Polyurethane Microparticles for Stimuli Response and Reduced Oxidative Degradation in Highly Porous Shape Memory Polymers

Abstract

Shape memory polymers (SMPs) have been found to be promising biomaterials for a variety of medical applications; however, the clinical translation of such technology is dependent on tailorable properties such as gravimetric changes in degradation environments. For SMPs synthesized from amino-alcohols, oxidation resulting in rapid mass loss may be problematic in terms of loss of material functionality as well as toxicity and cytocompatibility concerns. Control of gravimetric changes was achieved through the incorporation of small molecule antioxidants, either directly into the polymer matrix or included in microparticles to form a SMP composite material. With direct incorporation of small molecule phenolic antioxidant 2,2′-methylenebis(6-tert-butyl)-methylphenol (Methyl), SMPs displayed reduce strain recovery by more than 50% (Methyl) and increase elastic modulus from approximately 1.4 to 2.3 MPa, at the expense of the strain to failure being reduced from 45% to 32%. Importantly, such changes could not ensure retention of the antioxidants and therefore did not increase oxidative stability beyond 15 days in accelerated oxidative conditions (equivalent to approximately 800 days in porcine aneurysms) in all cases except for the inclusion of a hindered amine that capped network growth, which also resulted in shape memory reduction (only 80% recoverable strain achieved). However, the inclusion of antioxidants in microparticles was found to produce materials with similar thermomechanical (T_g migration below 1.0 °C) and shape recovery of 100%, while increasing oxidative resistance compared to controls (oxidation onset was delayed by 3 days and material lifespan increased to approximately 20–22 days in accelerated oxidative solution or beyond 1000 days in the porcine aneurysm). The microparticle composite SMPs also act as a platform for environmental sensing, such as pH-dependent fluorescence shifts and payload release, as demonstrated by fluorescent dye studies using phloxine B and nile blue chloride and the release of antioxidants over a 3 week period. The use of polyurethane-urea microparticles in porous SMPs is demonstrated to increase biostability of the materials, by approximately 25%, and ultimately extend their lifespan for use in aneurysm occlusion as determined through calculated in vivo degradation rates corresponding to a porcine aneurysm environment.

Graphical Abstract:

INTRODUCTION

The degradation of polyurethanes used in medical applications has been known for decades, with the original formulations undergoing hydrolysis of ester bonds in chain extending soft segments and subsequent formulations suffering from either hydrolysis or oxidation of alternative linkages, such as carbonates or ethers, respectively.^1,2 The carbamate linkage may eventually undergo hydrolysis, resulting in amine formation that is considered a carcinogen risk in the case of aromatic amines, although the actual risk from this is still not certain and the polyurethane species are widely utilized.^3–5 Degradation in polyurethane medical devices is still of great concern, either for medical functionality or to reduce regulatory risk associated with degradable materials and possible toxicity, such as with polyurethane foam implants.^1–5

Several notable shape memory polymer (SMP) systems have gained interest for possible translation toward clinical applications, such as Sokolowski’s thermoplastic polyurethane and Wilson’s thermoset amino-alcohol polyurethanes.^6–8 Degradation studies found rapid gravimetric changes as a result of amine oxidation in Wilson’s SMPs, with end products including aldehydes, lower amines, and oxalic acid, which can lead to renal failure at sufficient concentrations; initial studies confirmed biocompatibility but the degradation products are a source of long-term risk.^9–11 Tuning the rate of mass loss through control of biostability is an important method of reducing regulatory risk in cases such as cerebrovascular aneurysm occlusive treatments, and incorporation of antioxidants has previously been used to extend biostability in polymeric medical devices.^1,12,13 A goal of this work was to minimize matrix changes to retain biocompatibility of the SMP by forming SMP composites by incorporating polymeric particles with antioxidants. Compositing offers the opportunity to incorporate multiple functionalities simultaneously, such as mechanical and visualization. Previous work with metal and metal oxide composite SMPs resulted in enhanced imaging visibility in select cases, in addition to thermo-mechanical property changes, based upon loading thresholds migration as a function of additive concentration, which also altered SMP foam stability during synthesis by compromising pore integrity.^14–16 The use of polymeric particles was hypothesized to reduce property migration while allowing for incorporating antioxidants to control degradation or release as drug analogues.

Particle systems have been proposed and tested for a variety of fields. In medical applications, microparticles have been proposed for drug releasing systems, chemical sensors, and visualization.^17–21 A series of antioxidant microparticles, synthesized from a hydrolytically degradable vanillyl alcohol-containing copolyoxalate, was synthesized and tested in conjunction with dexamethasone.^22,23 However, these particles, with sizes ranging from ∼10–15 μm, would release ∼80% of the drug payload within 12 h.²⁴ While the authors did not examine the degradation products, this polymer will also degrade into the potentially toxic oxalic acid but were shown to function as peroxy radical scavengers.²⁵ Polyketal microparticles were tested for radical scavenging by delivering superoxide dismutase to a model of lung fibrosis, with the ketals displaying no inflammation response compared with the inflammation found using the standard poly(lactic-co-glycolic acid) (PLGA) particles and ultimately inhibiting lung fibrosis.²⁵ Formation of polyurethane microspheres with varied morphologies and diameters is primarily achieved through solution templating, using a variety of solvents, conditions, and monomers, which yields particles with sizes ranging from below 1 μm to nearly 500 μm.^26–31 Poly(urethane urea) shell particles from acetone/water mixtures were selected in this study, based on results from the previous characterization of degradation products from the materials as well as the possibility of loading radical scavenging antioxidants.^11,31

Numerous antioxidant systems have been studied in polymer systems, including hindered amine light scavengers (HALS), phenolic moieties, and phosphites in polyurethane systems across biomedical materials, although rarely for specifically increasing biostability as determined through gravimetric analysis.^32–34 Butylated hydroxytoluene (BHT) has been used successfully in polyethylene and its derivatives, with specific applications in medicine and medical materials such as improving oxidative damage that occurs during sterilization of total joint replacement polymeric implants.³⁴ These antioxidants may scavenge the radicals that preferentially attack the tertiary amines, thus reducing mass loss of the SMP. However, direct utilization of in-chain antioxidants or antioxidant degradation products requires the initial polymer chain degradation that may compromise the bulk integrity of the polymer. This can be avoided by incorporating antioxidants into the above-mentioned SMP composites.

Here we present a method for producing hollow poly-(urethane urea) microparticles with average diameters ranging from 2 to 10 μm. Particles were synthesized with or loaded with antioxidants prior to incorporation into SMPs. Characterization of these materials indicates that this method for including antioxidants is superior to the direct incorporation of small molecule antioxidants, as the composites do not display significant thermo-mechanical property changes or shape recovery kinetic differences. Additionally, the SMP composites were also demonstrated as a platform for sustained release and degradation environment detection as proof of concept for expanded material functionalities.

METHODS AND MATERIALS

Materials.

Materials were purchased from Sigma-Aldrich and used without modification unless otherwise stated. Hexamethylene diisocyanate (98%, HDI), isophorone diisocyanate (98%, IPDI), triethanolamine (99%, TEA), N,N,N′,N′-tetrakis (hydroxypropyl) ethylenediamine (98%, HPED), diethanolamine (99%, DEA), butylated hydroxytoluene (BHT), 1,2,2,6,6 tetramethyl piperidinol (Piper), and 2,2′-methylenebis(6-tert-butyl)-methylphenol (Methyl) (Figure S1), Phloxine B (PhB), and Nile blue chloride (Nile) were used as monomers. Ethanol (97%, EtOH) and isopropyl alcohol (99.5%, IPA) were used as solvents for cleaning the materials after synthesis and degradation testing, respectively. Cobalt chloride (CoCl₂) and hydrogen peroxide (H₂O₂, 50%) were used for degradation analysis. Ethyl isocyanate (98%) was used for model reactions.

General.

Fourier transform infrared spectroscopy (FTIR) attenuated total reflectance (ATR) was performed using a Bruker ALPHA infrared spectrometer (Bruker, Billerica, MA); 48 scans per spectra for both background and samples were used. Spectra data were collected in absorption mode with a resolution of 4 cm⁻¹. OPUS software was used to examine spectra, identify peaks, and perform baseline and atmospheric corrections. Spectral data collection was performed in triplicate to confirm results. X-ray photoelectron spectroscopy (XPS) (Omicron XPS with Argus detector) using a Mg/Al X-ray source was used to characterize particles and SMPs.

Gas chromatography mass spectrometry (GCMS, positive mode) was used to determine the extracted concentration using a Thermo Scientific DSQ II series Single Quadrupole GCMS (ThermoFisher Scientific, Waltham, MA) as well as the degradation products from the model compounds. A 10 μL sample volume was injected, with a dwell time of 4 s. The initial temperature was 50 °C, heated to 320 °C at a rate of 20 °C/min. The flow rate was 50 mL/min with a surge pressure of 0.44 psi using methanol.

Microparticle Synthesis.

Microparticle synthesis was achieved using a modified procedure previously reported by Kong et al.³¹ In an example reaction, IPDI (5 g, 22.5 mmol) was added to a water/acetone solution (95 g, 30/70 ratio water/acetone) as a single volume and mixed for 1 h at room temperature, 30 rpm. TEA (29.97 mmol) was added dropwise over a 5 min period to the solution and the solution was allowed to react from an hour. In total, 5 g of IPDI (22.5 mmol) was added over 10 min dropwise. The reaction was allowed to proceed overnight, yielding a viscous white suspension. The solution was centrifuged for 5 min at 4000 rpm, and the particles were rinsed in H₂O-acetone (30/70 ratio) three times before drying at 70 °C overnight. Particles were dried and characterized (yield 54%). Polyurea ¹H NMR characterization (d₆-DMSO): 0.92–1.03 (−CH₃, s), 1.11–1.67 (−CH₂−, d), 2.63–2.67 (−CH₂−, s), 3.01 (NH₂−CH−, m), 3.11−3.15 (−CH₂−NH−), 3.61–3.92 (−NH−CH−, m). Synthesis of antioxidant-containing particles was repeated with 50 wt % of antioxidants added as a single mass immediately after the alcohol, prior to the additional IPDI. Particles synthesized in this way are labeled Synth (i.e., BHT-S) and were characterized by XPS and FTIR. A general schematic for synthesis is presented in Figure 1.

Figure 1.

General overview of synthetic process, illustrating the microparticle synthesis in acetone/water, extraction of the particle core with the addition of antioxidants, and the synthesis of SMP foam using a two-step foaming method including the microparticles, which can be substituted for small molecule antioxidants.

Microparticle Extraction and Loading.

The thermoplastic core was extracted using IPA/acetone mixtures at 50 °C for 12 h with a 1 h sonication (solution was changed every 4 h).³¹ Particles were then centrifuged at 4000 rpm and separated from the effluent, and removal of the core was confirmed using differential scanning calorimetry (10 °C/min, first cycle, −80 to 200 °C) and X-ray diffraction experiments. Loading of the particles was performed by dissolving 20 g of antioxidant or fluorescent dye into 80 mL of 95% EtOH solution and adding 20 g of particles. The mixture was stirred and heated to 50 °C overnight. Particle solutions were then centrifuged, separated, washed three times using H₂O/acetone, and dried overnight at 50 °C under vacuum, to produce loaded microparticles (i.e., BHT-L). Particles were collected and washed the same way as described previously.

Shape Memory Polymer Synthesis.

Porous SMP scaffolds were synthesized using a traditional two-step polyurethane foam synthesis process, provided in more detail elsewhere.³⁵ In general, an isocyanate (NCO) premix was first made by adding 40% of the desired alcohols (molar % of total alcohols used; the exact percentage could be varied to control premix viscosity and thus scaffold pore size) to the entire amount of diisocyanate, with a ratio of NCO:OH of ∼2:1.³⁵ This premix was cured at 50 °C for 36 h. The alcohol (OH) premix was then synthesized, consisting of the remaining alcohols, antioxidants or particles (5 and 10 wt %), surfactants, catalysts, water, and a physical blowing agent.³⁵ The OH premix was combined with the isocyanate premix along with microparticles, using a high-speed shear mixer, followed by the addition of a physical blowing agent. The foam was cured at 90 °C for 20 min, followed by cold curing overnight at ambient conditions.

Light Microscopy, Confocal Microscopy, and Scanning Electron Microscopy (SEM).

SMP samples were cut to less than 1 mm thickness for microscopy. Using previously described methods, particles were characterized using light microscopy.³⁵ SMP samples were imaged using an Olympus Fluoview 1000 laser scanning confocal microscope (LSCM) (Olympus America Inc., Center Valley, PA) and were excited at 543 nm utilizing less than 1% maximum intensity.¹⁴ Foam cell structure was determined by cutting axial and transverse samples that were examined using SEM. Samples were mounted on a stage and sputter coated with gold using a Cressington Sputter Coater (Ted Pella, Inc., Redding, CA) for 60 s at 20 mA. Samples were then examined using a Joel NeoScope JCM-5000 SEM (Nikon Instruments Inc., Melville, NY).

Thermal Analysis.

Differential scanning calorimetry (DSC) was also used to measure both wet and dry T_g using a Q-200 DSC (TA Instruments, Inc., New Castle, DE). Samples of ∼(5.0 ± 1.0) mg were sealed in TZero aluminum pans and placed in the heating cell. Dry samples were cooled to equilibration at −40 °C, heated to 120 °C at 10 °C/min, cooled at 10 °C/min to −40 °C and heated to 120 °C at 10 °C/min. The half-height transition of the final heating cycle was the reported T_g. Wet samples were weighed and sealed in the same manner and were then heated from −40 to 80 °C at 10 °C/min.

Thermogravimetric analysis (TGA) was used to assess any changes in thermal degradation temperatures. A TA Q50 TGA (TA Instruments, New Castle, DE) was used to heat samples to 500 °C at 10 °C/min under a mixed atmosphere of oxygen and nitrogen (60 mL/min to 40 mL/min, respectively). Samples were examined for the onset of degradation temperature and major transitions in the mass loss profiles.

Shape Memory Characterization.

Shape recovery of cylindrical foam samples (6 mm diameter, 10 mm length, six samples per series) crimped over a wire was examined at 50 °C in reverse osmosis (RO) water to determine the volume recovery behavior (strain recovery).¹⁴ Samples were crimped to a minimal diameter (∼1.0 mm) using a SC150–42 Stent Crimper (Machine Solutions, Flagstaff, AZ). For radial compression, samples were equilibrated at 100 °C for 10 min and then radially compressed and cooled to room temperature. Samples relaxed for 12 h and were tested over the course of 30 min. ImageJ was used for analysis of the change in diameters over time.

Mechanical Testing.

Uniaxial tensile testing was performed on ASTM d638 IV samples using an Instron Tensile Tester with 500 N load cell. The extension rate was set to 5 mm/min at room temperature and ambient conditions. Seven samples were tested for each species. Elastic modulus, strain to failure, and ultimate tensile strength were calculated.

Fluorescence.

Fluorescence wavelengths were examined using a Fluoromax Fluorometer (Horiba UK Limited, Middlesex, U.K.). Selected excitation wavelengths were determined from the literature and confirmed by excitation scans; PhB was excited with approximately 515 nm; Nile was excited at 635 nm.¹⁴ Specific scanning and collection methodology is presented in detail elsewhere.¹⁴

Payload Release.

Release of antioxidants was studied using 2.0 g of SMPs in 50 mL of IPA at 37 °C. Samples were withdrawn and the solution was examined using GCMS. The solution was injected into the column (C18 silica, normal phase) and heated at 20 °C/min to 300 °C. UV–vis was used to confirm release rates.

Degradation Analysis.

Previously, the rate of degradation for control compositions was determined for porcine aneurysm conditions and related to the rate of degradation as demonstrated using accelerated oxidative solution (20% H₂O₂ with 0.1 M CoCl₂ at 37 °C). In vivo and in vitro sample comparison were characterized to determine a real time oxidative condition of 2% H₂O₂.¹¹ A uniformly applied acceleration scaling factor of 57.14 was found to model the real time degradation of the controls using a linear model of mass loss once the initiation period was excluded from analysis. Pore size and interconnectivity, and strut thickness were assumed to be identical and therefore were neglected.

RESULTS AND DISCUSSIONS

Microparticle Synthesis.

Synthetic conditions were varied by changing solvent ratio, temperature, and alcohol species in solution with IPDI. The size of the particles obtained with the three amino alcohols (HPED, TEA, DEA) was characterized using SEM (Figures S2–S11), with quantified metrics shown in Table 1. TEA was selected for further examination due to particle size and distribution as well as the suspected less hydrophilic network and more rapid release/degradation rates previously demonstrated in noncomposite porous SMPs.¹¹ While the smallest size distribution was achieved with HPED (D_w/D_n = 1.10), TEA at room temperature resulted in the smallest particle diameter, similar to those results reported by Kong and others using similar methods.^31,36 Results obtained by changing the temperature indicated that room temperature gave the smallest particle diameter of 2.43 ± 0.91 μm, a size that falls within the desired range for embolic applications and for particles that can be phagocytized by cells.^37,38

Table 1.

Impact of Synthetic Methods on Average Particle Diameter and Distribution

% acetone	% water	temperature (°C)	monomer	number average distribution diameter (Dn) (μm)	weight average distribution diameter (Dw) (μm)	Dw/Dn
70	30	50	TEA	6.79	10.69	1.57
70	30	50	DEA	7.25	8.74	1.29
70	30	50	HPED	9.54	10.48	1.10
70	30	0	TEA	6.66	9.12	1.37
70	30	20	TEA	2.80	4.18	1.49
70	30	80	TEA	6.42	9.12	1.42
30	70	20	TEA	6.66	9.72	1.46
50	50	20	TEA	4.53	7.05	1.56
60	40	20	TEA	4.69	8.29	1.77

Poly(urethane) urea (PU) microparticles contain a thermoplastic polyurea core with the PU cross-linked shell surrounding it. Removal of the core was achieved through sonication in solvent and confirmed using DSC and XRD (Figure 2), with the thermoplastic polyurea core displaying semicrystalline behavior as determined by a peak at 33.843° by XRD (lattice spacing of 5.47 Å by Bragg equation) and a melting peak at ∼170 °C, both of which are not present in the cross-linked shell after extraction. After core removal, the particles were loaded with the antioxidants or the fluorescent molecules. Loading during synthesis was viable only for BHT and Piper, as Methyl is not soluble in the polymer solution sufficiently. Loading percentages of up to 45% were attempted during synthesis and did not appear to alter particle composition or morphology, but the removal of BHT was a possibility during particle cleaning and so was not further examined. Loadings of ∼45% (wt) were achieved using physical means after synthesis.

Figure 2.

Scanning electron microscopy and XRD comparison of the areas of the original microspheres (PU with core) and those that have undergone extraction reveals a relative crystallinity of 3.3% for the starting material, which can be attributed to the thermoplastic polyurea core.

Composite SMPs Synthesis.

The spectra of the SMP foams (Figure 3) displayed the characteristic polyurethane bands including hydroxyl bands at 3250–3500 cm⁻¹, the methyl band at 2800–3000 cm⁻¹, the carbonyl peak at 1688 cm⁻¹ that denotes the urethane bond and the corresponding shoulder for the urea bond at ∼1640 cm⁻¹, and the skeletal carbon peak at 1252 cm⁻¹.³⁹ The lower wavenumbers for the urethane peak are due to the increased hydrogen bonding achieved using aliphatic monomers as well as the high urea content in the composition (due to water added during the foam blowing synthesis).¹¹ The Methyl and Thio SMPs display shoulders on the carbonyl at ∼1710–1730 cm⁻¹ that correspond to the ester bonds in the antioxidants (Thio) or the formation of a new urethane linkage (Methyl) (Figure S12). The increase in the relative height of the urethane carbonyl of the Piper SMP, compared with the control, indicates the formation of the urethane linkage. ¹H, ¹³C NMR confirmed the formation of urethane linkages for Methyl and Piper antioxidants; other examined species contained original structures and small concentrations of urea linkages. Mass spectroscopy (Figure S12) supported these findings, indicating that Methyl could form mono- and biurethane linkages, while Piper formed only a single urethane; BHT, Thio, and Triphen would only interact noncovalently.

Figure 3.

Spectroscopic characterization of poly(urethane urea) microparticles and SMP foams. XPS comparison of survey scan (A) and C 1s scan (B) demonstrates the greater incorporation of amines from amino alcohols (C 1s peak at ∼286 eV) and the presence of silicone from surfactants (survey scan below 200 eV). Fluorescence profiles, displaying minimal wavelength shift, that corresponds more closely to uncross-linked phloxine, compared with previous studies utilizing cross-linked phloxine in SMP foams (C) and the shift in the carbonyl peak indicative of linkages other than the carbamate/urea peaks due to small molecules added to the SMP (D).

XPS confirmed the incorporation of greater amounts of free hydroxyls (286.4 eV) associated with the leftover end groups from the synthesis as well as a shoulder peak at 288.7 eV associated with carbonyls.²⁸ No differences were found for antioxidants during loading, during synthesis, or postsynthesis (Figure S13).¹¹ The residual presence of tin, a foaming catalyst, and silicone, from the surfactants used, were found in all SMPs. The dual binding peak seen for the C 1s scan at 286.6 eV may also indicate a greater presence of C–N bonds as well as hydroxyls, as a result of greater incorporation of the alcohol into the SMP compared to the microparticle, which possesses a thermoplastic core and linear shell components (urea contributions). This is supported by the relative area of the N 1s peak displayed for the SMP compared with the particles in the survey scan. Additionally, the shoulder peak in the C 1s scan centered at 286.1 eV indicates a greater concentration of free alcohols, which is expected stoichio-metrically from the preferential consumption of the isocyanates to form the urea, followed by urethane, then branching reactions due to synthetic conditions (Figures S13–S15). The consistent shoulder at 288.5 eV, indicative of the carbonyls, is expected based on the urethane-urea nature of the materials. The qualitative decrease in peak strength for the SMPs may be due to the decrease in relative concentration of the urea/urethane carbonyl relative to the backbone.

For the particles themselves, the Cl 2s and Br 3p peaks associated with PhB were found as fairly weak signals, indicating that the dye was not just on the surface of the particles but contained within the shell as well. The decreased O 1s peak for the particles may be attributed to residual urea in the polymer shell compared to a urethane linkage, which may also explain the shoulders found in the C 1s scans as explained above. The dye particles did not display any changes in the N 1s peak, which was anticipated. The antioxidant-containing particles displayed the same shoulder at 286.6 eV as that of the other particles but is again not found in the SMP composites, providing further support for this being a urea carbonyl peak. Most importantly for these particles, the O 1s peak (531.7 eV) was found to be broader for the particles where antioxidants were loaded in after synthesis, compared to particles where the antioxidants were included during synthesis. For example, the Piper synth peak is sharp and centered at ∼531 eV, whereas Piper load is broader with a distinctive shoulder closer to ∼532.5 eV, a trend which is repeated in the BHT particles.

Fluorescence spectroscopy of SMP composites with PhB-containing particles was compared with SMPs doped with PhB and PhB in solution; emission spectra display a red-shift as a result of cross-linking and formation of a urethane linkage for PhB. The emission spectra peak for PhB in solution (1 mM in H₂O) was found to be 573 nm, 580 nm for the SMP composite, and 589 nm for the cross-linked PhB. This indicates that the PhB, enclosed in the microparticles, retains the functionality of the fluorescent molecule and that the payload is in fact incorporated in the particle rather than on the surface, thus providing further support for the XPS analysis results.

The average strut diameter, determined from SEM, was 21 ± 4 μm. The average aggregate diameter was 13 ± 10 μm, as displayed in Figure 4. It was determined through optical microscopy of fluorescent PhB particles, but aggregates were visible during SEM analysis as well; aggregate diameter seemed to be independent of composition. Qualitatively, pore membranes were found to be primarily intact with only pinholes between pores, with triangular struts throughout. Pore sizes were found to be consistent with the composite SMPs, with no statistical variation between species of the antioxidant or fluorescent molecule; qualitative evaluation of antioxidants, directly incorporated into the SMP foams, indicated a variation in foam cell formation resulting in widely variable pores, membranes, and bulk foam morphology. Antioxidant or dye species did not seem to impact pore morphology when included in microparticles (Figures S16 and S17).

Figure 4.

SMP foam with TEA-microparticles (control) (A) and BHT-microparticles (B). The agglomeration of microparticles can be seen to occur along the edges of pore membranes (C, D). Aggregate diameters (E) and pore sizes (F) determined using microscopy.

Thermomechanical Characterization.

Tensile testing comparisons revealed minimal differences between the microparticle-containing SMPs. However, SMPs with antioxidants directly incorporated into them were found to have widely varied mechanical behaviors (Table 2, Figure 5). For small molecule antioxidant additives, an initial analysis was performed using HDI-based foams (Table S1) before being repeated using TMHDI-based materials, with trends reproducible for both formulations. Methyl again displayed enhanced properties relative to the control and the other examined species. The elastic modulus of the control was 1.45 MPa, while that of Methyl was 2.32 MPa; strain to failure for the control was 45.2% while Methyl’s was 32.2%. Of the other examined small molecule antioxidants, all species displayed increased strain to failure and tensile strength, except for Triphen. The Triphen SMP displayed nearly twice the strain to failure (71.0%), while also possessing reduced tensile strength (1.41 MPa compared to 1.64 MPa for the control) and a lower elastic modulus (0.44 MPa). Microparticles do not change the mechanical properties significantly when compared with the control (no particles/additives) composition. The changes that are displayed are not statistically significant and may be attributable to the intrinsic variability in SMP foam morphology that is obtained through the foaming process.

Table 2.

Mechanical Properties of Control SMPs and Those Containing Small Molecule Antioxidants Only as Well as Microparticles Loaded with Antioxidants (n = 7)

species	elastic modulus (MPa)	strain to failure (%)
Control SMP	1.5 ± 0.2	45 ± 12
SMP composite with Piper microparticles	1.3 ± 0.3	53 ± 8
SMP with Piper small molecule additive	1.4 ± 0.2	57 ± 5
SMP composite with Methyl microparticles	1.2 ± 0.2	61 ± 26
SMP with Methyl small molecule additive	2.3 ± 0.4	32 ± 7
SMP Composite with BHT microparticles	1.5 ± 0.3	51 ± 6
SMP with BHT small molecule additive	1.4 ± 0.3	51 ± 6

Figure 5.

Shape memory behavior (both volumetric and strain recoveries) of porous SMPs with antioxidant small molecules (10%, A) and microparticles (10%, B) over 15 min in 37 °C RO H₂O; mechanical behavior (tensile) of SMPs with antioxidant small molecules (10%, C) and microparticles (10%, D).

For the small molecule antioxidant SMPs, the Methyl SMP displayed increased T_g (Table 3) from 55.3 °C (control) to 61.5 °C, which is attributed to the aromatic groups in the SMP backbone. Interestingly, Piper did not display an increase, but this may be due to the antioxidant being a pendant group rather than a cross-linker, with the increase in chain ends counterbalanced by the increase in backbone chain rigidity due to the pendant groups. The physical addition of the BHT (T_g = 57.2 °C) did not statistically alter the T_g. Thio and Triphen seemed to act as plasticizers causing the drop in the T_g of the SMP, with 41.0 and 38.9 °C, respectively. A melt trans-formation was found for Thio at approximately 40 °C (Figure S18), which corresponds to the melting point of the monomer. The other SMPs did not display melting points, despite all monomers displaying melting points within the examined range (BHT T_m ∼70 °C, Piper T_m ∼130 °C, Methyl T_m ∼125 °C, and Triphen T_m ∼22 °C).

Table 3.

Thermal Characterizations of Antioxidant SMPs (n = 3)

species	Tg (°C)	wet Tg (°C)	Td(onset temp) (°C)	Td (derivative peak) (°C)	Tm (°C) (monomer melt transition, °C)
control	55.3 ± 0.3	10.4 ± 1.0	239 ± 2	273 ± 3	-
SMP composite with Piper microparticles	55.5 ± 0.4	10.2 ± 0.1	251 ± 1	278 ± 2	− (130)
SMP with Piper small molecule additive	55.0 ± 0.3	11.5 ± 1.1	250 ± 1	279 ± 1	− (130)
SMP composite with Methyl microparticles	55.6 ± 0.2	10.4 ± 1.1	258 ± 1	300 ± 2	− (125)
SMP with Methyl small molecule additive	61.5 ± 0.2	14.2 ± 1.4	260 ± 3	299 ± 4	− (125)
SMP composite with BHT microparticles	55.3 ± 1.0	10.5 ± 0.7	261 ± 3	301 ± 2	− (71)
SMP with BHT small molecule additive	57.2 ± 0.8	10.2 ± 1.9	261 ± 1	303 ± 2	− (70)
SMP with Thio small molecule additive	41.0 ± 3.0	5.2 ± 0.5	236 ± 1	267 ± 1	39.5 ± 1.1 (40)
SMP with Triphen small molecule additive	38.9 ± 0.3	12.3 ± 1.3	247 ± 3	279 ± 4	− (22)

The use of the microparticles with antioxidants did not alter the T_g. Control microparticles were found to not have a significant T_g, an expected result due to the high concentration of urea linkages; microparticle behavior is discussed further by Kong et al.³¹ Dry T_g of the microparticle SMP composites was found to be ∼65 °C, regardless of additive species. Wet T_g was ∼28 °C. The microparticles themselves had T_g of ∼90 and 150 °C, consistent with results reported by Kong for similar materials.³¹ However, unlike Kong’s results, no peak at 200 °C was seen, which was expected for our system.³¹ Additionally, extraction of the particle core and addition of the TEA removed the T_g at 150 °C and added a T_g at ∼43 °C consistent with the thermal behavior for a similar polyurethane system.¹¹

Thermal degradation of the control composite SMP composition displayed an onset degradation of ∼240 °C (derivative local maxima of ∼275 °C), indicative of polyurethane thermolysis. Methyl and BHT antioxidants resulted in increased derivative peaks of thermal degradation temperatures (control at ∼273 °C, BHT at 303 °C, and Methyl at 299 °C). All antioxidant SMPs, except for the Thio, displayed an increased onset of thermal degradation, a trend which was repeated with the antioxidant-containing microparticles. This trend was reproduced for bare particles. Literature values for polyurethane thermal degradation display similar thermal degradation values in the range of 230–250 °C for aliphatic polymers; full thermal degradation curves are displayed in Figure S18.^11,40,41

Shape Memory Characterization.

All composite SMPs examined have similar shape recovery profiles compared to the control SMP, unlike the SMPs containing small molecule antioxidants. This desired outcome was achieved for the first time, by incorporating functional polymer microparticles; previous attempts have failed to deliver with regards to shape memory, with metal or metal oxide-based particles, including tungsten, zirconium oxide, barium sulfate, and iron oxide, altering the recoverable strains and strain recovery rates.^{14–16,35,42} With small molecule antioxidants, Methyl SMPs displayed little shape recovery (∼35% recoverable strain) (Figure 5) and Piper had slightly reduced shape memory (∼90% recoverable strain). The Methyl-loaded particles and the Piper-loaded particles have superior shape recovery behavior compared with the small molecules. Piper-loaded composite SMPs, the slowest recovering composition, are capable of 100% strain recovery over 15 min (with full recovery occurring at ∼10 min) with similar recovery kinetics to the control material. By comparison, Piper, added as a small molecule, results in only ∼80% recovered strain by 10 min and a final recoverable strain of ∼90%. The other SMP species did not have significantly altered shape memory properties. For the Methyl SMPs, the more rigid aromatic backbone having formed urethane linkages and thereby increasing backbone rigidity is the most plausible cause for reduced strain recovery. For Piper SMPs, the decrease in cross-link density due to increased chain ends (as a result of adding the antioxidant) will reduce the recoverable strain.⁸

As demonstrated through the fluorometry examination of PhB-containing particles (Figure 3), there are some reactions between isocyanates and surface groups of the microparticles, which in the case of PhB, result in urethane formation; in the case of metal and metal oxide additives, which were not characterized for surface chemistry, the incorporation of the particles into the polymer matrix during foam blowing may change the cross-linking density locally, but microscopic analysis did reveal pore instability and collapsed sections of the SMP foam when using such additives.¹⁴ These sections of increased density may dramatically have altered the shape recovery kinetics simply due to fracture of the more dense sections.^15,35 Regardless, with the presented polyurethane microparticles, superior incorporation without pore collapse and additional cross-linking at the particle surface is achieved and does not impact the strain recovery.

Payload Delivery and Stimuli Response.

Small molecule containing SMPs were found to release nearly all of the antioxidants in the matrix during cleaning, as determined by GCMS, and were therefore not examined for release studies. Nile blue chloride and PhB, in addition to BHT, Methyl, and Piper composite SMPs were examined for release of their respective molecule in vitro (Figure 6). The initial release of additives from the composite SMP was measured using GCMS (Table S2 and Figure S19) and UV–vis. At 1 h, which would be the equivalent of cleaning the SMP prior to device packaging and sterilization, the BHT particles appeared to have released the majority of the releasable payload. BHT-synth released more than 80% of the available antioxidant and BHT-load had released ∼50%. These 1 h results are similar to those reported by Yoo et al. using Sudan Blue in a poly(urea urethane) system.⁴³ By 1 week, 91% of the available BHT had been released from the BHT-synth and ∼75% from BHT-load. In an effort to improve loading and to examine the combination of small molecules and compositing, BHT and BHT (containing both microparticles and small molecule antioxidants) SMPs were examined and were found to release nearly 50% of the available antioxidant by 1 week. It is important to note that the solution concentration of the BHT and BHT SMPs was approximately twice of BHT-Load or BHT-Synth, demonstrating the potential of these methods for applications other than the biostability tuning of the SMP. By 1 week, the release from the SMPs appeared to have stabilized for simple diffusion driven release, further indicating the possibility of retaining some gravimetric resilience even after diffusion has occurred.

Figure 6.

GCMS profiles displaying extraction of small molecule antioxidants (A) and release of payloads from microparticle composites at 37 °C in H₂O (B), confirming that while the BHT antioxidants were primarily removed using solvent washes, sufficient retention of active molecules is achieved through the use of microparticle composite SMPs. Responsiveness of composite SMPs loaded with Nile blue (D) and PhB with Nile blue (E) (C) as determined through fluorescence emissions in 1 M solutions. The colorimetric response of SMPs is displayed with a visual change according to pH for dual loaded composite SMP (C).

The Piper and Methyl SMPs retained greater than 90% of the payload by 3 weeks, as did the particles containing PhB and Nile fluorescent dyes. Particles in solution displayed a similar, if the more rapid trend, compared with the composite SMPs, demonstrating the potential of the SMPs as a matrix material for sustained release. Tuning the hydrophobicity and size of the added molecule would obviously factor into the release profiles, and further work is needed to characterize interaction of the matrix and particle wall in terms of release (such as methods presented by Yoo et al.); for long-term sustained release or for oxidation-responsive conditions, the presented system seems to show promise.^43–45

The incorporation of fluorescent dye also presented an opportunity to incorporate an environmental detection functionality. Nile chloride emits a wide range of wavelengths depending on solvent and pH, and PhB is visible through blood as a way of monitoring the device’s location.¹⁴ SMP composites with PhB, Nile blue, or both were examined in 1 M solutions with KOH, HNO₃, RO H₂O, and H₂O₂. Responsiveness for Nile particles and foam was found with the acidic solution, with a strong emission peak displayed at 538 nm that is not present in basic or neutral solution, or in the presence of the peroxide. For the Nile composite SMP in a colorimetric assay, a shift in color was first noticed at a pH of 4 with a subtle color shift, with a more distinct change having occurred at pH 3. Fluorescence emission sweeps were used to determine the shift in emission peaks, from a sharper peak 576 nm and a broader peak at 471 nm for basic pH (pH = 12), to two more narrow peaks centered at 626 and 548 nm. With an optical tissue window in this region, it is hypothesized that it would be possible to remotely monitor the environmental conditions at the SMP-tissue interface through fluorescence microscopy.¹⁴

Gravimetric Analysis.

Gravimetric analysis was performed with the intent of determining SMP rate of degradation using conditions determined for the control materials in a porcine vascular aneurysm, with experimental testing using accelerated conditions used to determine real time equivalency (see Figure 7).¹¹ The selected antioxidants had a varied performance with regards to enhancing SMP biostability when incorporated as small molecules into the polymer matrix. When directly incorporated into the SMP as small molecules, the BHT was not retained in the SMP matrix, as confirmed through release studies, and the Methyl formed urethane linkages that could no longer act as radical scavengers. Triphen was also found to be removed during cleaning; Thio did not seem to be removed through diffusion but also did not improve oxidative biostability. BHT, Methyl, Triphen, and Thio SMPs all displayed degradation rates that match the control of SMPs. Piper was the only small molecule antioxidant that retained mass during the testing, with no gravimetric change recorded over the initial 3 days. After this initiation period, the degradation seemed to follow the behavior of the control materials, indicating that the Piper is preferentially oxidized and once consumed, oxidation begins on the amino-alcohol units of the SMP. Interestingly, when the diisocyanate species was altered to the methylated TMHDI, BHT directly incorporated during synthesis resulted in statistically increased mass retention, on the same order as Piper in the HDI series; Piper was not as effective in the TMHDI SMPs (Figure S20) compared to BHT.

Figure 7.

Gravimetric analysis of small molecule containing SMPs (5 wt %, A; 10 wt %, B) and composite SMPs (5% weight, C) in (20% H₂O₂ with 0.1 M CoCl₂ at 37 °C). Theoretical calculations of gravimetric behavior were performed to compare the real time control lifespan with small molecule and microparticle composite SMPs (D) for a 2% H₂O₂ solution (porcine aneurysm conditions) at 37 °C (n = 7).

However, while microparticles enhance the utility of same antioxidants, for the Piper series it was found to be no more effective than the addition of small molecules into the bulk. This is attributed to the maximum loading efficiency of 50% particle mass as well as the cleaning that is necessary for SMPs prior to any medical application. The 10% loading of particles in SMP composites would provide ∼0.85 g of antioxidant. Gravimetric testing revealed, as hypothesized, that the majority of the small molecule antioxidants diffused out of the matrix during cleaning, leaving on the matrix materials behind, which underwent oxidatively induced mass loss at the same rate as the control SMPs. Piper antioxidants delayed the onset of mass loss, as more of the active antioxidant was retained compared to the other formulations. After this initiation period (3 days in 20% H₂O₂ with 0.1 M CoCl₂), the rate of mass loss was found to approximately match that of the other formulations for the initial 50% mass. At this point, the rate of mass loss was reduced to ∼60% and extended the material life from 15 days to 25 days. The use of microparticles did allow for better retention of BHT and Methyl antioxidant species, thereby enhancing SMP biostability to approximately the same degree as the Piper SMPs.

Gravimetric results were supported by SEM imaging (Figure 8), where foam membranes were found to be relatively intact postsynthesis and after cleaning. Studies of solvent diffusion into porous polyurethanes have indicated that the outer pores will degrade primarily, as will the pore membranes, based upon the Fickian-derived diffusion model of moisture into polyurethane foams.⁴⁶ Poly(ether urethane) foam membranes, materials which are also oxidation susceptible, will form holes or begin to erode at existing pinholes and open the cell pore, and the foam struts will undergo pitting and surface erosion followed by the eventual collapse of the bulk matrix.^47,48 This reported behavior is similar to the presented SMPs, where the membranes erode away rapidly over the course of hours for the control SMPs, with complete membrane etching by 72 h and material collapse by 7 days; this behavior is also representative of in vivo oxidation of similar materials.¹¹ While these SMPs were not specifically imaged for surface blistering and strut cracking, these factors may also contribute to the material collapse; a lack of such features may be due to the rapid oxidatively induced surface erosion of the material.⁴⁹

Figure 8.

SEM images of the degradation of composite SMPs compared to control and selected small molecule antioxidant-containing SMPs over the course of 7 days in accelerated conditions (20% H₂O₂ with 0.1 M CoCl₂ at 37 °C) with red dots denoting peroxide, green dots denoting antioxidants and blue dots displaying microparticles: (A) day 0, (B) day 3, (C) day 7, with the corresponding schematic displaying the erosion of membranes and reduction of strut mass, followed by SMP collapse over the same time period (D–F). 5 wt % microparticles-containing Piper SMP degradation behavior of the SMP composites (D, the equivalent of 2% H₂O₂ at 37 °C). Control SMP and Piper SMP at day 0 (G), day 3 (H), and day 7 (I) is presented with the corresponding schematic of the same time (J–L) displaying release of antioxidants accompanying the erosion and revealing the microparticles within the SMP matrix. This ultimately allows the foam to retain mass and the pore membranes over the initial 3 days in accelerated oxidative solutions.

By contrast, the antioxidant SMPs displayed increased resistant to physical changes. Membranes were found to remain intact over the initial 72 h, followed by the same etching process as the control. In the first 72 h, polymer etching is still occurring but is counterbalanced by the diffusion of antioxidants from the SMP matrix, as well as the release of antioxidants from the etching process. The antioxidants will scavenge the peroxy radicals at the SMP-solution interface and reduce oxidation of the amino-alcohols, and ultimately this reduces the polymer chain scission initially. After consumption of the antioxidants, SMP degradation occurs in the same manner as the controls and will eventually result in the collapse of the foam structure. Qualitatively, membrane porosity increased moderately, while the strut diameter decrease was faster than that for the control compared to the antioxidant SMPs, with composite SMPs displaying the greatest oxidation resistance overall compared to the antioxidant small molecules. This may factor into control of the gravimetric changes, as the gradual erosion of pores and struts places microparticles in direct contact with the solution, allowing for more rapid antioxidant diffusion.

Real Time Oxidative Biostability Calculations.

Based upon previously obtained in vivo data, the rate of oxidatively induced mass loss for the control composition is known, as are the degradation products of the SMPs.¹¹ The amino alcohols are known to oxidize to N-oxides, followed by lower amines and corresponding aldehydes, which may further oxidize to carboxylic acids or are part of the tissue integration process, depending on location and foreign body response.¹¹ The use of TEA is designed to increase long-term cytocompatibility of the particles and particle degradation products, as TEA will not form oxalic acid as a potential byproduct, unlike HPED.¹¹ In this same study, the real time rate of degradation in a porcine aneurysm was determined to be most accurately modeled by 2% H₂O₂, with the lifespan of the control materials found to be ∼800 days.¹¹ The rate of control SMP mass loss was scaled using the accelerated experimental tests presented in the previous section, and approximations of real time oxidative mass loss were calculated.¹¹ For Methyl SMP composites, the lifespan of the material is ∼1120 days. Piper and BHT had predicted lifespans of ∼1200 days or greater. These results indicate that this method is promising for improving biostabilty, but further confirmation is needed through in vivo analysis.

From a toxicological risk perspective of the degradation products, increasing the time to total mass loss (decreasing the rate of degradation) achieves a lower production rate of degradation products and a reduced risk, as the overall exposure rate per day is reduced by extending the material lifespan by more than 30%.¹¹ For the SMP matrix, toxicity assessments were briefly undertaken using a literature review of the degradation products, and it appears the oxalic acid is the most concerning degradation product.^50,51 In selecting TEA as the alcohol monomer, complete particle degradation does not risk the generation of oxalic acid or glyoxal and overall toxicity risk from the SMP is reduced. For oxalic acid, 0.54 mg/kg body weight/day (37.8 mg/day for an average 70 kg human) is the accepted maximum threshold for no adverse effects.^50,51 In the presented control SMPs, this corresponds to ∼37 mg total; the use of Piper or BHT to increase oxidative stability would, therefore, produce SMPs that generate an average of ∼31 μg oxalic acid per day, 3 orders of magnitude below the threshold for concern. From the cytocompatibility analysis, it appears that oxalic acid is the most concerning degradation product. In selecting TEA as the alcohol monomer of the particles, complete particle degradation is not limited by the generation of oxalic acid or glyoxal.

CONCLUSIONS

The use of phenolic and HALS antioxidants for improving oxidative stability of SMPs was demonstrated, with only the HALS antioxidant retaining functionality when directly incorporated into the SMP. However, phenolic antioxidants could be retained in the SMP matrix when incorporated in microparticles. The strain to failure and the elastic moduli were impacted by the inclusion of antioxidants directly in the SMP. However, the composite SMPs displayed no statistical examined. The retention of the antioxidants was examined using in vitro release studies, which ultimately was used to determine that small molecule antioxidants are not well retained in the SMP matrix for the examined formulations, but retention could be greatly enhanced through the use of microparticles. This retention was demonstrated to be sufficient for increasing the lifespan of the SMP by 25% to 50%, according to calculated values, depending on the antioxidant and incorporation method. For the clinical use as a biomaterial, the incorporation of antioxidant-containing microparticles to reduce degradation was demonstrated to lower the exposure level per day of a degradation product to 3 orders of magnitude below the established limit of concern for oxalic acid, demonstrating the value of this method.