Effect of tungsten disulfide nanotubes on crystallization of polylactide under uniaxial deformation and annealing

Tungsten disulfide (WS2) nanotubes (NTs) are examined here as a filler for polylactide (PLA) for their ability to accelerate PLA crystallization and for their promising biocompatibility in relevant to biomedical applications of PLA-WS2 nanocomposites. In this work, we have studied the structural and thermal properties of PLA-WS2 nanocomposite films varying the concentration of WS2 NTs from 0 (neat PLA) to 0.6 wt%. The films were uniaxially drawn at 90 °C and annealed at the same temperature for 3 and 10 min. Using wide angle x-ray scattering, Raman spectroscopy and differential scanning calorimetry, we probed the effects of WS2 NT addition on the structure of the PLA films at various stages of processing (unstretched, stretching, annealing). We found that 0.6 wt% of WS2 induces the same level of crystallinity in as stretched PLA-WS2 as annealing in neat PLA for 10 min. These data provide useful insights into the role of WS2 NTs on the structural evolution of PLA-WS2 composites under uniaxial deformation, and extend their applicability to situations where fine tuning of PLA crystallinity is desirable.


Introduction
Polylactic acid or polylactide (PLA) is a polymorphic semicrystalline polymer synthesized from renewable sources that finds applications in diverse fields (food packaging, biomedical devices, drug delivery systems) due its properties of biocompatibility and biodegradability [1]. PLA biocompatibility derives from the fact that in physiological conditions, it degrades by hydrolysis into non-toxic lactic acid that is further metabolized by the human body into carbon dioxide and water as final products. As a result, PLA is widely used as a material for sutures, tissue scaffolds and supports [2][3][4][5]. Unfortunately, the applications of this polymer are often limited by its slow crystallization rate, low crystalline degree and low fracture toughness [6][7][8]. One approach to mitigating these issues is the incorporation of organic or inorganic fillers in the PLA matrix, which can increase the rate of crystallization of PLA and enhance its mechanical properties [9][10][11][12][13][14][15][16]. However, the biocompatibility of the filler and minimal loading concentrations with respect to the polymer matrix are essential requirements from a toxicological point of view, particularly in biomedical applications. In this framework, nanotubes (NTs) of tungsten disulfide (WS 2 ) are promising nanofillers for polymer matrices: their intrinsic superior mechanical properties (~150 GPa Young's modulus,~16 GPa strength,~14% elongation [17]) can reinforce polymers, especially under flow-induced crystallization due to their high aspect ratio (diameter~80-100 nm, length~2-3 μm [18]). They are also reported to possess lower in-vitro and in-vivo cytotoxicity compared to other nanoparticles such as silica or carbon black [18][19][20].
Despite these advantages, to the best of our knowledge, only a few studies are reported in literature on PLA-WS 2 nanocomposites, mainly focused on thermal properties of samples obtained by melt-mixing [21][22][23][24][25]. These studies show that WS 2 NTs remarkably influence the kinetics of nucleation and growth of the PLA during the cold crystallization. The mechanical properties of solvent cast PLA-WS 2 films measured at room temperature have been recently reported by Shalom et al. showing a large increase of elastic modulus (50% for 0.7 wt%), yield strength (about 70% for 0.7 wt%), and elongation (about 70% for 0.5 wt%) with increasing concentration of WS 2 NTs [24]. On the other hand only our previous work [25] reports on the crystallization process induced by stretching PLA-WS 2 films at temperatures above the glass transition of PLA (T g~5 5-65°C) but the effect of the WS 2 concentration on the crystallization of PLA was not analyzed at the time.
Therefore, in this work we have prepared PLA-WS 2 films with different amounts of WS 2 (0-0.6 wt%) with respect to PLA and we have studied the effect of WS 2 concentration on the strain-induced crystallization of PLA as a consequence of uniaxial stretching at 90°C and constant strain rate. We have also investigated the effect of post-stretching thermal annealing on the crystallization of PLA.

Experimental Materials
All the materials used in this study are commercially available. PLA (Ingeo 4032D, 1.2-1.6% D-isomer lactide) pellets were purchased from NatureWorks LLC (Minnetonka, MN). Tungsten disulfide nanotubes (WS 2 NTs) were purchased from Nanomaterials Ltd. (Israel) with diameter of about 80-100 nm and length of about 2-3 μm and they were used as received without any chemical modification. Chloroform was supplied by Sigma-Aldrich.
Preparation of unstretched and stretched films of PLA and PLA-WS 2 composites Figure 1 reports a scheme of the preparation process of unstretched and stretched films of neat PLA and PLA-WS 2 nanocomposites with different amounts of WS 2 NTs (0.1, 0.2, 0.3, and 0.6% by weight). The PLA pellets were dried under vacuum at 80°C for 4 h and then at 40°C for 3 days to remove moisture prior to the nanocomposite preparation. Suspensions containing the different amounts of WS 2 NTs were prepared in chloroform, sonicated for 30 min and then added to a PLA solution in chloroform (120 mg/mL) previously prepared by stirring at 40°C for about 3 h. The final PLA-WS 2 suspensions were stirred for 1 h (Fig. 1a). The PLA-WS 2 nanocomposite suspensions and neat PLA solution were poured into petri dishes of~10 cm diameter (Fig. 1b) and left at room temperature for 2 days to allow the majority of the solvent to evaporate. After 2 days, the films were dried under vacuum until the solvent was completely removed (no change in mass detected).
All the solvent cast films were hot pressed into1 00 μm thick films by a Carver laboratory press at 180°C for 10 min and rapidly quenched into an icewater bath to avoid crystallization (Fig. 1c). The weight average molecular weight (M w ) of the hot-pressed films (hereafter referred to as UNST) of PLA and PLA-WS 2 nanocomposites was comparable to that of the original PLA pellets (M w~1 00 K g/mol) as measured by gel permeation chromatography. Strips cut from the hot-pressed PLA and PLA-WS 2 films were stretched in a dynamometric apparatus at a temperature of 90°C. After stretching, some films were immediately removed from the instrument (samples called ST0 in Fig. 1d) while another set of films (with the same composition of the ST0 series) was kept at constant length in the stretching apparatus and thermally annealed at 90°C for 3 or 10 min (samples ST3 and ST10, Fig. 1e). All the stretched films (ST0, ST3 and ST10) were quenched in ice-water after removal from the instrument.

Methods of film characterization
The dispersion of the WS 2 NTs in UNST PLA-WS 2 films was studied by using a Zeiss Axio scope A1 microscope in transmission and bright field mode. The mechanical properties of the films were measured by uniaxial stretching experiments using an INSTRON dynamometric apparatus (mod. 4301) with 1 kN-load cell. Rectangular sections (width = 5 mm, length~50 mm) were cut from the hot-pressed films of each composition and mounted in the INSTRON apparatus kept at a temperature of 90°C. The films were drawn at a strain rate of 500 mm/min up to 100% deformation starting from an initial gauge length of 25 mm. Stressstrain curves were extracted from the forcedisplacement data and the elastic modulus was evaluated from the linear part of the stress-strain curves at deformation of 0.1% by averaging over ten replicas for each composition.
The structure and crystallinity of as-cast, hot pressed (UNST) and stretched (ST0, ST3, ST10) PLA and PLA-WS 2 films were probed measuring wide angle x-ray scattering (WAXS) at beamline 5-ID-D of the Advanced Photon Source (APS), Argonne National Labs. The incident X-ray beam with spot-size 250 μm × 250 μm was aligned perpendicular to the film surface. WAXS patterns were acquired on a Rayonix CCD detector that was located 200.83 mm from the sample. Six patterns were acquired at six different positions along the long side of each film (corresponding to the drawing direction for ST0, ST3, ST10, as an example, see inset in Fig. S2c, showing the position of the six acquisitions for the PLA-WS 2 0.2 wt% film) with an exposure time of 0.5 s using X-rays of wavelength 0.7293 Å. The six patterns were then averaged in one single scan per sample. Several replicas (3 or 4 films prepared with the same composition, drawing and annealing conditions) were measured for the stretched films (PLA ST0, ST3 and ST10 in Fig. S3, respectively; and the equivalent for the PLA-WS 2 0.1-0.6 wt% films in Fig. S4, S5, S6, S7). The scattered intensity per each pattern is reported on an absolute scale after taking into account the calibration factor provided by the beamline, the transmitted counts on the beamstop, the thickness of the sample and the background scattering from air. The 2D patterns were then integrated into 1D curves of intensity vs. the scattering vector q.
Raman spectroscopy was performed through a Renishaw InVia Reflex Raman spectrometer selecting the laser wavelengths of 514.5 nm (laser power of 50%) and using an objective of 50x of magnification. The range of investigated wavelength was 100-1800 cm − 1 . Each measurement is reported as an average of 20 subsequential accumulations with an exposure time of 10 s. The position, intensity and full width at half maximum (FWHM) of the Raman bands were evaluated from a Lorentzian fit of the spectra needed to resolve overlapping peaks (see Fig. S11a). The intensity and FWHM values of the bands were then normalized with respect to the PLA band centered at 1095 cm − 1 (associated to stretching OCO group [26]) in order to minimize the errors due to the different thickness of the samples. This band was chosen as internal reference because its intensity does not seem to change appreciably from one sample to the other independently on the crystallinity evaluated from WAXS, Raman and DSC analyses. For each type of sample three replicas per condition (composition, drawing, annealing time) were analyzed in their central region and the reported values of normalized intensity and FWHM of the investigated bands are the average of the measurements (obtained after fitting and normalization) made on the three replicas. The error bars correspond to the standard deviation.
The thermal properties were measured by Differential Scanning Calorimetry (DSC) carried out with a Perkin Elmer DSC7 instrument. About 10 mg of sample was placed in an aluminum pan under nitrogen flux and initially held at 30°C for 5 min, then heated up to 220°C with a rate of 5°C/min, held at 225°C for to 5 min and then cooled down to 30°C with a rate of 5°C/min. In order to investigate the effect of the stretching and annealing on the crystallinity of the samples, the first heating scan was analyzed. In detail, the cold crystallization temperature (T cc ) and the melting temperature (T m ) were determined from the minimum of the crystallization exotherm and from the maximum of the melting endotherm, respectively. The enthalpies of the cold crystallization (ΔH cc ) and melting (ΔH m ) were determined by integrating the areas under the corresponding peaks and used to estimate the percentage of crystallinity (X c ) existing before the DSC measurement and not induced by the latter. X c was calculated from the heating curves according to the eq. (1): where ΔH°m is the melt enthalpy of theoretically 100% crystalline PLA (93 J/g, [27]) and wt% indicates the weight fraction of WS 2 NTs in the nanocomposites.

Results and discussion
The UNST PLA-WS 2 nanocomposite films show a uniform dispersion of WS 2 NTs in PLA both at the lowest (0.1%, Fig. 2a) and highest (0.6%, Fig. 2b) concentration of NTs. The optical micrographs in Fig. 2 do not reveal presence of aggregates demonstrating that the preliminary sonication is efficient in unraveling the nanotubes and that they readily disperse in PLA without need of further functionalization [21].
The results of the uniaxial stretching experiments of PLA and PLA-WS 2 nanocomposites performed in the Instron chamber held at the temperature of 90°C are reported in Fig. 3. The stress-strain curves, recorded up to 100% strain, show a very similar behaviour for the PLA-WS 2 nanocomposites and neat PLA (Fig. 3a) for all the WS 2 concentrations examined here (0.1-0.6 wt%). A slight increase of~10% of the elastic modulus (from 3 to 3.5 MPa) is found for the films at WS 2 concentrations higher than 0.2% (Fig. 3b). The error bar length attributed to the modulus (corresponding to the standard deviation) indicates a good reproducibility between the specimens of the same type, which, in the case of the nanocomposites, also confirms the uniform dispersion of the WS 2 NTs in the PLA matrix (Fig. 2). The value of the elastic moduli of the films drawn at T = 90°C is lower than PLA samples drawn at room temperature (Fig. S1a-c) by 2-3 orders of magnitude (~3 MPa vs. 1.5 GPa, respectively). According to literature [28][29][30][31][32], at drawing temperatures above the glass transition the tensile strength and the elastic modulus of the PLA decrease, whereas the elongation at break significantly increases.
Wide angle x-ray scattering (WAXS) was used to detect variations in the crystallization status of the PLA-WS 2 nanocomposites after the different steps of preparation. The 2D and 1D diffraction patterns of the as-cast films of PLA and PLA-WS 2 at all concentrations of nanotubes show all reflections of the crystalline α-phase of PLA (Fig. S2) [33]. The diffracted intensity is uniformly distributed along rings (black trace, Fig. S2a-e), showing isotropic, crystalline morphology created upon quiescent crystallization conditions of solvent-casting procedure. The Bragg peaks of the WS 2 nanotubes are faint in PLA-WS 2 spectra at 0.1 wt% (Fig. S2b) but become clearly visible at 0.2 wt% (Fig. S2c). The hot press step (where the as-cast films were held at 180°C for 10 min) and subsequent fast quenching completely make the PLA matrix amorphous (red trace, Fig. S2a-e).
Indeed, only a broad diffuse scattering halo from PLA and the WS 2 reflections characterize the WAXS patterns of the UNST films (Fig. S2). The stretching and the subsequent annealing process induced a further change of the PLA structure from the amorphous UNST films to the stretched films ST0, ST3 and ST10 (Fig. 4 and Fig.  S8, S9, S10) characterized by a progressive reduction of the PLA amorphous component and corresponding increase of the α crystalline phase. Moreover, the intensity spots in the diffraction 2D patterns of the stretched PLA indicate that the polymer chains are aligned along the stretching direction x, drawn in the 2D pattern of PLA ST10 film (Fig. 4a). The same effect of the stretching and annealing was observed for all the nanocomposites with 0.1-0.6 wt% (see Fig. 4b and Fig. S8, S9, S10). In both cases of PLA (Fig. 4a) and the PLA-WS 2 nanocomposites all concentrations ( Fig. 4b and Fig. S8, S9, S10) increasing the annealing time from 0 (ST0) to 10 min (ST10) induces an increase of intensity of the PLA peaks at expense of the amorphous content. To give a quantitative estimation of the crystallinity in Fig. 5 we have plotted the ratio I peak /I valley between the intensity value of the main PLA reflection ((110)/(200), positioned at q = 1.16 Å − 1 ) and the intensity value at the valley between this and the (203) peak (positioned at q = 1.25 Å − 1 ) as a function of the WS 2 concentration (for all patterns shown in Fig. 4 and Fig. S8, S9, S10). The intensity ratio I peak /I valley is constant for the amorphous UNST films, but after stretching (ST0) it increases with the concentration of nanotubes first slowly up to 0.2 wt% and then with a sudden jump at 0.3 wt%, remaining almost constant the final concentration of 0.6 wt%. While the stretching itself does not induce noticeable crystallization in PLA (ST0, WS 2 = 0 wt%) (but probably just a chain alignment), the nanotubes promote crystallization by acting as nucleation agents [23]. The value of I peak /I valley for the conditions ST3 and ST10 is very similar to and larger than the corresponding ST0 value. This indicates that the crystallization of PLA has reached its maximum already after 3 min of annealing. In order to investigate the conformational changes due to the extensional stress and subsequent annealing process, all samples were characterized by Raman spectroscopy selecting the central region of each specimen where, in the stretched films, the mechanical stress should be maximum. The Raman spectra of PLA and PLA-WS 2 0.6% samples (UNST, ST0, ST3, ST10) normalized with respect to the band at 1095 cm − 1 (as described in the experimental section) are reported in Fig. 6. The first part of the spectra in the wavelength range 250-450 cm − 1 includes two weak bands of PLA (Fig. 6a); these two bands are present in the PLA-WS 2 0.6% spectrum (Fig. 6c) and partially overlap with two strong WS 2 bands detected at 354 and 420 cm − 1 (attributed to E 2g and A 1g WS 2 modes, respectively [34]). Their position is in accordance with literature data on WS 2 [34] and no peak shift is observed either for the presence of the PLA or upon stretching and annealing process conditions. This suggests the absence of chemical interactions between the filler and the polymer matrix and no influence of the processing conditions on the WS 2 structure. As expected, the intensity of the WS 2 bands increases with the WS 2 concentration. This result is confirmed by the quantitative analysis of signal intensities estimated through the fitting procedure explained in the experimental section and shown in Fig. S11. The second part of the Raman spectra (PLA, Fig. 6b and PLA-WS 2 0.6%, Fig. 6d) shows PLA bands in the range 700-1500 cm − 1 (where no band from WS 2 is present). We have explicitly marked the position of four bands (centered at wavenumbers 874, 924, 1095, 1454 cm − 1 ) on which we have focused the following discussion because some of these bands can be associated to the presence of crystallinity in the polymer [35][36][37][38]. Moreover, the band centered at 924 cm − 1 is observed only when PLA crystallizes in α [36,38] and α' [38] forms. These forms, indeed, present similar chain conformation (10 3 ) but α' is characterized by uniform conformational disorder [16]. In Table 1 we list the attributions of these bands to specific vibrations as reported in literature.
The normalized intensity of the bands (as evaluated from the fitting procedure shown in Fig. S11) centered at 924 cm − 1 and 1454 cm − 1 are reported as a function of WS 2 concentration in Fig. 7a and Fig. 7b, respectively. As expected for amorphous samples [36], all UNST samples present negligible values of the normalized intensity of the 924 cm − 1 crystalline band, while the behavior of the stretched series ST0, ST3, and ST10 is more complex. While the normalized intensity of the 924 cm − 1 band for PLA ST0 is similar to that of the PLA UNST (Fig. 7a), it slightly increases for the ST0 series for the nanocomposites with WS 2 up to 0.2 wt% and then increases more significantly above 0.2 wt%. This result indicates, first of all, the presence of chains in conformation 10 3 [36], typical of α and α' forms [39] in the ST0 series; hence some level of crystallization is induced by the stretching process without further annealing. Moreover, the increase of the Raman signal of the 924 cm − 1 band with the nanotube concentration confirms the nucleation activity of WS 2 NTs [21][22][23][24][25] towards the increase of crystallinity of PLA that makes the stretching process more effective. The Raman intensity of the 924 cm − 1 band for the series ST3 and ST10 is higher than the ST0 series and independent on the WS 2 concentration, suggesting that prolonged annealing (above 3 min) does not promote further crystallization.
A similar behavior was observed for the band at 874 cm − 1 as shown in Fig. S12a even though the effects are less pronounced than at the 924 cm − 1 band because the 874 cm − 1 band is also sensitive to the amorphous PLA phase [36].
The last band that we analyzed was the one located at wavenumber 1454 cm − 1 (Fig. 7b) where the intensity values are reversed with respect to the case of the 924 cm − 1 and 874 cm − 1 bands: the highest intensity is observed for the UNST series, followed by the ST0 and lastly by the ST3 and ST10 series which are very similar. This trend suggests that the 1454 cm − 1 band could be attributed to the amorphous content of PLA, but no other evidence has been found in literature to support this hypothesis. Further measurements at different points of the samples (we reiterate here that we only collected data in the central region of the films) may increase the statistics and reduce the magnitude of the error bar in Fig. 7b.
Regarding the position of the investigated bands, no significant Raman shift is observed as an effect of either WS 2 introduction or process conditions. This suggests that there are no interactions between the filler and polymer matrix that change the force constant of the bonds in vibration, or changes in the packing of the polymer chains in the crystal. Additionally, the FWHM of the investigated 924 cm − 1 band does not change with WS 2 wt%, suggesting that the crystalline domains in PLA and PLA-WS 2 nanocomposites samples are very similar in terms of size and defects. (Fig. S13a).
The thermal behaviour of the PLA and PLA-WS 2 samples was studied by DSC analysis, for additional information about the amorphous phase of PLA (presence of enthalpic relaxation, possible variation of T g , ability to cold crystallization) as a function of the drawing and annealing processes, and of WS 2 concentration. For this purpose, we focused our attention on the thermal transitions present in the first heating scan of the samples.
The thermograms of PLA and PLA-WS 2 0.6 wt% samples are compared in Fig. 8a and b; the temperatures and enthalpies associated to the principal thermophysical transitions are reported as a function of the WS 2 amount in Fig. 8c and d, respectively.
In the heating scans of both UNST PLA and PLA-WS 2 0.6 wt% samples, the thermal transitions of the polymer are observed: an endothermic peak at about 70°C, a cold crystallization exotherm between 100°C and 130°C, and a double peak melting endotherm between 155°C and 175°C.
The first endothermic peak at 70°C overlaps with the expected glass transition interval for PLA and is often associated to different endothermic transitions of the PLA amorphous phase (melting of mesophase [40][41][42], simultaneous devitrification and enthalpic relaxation of the rigid amorphous fraction [43], and physical aging [44]). Therefore, this peak is generally connected to the relaxation of the intermediate phase with some molecular ordering and it is more pronounced in samples characterized by a high percentage of the amorphous phase [26]. Indeed, the highest intensity of this peak is observed in UNST samples both in PLA and nanocomposites, that are found to be amorphous from WAXS (Figs. 4 and 5) and Raman spectroscopy (Figs. 6 and 7). Table 1 Assignments of the analyzed bands (labelled in Fig. 6b) to specific vibrations of PLA chains Vibration mode Associated to phase Ref.

C-COO stretching
Crystalline and amorphous [39] 924 CH 3 rocking and C-C bending Crystalline [39] 1095 OCO stretching - [26] 1454 asymmetric deformation modes of the CH 3 group - This first endothermic peak becomes less intense with stretching (ST0) and post-stretching annealing (ST3 and ST10) (Fig. 8a, b) due to the reduction of the amorphous phase by strain-induced crystallization [26]. Similarly, the enthalpy relaxation becomes less pronounced by increasing the WS 2 concentration: as WS 2 nanotubes favor polymer crystallization, the amount of amorphous phase (responsible for the enthalpic relaxation) decreases with increasing WS 2 concentration. Due to the overlap with enthalpic relaxation, we could make only an approximate estimation of the values of the glass transition temperature, hence we refer to it as "apparent" -T g(app) (measured at the inflection point of the rising edge of the endothermic peak). T g(app) remains almost constant for all the samples independently of the amount of WS 2 and of the process (see Fig. 8c) suggesting that the filler introduction into polymer matrix has no measurable stiffening effect on the amorphous phase. This result is not surprising, given the low quantity of WS 2 used. By continuing to heat the UNST films of PLA and PLA-WS 2 , an exothermic peak is observed in the range 100°-130°C where the cold crystallization of the amorphous phase takes place (Fig. 8a, b). The cold crystallization temperature T cc is plotted as a function of WS 2 concentration in Fig. 8c; for the UNST samples, it slightly decreases with the concentration of WS 2 , in good agreement with the literature [23]. The incorporation of WS 2 NTs into the PLA influences the formation of the nuclei during crystallization of PLA by decreasing the free energy of nucleation [23]. Among the ST series, the cold crystallization is detected only for the ST0 and for the ST3 samples until a WS 2 concentration of 0.3 wt%. It is absent for ST3 of PLA-WS 2 0.6% and in all ST10 of PLA-WS 2 , even though a melting peak is detected in the thermograms of all the ST samples. This suggests the presence of pre-existing crystals from processing according to WAXS and Raman data.
The presence or absence of cold crystallization in the heating scans of ST samples is associated with the ability of the stretching and annealing process to promote the formation of crystals due to chain orientation and changes in the amount of WS 2 NTs. When the thermomechanical process is effective, the fraction of polymer chains in the amorphous state decreases. As a consequence, the cold crystallization is reduced or even suppressed [26] as in the case of all ST10 samples and for PLA-WS 2 ST3 with wt% higher than 0.3% (no T cc measured, Fig. 8c). For these latter samples the major role towards crystallization is played by strain and not by annealing.
As a further consideration of Fig. 8c, the value of T cc decreases from the UNST to the ST series, as a consequence of the stretching and of the annealing [45]. For the ST series, a second weak exothermic peak (highlighted with a star in Fig. 8a), centered around 155°C, is often visible before the melting peaks, due to further crystallization occurring as the heating proceeds. This peak is characteristic of presence of PLA crystallites in α' phase which change into α phase during the heating scan by solid-solid transition. Indeed, the more disordered α' structure, generally obtained at lower temperatures than the α form (below 120°C), is characterized by lower thermal stability than α form [16] and can be converted into α by annealing at high temperature [38].
The range of temperatures between 160°C and 180°C is where the melting of the crystals takes place (Fig. 8a,  b). While a single melting peak (corresponding to a single melting temperature T m ) is visible for the ST samples, a double peak (corresponding to two melting temperatures T m1 and T m2 with T m1 < T m2 ) is detected for UNST samples. According to similar observations in melt-processed PLA-WS 2 nanocomposites, the first endothermic peak (T m1 ) is due to the fusion of smaller and metastable crystals formed during the DSC heating process itself (giving the slow heating rate of 5°C/min). Such crystals recrystallize at higher temperature leading to the second melting peak (T m2 ) [22,23]. The absence of a double peak due to a process of melt-crystallization in ST samples suggests that the stretching process favors the formation of more perfect and stable crystals independently of the presence of the nanotubes. The values of T m1 , T m2 , T m and ΔH m are similar for all the samples (Fig. 8c, d) and are not influenced by the WS 2 concentration, in good agreement with data reported in [22,24]. As the results of Raman experiments also point out, this is a further indication that the size and degree of perfection of crystallites of PLA in PLA-WS 2 composites is similar for all the stretched samples.
The degree of crystallinity (X c ) was evaluated from the DSC thermograms using eq. (1) in the Methods section ( Table 2). For PLA (wt% = 0) and PLA-WS 2 with WS 2 NTs concentration up to 0.3% the X c doubles for the effect of annealing (see values for ST0 vs. UNST) and increases four times after 3 min of annealing (ST3) reaching a value of about 40%; a longer annealing does not further modify the X c (ST10 similar to ST3). For the PLA-WS 2 0.6% the stretching itself is enough for the sample to reach the maximum crystallinity.
The data of Table 2 are displayed in Fig. 9 and correlated with the Raman data, specifically with the ratio between the intensity of the 924 cm − 1 and 1454 cm − 1 bands I 924 /I 1454 as an index of the amount of the crystalline phase with respect to the amorphous one as measured by Raman spectroscopy. As expected, the lowest values of X c and of the ratio I 924 /I 1454 are measured for the UNST samples that show an amorphous structure from WAXS (Figs. 4 and 5). Both the stretching process and inclusion of WS 2 cause an increase of X c and I 924 / I 1454 . A further increment of X c and I 924 /I 1454 after the annealing is detected for PLA and PLA-WS 2 nanocomposites with WS 2 wt% up to 0.3%, whereas the stretching itself suffices to induce the highest X c and I 924 /I 1454 values in the PLA-WS 2 0.6% .

Conclusions
We studied the crystallinity and the thermal behaviour of stretched and annealed nanocomposite films of PLA and PLA-WS 2 containing WS 2 nanotubes at concentrations from 0.1% to 0.6% by weight. The samples were prepared by stretching amorphous films up to 100% strain held at 90°C and annealed at the same temperature for different times: 0, 3 and 10 min. WAXS, Raman spectroscopy and DSC analyses showed that increasing the nanotube concentration enhances the crystallinity of PLA induced by the stretching process. In particular, WAXS and Raman data proved that a WS 2 NTs concentration of 0.3 wt% leads to more than twofold increase of crystallinity with respect to neat, stretched PLA. The effect of the additional annealing step on the crystallinity is more evident for samples with WS 2 content lower than the maximum concentration Table 2 Percentage of crystallinity (X c ) evaluated from DSC data (Fig. 8)