High performance shape memory polymer networks based on rigid nanoparticle cores

Abstract

Smart materials that can respond to external stimuli are of widespread interest in biomedical science. Thermal-responsive shape memory polymers, a class of intelligent materials that can be fixed at a temporary shape below their transition temperature (T_trans) and thermally triggered to resume their original shapes on demand, hold great potential as minimally invasive self-fitting tissue scaffolds or implants. The intrinsic mechanism for shape memory behavior of polymers is the freezing and activation of the long-range motion of polymer chain segments below and above T_trans, respectively. Both T_trans and the extent of polymer chain participation in effective elastic deformation and recovery are determined by the network composition and structure, which are also defining factors for their mechanical properties, degradability, and bioactivities. Such complexity has made it extremely challenging to achieve the ideal combination of a T_trans slightly above physiological temperature, rapid and complete recovery, and suitable mechanical and biological properties for clinical applications. Here we report a shape memory polymer network constructed from a polyhedral oligomeric silsesquioxane nanoparticle core functionalized with eight polyester arms. The cross-linked networks comprising this macromer possessed a gigapascal-storage modulus at body temperature and a T_trans between 42 and 48 °C. The materials could stably hold their temporary shapes for > 1 year at room temperature and achieve full shape recovery ≤ 51 °C in a matter of seconds. Their versatile structures allowed for tunable biodegradability and biofunctionalizability. These materials have tremendous promise for tissue engineering applications.

A thermal-responsive shape memory polymer (SMP) can be imparted with a “permanent” shape above a critical transition temperature (T_trans) when it is cast in a mold. For polymers, the T_trans is either glass transition temperature T_g or melting temperature T_m. Such a permanent shape, formed at the elastic state of the material without external stress, is retained (memorized) as the SMP cools to a temperature below its T_trans. The SMP can be deformed into a desired “temporary” shape by force at T > T_trans, and this strained configuration can be fixed as the temperature cools below the T_trans. When a thermal stimulus above the T_trans is reapplied, however, the SMP recovers to its less strained permanent shape. This unique shape memory behavior has captured the imagination of the biomedical community as scientists strive to design smart implants and tissue scaffolds that can be delivered in a minimally invasive configuration and be subsequently reverted to a preprogrammed permanent shape in vivo.

Since the first demonstration of degradable SMPs for potential tissue engineering applications (1, 2), many semicrystalline and amorphous polyester and polyurethane SMP networks have been reported. Adjustment of thermomechanical properties and degradability of these SMPs was accomplished by copolymerizing multiple monomers (3), incorporating inorganic elements into polymer backbones (4, 5), or applying different cross-linking methods (1, 5). The intertwined correlation between the composition and structure of an SMP network and its derived physical properties, however, has made it difficult to achieve a combination of T_trans, degradation profile, and mechanical strength suitable for biomedical applications in a single material. Existing biocompatible SMP networks contain either untethered polymer chains resulting in plastic deformations and broad transitions or excessive chain–chain interactions requiring extra energy to overcome. Consequently, they require harsh temperatures to fix temporary shape (< 0 °C) (3, 6–8) or trigger shape recovery (> 70 °C) that is often slow and incomplete (9–11). Moreover, few existing SMPs possess tunable biofunctionalizability and adequate mechanical strength at body temperature (12).

The shape memory capacity of polymers lies in the entropy-driven tendency for polymer chains to adopt a randomly coiled configuration. The intrinsic mechanism for shape memory behavior of polymers is the freezing and activation of the long-range motion of polymer chain segments below and above T_trans, respectively. To achieve complete freezing of chain segment motion and thus prevent chain recoiling below T_trans (temporary shape fixation) and full activation of chain recoiling above T_trans (shape recovery), a homogenous SMP network consisting of identical chains with tunable chain–chain interactions would be ideal. We hypothesize that a network cross-linked from a well-defined star-branched macromer containing a rigid nanoparticle core could meet such requirements (Scheme 1). The rigid, symmetric core defines the spatial distribution of polyester arms upon cross-linking and decreases excessive chain–chain interactions as often occurred in linear polyester networks. The multiple reactive ends of the macromer are designed to achieve adequate mechanical strength via high-density cross-linking and desired bioactivity via selective end-group functionalization. Here we report such a nanoparticle-based homogeneous SMP network that exhibits an extraordinary combination of stable temporary shape fixation and rapid and full shape recovery slightly above physiological temperature with excellent mechanical properties.

Scheme 1.

Depiction of a nanostructured SMP network.

Results and Discussion

Macromer Design: Polyhedral Oligomeric Silsesquioxane Core Versus Organic Core.

Previous studies on dendritic and hyperbranched polymers suggest that the core architecture (size and rigidity), molecular weight, and chain end composition of branched polymer systems could profoundly affect their physical properties (13, 14). Here we chose a polyhedral oligomeric silsesquioxane (POSS) nanoparticle as the core to prepare a star-shaped macromer building block for the SMP network (Fig. 1A). This design is motivated by (i) the well-defined cubic geometry of POSS that enables the grafting of up to eight identical polymer arms, (ii) the capability of the rigid POSS nanoparticle in controlling the grafted polymer chain motions on a molecular scale (15), and (iii) the demonstrated biocompatibility of POSS (16). Polylactides (PLAs) of various lengths were grafted to the octahydroxylated POSS core (Fig. S1) via ring-opening polymerization of D,L-lactide (Fig. 1A) to give star-branched macromer POSS-(PLA_n)₈ (n = 10,20,40; the number of lactide repeating units) in near quantitative yields and low polydispersity (Table S1). End-group titration by protein nuclear magnetic resonance (¹H NMR) (17, 18) confirmed the successful grafting of eight PLA arms to each core (Table S2 and Fig. S2). The hydroxyl end groups of the macromer were then reacted with hexamethylene diisocyanate (Table S3) to form cross-linked POSS-SMP via urethane linkages.

Fig. 1.

Preparation and thermal-mechanical properties of SMPs containing POSS (POSS-SMP) versus organic (Org-SMP) cores: (A) synthesis and cross-linking of macromers; (B) storage modulus (E′)-temperature and loss angle (Tan δ)-temperature (denoted by black arrows) curves of POSS-SMP-20 versus Org-SMP-20; (C) recovery rates of POSS-SMP-20 (red arrows) versus Org-SMP-20 (blue arrows) from an identical rolled-up temporary shape (Left) to fully extended rectangle (30.0 mm × 6.0 mm × 0.5 mm) in water at 51 °C.

Mono- and difunctional POSS nanoparticles have been previously utilized in SMP designs wherein interactions between the particles themselves (crystallization tendency of POSS) were exploited to form percolating physical cross-links within a chemically cross-linked system (4, 5, 19). Although the competitive crystallizations between POSS and polymer domains resulted in unconventional thermoplastic properties, neither the temperatures nor the broadness of the thermal transitions were suitable for biomedical applications. Our hope was that the octafunctional POSS-based macromers could be cross-linked to form an amorphous network wherein the rigid POSS cores impart controlled interactions between the tethered PLA arms. In addition to synthesizing SMPs built from a rigid POSS core, we prepared SMPs comprising a flexible organic core (Fig. 1A). Functionalized with the same PLA arms (Table S2, Figs. S3 and S4) as the POSS-SMPs, this all-organic SMP network (Org-SMP) allowed for comparative studies to determine the impact of core structure on physical properties.

Impact of Core Structures on Thermal-Mechanical and Shape Memory Properties.

We first compared the thermal-mechanical properties of the SMPs comprising the rigid POSS core with those containing the flexible organic core. As representatively shown in Fig. 1B, POSS-SMP-20 and Org-SMP-20, cross-linked from POSS-(PLA20_n)₈ and Org-(PLA₂₀)₈, respectively, both possessed gigapascal (GPa)-storage moduli at body temperature. Further, they exhibited similar temperature-dependent viscoelastic properties, with the storage modulus sharply descending from the GPa-glassy state to a megapascal-elastic plateau around their respective glass transitions, both within a narrow transition temperature range (width at the half-peak height of the transition, or WHPH, < 10 °C). The narrow glass transitions observed in both systems support our hypothesis that a homogenous network cross-linked from star-branched macromer building blocks containing identical polymer chains could respond to the thermal stimuli more uniformly than structurally ill-defined networks. The glass transition temperature (T_g) of POSS-SMP-20, however, was > 10 °C lower than that of Org-SMP-20, and the storage modulus drop around the T_g was more pronounced for POSS-SMP-20. Given that POSS-(PLA₂₀)₈ and Org-(PLA₂₀)₈ had similar polymer chain compositions, molecular weights, and polydispersity (Table S1) and identical numbers of urethane cross-linking sites (Table S2), the observed difference in thermal-mechanical properties can be attributed to the different size and rigidity of POSS versus the organic cores. Indeed, grid search analysis (Fig. S5) revealed more torsional freedom at each polyol branching point in the bulkier POSS core (1,103-ų molecular volume) than in the organic core (539-ų molecular volume), suggesting that the PLA arms could distribute more homogenously in the cross-linked POSS-SMP network with less excessive chain entanglement.

We then examined the shape memory performance of the SMPs comprising different cores. Both POSS-SMP-20 and Org-SMP-20 could be stably fixed at a temporary shape within seconds upon cooling to room temperature, indicating complete freezing of chain segment motions below the T_g in both networks. At 51 °C, the rate of shape recovery of POSS-SMP-20 (< 3 s) was much faster than that of Org-SMP-20 (> 20 s) (Fig. 1C and Movie S1), whereas at 73 °C, they recovered at a similar rate (< 1 s; Movie S2). These observations are consistent with the notion that it is the increased polymeric segment motions (entropy elasticity) above the T_trans (T_g in this case) that drive the shape recovery process. The observed differential shape memory performances support our hypothesis that the nanostructured molecular network imposed by the POSS core could translate, on a macroscopic scale, into more rapid shape recovery at a lower triggering temperature.

Tuning Thermal-Mechanical and Shape Memory Properties of POSS-SMPs.

To explore the possibility of tuning thermal-mechanical properties and biodegradation rates of POSS-SMP, macromers containing various PLA arm lengths (n = 10,20,40) were prepared and cross-linked. Differential scanning calorimetry (DSC) revealed a single narrow endothermic transition for each cross-linked POSS-SMP with no crystalline phase transitions detected (Fig. 2A). This observation, as well as observed optical transparency, supports an amorphous network structure where POSS cores are well-dispersed rather than crystallizing as in the case of the SMPs containing mono- or difunctional POSS (4, 5, 19). Unlike linear high molecular weight poly(D,L-lactide) (T_g ∼ 54 °C) (20), the T_gs exhibited by POSS-SMPs, 42.8–48.4 °C, are more suitable for biomedical applications. Conventional cross-linked elastomers and the Org-SMPs (Fig. S6) have T_gs that decrease as chain lengths between cross-linking points increase. By contrast, the T_gs of POSS-SMPs (, Table 1) increased as the PLA chain length of the macromer increased, suggesting a more profound impact of the rigid POSS core on the chain–chain interactions within the cross-linked network than the flexible organic core. Dynamic mechanical analysis (DMA) revealed a similar relationship of to chain length (Fig. 2B), suggesting that the effect of POSS on chain–chain interaction becomes less significant with longer PLA chains.

Fig. 2.

Thermal-mechanical properties of POSS-SMPs with varying PLA arm lengths: (A) DSC heat flow (ΔH)-temperature curves and the transparent appearance (Inset); (B) storage modulus (E′)-temperature and loss angle (Tanδ)-temperature curves (denoted by black arrows); (C) one-way shape memory cycles. Starting with 0% tensile stain at 85 °C, all specimens were subjected to consecutive cycles of tensile deformation (1), cooling (2), unloading of tensile stress (3), and recovering (4). The first 4 cycles of each specimen are representatively shown.

Table 1.

Summary of thermal-mechanical properties of POSS-SMPs

Sample name	(°C)*	(°C)†	Tan δ‡	E′37 °C (MPa)§	E′85 °C (MPa)¶	WHPH (°C)∥	Mc (Dalton)**	Rf (%)††	Rr (%)††
POSS-SMP-10	42.8	51.8 ± 0.4	2.34 ± 0.04	2,027.0 ± 38.3	4.2 ± 0.1	9.4 ± 0.3	876.9	96.0	100
POSS-SMP-20	45.4	56.0 ± 0.8	2.68 ± 0.06	2,286.8 ± 62.7	2.3 ± 0.1	8.7 ± 0.2	1,563.3	91.6	100
POSS-SMP-40	48.4	57.9 ± 0.5	2.69 ± 0.04	2,234.7 ± 17.1	1.2 ± 0.1	9.1 ± 0.3	2,996.3	100	95.2 (100)‡‡

^*Glass transition temperatures as determined from the DSC scans.

^†Glass transition temperatures as determined from the Tanδ-temperature curves.

^‡Peak value of Tanδ-temperature curves.

^§Storage moduli of the glassy state determined at 37 °C.

^¶Storage moduli of the elastic state determined at 85 °C.

^∥Peak WHPH of the Tanδ-temperature curves.

^**The number-averaged molecular weight between the cross-linking points of the polymer network (M_c) calculated from E^′ = ρRT/M_c, where ρ is the density of the polymer network (ρ = 1.27 g cm^-3), R is the gas constant (R = 8.314), T is an elastic state temperature (T = 358 K), and E′ is storage modulus measured at this temperature (E^′85 °C).

^††Shape fixing ratio (R_f) and shape recovery ratio (R_r) calculated from the second one-way shape memory cycle shown in Fig. 2C.

^‡‡The R_f value shown in the parentheses was calculated from the third cycle.

We next sought to investigate how the strengths of these POSS-SMPs vary as a function of PLA chain length and temperature by using DMA. We found that all three POSS-SMPs possessed similar glassy state storage moduli, > 2.0 GPa, at body temperature (, Table 1). Interestingly, this value is ideal for cortical bone replacement materials (21). It would allow POSS-SMPs to be used for weight-bearing in vivo applications. On the other hand, the storage modulus of POSS-SMP in the elastic state (e.g., ), which is determined by the density of cross-links (22), decreased as the PLA arm length increased.

An exciting revelation from these experiments was the observation of extremely narrow glass transitions (WHPH < 10 °C, Table 1) accompanied by sharp storage modulus changes of up to 3 orders of magnitude for POSS-SMPs. By contrast, previous SMP networks typically exhibited wide glass transitions (WHPH > 20 °C) with no more than 2 orders of magnitude modulus changes around the T_trans (9–11, 23). The steep and narrow thermomechanical transitions exhibited by POSS-SMPs make it possible to achieve both temporary shape fixing and permanent shape reversion within a narrow physiologically relevant temperature range. We realized this notion by stably fixing various temporary shapes at room and body temperatures for > 1 year and instantly (within a matter of seconds) recovering their permanent shapes ≤ 51 °C (Fig. S7 and Movies S3–S5). Clinical thermal treatments employing a combination of such mild temperature and short exposure time were shown to be well tolerated by human sclera (24, 25), epidermis (24, 25), and bony tissues (26–28). These properties are critical for implantation applications, wherein the implant material must be delivered in a stable minimally invasive shape and subsequently reverted to a permanent shape in vivo. Quantitative assessment of the shape memory performance through stress-controlled one-way shape memory cycles (4, 29) verified that all POSS-SMPs exhibited a high shape-fixing ratio (R_f > 91%) and shape recovery ratio (R_r≈100%) (Fig. 2C and Table 1), with POSS-SMP-40 achieving ∼100%R_f and R_r after the second cycle.

Finally, we found that the temperature window for deformation recovery (Fig. 2C) correlated well with the glass transition process (Fig. 2B), supporting that entropy elasticity during glass transition was the driving force for shape recovery. The number-averaged molecular weight between the cross-linking points (M_c), derived from Flory’s affine network model (22), also correlated well with the corresponding PLA arm lengths of the macromer (Table 1). This observation suggests that almost all PLA arms were tethered by urethane cross-links and could participate in the elastic deformation and recoiling, thereby contributing to the rapid and complete shape recovery.

Biofunctionalization of POSS-SMP.

To demonstrate the possibility of presenting biological signals on POSS-SMP without compromising its desired thermal-mechanical properties, we covalently coupled a model integrin-binding peptide Arg-Gly-Asp-Ser known to mediate biomaterial–cell interactions (30, 31) by using a 2-step modification strategy. First, POSS-(PLA₂₀)₈ was cross-linked in the presence of 3-azido-1-propanol (step 1, Fig. 3A) to introduce azido end groups in POSS-SMP-20-Az [see Fig. S8 for Fourier transform infrared (FTIR) spectroscopy characterizations]. POSS-SMP-20-Az exhibited storage moduli of 2.1 GPa at 37 °C and 1.2 MPa at 85 °C (Fig. 3B), with a slightly reduced comparing to POSS-SMP-20, indicating that the incorporation of the azido end groups at the cost of 25% reduction in urethane cross-links did not deteriorate the thermal-mechanical properties. No notable difference in temporary shape fixing but slightly slower shape recovery were observed for POSS-SMP-20-Az at 51 °C, presumably because of the reduction in the number of tethered chains participating in the elastic deformation and recovery. POSS-SMP-20-Az was then coupled with a fluorescently labeled alkyne-functionalized Arg-Gly-Asp-Ser, (4-pentynoic acid)-Gly-Arg-Gly-Asp-Ser-K(FTIC)-COOH, by using high-fidelity “click” chemistry (step 2, Fig. 3A) (32). The covalent attachment of the fluorescently labeled peptide was confirmed by fluorescence microscopy (Fig. 3C). No change in thermal-mechanical properties except for a minor reduction in (< 1 °C, Fig. 3B) was detected upon the attachment of the peptide to POSS-SMP-20-Az. This strategy can be extended to introduce a wide range of bioactive molecules to POSS-SMP while maintaining its desired physical properties. Finally, we demonstrated that POSS-SMPs can be engineered for varied hydrolytic degradation rates, e.g., 50% weight loss in 3–9 months, through the alteration of PLA chain length (Fig. S9). Together, these features make it possible to prepare POSS-SMP–based tissue scaffolds and implants tailored for patient- and defect-specific biochemical environment and tissue repair/regeneration rate.

Fig. 3.

Chemical modification of POSS-SMP with a bioactive peptide: (A) Synthetic scheme illustrating the introduction of azido groups during the covalent cross-linking of POSS-(PLA₂₀)₈ and subsequent conjugation of fluorescently labeled integrin-binding peptide to POSS-SMP via “click” chemistry. (1) 100 ppm DBTDL, CH₂Cl₂, argon, r.t., 12 h; 75 °C, argon, 24 h; 75 °C under vacuum, 48 h. (2) Aqueous solution of CuSO₄ (2.5 mM) and L(+)-ascorbic acid sodium salt (7.5 mM), r.t., 24 h. (B) Storage modulus (E′;)-temperature curves and loss angle (Tan δ)-temperature curves (denoted by black arrows) of POSS-SMP-20, POSS-SMP-20-Az, and POSS-SMP-20-Peptide. (C) Differential interference contrast (DIC) and fluorescent (Fl) micrographs confirming the covalent conjugation of the fluorescently labeled peptide via click chemistry. In the negative control (Left), POSS-SMP-20-Az was exposed to the fluorescently labeled peptide in the absence of ascorbic acid under otherwise identical reaction conditions.

Conclusions and Perspectives

We have prepared biodegradable SMP networks with an unprecedented combination of excellent mechanical properties (E^′ > 2 Gpa) and stable temporary shape fixing at room and body temperatures, fast (< 3 s) and complete shape recovery within a narrow physiologically relevant temperature range (< 51 °C), and tunable bioactivities for biomedical applications. The key structural feature of the network is the well-defined star-branched macromer building blocks containing identical polymer chains that enabled high-density cross-linking, selective biofunctionalization, and more uniform response of the polymer chains to thermal stimuli. Compared to the all-organic polyhydroxyl core, the bulkier and more rigid POSS nanoparticle core is more effective in minimizing excessive global entanglement of the tethered network chains and in maximizing their participation in the shape memory process. The strategic use of well-defined nanoparticles to mediate polymer chain–chain interactions and the bottom-up approach towards the control over the structure, mechanical properties, and chemical functionalities may also be useful in other situations wherein the optimization of multiple properties is desired.

A major roadblock in translating scaffold-based tissue engineering into clinical practice is the lack of materials combining tissue-like mechanical and biochemical properties with clinically relevant deployability to enable their safe delivery and integration with target tissue (33–36). The POSS-SMPs reported here have great potential as self-fitting tissue scaffolds and implants where their unique properties could address unmet medical challenges, for instance, in the reconstruction of skeletal and craniofacial defects that are characterized with complex and irregular geometries, particularly at weight-bearing locations. Conventional prefabricated weight-bearing scaffolds (e.g., stiff polymers and ceramics) do not readily conform to such defects. On the other hand, injectable formulations that could penetrate into such defects and solidify in situ are known for health concerns because of the exothermic solidification process (e.g., leading to tissue necrosis) and potential leaks. A POSS-SMP scaffold that is cast in vitro with the desired size/shape on the basis of the MRI or radiographic scans of the defect can potentially overcome such limitations by enabling its delivery in a less invasive compressed configuration and subsequently conforming to the defect upon brief and safe thermal triggering (e.g., via catheter heating). Coupled with its excellent mechanical properties at body temperature, the more securely anchored POSS-SMP implant may also reduce the need for auxiliary metallic fixators, which often require a second surgery to remove and obscure postoperative radiographic monitoring of the osteointegration of the implant. Finally, the tunable biofunctionality and degradability of POSS-SMPs opens the possibility of locally delivering therapeutics expediting the healing of the defect while enabling the implanted scaffold to “vanish” after fulfilling its function.

Materials and Methods

Detailed methods and results on molecular modelling of POSS and organic cores, gel permeation chromatography (GPC), NMR, FTIR, and high resolution mass spectrometry (HRMS) characterizations, end-group titration of POSS-(PLA_n)₈ and Org-(PLA_n)₈ macromers, formulations for preparing POSS-SMP and Org-SMP, and thermal-mechanical properties of Org-SMP can be found in SI Text. Detailed protocol for stress-controlled cyclic thermal-mechanical testing to quantify shape memory properties is also included in SI Text, and the movies demonstrating the shape memory processes are available online.

Synthesis of Macromer Building Blocks POSS-(PLA_n)₈.

Macromer building blocks POSS-(PLA_n)₈ (n = 10,20,40) were synthesized from an octahydroxylated POSS core as summarized in Fig. S1 and Table S1.

Octa(3-hydroxypropyldimethylsiloxy)octasilsesquioxane (POSS core).

The POSS core was synthesized according to the literature with slight modifications. PSS-Octakis(dimethylsilyloxy) substitute (≥97%, Aldrich, 5.012 g, 4.923 mmol) was dissolved in 25 mL dry toluene and bubbled with argon for 20 min upon the addition of ally alcohol (98.5%, Aldrich, 3.430 g, 59.07 mmol). Platinum (0)-1,3-divinyl-1,1,3,3-tetramethyldisioxane complex [Pt(dvs), 3 wt % solution in xylene, Aldrich], diluted by dry toluene into 2.0-mM solution, was then introduced to the reaction mixture (1.5 mL, 0.003 mmol) by a syringe. The reaction proceeded for 1 h at room temperature (r.t.) under argon atmosphere and another 1.5 h under reflux at 90 °C. Upon cooling to r.t., the bottom phase of the mixture was separated and concentrated under vacuum. The crude product was washed with toluene (50 mL, 3 times) to yield white solid octahydroxylated POSS core in > 90% yield. ¹H NMR (400 MHz, CDCl₃): δ3.57 (16H, t, J = 7.2 Hz), 2.85 (b), 1.63 (16H, p, J = 8.0 Hz), 0.60 (16H, t, J = 8.5 Hz), 0.15 (48H, s) ppm. HRMS for C₄₀H₁₀₄O₂₈Si₁₆Na[M + Na⁺]: calculated 1,505.2921; observed 1,505.2928.

Star-Shaped Macromer POSS-(PLA_n)₈.

PLAs of varying lengths (n = 10,20,40) were grafted to the POSS core via ring-opening polymerization of D,L-lactide. For the synthesis of POSS-(PLA₂₀)₈, POSS core (0.604 g, 0.408 mmol) and D,L-Lactide (≥97%, 4.707 g, 3.266 mmol) were degassed at r.t. for 1 h in a 25-mL Kjeldahl reaction flask and then melted at 125 °C . Tin(II) 2-ethylhexanoate [Sn(Oct)₂, 95%, 0.265 mg, 0.654 µmol] was introduced by a syringe and the melt was reacted at 130 °C for 20 h. Upon cooling, the resulting solid was dissolved in chloroform (60 mL) and precipitated in hexane (200 mL). The precipitation purification was repeated 3 times before the product was dried in vacuum oven at r.t. for 24 h and then at 90 °C for 48 h. Transparent solid POSS-(PLA₂₀)₈ was obtained in near quantitative yield (95–99%) with a polydispersity index of 1.19. ¹H NMR (400 MHz, CDCl₃): δ5.24–5.12 (152H, m), 4.42–4.25 (8H, b), 4.15–4.00 (16H, b), 1.68–1.49 (496H, m), 0.65–0.44(16H, b), 0.16–0.05 (48H, s) ppm. ¹³C NMR (100 MHz, CDCl₃): δ169.78–169.28, 69.31–69.13, 67.79, 66.83, 22.27, 20.65, 16.97–16.77, 13.47, -0.32 ppm. Macromers POSS-(PLA₁₀)₈ and POSS-(PLA₄₀)₈ were synthesized and purified in a similar fashion with near quantitative yield. Representative ¹H NMR spectra and summary of GPC characterizations of the purified macromers are shown in Fig. S1 and Table S1.

Synthesis of Macromer Building Block Org-(PLA_n)₈.

Macromer building block Org-(PLA_n)₈ was synthesized from a multihydroxylated organic core as shown in Fig. 1A.

Acetal-protected organic core.

4-Dimethylamino-pyridinium-p-toluenesulfate (DPTS) and 2,2,5-trimethyl-1,3-dioxane-5-carboxylic acid (TMDC) were prepared according to the literature. TMDC (10.79 g, 61.97 mmol), di(trimethylolpropane) (3.30 g, 13.18 mmol), and DPTS (3.88 g, 13.18 mmol) were dissolved and stirred in 200 mL anhydrous pyridine under N₂. Dicyclohexylcarbodiimide (14.06 g, 68.17 mmol) was added after the solution became clear and stirred at r.t. for 20 h under N₂. The mixture was filtered, and the filtrate was concentrated under vacuum. The resulting light-yellow solids were redissolved in hexane and filtered, and the filtrate was concentrated in vacuum for further purification by flash chromatography (silica gel, Merck grade 9384, 230–400 mesh; ethyl acetate/hexane 2∶3) to yield 4.5 g colorless oil (R_f = 0.3, 40%). ¹H NMR (400 MHz, CDCl₃): δ4.15–4.12 (8H, d, J = 12.0 Hz), 4.07 (8H, s), 3.62–3.59 (8H, d, J = 12.0 Hz), 3.29 (4H, s), 1.49–1.43 (4H, q, J = 7.4 Hz), 1.39 (12H, s), 1.33 (12H, s), 1.12 (12H, s), 0.87–0.84 (6H, t, J = 7.4 Hz) ppm. ¹³C NMR (100 MHz, CDCl₃): δ 173.97, 98.27, 71.11, 66.25, 64.31, 42.56, 42.30, 25.37, 23.15, 22.32, 18.80, 7.72 ppm. HRMS for : calculated 892.5270; observed 892.5273.

Organic Core.

Acetal-protected organic core (3.2 g, 3.66 mmol) was deprotected with 15.0 g resin (Amberlite IR-120, H+ form, 16–45 mesh) in 75 mL methanol for 15 h. After removing the resin by filtration, the filtrate was concentrated in vacuo to give colorless oil, which was redissolved in 16 mL methanol and precipitated in 100 mL anhydrous ethyl ether. The precipitate was filtered and dried under vacuum over P₂O₅ to give white powder (2.2 g, 84%). ¹H NMR (400 MHz, CD₃OD): δ 4.06 (8H, s), 3.70–3.67 (8H, d, J = 10.9 Hz), 3.62–3.59 (8H, d, J = 10.9 Hz), 3.37 (4H, s), 3.31 (s) 1.54–1.49 (4H, q, J = 7.4 Hz), 1.16 (12H, s), 0.94–0.90 (6H, t, J = 7.4 Hz). ¹³C NMR (100 MHz, CD₃OD), δ 175.11, 70.76, 64.71, 63.99, 50.66, 42.12, 23.01, 16.20, 6.75 ppm. HRMS for C₃₂H₅₉O₁₇[M + H⁺]: calculated 715.3749; observed 715.3752.

Macromer Org-(PLA_n)₈.

Organic macromers were prepared in the same way as POSS-(PLA_n)₈. Representatively, organic core (1.073 g, 1.50 mmol) and D,L-lactide (17.35 g, 120.38 mmol) were reacted at 130 °C with Sn(Oct)₂ (25.02 mg, 61.75 µmol) to prepare Org-(PLA₂₀)₈ in near quantitative yield. 1H NMR (400 MHz, CDCl₃): δ 5.25–5.12 (159H, m), 4.40–4.28 (16H, m), 4.24–4.20 (8H, b), 4.03 (8H, b), 3.25 (4H, s), 2.84–2.60 (OH, b), 1.75–1.71 (4H, m), 1.69–1.45 (m), 1.23 (12H, b), 0.90–0.81 (6H, m) ppm. ¹³C NMR (100 MHz, CDCl₃): δ175.31–175.19, 169.90–169.36, 72.66, 69.38–69.19, 66.87, 66.82, 46.65, 42.33, 20.70, 20.25, 16.94–16.87, 15.95, 7.68 ppm. Representative ¹H NMR spectra and summary of GPC characterizations of the purified macromers are shown in Fig. S3 and Table S1.

Preparation and Characterization of POSS-SMPs, Org-SMPs, and Azido-Functionalized POSS-SMP-20-Az.

In a typical procedure, POSS-(PLA_n)₈ or Org-(PLA_n)₈, hexamethylene diisocyanate (HDI, ≥98.0%, Fluka), and 3-azidopropan-1-ol were mixed (molar ratios shown in Table S3) in 2.5 times (wt/wt) dichloromethane. A catalytic amount (100 ppm) of dibutyltin dilaurate (DBTDL, ≥95%, Aldrich) was added. The solution was stirred for 2 h at r.t. before being poured into Teflon molds. The solvent was evaporated at r.t. overnight under Ar, and the material was further cross-linked at 75 °C under Ar for 24 h. The final product was heated at 75 °C under vacuum for 48 h to remove residue volatiles. The complete conversion of HDI to urethane cross-links was confirmed by the disappearance of the FTIR absorption at 2,280 cm^-1 (for isocyanate) upon cross-linking (Fig. S8). To determine the efficiency of cross-linking, POSS-SMPs were extracted in chloroform (100 mL/g) for 12 h and then dried under vacuum for 24 h. Gel content, defined by the ratio of dry weight before and after the solvent extraction, was calculated (Table S3).

Preparation of POSS-SMP-20-Peptide.

A specimen of POSS-SMP-20-Az (30.0 mm × 6.0 mm × 0.5 mm) was immersed into a 50-mL aqueous solution of (4-pentynoic acid)-Gly-Arg-Gly-Asp-Ser-Lys(FTIC)-COOH (BiomerTechnology; 1.0 mg/mL), to which 0.8 mL CuSO₄ aqueous solution (2.5 mM) was added. The mixture was degassed under argon for 1 h before 0.8 mL degassed solution of L(+)-ascorbic acid sodium salt (7.5 mM) was injected. The reaction was carried out at r.t. under argon for 24 h. The peptide-modified specimen was washed with water and ethanol for 1 h, respectively, and dried under vacuum.

DMA.

The dynamic mechanical properties of the POSS-SMPs and Org-SMP were determined on a Q800 DMA (TA Instruments) equipped with tensile film clamps. Specimens with dimensions of 30.0 mm × 6.0 mm × 0.5 mm were used for testing. The temperature was ramped from r.t. to 110 °C at a heating rate of 2.0 °C/ min. A 0.02% strain amplitude and 1.0-Hz frequency were applied. Three specimens were tested for each sample.