Photoinduced topological transformation of cyclized polylactides for switching the properties of homocrystals and stereocomplexes

Cyclized poly(L-lactide) and poly(D-lactide) (Mn ∼ 3 kDa) incorporating an o-nitrobenzyl group as a photocleavable linker were synthesized and photoirradiated for topological transformation to form photocleaved linear polylactides. By DSC, Tm of the cyclized stereocomplex (167 °C) decreased by more than 40 °C from that of the linear prepolymers (209 °C) despite their essentially identical molecular weights. Upon the photocleavage, the resulting linear stereocomplex showed almost the same Tm (211 °C) as that before the cyclization. The enthalpy of melting of crystals having an infinite thickness, i.e. ΔHm(100%), and the surface free energy (σe) were determined by the combination of WAXD, SAXS, and DSC. Both ΔHm(100%) and σe were considerably smaller for the cyclized polylactide homocrystals and stereocomplexes than those of the linear prepolymers and photocleaved products. These suggest that the absolute enthalpy of the melt state is lower, and the crystalline–amorphous interface is more stable for the cyclized polylactides arising from the topology.


Introduction
Stimuli-responsive polymers have attracted great interest and been developed for numerous applications. 1 For example, poly (N-isopropylacrylamide) is known for a thermal responsive polymer that shows a lower critical solution temperature at around 32°C by dehydration. 2 pH-responsive polymers are often employed for lithography 3 and drug delivery systems (DDS) 4 by switching their solubility through deprotection and (de)protonation. Furthermore, DDS 5 and self-healing materials 6 using redox-responsive polymers were reported. However, these responses require a stimulus on essentially every monomer unit, and thus a considerable amount of heat, acid/base, or oxidant/reductant is indispensible. In order to establish a notably more efficient stimuli-responsive polymer system, we expected cyclic polymers to have substantial potential because only one cleaving reaction per polymer molecule leads to cyclic-to-linear topological transformation to trigger changes in the properties. Furthermore, the molecular weight and chemical structure of the main chain are unaffected through the topological transformation. In this regard, cyclic polymers were reported to show significantly enhanced material properties compared to relevant linear polymers, including the stability of micelles. 7 The reversible linear-cyclic topological transformation of polystyrene and poly(ethylene oxide) was reported by making use of thiol-disulfide conversion, 8 hydrogen bonding, 9 and dimerization of porphyrin. 10 In addition, photocleavage of cyclic morpholino 11,12 and oligonucleotide 13 was sometimes employed to control gene expression. Nevertheless, topological transformation has never been applied to stimuli-responsive polymeric materials.
We previously synthesized cyclic poly(L-lactide), PLLA (2L), and poly(D-lactide), PDLA (2D) (number average molecular weight (M n ) ∼3 kDa, Fig. 1a), as well as cyclic stereoblock polylactides, and reported the melting points (T m ) of their homocrystals and stereocomplexes 14,15 using differential scanning calorimetry (DSC). 16 The linear and cyclic stereocomplexes showed distinguishable T m , and we expected that this stereocomplex system can be applied for a topology-dependent stimuli-responsive polymeric material. Meanwhile, Waymouth and co-workers reported DSC, wide-angle X-ray diffraction (WAXD), and small-angle X-ray scattering (SAXS) of cyclic PLLA and PDLA having a relatively large molecular weight (≥26 kDa) synthesized by zwitterionic ring-opening polymerization. 17 These cyclic polylactides exhibited similar thermal and crystallographic properties to their linear counterparts. In regard to this, polylactide stereocomplexes are known to form folded chain crystals of molecular weights larger than several thousands. 18 The folded chain crystals likely resulted in dimin-ished and/or complicated influences by the polymer topology. In addition, the optical purity of these cyclic polylactides was modest due to partial racemization during the ring-expansion polymerization.
In the present work, cyclic PLLA and PDLA with an M n of approximately 3 and 1.5 kDa incorporating an o-nitrobenzyl group (NB) as a photocleavable linker 19 were synthesized via highly optically pure polymerization (Fig. 1b). These homopolymers and their stereocomplexes were subjected to WAXD, SAXS, and DSC measurements to study the effects of the polymer topology on the extended chain crystals. Furthermore, cyclic-to-linear topological transformation by photoirradiation was performed. T m of the stereocomplex, which decreased upon cyclization by 42°C (T m before the cyclization, 209°C; T m after the cyclization, 167°C), was restored to 211°C by photocleavage. The difference in T m arose from the crystal thickness (l c ) in the lamellar structure, 20 directly reflecting the topology of the polymers. Furthermore, the enthalpy of melting of crystals having an infinite thickness, i.e. ΔH m (100%), and the surface free energy (σ e ) of the crystalline layers were significantly smaller for the homocrystals and the stereocomplex of the cyclic polylactides. Coupled with the recently attracted bio-based, bio-degradable, and carbonneutral features of polylactides, the switch of the properties by topological transformation should find various applications.

Results and discussion
Synthesis of photocleavable cyclic PLLA (5L) and PDLA (5D) with a NB group A NB group, which is one of the most widely used photolabile protecting groups due to its prompt removal, was chosen as the photocleavable linker. 19 The cyclic-to-linear topological transformation was expected to be enabled by the molecular design that incorporates the photocleavable NB linker in the main chain of the cyclic polymers (Fig. 1c). Cyclic PLLA (5L) and PDLA (5D) with a photocleavable linker were prepared using a similar method to that of 2L and 2D, which were reported previously ( Fig. 1a and b). 16 In this regard, the employed polymerization is known to form highly optically pure polylactides without degradation in the stereochemistry. 21 The intermediate and final products were fully characterized using 1 H NMR, SEC, and MALDI-TOF MS (Fig. S1-S3 †).

Photocleavage
To achieve selective photoinduced cyclic-to-linear topological transformation, the conditions for the photocleavage reaction were investigated. The photoirradiation experiments were first conducted with the most available linear polylactides with a photocleavable linker at the chain end (3L and 3D) under various conditions to form 7L and 7D, respectively (Fig. 1d). The experiments were performed using a wavelength of 365 nm and an intensity of 700 mW cm −2 . 19 A polylactide has a very weak absorption at 365 nm; the main chain should not be directly decomposed by photoirradiation. 22 Good solvents for polylactide homopolymers in the absence of absorption around 365 nm such as CH 2 Cl 2 , CHCl 3 , THF, toluene, CH 3 CN, 1,2-dichloroethane, and EtOAc were used. 22 At a polymer concentration of 0.5 mg mL −1 , 30 min of irradiation caused decomposition and crosslinking of the main chains, as observed by SEC for all of these solvents (Fig. S4 †). The photocleavage of the NB group generated unstable nitroso compounds, which spontaneously collapsed to form radicals, leading to the radicalmediated degradation of the polylactide main chain. 19 Therefore, a radical scavenger (butylated hydroxytoluene, BHT) was added, and the decomposition and crosslinking were found to be effectively suppressed (Fig. S4b, † bottom).
On the basis of these results, the photocleavage of cyclic 5L and 5D was investigated. The SEC traces after photoirradiation (6L and 6D) in the presence of BHT exhibited suppression of the main chain degradation (Fig. S5 †). The photocleavage was observed after 2 h by noticeable peak top shifts toward the smaller elution volume. The peak molecular weight of 6L (M p = 5.0 kDa), which represented the hydrodynamic volume, was nearly equal to that of linear prepolymer 4L (M p = 5.2 kDa), in agreement with previously reported linearization reactions. 8,23 Moreover, the SEC trace of 6L also has a shoulder peak at M p = 11 kDa, which is comparable to that of a dimer of 6L (2 × 5.0 kDa = 10 kDa). A similar shoulder also appeared in 6D (Fig. S5 †). The dimers presumably formed by the intermolecular coupling of the nitroso groups, which were resulted by the photocleavage (Fig. S6 †). 24 During the photocleavage process, the nitroso group remained at the terminus of the polylactide chains and likely formed the dimer by the nitrosonitroso association. 25 Furthermore, trimer or higher multimer formation was not detected, suggesting that the dimerization occurred through this mechanism. This hypothesis was also supported by the photocleavage reaction of linear PLLA possessing a NB group (3L) in Fig. S4. † The dimerization of photocleaved 3D in the presence of BHT was not observed because 3D had a molecular structure to detach the nitroso group from the main chain upon the photocleavage (Fig. 1d). By comparing the 1 H NMR spectra of 5L and 6L, the signals for the NB group ("f" at 6.38-6.46 ppm, "d" at 6.91 ppm, and "e" at 7.63 ppm) clearly decreased after photoirradiation (Fig. S7 †). The small peaks remaining in the aromatic region of 6L could be due to the nitroso-nitroso association products. 25 A decrease of signal "o" in 6L (Fig. S7 †) suggests that this photocleavage process is accompanied by some other side reactions on the alkene due to the reactivity of the intermediate radicals and/or the nitroso group. Based on these results, it was concluded that the cyclic-to-linear topological transformation was successfully complete. The photocleavage in the solid state was also studied. Despite many attempts under various conditions, the decomposition of the polylactides occurred before the completion of the intended photocleavage reaction. Thus, the solid-state topological conversion is currently investigated.

X-ray structures
The X-ray structural analysis of the solvent-cast samples of the previously reported non-photocleavable linear and cyclic polylactides without a NB group (1L, 1D, 2L, 2D, 1L/1D, 1L/2D, and 2L/2D) 16 was first conducted. The WAXD profiles of individual 1L, 1D, 2L, and 2D showed diffraction peaks at 2θ = 15, 17, and 19°that are ascribed to the αor α′-forms of the polylactide homocrystals (Fig. S8, † left). 26 The degree of crystallinity (χ c ) of the homocrystals was estimated by the profiles (1L, 63%; 1D, 58%; 2L, 64%; 2D, 50%).  (Table S1 †). 27 The direction of the thickness of the polylactide homocrystals is the same as the c axis of the unit cell, which is 2.886 nm consisting of ten lactic acid units. 28 Thus, based on the above-mentioned l c , the number of lactic acid units and its molecular weight in the crystal layers were 1L, 28 units for M n = 2.0 kDa; 1D, 33 units for M n = 2.3 kDa, 2L, 14 units for M n = 1.0 kDa, and 2D, 12 units for M n = 0.9 kDa (Table S1 †). On the basis of M n determined by NMR (∼3 kDa), these results suggest that linear 1L and 1D form extended chain crystals with the residual lengths of the polymer chains existing in the amorphous layers (Fig. 2a). 29 In the case of 2L and 2D, the polymer chain supposedly formed the flattened conformation shown in Fig. 2b, and thus l c became nearly half of that of 1L and 1D. 30 Furthermore, the SAXS peak became dull for cyclized 2L and 2D (Fig. S9 †). This indicates that the crystal lattice was distorted likely due to the difficulty in the alignment of the polymer chains restricted by the cyclic topology.
The SAXS profiles of the enantiomeric blends are also shown in Fig. S9, † and the values of L, l c and l a were estimated (1L/1D, L = 12 nm, l c = 10 nm, l a = 2.1 nm; 1L/2D, L = 8.8 nm, l c = 7.2 nm, l a = 1.6 nm; 2L/2D, L = 7.0 nm, l c = 5.4 nm, l a = 1.6 nm) as shown in Table S1. † 27 The direction of the thickness of a polylactide stereocomplex is the same as the c axis of the unit cell, which is 0.87 nm consisting of three lactic acid units. 18,31,32 Thus, based on the above-mentioned l c , the number of lactic acid units and the molecular weights in the crystal layers were 1L/1D, 34 units for M n = 2.5 kDa; 1L/2D, 25 units for M n = 1.8 kDa, and 2L/2D, 17 units for M n = 1.3 kDa (Table S1 †). Given that M n ∼ 3 kDa, these results suggest that, similar to the homocrystals, 1L/1D forms extended chain crystals with the residual lengths of the polymer chains existing in the amorphous layers (Fig. 2a). In the case of 2L/2D (l c = 5.4 nm), the polymer chains form the flattened conformation shown in Fig. 2b, and thus l c became nearly half of that of 1L/1D (l c = 10 nm). 30 In the case of the linear/cyclic blend (1L/2D), the L and l c values were in between the linear/linear and cyclic/cyclic blends.
The WAXD peaks at 2θ = 12, 21, and 24°, attributing to a stereocomplex, were observed for 4L/4D, 4L/5D, and 5L/5D (Fig. 3). 26 Additional diffractions at 2θ = 13 and 18°were detected for the stereocomplexes involving the cyclic polylactides (4L/5D and 5L/5D). We assume that the photocleavable linker in the cyclic polymer that is incorporated in the crystalline layer caused these diffractions (Fig. 2b). In contrast, the photocleavable linker in the linear polymers, which was attached at the chain end, was presumably in the amorphous layer (Fig. 2a). Based on the WAXD profiles, the χ c values of the stereocomplexes were estimated as shown in Table 1 (4L/4D, 69%; 4L/5D, 54%; 5L/5D, 61%). In this regard, the linear/ linear blends showed several percent higher χ c values than the cyclic/cyclic blends for both photocleavable (4L/4D, 69%; 5L/5D, 61%) and non-photocleavable (1L/1D, 79%; 2L/2D, 69%) polylactides. The χ c value for the linear/cyclic blends was the lowest (4L/5D, 54%; 1L/2D, 62%) likely due to the difficulty of crystallization arising from the unmatched topologies of the enantiomeric pairs. By comparing the lamellar structures estimated from the SAXS profiles of 4L/4D (L = 12 nm, l c = 9.4 nm, l a = 2.5 nm) and 5L/5D (L = 7.6 nm, l c = 5.1 nm, l a = 2.4 nm) (Table 1), a significant decrease of L and l c due to the cyclization was revealed. Therefore, the polylactides possessing a photocleavable linker (4L, 4D, 5L, and 5D) and their blends have similar effects on the crystallographic properties via cyclization 30 as the polylactides without a photocleavable linker (1L, 1D, 2L, and 2D) and their blends.  Lastly, the photoirradiated PLLA/PDLA blend (6L/6D) was also subjected to the WAXD and SAXS measurements. The crystal form was unchanged; the diffraction peaks at 2θ = 12, 21, and 24°in the WAXD profile (Fig. 3) were ascribed to the stereocomplex. The χ c value was estimated to be 49% (Table 1). The additional peaks at 2θ = 13 and 18°, which were found in cyclic/cyclic 5L/5D before the photocleavage as well as in linear/cyclic 4L/5D, disappeared indicating that these diffractions were indeed from the photocleavable units in the crystalline layers. Despite the same crystal form determined by WAXD, SAXS revealed an increase of L and l c upon photoirradiation. The estimated values for 6L/6D (L = 15 nm, l c = 13 nm) were more than double in comparison with those of 5L/5D (L = 7.6 nm, l c = 5.1 nm) (Table 1), indicating that the cyclic polymers were photocleaved to likely form extended chain crystals (Fig. 2). When compared with 4L/4D (L = 12 nm, l c = 9.4 nm), those for 6L/6D were found to be significantly larger. This is probably because some degree of dimerization through the nitroso-nitroso association occurred (Fig. S6 †). 25 The number of lactic acid units and their molecular weight in the crystal layers were 45 units and M n = 3.2 kDa, respectively. Furthermore, l a for both homocrystals and stereocomplexes was relatively large for the polylactides containing the photocleavable linker (4L, 4D, 5L, 5D, 4L/4D, 4L/5D, and 5L/5D) compared with those without the photocleavable linker including the one after the photocleavage (1L, 1D, 2L, 2D, 1L/1D, 1L/2D, 2L/ 2D and 6L/6D), as shown in Tables 1 and S1. † These results were likely due to the volume of the photocleavable linkers in the amorphous layer. Moreover, the SAXS peak of photocleaved 6L/6D was somewhat sharpened in comparison with that of cyclic 5L/5D (Fig. 4), suggesting that the order of the crystal lattice is restored to some extent after the photocleavage.

Melting points (T m )
The T m values of the non-photocleavable linear and cyclic polylactides were previously reported. 16 In the present study, those of the linear and cyclic polylactides with a photocleavable linker and the photocleaved stereocomplex were measured by DSC to investigate the effects of the topological conversion. The homocrystals of linear 4L and 4D had a T m of 142°C, whereas cyclic 5L and 5D showed a lowered T m of 128 and 125°C, respectively (Table 1 and Fig. 5). 15 The blend of linear 4L/4D showed a T m of 209°C, which was higher than the respective homopolymers due to typical stereocomplexation. 15 . After the photocleavage, T m of 6L/6D increased to 211°C, which was more than 40°C higher than that of the cyclized 5L/5D (167°C) and comparable to that before the cyclization (4L/4D, 209°C). These results suggest that T m of the stereocomplexes of the present polylactides is switchable by approximately 40°C through controlling their polymer topology.

Enthalpies of melting (ΔH m , ΔH m (100%))
The ΔH m values were calculated from the DSC thermograms (Fig. 5) and converted to those of crystals having an infinite thickness, i.e., ΔH m (100%), based on the crystallinity determined by WAXD (Fig. 3). 34 Interestingly, we found that the homocrystals consisting of a cyclic polymer (5L and 5D) have significantly smaller ΔH m (100%) (44 J g −1 ) than those of the linear counterparts (4L, 60 J g −1 ; 4D, 68 J g −1 ) as shown in Table 1. This difference presumably arose from the absolute enthalpy of the respective melt states rather than those of the crystalline states because the line widths of the WAXD peaks, thus crystallite size and lattice distortion, were comparable (Fig. 3). In other words, the absolute enthalpy of cyclic 5L and 5D in the melt state is expected to be lower by roughly 20 J g −1 than that of linear 5L and 5D due to the topology. In this regard, linear polymers are relatively easy to entangle in the melt state, possibly leading to a higher absolute enthalpy, while cyclic polymers are less likely. 35 Moreover, the ΔH m (100%) values of the present polylactides (4L, 60 J g −1 ; 4D, 68 J g −1 ; 5L, 44 J g −1 ; 5D, 44 J g −1 ) were much smaller than those used for PLLA with high molecular weights (100 and 135 J g −1 for linear PLLA (M n (PS) = 360-450 kDa), where M n (PS) indicates that M n determined by SEC calibrated using standard PS) as shown in Table 1. 36 Presumably, this was also caused by the high degree of entanglement in the melt state due to the length of the polymers. The trend for the ΔH m (100%) values of the stereocomplexes is parallel to that of the homocrystals; the cyclic/cyclic blend (5L/5D, 43 J g −1 ) as well as the linear/cyclic blend (4L/5D, 41 J g −1 ) has notably smaller ΔH m (100%) than the linear/linear blend (4L/4D, 87 J g −1 ) and that after the photocleavage (6L/6D, 89 J g −1 ). In addition, the ΔH m (100%) value for a stereocomplex calculated based on the previous report using longer PLLA (M v = 27 kDa) and PDLA (M v = 25 kDa) 34 was also considerably larger (146 J g −1 ). Furthermore, photocleavable linear PLLA (4L′), linear PDLA (4D′), cyclic PLLA (5L′), and cyclic PDLA (5D′) with a molecular weight of approximately 1.5 kDa were also synthesized and subjected to DSC measurements in order to study the molecular weight dependence. No T m was found for any of these individual polylactides likely due to the insufficient length for crystallization. However, the blends of linear/linear 4L′/4D′, cyclic/cyclic 5L′/5D′, and photocleaved 6L′/6D′ had T m of 138, 127, and 155°C with ΔH m of 29, 6, and 16 J g −1 , respectively. This trend concerning the polymer topology is the same for the polylactides with a molecular weight of approximately 3 kDa, but a lower T m and smaller ΔH m were likely due to the smaller molecular weight.

Surface free energy (σ e )
σ e of the interface between the crystal and amorphous was calculated by the Gibbs-Thomson equation (1) using the reported values of the densities (homocrystal, 1.285 g cm −3 ; 37 stereocomplex, 1.27 g cm −3 ) 32,38 and equilibrium melting points, T°m (homocrystal, 215°C; 39 stereocomplex, 279°C). 40 The σ e value for linear 4L (54 mJ m −2 ) and 4D (55 mJ m −2 ) was close to that of high-molecular-weight linear PLLA (M n (PS) = 360-450 kDa 36 (52-64 mJ m −2 )), which has the chain-folding surfaces of the crystalline layers. On the other hand, σ e for cyclic 5L (24 mJ m −2 ) and 5D (28 mJ m −2 ) was significantly smaller. This suggests that cyclic 5L and 5D form a wellordered looped structure at the crystal surfaces (Fig. 2b) to help stabilize the crystalline-amorphous interface compared to linear 4L and 4D having chain ends (Fig. 2a). On the other hand, the folding segments of high-molecular-weight PLLA were probably not as organized as those of cyclic 5L and 5D, resulting in high σ e values (52-64 mJ m −2 ). The less ordered folding segments of high-molecular-weight PLLA are also suggested by the thicker amorphous layers (l a = 11.9-20.6 nm in Table 1). The higher stability of the interface was also determined for the stereocomplex of cyclic 5L/5D (28 mJ m −2 ) compared to linear 4L/4D (66 mJ m −2 ) and the reported blend of linear PLLA (M v = 27 kDa) and PDLA (M v = 25 kDa) (66 mJ m −2 ). 34 The linear/cyclic blend of 4L/5D (35 mJ m −2 ) was comparable to cyclic/cyclic 5L/5D (28 mJ m −2 ). On the other hand, considerably higher σ e was found for photocleaved 6L/6D (91 mJ m −2 ), probably caused by the non-uniform chemical structure of the photoreacted chain ends and/or the partial intermolecular coupling through the nitrosonitroso association. Interestingly, Müller and coworkers studied the crystallization of linear and cyclic poly(ε-caprolactone) and found higher T m and larger σ e for the cyclic polymers, 41 in contrast to the present results of the polylactides. The different crystal systems were expected to cause the opposite topology effects from the respective polymers. For reference, they also reported higher spherulitic growth rates for cyclic poly(ε-caprolactone) than linear analogues. 41,42 Conversely, cyclic poly(tetrahydrofuran) 43 and cyclic polyethylene 44 showed a slower rate than their linear counterparts. Presumably, the higher growth rate for cyclic poly(ε-caprolactone) was due to faster diffusion. 45 On the other hand, the unfavorable entropic process governed the crystallization of cyclic poly(tetrahydrofuran) and cyclic polyethylene, resulting in slower rates. The elucidation of the relationship between the polymer topology and T m and σ e of these crystal systems is strongly desired and currently in progress.

Conclusions
Photocleavable cyclic PLLA and PDLA possessing a NB group were synthesized and subjected to a photocleavage reaction to transform the topology. The WAXD and SAXS measurements of the present polylactides, along with previously reported nonphotocleavable polylactides, were performed. Despite WAXD indicating that the essentially crystallographically identical homocrystals and stereocomplexes are formed by the linear and cyclic polylactides, SAXS of the cyclic polylactides showed that L and l c are roughly half of the linear counterparts due to the flattened ring conformation. Moreover, the photocleaved PLLA/PDLA blend showed L and l c values comparable to those before the cyclization. DSC showed that T m of the stereo-complex reflects l c ; more than 40°C of T m was controlled through the linear-cyclic topological transformation. The combination of WAXD, SAXS, and DSC revealed that ΔH m (100%) is smaller for the homocrystals and stereocomplex of the cyclic polylactides than the respective linear counterparts. Furthermore, the interface between the crystalline and amorphous phases was significantly more stable for the cyclic polylactides, shown by smaller σ e .
The present results provide new insights into the structure and properties of crystalline polymeric materials. Novel designs based on the cyclic polymer topology and simple photoirradiation would lead to versatile control on the properties such as T m , ΔH m (100%), and σ e , which cannot be achieved by traditional means. Moreover, a photocleavage reaction to increase T m by approximately 40°C is appropriate for the efficient processing of polylactide materials. Altogether, the present methodology is expected to lead to the development of novel topology-based stimuli-responsive materials, which require only one reaction per polymer molecule and allow for switching the properties without altering the chemical structure and molecular weight of the polymer.
Measurements 1 H NMR spectra were recorded at ca. 20°C on a JEOL JNM-AL300 spectrometer operating at 300 MHz. CDCl 3 was used as a solvent, and the proton chemical shifts were reported relative to the signal of tetramethylsilane. SEC measurements were performed at 40°C on a Tosoh model CCPS equipped with a TSK G3000HXL column and a refractive index detector model RI 8020. CHCl 3 was used as an eluent at a flow rate of 1.0 mL min −1 . Linear polystyrene standards were used for calibration, and M p of the sample was reported as a linear polystyrene equivalent. MALDI-TOF mass spectra were recorded on a Shimadzu AXIMA Performance spectrometer equipped with a nitrogen laser (λ = 337 nm). The spectrometer was operated at an accelerating potential of 20 kV in linear positive ion mode with pulsed ion extraction. A CH 2 Cl 2 solution (3.0 µL) of a polymer sample (1.0 mg mL −1 ), an acetone solution (3.0 µL) of trans-2-[3-(4-tert-butylphenyl)-2-methyl-2propylidene]malonitrile (DCTB) (10 mg mL −1 ), and an acetone solution (3.0 µL) of sodium trifluoroacetate (10 mg mL −1 ) were deposited onto a sample target plate in sequence. Mass values were calibrated using the four-point method using peaks from SpheriCal ® , monodisperse dendritic calibration standards that had m/z = 1715.68, 2255.88, 2796.08, and 3422.35 for polymers in the molecular weight range of 1200-3800 and m/z = 3636.44, 4816.89, 5997.34, and 7263.87 for polymers in the molecular weight range of 3400-8000. SEC fractionation was performed on a JAI LC-908 equipped with JAIGEL-2H and JAIGEL-3H columns, and CHCl 3 was used as an eluent at a flow rate of 3.5 mL min −1 . WAXD measurements were performed on a Rigaku RINT-2100 spectrometer with CuKα radiation (λ = 0.15418 nm) operating at 40 kV and 40 mA. The optical system with a pinhole geometry consisted of a graphite monochromator and scintillation counter. The intensity data were collected in the 2θ range between 3 and 60°. SAXS measurements were conducted on a Rigaku NANO-Viewer system with CuKα radiation (λ = 0.15418 nm) operating at 40 kV and 20 mA. The SAXS intensity measured by pin-hole collimation was corrected for parasitic scattering. The intensity data were collected in the 2θ range between 0.1 and 2.5°. A powder sample was filled in the center space of a 0.8 mm-thick stainless folder and sealed both faces with amorphous polyethylene thin films of less than 10 µm thickness. DSC measurements were performed on a Seiko DSC 6200, and all samples were cast from CH 2 Cl 2 solutions. Thermograms were recorded during the first heating cycle at a rate of 5 or 10°C min −1 under a nitrogen gas flow of 50 mL min −1 .