Biomacromolecules 2003, 4, 216-223
216
Characterization of Polylactide-b-polyisoprene-b-polylactide Thermoplastic Elastomers Esther M. Frick, Andrew S. Zalusky, and Marc A. Hillmyer* Department of Chemistry, University of Minnesota, Minneapolis, Minnesota 55455 Received August 1, 2002; Revised Manuscript Received December 20, 2002
Model polylactide-b-polyisoprene-b-polylactide (PLA-PI-PLA) triblock copolymers were prepared by an efficient protocol starting with R,ω-dihydroxy polyisoprene (HO-PI-OH). Using a moderately electrophilic Al(O-i-Pr)3 catalyst and carefully controlling the ratio of Al to HO-PI-OH avoided gel formation and resulted in acceptable lactide polymerization rates. The triblock copolymers were free of homopolymer or diblock contaminants as determined by chromatographic and spectroscopic analyses. Three representative PLA-PI-PLA materials were prepared with spherical, cylindrical, and lamellar morphologies as confirmed by small-angle X-ray scattering and transmission electron microscopy. We employed dynamic mechanical analysis and tensile testing to assess the viscoelastic and mechanical behavior. The morphology largely determined the tensile properties of these materials, with the Young’s modulus and ultimate tensile strength following predicted trends. Excellent elongations were achieved especially for the PLA-PI-PLA sample with the cylindrical morphology, and the PLA-PI-PLA sample with the spherical morphology showed the best elastomeric recovery. Microphase alignment and pull rate significantly influenced the resultant tensile properties. Introduction Polylactide (PLA) is a thermoplastic with various applications in the medical and pharmaceutical fields due to its biodegradability and biocompatibility.1 Of equal appeal is the fact that lactic acid, a precursor to polylactide, is derived from renewable resources such as corn. Furthermore, PLA is recognized for its excellent crease and crimp properties, oil resistance, and superior barrier properties.2 This combination of factors has contributed to the motivation behind the recent large-scale production of polylactide, an indication of its increasing commercial potential.3 PLA is most commonly synthesized by the ring-opening polymerization of lactide, the cyclic dimer of lactic acid. The stereochemistry along the polymer backbone largely dictates the properties of PLA. Polymerization of L-lactide (S,S) or D-lactide (R,R) results in isotactic, semicrystalline (Tm ≈ 180 °C) poly(L-lactide) (PLLA) or poly(D-lactide) (PDLA), respectively. Polymerizing the racemic mixture typically yields atactic (Tg ≈ 60 °C), amorphous poly(D,L-lactide) (PLA).2,4 Due to their similar high modulus characteristics, PLA is often compared to polystyrene (PS). However, like PS, the modest impact resistance of polylactide has limited its widespread applicability. To combat this problem, plasticization, blending, and block copolymerization strategies have been investigated for the formation of new tough PLA composites.5,6 To improve the properties and expand the applicability of polylactide, we have been specifically interested in the synthesis of polylactide containing ABA triblock copolymers, * To whom correspondence should be addressed (e-mail: hillmyer@ chem.umn.edu).
where A is polylactide and B is a low Tg, polylactide immiscible, amorphous polymer.7 These materials typically have improved physical characteristics over the parent homopolymer (e.g., improved impact strength) and can behave as thermoplastic elastomers (TPEs). Common TPEs generally exhibit high tensile strengths and large reversible elongations similar to conventional vulcanized rubbers.8 This behavior is due to the “physical cross-linking”8,9 that results from the microphase separation of the immiscible polymer segments. The product of the Flory-Huggins interaction parameter for the specific monomer pairs (χ) and the overall degree of polymerization (N) determines the degree of microphase segregation.10 At specific volume fractions, microphase separation results in hard (high Tg) microdomains that behave as cross-links in a soft (low Tg) matrix. The absence of chemical cross-links allows TPEs to be processed like conventional thermoplastics. Polystyrene-b-polybutadiene-b-polystyrene (SBS) and polystyrene-b-polyisoprene-bpolystyrene (SIS) triblock copolymers are TPEs commonly used as blend compatibilizers, elastomers, asphalt modifiers, and pressure-sensitive adhesives.8 Though TPEs are well developed, the synthesis of biocompatible, biodegradable, renewable resource derived TPEs would expand their utility in the medical, pharmaceutical, and agricultural fields.8,11 Furthermore, the application of polylactide TPEs in toughened polylactide composites is an exciting and emerging research area. We previously reported the synthesis and characterization of polylactide-b-polyisoprene-b-polylactide (PLA-PI-PLA) linear ABA triblock copolymers (Figure 1).12 The large interaction parameter between PI and PLA results in microphase separation at relatively low molecular weights (i.e.,
10.1021/bm025628b CCC: $25.00 © 2003 American Chemical Society Published on Web 02/11/2003
Thermoplastic Elastomers
Figure 1. Polylactide-b-polyisoprene-b-polylactide (PLA-PI-PLA) triblock copolymer. The PLA was prepared from D,L-lactide and is amorphous. The PI block contains >90% of the 4,1-regioisomers as determined by 1H NMR spectroscopy.
small values of N).6g At the appropriate compositions, these new materials have potential as TPEs, and our reported controlled synthetic protocol yields materials free of diblock and homopolymer, which is imperative to achieve desired elastomeric properties.8 We used simple manipulation of the initiation and termination steps in the anionic polymerization of isoprene to achieve narrow molecular weight R,ωdihydroxy functionalized polyisoprene (HO-PI-OH).13 Using HO-PI-OH as a macrointiator for the polymerization of lactide resulted in a straightforward route to model PLAPI-PLA.14 With PI as the midblock we can readily compare our new materials to other PI-containing TPEs. Moreover, the entanglement molecular weight, Me, is relatively low for polyisoprene (Me ) 6.4 kg/mol),15 and since “slippage” of the entangled chains during deformation is implicated in delayed brittle domain fracture, a moderate molecular weight (Mn ≈ 35 kg/mol) PI midblock is sufficient for high-strength TPEs.16 One synthetic limitation of our previously reported scheme was the low concentration of lactide used in the polymerization reactions. To avoid gelation during the formation of the aluminum alkoxide macroinitiator,17 the previously reported triblock copolymers were synthesized in very dilute solutions ([lactide]0 ≈ 0.1 M).12 To avoid the use of excess solvent for large-scale syntheses (required for mechanical property studies), we turned to an alternative controlled protocol. By using a less electrophilic aluminum catalyst, namely, aluminum triisopropoxide (Al(iOPr)3), instead of triethylaluminum (AlEt3), as well as a modified synthetic protocol, we were able to prepare PLA-PI-PLA triblock copolymers in larger quantities and at higher reaction concentrations.18 We herein report an efficient large-scale synthesis of PLA-PI-PLA thermoplastic elastomers and their morphological, rheological, and mechanical characterization. Experimental Section Materials. D,L-Lactide (Aldrich) was recrystallized from ethyl acetate, dried at room temperature under vacuum for 20 h, and stored under argon in an inert atmosphere drybox. Aluminum triisopropoxide, Al(iOPr)3 (Aldrich), was purified by a previously reported method in order to obtain the more reactive aggregate, namely, the A3 trimer.18 In short, the solid Al(iOPr)3 was distilled (10-3 mTorr, 150 °C) into a 5 mL glass ampule that was sealed to ensure inert conditions. The ampule was immersed in a 150 °C oil bath for 4 h and subsequently quenched in liquid nitrogen. After slowly warming to room temperature, the product was a viscous, yellow-tinted liquid presumed to contain predominately the active A3 aggregate. A stock solution of the isolated Al(iOPr)3 trimer in dry toluene (0.1 M) was made and stored
Biomacromolecules, Vol. 4, No. 2, 2003 217
in an inert atmosphere. Toluene used for all polymerizations was obtained through an in-house solvent purification line.19 All other solvents were purchased from Aldrich and used as received. Block Copolymer Synthesis. The following synthesis is for triblock L (Table 1) and is representative of the synthetic protocol followed to obtain the materials referred to in this paper. The polymerization was conducted in a 1-L, roundbottom, three-port glass reactor containing baffles to maximize vigorous mixing. The reactor was equipped with a Dean-Stark trap, a condenser, a nitrogen inlet, and a rubber septum. Overhead, motorized stirring was employed to ensure adequate mixing, and the middle port of the reactor was reserved for the glass shaft used for stirring. Twenty grams (6.1 × 10-4 mol) of previously prepared R,ω-hydroxyl polyisoprene (HO-PI-OH)12 was dissolved in 250 mL of toluene and added to the reactor. The reactor was flushed with nitrogen, sealed, and heated to reflux (∼120 °C) under nitrogen. Any water contamination was minimized by removal of ∼20 mL of solution in the Dean-Stark trap (azeotropic removal). A 1.9 mL (1.9 × 10-4 mol) portion of previously prepared 0.1 M Al(iOPr)3 stock solution was collected and removed from the drybox in an airtight syringe and immediately injected into the reaction through the septum. The solution was stirred at reflux for 20 h to give sufficient time for the aluminum alkoxide macroinitiator to form. Any 2-propanol byproduct generated during the reaction was assumed to be isolated and removed in the Dean-Stark trap (∼30 mL) based on the purity of the resultant triblock copolymers. Under a positive nitrogen pressure, the septum was removed from the reactor and 24.0 g (0.17 mol) of lactide was quickly added ([L]0 ≈ 1 M). The reactor was resealed with a greased glass stopper, and the reaction was allowed to stir at reflux for 4 h. The solution was cooled to room temperature and precipitated in cold methanol. After being stirred in ∼1 L of methanol overnight, the white precipitate was vacuum filtered and dried at room temperature for 10 h and 60 °C for 10 h in a vacuum oven. The isolated, dried product was weighed, and the mass yield was ca. 85% (assuming complete recovery of the PI). Triblock copolymers were always stored in the freezer (T ≈ 0 °C). Size Exclusion Chromatography (SEC). SEC samples were prepared using HPLC grade THF (ca. 10 mg/1 mL of THF). SEC measurements were taken on a Hewlett-Packard 1100 series liquid chromatograph (LC) using THF as the mobile phase and calibrated to polystyrene standards (Polymer Laboratories). The analysis was performed at 40 °C. The LC was equipped with a Hewlett-Packard refractive index detector and three Jordi polydivinylbenzene columns of 104, 103, and 500 Å pore sizes. 1 H and 13C NMR Spectroscopy. Samples for 1H NMR and 13C NMR analysis were dissolved in CDCl3 (Cambridge) (ca. 50 mg/1 mL of deuterated solvent). All spectra were acquired on a Varian Inova VI-500 spectrometer (delay time ) 3 s, transients g32). Dynamic Mechanical Spectroscopy (DMS). DMS specimens were prepared in stainless steel molds (0.5 mm in thickness) by compression (2000 psi) between two Teflon-
218
Biomacromolecules, Vol. 4, No. 2, 2003
Frick et al.
Table 1. PLA-PI-PLA Triblock Copolymer Characterization Data triblock
Mn(PLA)-Mn(PI)-Mn(PLA)a (kg/mol)
fPLAb
PI PDIc
block PDIc
morphologyd
De (nm)
Tg(PI)f (°C)
Tg(PLA)f (°C)
S C L
5.1-35-5.1 9.0-33-9.0 14-33-14
0.18 0.28 0.40
1.07 1.08 1.08
1.12 1.12 1.10
PLA spheres PLA cylinders lamellar
29 38 45
-62 -61 -63
52 61 60
a Calculated from 1H NMR spectroscopy. b Volume fraction of PLA calculated using F 3 3 15,22 (PLA) ) 1.154 g/cm and F(PI) ) 0.830 g/cm at 140 °C. Determined by SEC (see Experimental Section). d Determined by SAXS and TEM. e Principal domain spacing ) D ) 2π/q* (SAXS): T ) 120 °C for triblocks C and L, T ) 90 °C for S. f Determined by DSC (see Experimental Section).
c
Table 2. Tensile Data for Triblock Copolymersa triblock
Young’s modulus E ( sd (MPa)
yield strain Y ( sd (%)
yield stress σY ( sd (Mpa)
strain at break B ( sd (%)
ultimate tensile strength σB ( sd (MPa)
S C L
6.7 ( 0.6 3.8 ( 11 150 ( 26
55 ( 9.1 5 ( 0.8 4 ( 0.5
1.5 ( 0.4 2.9 ( 0.7 5.5 ( 1.0
200 ( 40 650 ( 70 450 ( 60
3.1 ( 0.9 9.2 ( 1.9 10.1 ( 1.8
a All data taken following ASTM D178 (see Experimental Section for details). All data reported are averages of five samples and sd is the standard deviation of these measurements.
lined plates at elevated temperatures (T ) 80-140 °C, depending on PLA molecular weight). The viscoelastic properties of the samples were measured using a Rheometrics RSAII solids analyzer equipped with a shear sandwich fixture. Nitrogen purge gas was used during measurements in order prevent oxidative damage to the samples during deformation. Initial strain () and frequency (ω) scans were conducted to ensure that the temperature ramp experiments were taken within the linear viscoelastic regime. Temperature sweeps were taken at ramp rates of 2 °C/min (ω ) 0.5 rad/ s, ) 1%). Frequency sweeps were also conducted at elevated temperatures (T ) 70-100 °C, ) 1%). Differential Scanning Calorimetry (DSC). DSC measurements were acquired using a Perkin-Elmer Pyris 7 DSC. All samples were initially heated from room temperature to 100 °C at 50 °C/min and held at 100 °C for 5 min to erase the thermal history of the samples. The samples were then cooled to -100 °C at 50 °C/min and held at -100 °C for 5 min. DCS data were taken from -100 to 100 °C at a ramp rate of 10 °C/min. Small-Angle X-ray Scattering (SAXS). Unless otherwise noted in the text, samples used for the SAXS data presented in this paper were freshly precipitated, powder samples. The samples underwent no pressing or thermal treatment after vacuum oven drying at room temperature and 60 °C. SAXS data was obtained on a custom-built beamline. Cu KR X-rays (wavelength ) 1.542 Å) were generated by a Rigaku RU200BVH rotating anode X-ray machine fitted with a 0.2 × 2 mm2 microfocus cathode and Franks mirror optics. The sample temperature was controlled inside the evacuated sample chamber by an electrically heated brass block that could be water cooled. Two-dimensional diffraction patterns were recorded using a Siemens multiwire area detector and corrected for detector response before analysis. Azimuthally isotropic two-dimensional scattering patterns were averaged to the one-dimensional form of intensity versus scattering wave vector q ) 4λ-1sin(θ/2), where λ and θ are the radiation wavelength and scattering angle, respectively. Scattering at very low q attributed to anomalous scattering around the beam stop was removed from the data. Transmission Electron Microscopy (TEM). TEM images were obtained using a JEOL 1210 TEM instrument
operating at 120 kV. Thin film samples were prepared by spin-coating using the following method. Approximately 50 mg of triblock copolymer was dissolved in 1 mL of toluene. The solutions were then spun directly onto carbon-coated copper grids (Ted Pella) at 2000 rpm using a home-built spin-coating apparatus. Contrast between the PI and PLA blocks was achieved by exposing the samples to vapor from a 4% aqueous solution of osmium tetroxide (Ted Pella) for approximately 10-15 min. Tensile Testing. Tensile testing samples were prepared by pressing the triblock copolymers in “dog-bone” stainless steel molds. The dimensions of the tensile bars are given in Figure 8. The samples were pressed at elevated temperatures (70-140 °C, depending on PLA molecular weight) and subjected to a series of pressures (500, 1000, 1500, 2000, and 2500 psi, 5 min at each pressure) to minimize the occurrence of air bubbles. Tensile measurements were taken on an Instron 5500 following ASTM D178 with a pull rate of 10 mm/min unless otherwise noted in the text. At least five samples were measured and averaged for the data reported in Table 2. Results and Discussion Synthesis. Our previously reported synthetic protocol required very dilute reaction conditions in order to avoid gelation during the macroinitiator formation step.12 To conduct more concentrated reactions, aggregation of the active metal alkoxide macroinitiator chain ends would have to be prohibited. Consequently, we turned to a less electrophilic metal catalyst. Though there are many catalysts known to facilitate the metal-catalyzed ring-opening polymerization of lactide, we chose an aluminum system since many alkyl and alkoxide aluminum catalysts are commercially available and have been studied in detail for these polymerizations.20 In addition, a recent kinetic study on aluminum alkoxide catalysts used for lactide polymerizations elucidated detailed kinetic and mechanistic aspects of these reactions.21 All triblock copolymer polymerizations reported in this paper were catalyzed by aluminum triisopropoxide (Al(iOPr)3). To avoid gelation complications, we considered the macroinitiator formation reaction step as a simple “A2 + B3”
Biomacromolecules, Vol. 4, No. 2, 2003 219
Thermoplastic Elastomers
polycondensation reaction. In our case, the HO-PI-OH precursor polymer was labeled “A2” (A ) -OH) and Al(iOPr)3 catalyst “B3” (B ) -iOPr). For these types of reactions, simple gel-point calculations can be used to determine the [B] to [A] ratio necessary to avoid a gelled network using eq 117 rB )
[B] 1 ) [A] (pB,c)2(f - 1)
(1)
where pB,c is the critical extent of conversion of the B functionality at the gel point, rB is the molar ratio of B functional groups to A functional groups, and f is the functionality of the branched monomer, which in this reaction is three. At 100% conversion of the B functionality (pB,c ) 1), eq 1 gives rB ) 0.5. In other words, if all of the -iOPr groups were to react during macroinitiator formation, a ratio of [iOPr]0/[OH]0 of 0.5 would be the minimum value of rB required in order for a gel to occur. Therefore, reactions using a value of rB less than 0.5 would theoretically ensure the absence of a gel. However, decreasing rB requires reduced amounts of catalyst (i.e., Al) in the reaction mixture, and very low levels of Al naturally lead to slow reaction rates. In test polymerizations we observed little or no conversion for reactions with significantly low rB e 0.1 even when employing extended reaction times (i.e., many days). Therefore, a proper balance had to be established, and we found that using rB ≈ 0.45, afforded “gel-free” polymerizations, good reaction rates, and high conversions at relatively high solution concentrations ([lactide]0 ) 1-2 M).18 Using the aforementioned protocol (See Experimental Section for details), all PLA-PI-PLA triblock copolymers were synthesized in a controlled manner and exhibited narrow molecular weight distributions and predictable molecular weights (Table 1). The triblocks were typically prepared on a 40-50 g scale and were free of PI or PLA homopolymer contamination as determined by SEC and 1H NMR spectroscopy.12 Morphology. We targeted the three most common equilibrium morphologies, namely, spheres (typically packed on a lattice), hexagonally packed cylinders, and lamellae. We expected well phase separated systems even for our lowest molecular weight sample due to the large polymer interaction parameter (χ) for PLA and PI. χN values greater than 100 for all of the samples at 140 °C were calculated.23 A combination of TEM and SAXS was used to determine the sample morphologies. All one-dimensional SAXS data showed a principal peak (See Table 1) and higher order reflections. Considering the volume fraction of PLA in these samples (Table 1), the scattering was consistent with PLA spheres in a matrix of PI for triblock S, hexagonally packed cylinders of PLA in a matrix of PI for triblock C, and alternating lamellae of PLA and PI for triblock L (See Figure 2). For sample S, the scattering is most consistent with disorganized spheres of PLA in a matrix of PI. Modeling of the SAXS data for sample S using a spherical form factor and the PercusYevick model,24 we determined a sphere radius of 8 nm (this is consistent with the TEM measurements below). As
Figure 2. 1-D SAXS patterns for triblocks S (T ) 90 °C), C (T ) 120 °C), and L (T ) 120 °C). The triangles are at 2q*, 3q*, 4q*, and 5q* for L and x3q*, x4q*, x7q*, x9q*, and x12q* for C.
expected the D-spacing increased with increasing PLA molecular weight using almost identical PI center blocks (Table 1). Additionally, DSC confirmed the expected microphase separation showing Tg values for PI (ca. - 60 °C) and a Tg between 52 and 61 °C for PLA (Table 1).20e Triblock S gave a slightly lower Tg value for PLA than typically reported values for PLA homopolymer, no doubt due to the relatively low molecular weight of the PLA segment in this material. For the C and S samples, TEM analysis confirmed the morphology as determined by SAXS. Thin film samples (thickness ca. 50 nm) of C and S obtained by spin casting from toluene and stained with OsO4 (selective for PI) were subjected to TEM (Figure 3). In Figure 3a, spheres of PLA in a PI matrix are obvious for sample S and no regular packing is easily identifiable. This was processed using Scion Image to extract an average sphere diameter of 9 nm.25 This result is consistent with the SAXS assignment for triblock S. Figure 3b shows the TEM image of the spin-coated film for triblock C. Cylinders of PLA are evident in this image, and the diameter of the cylinders (diameter ≈ 23 nm) is consistent with the SAXS data (diameter ≈ 24 nm). Spincoating of sample L gave samples that could not be adequately analyzed by TEM. This is most likely due to a stacking of the lamellae parallel to the substrate.26 Dynamic Mechanical Analysis. The elastic (G′) and loss (G′′) components of the dynamic shear modulus were measured as a function of temperature. Figure 4 shows G′ and G′′ as a function of temperature from 25 to 225 °C for the three triblock copolymers.27 All triblock copolymers showed a predicted decrease in G′ around the Tg of the PLA component. No order-disorder transitions (ODTs) were observed for these samples as expected.6g The relatively temperature independent plateau (changes in G′ of less than 4% from 100 to 200 °C) seen especially for triblocks C and
220
Biomacromolecules, Vol. 4, No. 2, 2003
Frick et al.
Figure 4. DMS temperature sweeps for triblocks S, C, and L (temperature ramp rate ) 2 °C/min, frequency (ω) ) 0.5 rad/s, strain () ) 1%): G′, filled circles; G′′, open circles.
Figure 3. TEM images of (a) triblock S and (b) triblock C. The samples were cast in toluene and spin-coated directly onto carboncoated copper grids. The PI phase has been selectively stained with OsO4 and appears dark in the images.
L corroborates previous SEC and 1H NMR analyses that these triblocks are free of significant diblock contamination.28 The negative slope for the plateau region in the triblock S data is attributed to the very short PLA segments leading to “weaker” hard domains (especially at high temperature or lower values of χN where the interfacial width becomes significant). In addition to expected rheological responses, the triblocks also displayed a viscoelastic feature around 70-80 °C (Figure 4). This transition is most clearly seen for triblocks C and L. The transition is both reproducible and reversible; the change in modulus is observed upon heating and cooling and is independent of sample preparation history. Diblock contamination would presumably result in one continuous transition over a broadened temperature range instead of two distinct transitions.8,28a Viscoelastic changes seen due to morphological changes in the sample (OOT) during heating were also ruled out since we did not observe other morphologies in analogous SAXS analyses (40-100 °C).29 On the
basis of reported rheological data for PI, we do not believe this is due to the transition from the rubbery plateau of the PI midblock to terminal flow behavior.30 At this point we are unable to unequivocally determine the nature of this transition; however, it simply may be that over this temperature range (70-80 °C) and analysis frequency (0.5 rad/s) we are fortuitously probing the transition between block copolymer chain dynamics (i.e., entanglements) and microstructural dynamics.31 Tensile Properties. The tensile properties of the PLAPI-PLA triblock copolymers were characteristic, for the most part, of traditional TPEs. At low strains, a linear response was observed in the stress-strain curve for all of the samples (Figure 5). Representative stress-strain plots for the triblock copolymers pulled to ultimate tensile failure are shown in Figure 6, and Table 2 gives the relevant tensile data. The increase in Young’s modulus is expected with the increasing PLA volume fraction in the samples.32 The ultimate tensile strengths (σB) of the samples increase with increasing PLA content as expected and are lower than reported σB values for common commercial PS-based TPEs (ca. 30 MPa). Beyond the low strain elastic region the materials (particularly triblocks C and L) show yielding behavior that is not uncommon for TPEs and indicative of a semicontinuous PLA phase dispersed in a PI matrix (Figure 6, Table 2).32 Relatively high elongations were achieved for all samples (Table 2). The recovery or “true elasticity” of all of the PLA-PIPLA triblock copolymers was assessed by subjecting the
Thermoplastic Elastomers
Figure 5. Representative stress-strain curves for triblocks S, C, and L pulled to 20% strain (ASTM D178).
Biomacromolecules, Vol. 4, No. 2, 2003 221
Figure 7. Three consecutive loading (open symbols) and unloading (filled symbols) cycles for triblock S pulled to ca. 9% strain: first cycle, circles; second cycle, squares; third cycle, triangles.
Figure 6. Representative stress-strain curves for triblocks S, C, and L pulled to ultimate tensile failure (ASTM D178).
samples to cycles of loading and unloading at different degrees of applied strain. We observed during these experiments that independent of the applied deformation, all of the triblock copolymers exhibited some “stress-softening”, often seen for common PS-based TPEs.8,33 This behavior is indicated in the stress-strain data by a lower modulus upon the second applied load, and no evidence of the yielding behavior is observed in the first cycle. Successive cycles usually closely resemble the response seen for the second cycle as exemplified for triblock S in Figure 7. This behavior has been explained by the initial breaking up of the hard domains into smaller domains (in the case of the C and L samples) upon the first applied load.34 These domains remain “broken” for the duration of the cyclic recovery experiment, but in related examples “healing” and re-formation of larger domains over time and with different annealing techniques have been observed.35 We quantified the recovery of our PLA-PI-PLA samples by noting the residual strain in the sample after the applied load was completely released. At low strains (
