alumoxane particles that simply serve as inert fillers.15 It is perhaps surprising, therefore, that the stearic acid-alumoxane (where no covalent interaction with the polymer is possible) shows a considerable increase

Figure 9. Schematic representation of the reaction of carboxylatealumoxane with M(acac)3.

in the composite strength as compared to the activated alumoxane (in which covalent cross-linking is possible). A consideration of the SEM and EDX images for the stearic acid-alumoxane/polymer composites suggests that the enhanced performance of the hydrophobic stearic acid derived nanoparticles is due to the uniform dispersion of an inert filler within the hydrophobic polymer matrix. A possible explanation of this phenomenon is that well-dispersed nanoparticles in small cluster sizes immobilize polymer chains, resulting in reinforcement and increased moduli. In comparison to blank polymer resin, composites formed from mixed and hybrid alumoxane containing a long carbon chain to enhance dispersion and reactive double bonds available for covalent bonding to polymer showed significant improvements in flexural modulus (Figure 5a, Table 1). Thus, the combination of both stearic acid and diacryloyl lysine substituents in the mixed alumoxane at 2 wt % loading shows an 89% improvement in flexural modulus compared to blank polymer resin. However, the hybrid acryloyl undecanoic amino acid-alumoxane containing a combined surfactant and reactive double bond substituent (Figure 2c) offers a far superior approach to the mixed ligand alumoxane with a 143% improvement in flexural modulus over polymer resin at 2 wt % loading (Table 1). In a more detailed analysis, a 1 wt % loading of acryloyl undecanoic amino acid-alumoxane nanoparticles improved the flexural modulus by a factor of almost 3.5 over polymer alone to 5410 ( 460 MPa (Figure 5a). The hybrid alumoxane in SEM showed a much better dispersion than activated alumoxane, likely explained by the long hydrophobic carbon chain in the functionality. Although the hybrid alumoxane dispersion was not quite as good as the surfactant alumoxane, mechanical reinforcement was significantly improved. This might be explained by the reactive double bond in the functional group available for cross-linking with the polymer matrix. Thus, alumoxane nanoparticles modified for enhanced dispersion in a hydrophobic matrix and activated for covalent interaction with the polymer backbone provided the most significant increases in mechanical properties over polymer resin alone. These reported values of flexural modulus and fracture strength represent significant improvements of polymer flexural properties; however, they are still only one-third of the flexural mechanical properties of human cortical bone. We have previously reported that doped alumoxane may be prepared by a metal exchange reaction between the alumoxane and an acetylacetonate complex (Figure 9).32 We are interested in the potential effects of metal dopants on the strength of the polymer composites. Calcium, a natural



component of bone, is of particular interest to us as it may enhance assimilation into natural bone tissue in a clinical application. Although ionic species may be problematic for implant materials, we do not expect free calcium ions to be released from the alumoxane nanoparticles in the composite. As may be seen from Table 1, the presence of calcium dopant (approximately 10 wt %) does not have a significant effect on the mechanical properties of the composite. In comparison with other nanocomposite studies described earlier, our study investigated the effect of spherical nanoparticles incorporated into a biodegradable polymer as opposed to apatite crystals or MMT platelets. The shape and small size of functionalized alumoxane nanoparticles provided more space for interaction between reactive functional groups and the polymer matrix. Furthermore, small particles with surfactant groups easily dispersed into the polymer matrix. Flexural and compressive properties were examined as these measurements are more relevant to load-bearing hard tissue applications than tensile properties. Our study is also unique in terms of the 3-fold increase seen in flexural modulus of the nanocomposite compared to the polymer alone. This significant increase has not been observed elsewhere in the literature for a biodegradable nanocomposite for hard tissue applications. Finally, we should make comment on the method of synthesis of the alumoxane nanoparticles. Ordinarily, the synthesis of carboxylate-alumoxanes has involved the reaction of boehmite with the appropriate carboxylic acid at slightly acidic conditions. Workup involves washing or dissolution and evaporation from a suitable solvent. However, for some preparations, we have observed that in addition to chemically reacted carboxylate, a small quantity of physically absorbed carboxylic acid may remain.31 In most applications, such residual acid is not of consequence; however, for biomedical applications, such residue is of concern. We have therefore prepared the alumoxanes using a multistep purification (see Methods) that will remove residual unreacted carboxylic acids. In a comparison of the physical properties of the polymer composites for alumoxanes made by both routes (Table 2), there appears no disadvantage to employing the multistep purification, whereas advantages of alumoxane purity are significant. Conclusions A biodegradable, biocompatible polymer was mechanically reinforced by a ceramic nanoparticle chemically modified to achieve optimal dispersion and interaction within the polymer matrix. A hybrid alumoxane nanoparticle with a long carbon chain (surfactant) and a reactive double bond (activated) dispersed in PPF/PPF-DA showed over a 3-fold increase in flexural modulus over polymer resin alone. The hybrid nanocomposite showed improved dispersions compared to activated nanocomposites and improved mechanical reinforcement over surfactant alumoxane nanocomposites, demonstrating the importance of these two elements of functional groups for mechanical reinforcement. This example of strengthening of a biodegradable polymer represents a promising development in biomaterials research for bone tissue engineering applications.

Acknowledgment. This work was supported by the National Institutes of Health (R01 AR48756) (A.G.M.), the National Science and Engineering Initiative of the National Science Foundation (EEC-0118001), the Center for Biological and Environmental Nanotechnology (EEC-0118007), and the Robert A. Welch Foundation. R.A.H. acknowledges the support of the Rice Century Scholars Program, and A.S.M. acknowledges the support of the NIH Biotechnology Training Grant (5 T32 GMO 08362). Glossary HA ) hydroxyapatite PPF ) poly(propylene fumarate) PPF-DA ) poly(propylene fumarate)-diacrylate NEt3) triethylamine Ca(acac)2 ) calcium acetylacetonate HCl ) hydrochloric acid NaOH ) sodium hydroxide BAPO ) bis-(2,4,6-trimethylbenzoyl) phenylphosphine oxide CH2CH2 ) methylene chloride Et2O ) diethyl ether FTIR-ATR ) Fourier transform infrared-attenuated reflectance spectroscopy SEM ) scanning electron microscopy EDX ) energy-dispersive X-ray analysis

References and Notes (1) Hall, M. J.; Owings, M. F. 2000 National Hospital Discharge SurVey; CDC-National Center for Health Statistics: Atlanta, GA, 2002. (2) Petty, W.; Spanier, S.; Shuster, J. J.; Silverthorne, C. J. Bone Joint Surg. Am. 1985, 67, 1236-1244. (3) Petty, W.; Spanier, S.; Shuster, J. J. J. Bone Joint Surg. Am. 1988, 70, 536-539. (4) Bobyn, J. D.; Mortimer, E. S.; Glassman, A. H.; Engh, C. A.; Miller, J. E.; Brooks, C. E. Clin. Orthop. 1992, 79-96. (5) Yaszemski, M. J.; Payne, R. G.; Hayes, W. C.; Langer, R.; Mikos, A. G. Biomaterials 1996, 17, 175-185. (6) Sedlin, E. D.; Hirsch, C. Acta Orthop. Scand. 1966, 37, 29-48. (7) Thompson, J. B.; Kindt, J. H.; Drake, B.; Hansma, H. G.; Morse, D. E.; Hansma, P. K. Nature 2001, 414, 773-776. (8) Peter, S. J.; Miller, S. T.; Zhu, G.; Yasko, A. W.; Mikos, A. G. J. Biomed. Mater. Res. 1998, 41, 1-7. (9) Fisher, J. P.; Vehof, J. W.; Dean, D.; van der Waerden, J. P.; Holland, T. A.; Mikos, A. G.; Jansen, J. A. J. Biomed. Mater. Res. 2002, 59, 547-556. (10) Peter, S. J.; Nolley, J. A.; Widmer, M. S.; Merwin, J. E.; Yaszemski, M. J.; Yasko, A. W.; Engel, P. S.; Mikos, A. G. Tissue Eng. 1997, 3, 207-215. (11) Jansen, B. J. P.; Tamminga, K. Y.; Meijer, H. E. H.; Lemstra, P. J. Polymer 1999, 40, 5601-5607. (12) Kim, J. U.; O’Shaughnessy, B. Phys. ReV. Lett. 2002, 89, 238301/ 238301-238301/238304. (13) Whitesides, G. M.; Mathias, J. P.; Seto, C. T. Science 1991, 254, 1312-1319. (14) Becu-Longuet, L.; Bonnet, A.; Pichot, C.; Sautereau, H.; Maazouz, A. J. Appl. Polym. Sci. 1999, 72, 849-858. (15) Vogelson, C. T.; Koide, Y.; Alemany, L. B.; Barron, A. R. Chem. Mater. 2000, 12, 795-804. (16) Alexandre, M.; Dubois, P. Mat. Sci. Eng. R 2000, 28, 1-63. (17) Liu, Q.; de Wijn, J. R.; van Blitterswijk, C. A. Biomaterials 1997, 18, 1263-1270. (18) Xu, R.; Manias, E.; Snyder, A. J.; Runt, J. J. Biomed. Mater. Res. 2003, 64, 114-119. (19) Rhee, S. H.; Choi, J. Y. J. Am. Ceram. Soc. 2002, 85, 13181320. (20) Yoo, J. J.; Rhee, S. H. J. Biomed. Mater. Res. 2004, 68A, 401410. (21) Zerda, A. S.; Lesser, A. J. J. Polym. Sci. Pol. Phys. 2001, 39, 11371146.

1998


(22) Lee, J. H.; Park, T. G.; Park, H. S.; Lee, D. S.; Lee, Y. K.; Yoon, S. C.; Nam, J. D. Biomaterials 2003, 24, 2773-2778. (23) Liu, Q.; de Wijn, J. R.; van Blitterswijk, C. A. J. Biomed. Mater. Res. 1998, 40, 490-497. (24) Shung, A. K.; Timmer, M. D.; Jo, S.; Engel, P. S.; Mikos, A. G. J. Biomater. Sci. Polym. Ed. 2002, 13, 95-108. (25) Timmer, M. D.; Ambrose, C. G.; Mikos, A. G. J. Biomed. Mater. Res. 2003, 66A, 811-818. (26) Callender, R. L.; Harlan, C. J.; Shapiro, N. M.; Jones, C. D.; Callahan, D. L.; Wiesner, M. R.; MacQueen, D. B.; Cook, R.; Barron, A. R. Chem. Mater. 1997, 9, 2418-2433. (27) Praill, P. F. G. Acylation Reactions; Pergamon Press Ltd.: Oxford, U.K., 1963.

Horch et al. (28) Havilcek, L. L.; Crain, R. D. Practical Statistics for the Physical Sciences; American Chemical Society: Washington, DC, 1988. (29) Vogelson, C. T.; Barron, A. R. J. Non-Cryst. Solids 2001, 290, 216223. (30) Koide, Y.; Barron, A. R. Organometallics 1995, 14, 40264029. (31) Landry, C. C.; Pappe, N.; Mason, M. R.; Apblett, A. W.; Tyler, A. N.; Macinnes, A. N.; Barron, A. R. J. Mater. Chem. 1995, 5, 331341. (32) Kareiva, A.; Harlan, C. J.; MacQueen, D. B.; Cook, R. L.; Barron, A. R. Chem. Mater. 1996, 8, 2331-2340.

BM049768S

Nanoreinforcement of Poly(propylene fumarate) - American Chemical

Recommend Documents