The main results on the physicochemical and thermal behaviors of the analyzed random copolymer P(DLLA-co-CL) have been the number average molecular weight [Mn] calculated by 1H-NMR and Tg obtained by DSC. In particular, the Mn was calculated through two different approaches: the first one, taking into account the terminal groups, produced a value of 21,000 g/mol, while the ratio between the caprolactone [CL] units and the initiator produced a value of 28,000 g/mol. From thermal analysis, we obtained a Tg of about 24°C.
Moreover, the measured PDLLA content obtained from 1H-NMR was 89.8 mol% in contrast with the value of PDLLA which was 86 mol% given by the supplier. This difference can be considered a small one and in the range of the possible deviation in different batches. In fact, the ratio of the lactic acid [LA] and CL signals allows a quite high accuracy for this calculation with a dispersion of 0.3% in three repetitions. We think that the supplier has provided an average value that could change from batch to batch without reporting the exact value for each batch.
The 1H-NMR spectrum for P(DLLA-co
-CL) is reported in Figure , where also the general chemical structure for this copolymer, assuming monofunctional initiator R, is described. This scheme does not imply a di-block structure. The analysis of the 1H-NMR spectrum was performed using Kasperczyk's work as reference [25
]. Characteristic signals for polymerized caprolactone and polymerized lactide are observed. The multiplet from 5.05 to 5.25 ppm is assigned to methine proton of polymerized lactide (f
), with some rests of unpolymerized lactide at approximately 5.03 ppm. Almost undetectable, a negligible signal at approximately 4.35 ppm for terminal LA units appears. At approximately 4.23 ppm and 2.63 ppm, two small signals are due to unreacted ε-caprolactone. Calculations allow the determination of the amount of unreacted ε-caprolactone and unreacted lactide as less than 0.6 wt.% and less than 0.2 wt.%, respectively. The multiplet from 4.08 to 4.18 ppm is due to the CL proton a
that linked to a LA molecule, while the triplet at 4.05 ppm indicates that the CL proton a
linked to another CL molecule. The triplet at 3.74 ppm is related to the CL proton a
for terminal CL molecules (-CH2
-OH). The multiplet between 2.34 to 2.44 ppm is due to the CL proton e
that linked to a LA molecule, while the triplet at 2.30 indicates that the CL proton e
linked to another CL molecule. For the rest of the spectrum, multiplets at 1.66 ppm and 1.39 ppm are related to the CL protons b, d
, and c
, respectively, and the multiplet at 1.56 ppm, to the LA methyl protons g
. So, the ratio of the LA signals to the CL signals results in a molar composition LA/CL of the copolymer of 89.8:10.2 mol% (corresponding to 84.8:15.2 wt.%).
Figure 1 1H-NMR spectrum of the P(DLLA-co-CL) used in this work. CL refers to polymerized caprolactone units, and LA, to polymerized lactide units. The general chemical structure of the copolymer P(DLLA-co-CL) is reported on top (R = polymerization initiator). (more ...)
If the copolymer is a di-block copolymer, the ratio of the signal of polymerized CL linked to LA molecules to the signal of terminal CL should be 1, and in our case, it is approximately 8.9. Furthermore, the ratio of the signal due to CL linked to LA to the signal of CL linked to CL is approximately 3.15, indicating the preponderance of isolated CL units in the polymer backbone. From these results, it is clear that the chemical structure of the copolymers approaches more likely the structure of a random copolymer. As the molar content of CL in the copolymer is low, 10.2%, it is reasonable to presume that CL units are isolated in between PLA units (-LA-CL-LA-) or form blocks of double CL units (-LA-CL-CL-LA-), with the existence of longer CL blocks being negligible. Then, from the signals due to CL linked to LA and to CL linked to CL, a 68 mol% of isolated CL units and 32 mol% of double CL units are calculated. Once the total CL and CL-CL units are determined, it is possible to calculate the mean length of the LA blocks which results to 12.
Summarizing the 1H-NMR results, the P(DLLA-co-CL) copolymer used in this study has the structure of a predominantly random copolymer with most of the CL units isolated in the copolymer backbone, therefore causing its inability to crystallize, and with blocks of polymerized LA units that are also unable to crystallize, producing an amorphous copolymer. Neither melting nor crystallization was found in the DSC thermogram (not shown), indicating the amorphous nature of the copolymer. The amorphous structure was also confirmed by SAXS (data not shown).
The amorphous state of the copolymer is confirmed also by Raman spectroscopy. In fact, as reported by Kirster et al. [26
], the presence of a broad band at 868 cm-1
and the absence of the 1,107-cm-1
narrow peaks are discriminant to characterize the amorphous state of PCL. The spectrogram reported in Figure is in good agreement with this analysis. Above the Raman spectrogram, the values of the main peaks for the CL monomer are indicated, while below the line, the main characteristic peaks for the DLLA monomer are reported. In our case, the Raman line at 868 cm-1
is clearly detected and the peaks at 1107 cm-1
and 912 cm-1
are not detected. Moreover, the large band in the region from 1,300 to 1,360 cm-1
including the Raman line at 1,338 cm-1
confirms the presence of DLLA units and so the amorphous state of the copolymer.
Raman spectrogram of P(DLLA-co-CL).
When the copolymer was spin-cast in a film from a chloroform solution, an unexpected nanostructured phase separation was obtained. In fact, the AFM results reported in Figure indicate the formation of a nanostructure with spherical domains having an average diameter of 48 nm.
Images of the solvent spin-cast copolymer film at room temperature. (a) 3 × 3-μm TM-AFM height and (b) phase images.
The nanostructuration observed is coherent with the computer simulation of the phase diagram of random copolymers carried out by Houdayer and Muller [27
]. Based on our knowledge, this is the first time that the nanostructuration of a P(DLLA-co
-CL) copolymer is reported. Moreover, very few experimental demonstrations of the nanostructuration of random copolymers have been reported in the scientific literature [28
]. Taking into account the small amount of PCL, less than 15 mol%, it is assumed that spherical CL-enriched domains have been obtained. In this case, we consider that, because of the chemical nature of the copolymer, the higher affinity of chloroform for CL than for LA (as obtained by the solubility parameters calculated by the Hoftyzer and van Krevelen theory [24
]) has favored the phase separation of CL-enriched domains in a matrix of pure LA or of LA with a lower CL content.
The same spherical morphology has been detected by TEM analysis, as shown on Figure , where spheres with a diameter of about 120 nm are observed. Whereas the TEM image distinguishes between areas with different chemical composition, the AFM image distinguishes between areas with differences in rigidity, leading to the different size determined by both techniques. Moreover, it is known that nanostructuration is strongly dependent on the substrate type and film thickness [23
]. In this particular case, the TEM analysis was performed on a 150-nm-thick film on a carbon substrate, while the AFM analysis was performed on a 300-nm-thick film on a glass substrate.
TEM images of the solvent cast copolymer film at room temperature. Scale bar, 500 nm.
After 3 h of annealing treatment at 65°C under vacuum, the spherical domains increased their dimensions (Figure ), while the fraction of the spherical domains, calculated from the AFM images, remain almost constant (ca
. 7%). In this case, the spherical domains present an average diameter of 76 nm which is clearly higher than the average diameter of the CL-enriched domains obtained before annealing. This fact indicates that the spherical morphology obtained at an ambient condition represents a non-equilibrium nanostructure that is able to modify itself when the diffusion process is activated by an annealing treatment. This means that the morphological structure obtained is able to reach a free-energy minimization, resulting in the formation of ordered structures as in the case of block copolymers [30
Images of the 3-h-annealed copolymer at 65°C. (a) 3 × 3-μm TM-AFM height and (b) phase images.
Moreover, the diameters of the spherical domains in the case of the room temperature-nanostructured copolymer follow a Gaussian statistical distribution (Figure ). As shown by Teraoka [31
], given two different points, r1 and r2, the Gaussian distribution indicates a transition probability for r2 to move into a small volume around r1, justifying in our case a stable phase separation of CL-enriched spherical domains in a LA-enriched matrix. Instead, an exponential curve of third order is required in order to fit the experimental data of the 3-h-annealed copolymer, confirming the strong changes in the phase distribution of the two samples.
Statistical distribution. Statistical distribution of the spherical domains for spin-cast P(DLLA-co-CL) at room temperature (white squares) and for 3-h-annealed copolymer (black squares).
For longer annealing times, the nanostructured domains collapse and a disordered homogeneous structure is formed (not shown). It turns out that the phase segregation is characterized by a non-equilibrium geometrical rearrangement of the interfaces which tends to aggregate, minimizing the surface energy, and evolve to a dissolution of the nanostructured domains in the PDLLA-rich phase.