The chemical modification of multilayer carbon nanotubes has already been described previously [24]. Herein, a brief description. 2 g of MWCNT was added to 400 mL of sulfonitric solution H₂SO₄/HNO₃ (3:1). The dispersion was kept under stirring and refluxing, at 110˚C, for 3 hours. After that, the medium was diluted (four times) by distilled water accommodated in dialysis membranes and immersed in distilled water until reaching pH 6.0. At the end, the content was conducted to evaporate until constant weight. The black powder was labeled as oxidized-CNT (o-CNT).

Hybrid fillers of o-CNT and hydroxyapatite (HAp) were prepared through acetone dispersion, at different proportions. The first one consisted of blending by handling o-CNT-HAp (1:2) and the second one of o-CNT-HAp (2:1). These proportions were thought in order to assess how its constitution impact on the physico-chemistry properties of the composites. For both, the HAp dispersion was dripped slowly to the o-CNT dispersion, at 25˚C, under magnetic stirring, for 24 hours. After that, the content was submitted to evaporation and a dark brown powder was attained being labelled o-CNT-HAp(1:2) and o-CNT-HAp(2:1), respectively.

PCL and composite films were obtained by its dissolution in methylene chloride (10%, m/v) following the addition of each filler MWCNT, o-CNT, HAp and o-CNT-HAp(1:2) (0.05 wt.%), maintaining under stirring for 24 h. After that, the dispersion was poured onto Petri dish, kept for complete solvent evaporation to attain a film. For comparison, a film of PLA and PLA plus o-CNT-HAp(2:1) (0.05 wt.%) was prepared as mentioned above. Summarizing, the samples were labeled as follows: PCL, PCL-MWCNT, PCL-o-CNT, PCL-HAp and PCL-o-CNT-HAp(1:2); PLA and PLA-o-CNT-HAp(2:1). In all cases, the low content of filler was thought to be enough to attain better dispersion and promote changes in physico-chemical properties of the polymer matrix.

Fourier-transform infrared spectroscopy (FTIR) was performed using a Perkin Elmer Frontier equipment in the of 4000 - 400 cm⁻¹, with 120 scans and a resolution of 4 cm⁻¹.

The crystallographic structure was evaluated using a wide-angle X-ray diffractometer Rigaku Miniflex, employing CuKα radiation with a wavelength λ = 1.5418 Å, a Ni filter, a current of 20 mA, a voltage of 20 kV, 2θ between 2˚ - 50˚ and resolution of 0.05˚.

Molecular relaxation of PCL and composites was performed through spin-spin relaxation, transverse relaxation time (T₂), by magic sandwich echo-free induction decay (MSE-FID) by using time domain nuclear magnetic resonance on a MARAN Ultra 0.54 T (23.4 MHz for 1H) equipment, with a probe diameter of 18 mm and WinFit 2.4-point adjustment software. The analysis was performed at temperature of 30˚C ± 2˚C, with a 2-Eco-Solid pulse sequence, a pulse duration of 90˚ in 7.5 μs, with 2048 points spaced 0.5 μs apart. The number of accumulations corresponds to 64, with a recycling time of 1 second and a receiver gain of 8%. The adjustment function for the MSE-FID signals was determined by Equation (1): A_R, amplitude or fraction of the rigid region; A_M, amplitude or fraction of the moving region; T₂*, transverse relaxation time of each of the fractions obtained by the equipment; μ b , centroid of the Gaussian function; K, offset or baseline of the relaxation signal that compensates for the influence of noise during nonlinear tuning. The signal is governed by two components. The first (on the left) is related to highly rigid ¹H nuclei and is governed by a Gaussian function. The second (on the right) is characterized by more mobile hydrogens that exhibit higher T₂H values and an exponential decay according Neto et al. [25]. The fraction of each domain rigid domain (A_R) and mobile domain (A_M) was calculated according to Equation (2). T₂ relaxation analysis was selected for PCL and its respective composites since better resolution was achieved.

Molecular motion of PLA and composite was performed in the equipment described above evaluating the longitudinal relaxation time (T₁), also known as spin-lattice relaxation, which result of the existence of transient magnetic moments produced by the rotational and translational movements of neighboring molecules [25][26]. Two replicates were evaluated.

Samples thermal stability was conducted in a TA Instruments Q-500 equipment, in the range of 30˚C - 700˚C, at 10˚C/min, in a nitrogen atmosphere. T_onset, T₁₀, T_max and residue were evaluated.

Differential scanning calorimetry was performed using a TA Instrument Q1000. Five thermal cycles were performed according to ASTM D3418. In the first cycle, the sample was heated from −80˚C to 200˚C at 10˚C/min, using nitrogen as the carrier gas, and held at this temperature for 2 minutes to eliminate the thermal history. A cooling cycle was then performed to −80˚C at the maximum speed of the equipment. A second heating cycle was performed under the same conditions as the first. In the fourth cycle, a second cooling cycle was performed from 200 to −80˚C at 10˚C/min, where it was possible to determine the cooling crystallization temperature (T_c). The fifth cycle was conducted under the same conditions as the first, and the crystalline melting temperature (T_m) was registered. Degree of crystallinity was calculated based on Equation (3) where ΔH_m, experimental enthalpy of fusion and Δ H m 0 , the enthalpy of fusion of 100% crystalline PCL, 139.5 J/g [27] and PLA 93.6 J/g [28], Φ corresponds to the content of the added particle.

Rheology evaluation was conducted by using a TA rheometer, model AR-2000, 25 mm in diameter with parallel plate geometry at 60˚C. The procedure was performed in an inert atmosphere under dynamic conditions. In order to observe the linear viscoelasticity region of the films, deformation tests were performed at a frequency of 1 Hz. Storage and loss moduli, as well as complex viscosity, were also determined. Two replicates were evaluated.

Figure 1 shows samples’ spectra. PCL showed absorptions at 2950 cm⁻¹ (asymmetric stretching of CH₂), 2865 cm⁻¹ (symmetric stretching of CH2), 1730 cm⁻¹ (symmetric stretching of C=O), 1294 cm⁻¹ (stretching of C-O and C-C in the crystalline phase), 1234 and 1167 cm⁻¹ (symmetric and asymmetric stretching of C-O-C), 1106 and 1049 cm⁻¹ (O-C vibrations) in agreement with reported by Adeniyi et al. [29]. PLA absorptions were ester carbonyl (C=O) stretching at 1750 cm⁻¹, C–H bending modes at 1450 and 1340 cm⁻¹, and the C–O–C stretching vibrations in the region between 1260 and 1050 cm⁻¹ which are in accordance to published by Justino Netto et al. [30]. To guide the evaluation of the hybrid fillers, spectra of them and their precursors are described. HAp absorptions were found at 650, 1631 and 3435 cm⁻¹ (hydroxy group), 1037 and 1095 cm⁻¹ (phosphate crystalline arrangement) and 560, 601 and 961 cm⁻¹ (phosphate group) which were endorsed by several authors [31]-[33]. For o-CNT, multi absorptions were detected namely intense and enlarged peak at 3495 cm⁻¹ (hydroxy group), 1725 cm⁻¹ (carbonyl stretching), 1624 and 1497 cm⁻¹ (C=C ring stretching), 1435 and 995 cm⁻¹ (C-O-H, in plane and out of plane respectively), 1232 cm⁻¹ (C-C-O stretching), 1139 and 1067 cm⁻¹, vibrations of O=S=O(OH), 889 and 846 cm⁻¹ (C-H aromatic ring vibrations) following what was published by Puliyasseri et al. and Saipanya et al. [34][35]. The spectrum of each hybrid filler shows fidelity to the proportions of HAp and o-CNT. For o-CNT-HAp(1:2), 3416, 2973, 2930, 2859, 1740, 1622, 1095, 1037, 964, 865, 634, 605 and 564 cm⁻¹ were highlighted. For o-CNT-HAp(2:1), absorptions 3602, 3552, 2975, 2923, 2852, 1740, 1621, 1420, 1150, 1093, 1013, 958, 657, 599 and 567 cm⁻¹ were registered. With respect to the PCL-o-CNT-HAp(1:2) and PLA-o-CNT-HAp(2:1) composites their spectra display the predominant absorptions of polymer-base which could be associated with the hiding of fillers’ absorptions.

Figure 1. Infrared spectra of fillers and composites.

Figure 2 presents the samples’ diffraction patterns. PCL exhibits two diffraction angles (2θ) at 21.5 (110) and 23.8˚ (200) endorsed by publication of Doostmohammadi et al. [36]. PLA displays diffraction angle at 16 - 17 representing its α-crystalline form as reported by Tamburini et al. [37]. HAp presented diffraction angles at 25.9˚ (201), 26.5˚ (002), 30.2˚ (211), 31.9˚ (112), 33.0˚ (300), 34.2˚ (202), 40.0˚ (410), 46.9˚ (222) and 49.6˚ (213) finding correspondence with the results reported by Targońska et al., Cao et al. and Nurbaiti et al. [32][33][38]. Before commenting the o-CNT diffraction patterns, it is necessary to emphasize that MWCNT precursor presented diffraction angles at 25.9˚ (002) and 43.1˚ (100) being the first one associated with its concentric graphite structure. Interesting to note that o-CNT revealed the first 2θ angle displaced to 23.5˚ showing variation of d_spacing from 3.44 to 3.77 nm. This increase in interplanar spacing is indicative of the size and structure of the crystals, justified by the presence of hydroxyls, carbonyls, and sulfonic groups possibly adhered to the surface of the o-CNT, after the chemical modification of MWCNT [39]. PCL-o-CNT-HAp(1:2) exhibits 2θ angle at 21.3, 21.9 and 22.6˚ related to PCL besides 23.6˚ which could be overlapping of 2θ of PCL and o-CNT; a low intense 2θ around 30˚ could be attributed to HAp apatite crystalline plane. Irani et al. developed scaffold by co-electrospun of PCL, gelatin, chitosan and carbon nanotubes (CNT, 0.1, 0.2 and 0.4 wt.%) for tissue engineering. X-ray diffraction evaluation did not detect CNT-related 2θ angles which were associated to its low concentration [40]. PLA-o-CNT-HAp(2:1) shows diffraction pattern similar to PLA alone which could be attributed to the low filler content.

Figure 2. DRX patterns of fillers and composites.

The application of magnetic field in the study of molecular motions is considered a powerful technique to evaluate relaxation times. Different nuclear magnetic resonance processes named longitudinal and transverse relaxation times allow to access the molecular mobility of polymer, fluid and so on. Herein, longitudinal magnetization (T₁) and transverse magnetization (T₂) were applied to evaluate the effect of fillers on the polymer molecular motion. Table 1 presents data (T₂ and % of area) of transverse magnetization of PCL and composites. In general, for all composites, T₂ increased when faced up to PCL alone indicating increase of molecular mobility. When it compares the percentage of rigid and mobile domains it was noticed that were invariable. Only the PCL-o-CNT composite got away from that trend. Although T₁ is determined by distinct physical process when compared to T₂ it is also valid for evaluation of polymer relaxation. T₁ was applied to assess the relaxation mode of PLA and PLA-o-CNT-HAp(2:1). Figure 3 presents the domain curves of PLA and composite. Both showed relaxation curves with two domains: flexible (shorter relaxation time) and rigid (longer relaxation time). At relaxation times below 100.000 ms, the domain was related to the mobility of the amorphous phase chain; the second, more intense peak, in the region of 100,000 - 2500.000 ms, was related to the conformation of more rigid, packed chains, or those with greater mobility hindrance. The influence of o-CNT-HAp(2:1) on the molecular mobility of the PLA phases was sharply evidenced by the enlargement of the domain curves and their shift to lower and higher values along the time axis. The lowering of T₁ indicates higher chain mobility while on the contrary its increase reveals lower chain mobility. Catauro et al. synthesized nanocomposites of SiO₂ with PCL (0, 6, 12, 24 and 50 wt.%) by sol-gel process. The occurrence of H-bonds among carbonyl groups of PLA chain and Si-OH was evaluated by infrared spectroscopy (FTIR) and solid-state nuclear magnetic resonance (NMR). By ¹³C-CPMAS-NMR (cross-polarization magic-angle spinning nuclear magnetic resonance), two resonance peaks were detected which were associated to the PCL crystalline and amorphous phases. Still, ¹H and ¹³C-CPMAS-NMR applied to SiO₂/PCL composites revealed enlargement of carbonyl peak with the increase of PCL content which it was considered as non-covalent interaction among PCL oxygens and the -OH moieties of silicon dioxide [41]. Through molecular dynamics simulations, Çelik et al. intended to find a guide for design and optimization of molecular-level interaction of polymer-nanoparticle in composites of poly(lactic-co-glycolic acid) (PLGA) and poly(ε-caprolactone) (PCL) with oxidized carbon nanotubes (GO) and reduced oxidized carbon nanotubes (rGO). A series of evaluations such as hydrogen bond occupancy, radius of gyration, potential and binding energies, radial distribution functions, and solvation free energy were experienced. The authors found that 75:25 PLGA-GO and 75:25 PLGA-rGO were the most stable interaction profiles among PLGA variants, while PCL-GO was the most stable [42]. In our study, both polymers are biodegradable polyesters in which carbonyl groups could form hydrogen bonds with reactive groups onto o-CNT. In the case of PCL-o-CNT-HAp(1:2), this type of interactions was avoided or in lower extent owing to the lower availability of the reactive groups. In Table 1, when o-CNT is added to PCL, there is great variation on T₂ and domain percentage of the polymer. The addition of HAp to PCL produces no changes on the PCL relaxation parameters. The incorporation of o-CNT-HAp(1:2) (0.5%) causes a negligible change in T₂ and none in the domain percentages which could indicate that in the o-CNT-HAp(1:2) there is low availability of the carbonyl and hydroxyl groups of o-CNT leading to lower level of polymer-filler interaction in the PCL-o-CNT-HAp(1:2) composite. The spatial degree of freedom of the methylene and chains entanglement of the PCL could be additional factors to affect polymer-filler interaction considering the availability of oxygen atoms to form hydrogen bond with carbonyl and hydroxyl groups in o-CNT. On the contrary, for PLA-o-CNT-HAp(2:1), in the repeat unit of PLA only a secondary carbon linked as follows –H-C-CH₃ appeared between two ester groups resulting in no variation on the degree of freedom. If compared to PCL, even tangled the PLA chains favor the formation of hydrogen bond between the oxygen atoms of the ester groups, near to each other, and carbonyl and hydroxyl groups in o-CNT. This could endorse the more effective action of o-CNT-HAp(2:1) on the PLA relaxation process. As a suggestion, Figure 4 shows step-by-step the structural constitution of the fillers. o-CNT possesses several carbonyl groups of carboxyl acid/ketone and hydroxyl groups onto its surface while HAp owning two hydroxyl anion for charge equilibrium. When o-CNT and HAp are physically mixed at 1:2 proportion most reactive groups onto o-CNT are hidden. On the contrary, at 2:1, o-CNT most reactive groups are available. Then, when mixed, the pair PLA and o-CNT-HAp(2:1) greater possibility of chemical interaction arises while on the contrary in the pair PCL and o-CNT-HAp(1:2) is more difficult.

Table 1. T₂ relaxation time of PCL and composites.

Figure 3. T₁ relaxation domain curves of PLA and PLA-o-CNT-HAp(2:1).

Figure 4. Fillers structural constitution: speculative schematic representation.

Figure 5 exhibits the loss mass and derivative curves. Either PCL and composites or PLA and PLA-oCNT-HAp(2:1) presented only one stage of mass loss. Except for PLA-oCNT-HAp(2:1), for the others samples a unique derivative peak was noticed. Table 1 displays the thermogravimetric data. For all fillers tested, PCL composites revealed decrease on T_onset, T₁₀ and T_max. Inversely, the presence of oCNT-HAp(2:1) into PLA matrix denotes increase on T_onset, T₁₀ and T_max. Fillers can play a role in increasing or decreasing polymer matrix thermal stability in polymeric composites. Mohammed et al. (2025) studied the incorporation of silica (0 - 40 wt.%) on the properties of polycaprolactone/poly(vinyl chloride) (PCL/PVC). The authors registered multi-step decomposition and deduced that silica residue improved the thermal stability at high temperatures [43]. Poljacek et al. evaluated the action of PCL and nanosilica when incorporated in PLA. The addition of 1 and 3 wt.% of nanosilica in PLA did not alter its thermal while for PLA/PCL blends an increase of thermal stability was noticed [44]. Espirito Santo et al. investigated the influence of biosilica (extracted from Dragmacidon reticulatum) at different proportions (10, 15 and 20 wt.%) on PCL samples from electrospun fibers and casting films. For all composites, the thermal decomposition occurred at lower temperatures which denoted decrease of thermal stability [3]. Taha et al. reported a study on hybrid composites of PLA filled with zinc oxide (ZnO) nanoparticles and varied content of oxidized carbon nanotubes (GO) (3, 4 and 6 wt.%). Significant enhancement of thermal stability was observed at high content of GO which was attributed to the filler network which effectively hindered the release of volatile degradation matters [45]. Table 2 displays the thermogravimetric data of the PCL and PLA materials. For PCL-based composites, T_onset, T₁₀ and T_max showed a tendency to decrease but the decrement was more evidenced for the PCL-oCNT-HAp(1:2) which could be imputed to some filler catalytic action. Reversely, for PLA-oCNT-HAp(2:1), T_onset, T₁₀ and T_max were significantly higher denoting upper thermal stability which endorses the hypothesis of better chemical interactions between polymer matrix and filler. Summarizing, the thermogravimetric data can suggest that either polymer structure or filler composition had influence on the thermal stability of the composites. If compared PCL-oCNT-HAp(1:2) and PLA-oCNT-HAp(2:1) composites, in each one polymer matrix and filler had inverse action on their thermal stability. Herein, it was possible to infer that the composite PLA-oCNT-HAp(2:1) possesses high thermal stability which could be imputed to the occurrence of chemical interaction between constituents. On the contrary, thermal stability tends to decrease for all PCL composites probably owing to a gap of chemical interaction between polymer and fillers.

Figure 5. Mass loss and derivative curves of PCL and PLA materials.

Table 2. Thermogravimetric data of PCL and PLA materials.

Figure 6 shows the third heating curves of PCL and PLA materials. For the third heating curves of PCL, it was detected the melting temperature (T_m). On the other hand, their counterpart for PLA materials showed T_g and T_m besides heating crystallization temperature (T_c). Table 3 displays the calorimetric data of the PCL and PLA materials. With respect to PCL, its composites did not show significant variations on the calorimetric data. The values of T_c and T_m are in accordance with those reported by Mohammed et al. [43]. Similarly, for all composite, X_c was invariable. For PLA, T_g and T_m value are agreement with what was published by Oliver-Cuenca et al. [46]. In the same article, it is described that PLA homopolymer possesses four polymorphic structures named α, α', β, and γ. α-phase is the most common and stable crystalline structure. Α'-phase can occur at lower crystallization temperatures. β-phase is formed by mechanic stress combining high temperature and high drawing ratio. γ-form is obtained through specific crystallization condition, epitaxial crystallization. Herein, none of the conditions mentioned for the appearance of polymorphism was applied, then the double T_m peaks were associated to the formation of crystals of different sizes. With respect to PLA crystallization temperature (T_c), its value was endorsed by the article of Akdevelioğlu et al. [28]. For PLA-oCNT-HAp (2:1), it was observed that the T_g and T_c of PLA were shifted to higher temperature, a unique melting temperature (T_m) was detected while the degree of crystallinity (X_c) diminished. The effects on the PLA calorimetric characteristics by addition of PCL (0 - 50 wt.%) and nanosilica (1 and 3 wt.%) were investigated by Poljacek et al. When alone, the incorporation of nanosilica in PLA promoted an increase of degree of crystallinity (X_c). In blend, with or without addition of nanosilica, for PLA, X_c did not show any tendency while X_c of PCL phase decreased when its content was 30 wt.%; above this amount X_c stabilized. With respect to the crystallization temperature (T_c) and melting temperature (T_m), both were invariable for PLA and PLA/PCL blends, with or without the presence of nanosilica [44]. Lomakin et al. studied composites of PLA with layered nanomodifiers-graphene nanoplates and sodic montmorillonite (0, 1, 5 and 10 wt.%). The calorimetric data revealed that T_g and T_m were practically invariable. Independent on the fillers amount, both ones showed increase on T_c and X_c which the authors associated to the role as nucleating agent promoting PLA crystallization [47]. Herein, it was evidenced that not only the amount of filler but its chemical nature and constitution besides the polymer chemical structure address the interaction polymer-filler in polymer composite. In the case of PCL-o-CNT-HAp(1:2), no variation in calorimetric data could be imputed to the lack of interaction PCL with o-CNT-HAp(1:2) because of quasi total occlusion of the carbonyl and hydroxy groups in o-CNT by HAp hindrance. For PLA-o-CNT-HAp(2:1), the existence of hydrogen bond between carbonyl and hydroxyl groups in o-CNT and oxygen atoms in PLA chains favors the increase of T_c but create an obstacle to the diffusion of PLA chains to the crystallization centers decreasing the X_c. Summarizing, the DSC results emphasized that structural arrangement of the fillers o-CNT-HAp(1:2) and o-CNT-HAp(2:1) was the reason for the differences found in the calorimetric results. PLA and o-CNT-HAp(2:1) attained a certain degree of chemical interaction which is absent or in lower extent that found for PCL and o-CNT-HAp(1:2). TDNMR and TGA endorsed these findings.

Figure 6. Precursors and composites heating curves (third cycle).

Table 3. Precursors and composites calorimetric data.

Figure 7 presents the moduli (storage and loss) and complex viscosity curves. For all samples, storage modulus (G') increased with frequency. The curve of PCL-CNT is separated from the others. The curves of PCL, PCL-o-CNT, PCL-HAp and PCL-o-CNT-HAp(1:2) are superimposed while the curves of PLA and PLA-o-CNT-HAp(2:1) are also overlapped. For loss modulus (G"), the curves of PCL, PCL-CNT, PCL-HAp and PCL-o-CNT-HAp(1:2) are superimposed. The curves of PLA and PLA-o-CNT-HAp(2:1) are overlapped and laid out as intermediate related to the others. Immediately below, the curve of PCL-o-CNT can be seen. With respect to |η*|, the curves of PLA and PLA-o-CNT-HAp(2:1) are upper than others being the latter with higher viscosity value. Following, the curves of PCL, PCL-CNT, PCL-HAp appeared practically superimposed while immediately below the curve of PCL-o-CNT is seen. In the range of frequency studied, PCL, PCL-CNT, PCL-HAp and PCL-o-CNT-HAp(1:2) behaved as a Newtonian fluid while PLA and PLA-o-CNT-HAp(2:1) resembled as pseudoplastic fluid. Table 4 displays the G', G" and G"/G' ratios of the samples at 10¹ and 10² rad/s. Either PCL and composites or PLA and composites the values of G' and G" reflected the response time of the materials. In this type of measurement, both increased at high frequency once polymer chains, microstructures, distribution and dispersion of fillers exerted influence on chain relaxation time and flowability. Herein, at high frequency, the chain relaxation time was not achieved and then the material response is rigidity. As the amount of filler was very low either in PCL or in PLA, the results in the rheological evaluation were driven mainly among neighboring polymers chains through restrict mobility to each other.

This explanation was validated by the values of ratio G"/G'.

Figure 7. Precursors and composites moduli and complex viscosity curves.

Table 4. G', G" and G"/G ratio at different frequencies.