*2.4. Statistical Analysis*

The test data were evaluated through analysis of variance (ANOVA) using STATGRAPHICS Centurion XVI v 16.1.03 from StatPoint Technologies, Inc. (Warrenton, VA, USA). Fisher's least significant difference (LSD) was used at the 95% confidence level (*p* < 0.05). Mean values and standard deviations were also calculated.

#### **3. Results and Discussion**

#### *3.1. Morphology of the Electrospun PCL Fibers and Films*

As it can be seen from the observation of Figure 1a,b, a narrow distribution of fiber diameter with an average at 2.75 ± 0.4 μm in PCL fibers, and 2.25 ± 0.7 μm in PCL/PdNP fibers was observed. The surface of the formed fibers was seen to be smooth and without beaded regions. The diameters of the fibers produced by electrospinning primarily depended on the spinning parameters, the most crucial being the solution concentration [34]. The smaller average diameter of the PCL/PdNP fibers can be attributed to an expected increase in conductivity, in agreemen<sup>t</sup> with previous works making use of metallic nanoparticles. Thus, in the case of nanoparticles of ZnO, the authors hypothesized that the solution was seen to have a larger charge capacity, and then to be driven by a stronger electric force along the fibers; therefore, smaller fiber diameters were obtained [35].

The SEM images of the films' cross-sections, shown in Figure 1c,d, indicated the presence of compact structures that resulted from the annealing post-processing step, which was in good agreemen<sup>t</sup> with previous works [20–22]).

**Figure 1.** Scanning electron microscopy (SEM) images of the surface view and the cross-section of the polycaprolactone (PCL) fibers, with and without palladium nanoparticles (PdNPs), and their respective annealed films: (**a**) Surface view of the neat PCL fibers; (**b**) Surface view of the PCL/PdNP fibers; (**c**) Cross-section of the neat PCL film; (**d**) Cross-section of the PCL/PdNP film.

Additionally, FIB-SEM was used to cross-section the internal structure of the electrospun fibers. The detailed examples of the FIB-SEM images of PCL /PdNP fibers collected during FIB sectioning are presented in Figure 2, showing that Pd particles are incorporated in the two fibers. The cross-sectional images enabled the visualization of the PCL/PdNP fibers in 3D, as shown in Figure 3. The Pd nanoparticles were seen to be distributed along the PCL fibers, forming small agglomerates at different parts within the fiber cross-sections. Finally, the concentration of particles was also estimated per the given volume of the piece of fiber analyzed, and for the PCL fiber 1, this was 1.1%, and for PCL fiber 2, this was 0.9%. These observations sugges<sup>t</sup> that the PdNP are better dispersed and distributed across the PCL fibers, and that the agglomerates may account for most of the Pd in the nanocomposites.

**Figure 2.** The cross-sectional SEM image of two PCL fibers with Pd nanoparticles after (FIB-SEM) sectioning. (**a**) PCL fiber 1 at the bottom with visible bright nanoparticles and (**b**) PCL fiber 2 at the top of the image with visible bright nanoparticles.

**Figure 3.** (**a**) The side view of 3D reconstructions of PCL fiber 1 (in semi-transparent yellow) and Pd nanoparticles (in white), including the SEM image (obtained from FIB sectioning) in the middle of the reconstruction, (**b**) 3D reconstruction of PCL fiber 1 with Pd nanoparticles inside, (**c**) 3D reconstruction of particles only inside of PCL fiber 1. The binding box size for this reconstruction had the following dimensions 3.105 × 1.185 × 1.526 μm.

In spite of the very revealing FIB-SEM results, the SEM technique is thought to be inadequate for resolving highly dispersed Pd nanoparticles within the polymer matrix [36]. In order to check for this, additional TEM experiments were conducted on the samples.

The additional TEM experiments displayed in Figure 4 indicate that the Pd nanoparticles, in agreemen<sup>t</sup> with the FIB-SEM experiments, exhibit a significant degree of aggregation within the fiber. Due to attractive forces (Van der Waals and others), particles tend to agglomerate, even in suspension, unless stabilized by equivalent repulsive forces such as surface charge or steric effects. Thus, the smaller the particle size, the greater the relative attractive forces per unit mass. This means that it becomes progressively more difficult to disperse nanoscale materials as the size decreases [37]. In any case, TEM also revealed the presence of some highly dispersed and distributed nanoparticles within the cross-section of the biopaper film. The smallest particles were seen to have diameters of ca. 6 ± 2 nm, and they seemed evenly distributed throughout the fibers/film. Our prior studies of PHB/PdNPs electrospun fibers, also showed a similar dispersion of Pd nanoparticles, with some clear agglomeration zones within the fibers [20].

**Figure 4.** Scanning electron microscopy (TEM) images taken (**a**) directly on electrospun polycaprolactone (PCL) fibers containing palladium nanoparticles (PdNPs) and on (**b**) microtomed sections of their corresponding annealed film.

#### *3.2. FTIR Analysis of the PCL Electrospun Fibers and Films*

The FTIR spectra of the electrospun neat PCL fibers and film, and the PCL/PdNP fibers and film are shown in Figure 5.

The PCL spectrum displays the characteristic peaks of C=O stretching vibrations at 1726 cm<sup>−</sup>1, CH2 bending modes at 1361, 1397, and 1473 cm<sup>−</sup>1, and CH2 asymmetric stretching at 2942 and symmetric stretching at 2862 cm<sup>−</sup>1. The C-O-C stretching vibrations yield peaks at 1042, 1107 and 1233 cm<sup>−</sup>1. The bands at 1160 and 1290 cm<sup>−</sup><sup>1</sup> are assigned to C-O and C-C stretching in the amorphous and crystalline phases, respectively [38–40].

The overall PCL spectrum, including the main bands ascribed to PCL, such as the peaks at 2949 and 2865 cm<sup>−</sup><sup>1</sup> from methylene (CH2) groups, and the strong carbonyl (C=O) peak centered at 1720 cm<sup>−</sup>1, were not seen to be affected either by incorporating PdNP, nor by the post-processing step, suggesting a lack of changes across the polymer molecular backbone.

**Figure 5.** ATR-FTIR spectra of the electrospun PCL and the PCL/PdNP fiber mats and annealed films.

#### *3.3. Thermal Properties of the PCL Electrospun Fibers*

Table 1 shows the thermal properties, melting and crystallization points and enthalpies, in the first heating run and the subsequent crystallization run from the melt for the PCL and PCL/PdNP fibers mats. With the exception of the melting point, the rest of the thermal features were very similar for the neat PCL and nanocomposite fibers. The melting temperature for pure PCL is typically reported

at 60 ◦C, and the glass transition temperature is −60 ◦C. As discussed, in the melting point, there was an increase of ca. 5 ◦C in comparison with the mean PCL, which must be explained by the addition of PdNPs to the polymer. This may due to the interaction of the polymer chains with the surface of the particles, which can change the chain kinetics in the region immediately surrounding the nanoparticles [41]. Similar results were reported by Bajsi´c et al. [42], where the melting point of the PCL/TiO2 composites was found to increase slightly with an increasing load of TiO2 microand nanoparticles.

**Table 1.** Thermal properties obtained by DSC in terms of melting temperature (Tm), normalized melting enthalpy (ΔHm), crystallization temperature (Tc), and crystallization enthalpies (ΔHc) for PCL and PCL/PdNP fibers.


Thermogravimetric analysis (TGA) was carried out to evaluate the degradation temperature of the PCL and PCL/PdNP fibers, including the curves of the first derivative analysis (blue lines) (see Figure 6), and the results are summarized in Table 2.

From Figure 6, it can be observed that PCL and PCL/PdNP initiated degradation at 342 and 355 ◦C, respectively, exhibiting two transition peaks: The first transition peaks were at 388 and 391 ◦C, and the second transition peaks were at 449 and 447 ◦C, respectively. The residual material of PCL and PCL/PdNP had a slight difference of ca. 0.8%, which is ascribed to the Pd that is present in the sample.

The data in the work reported here indicate that adding 1 wt % PdNP resulted in a slightly higher degree of thermal stability for the composite. Previous studies showed that other metallic nanofillers can impact the degradation temperature of PCL in different ways. Thus, Wang et al. demonstrated that the thermal stability of PCL was depressed by the incorporation of Fe3O4/GO nanoparticles, most likely due to the filler acting as a catalyst for polymer degradation [43]. Castro-Mayorga et al. also observed that the degradation temperature of poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBVs containing ZnO nanoparticles showed lower degradation temperatures than that of pure PHBV3. This was attributed to the high thermal conductivity and catalytic properties of the ZnO nanoparticles [19]. Other studies sugges<sup>t</sup> that a temperature drop can also be explained by the fact that nanoparticles can weaken to some extent the interactive force of polymer inter-chains, and hence assist the thermal decomposition of the nanocomposite [44]. However, in the current work, it was observed that the nanofiller induced a somewhat improved degree of thermal stabilization, which may be ascribed to a better adhesion between the nanoparticles and the polymer matrix, resulting in both a hindered diffusion of volatile decomposition products, and/or the sorption of these over the filler surface, for the nanocomposites. Thus, the effect of a filler in thermal stability is in fact thought to depend on the type, content, interfacial interaction, and the degree of dispersion and distribution of this into the polymer matrix [45]. Thus, Ag [45], Fe, and Zn nanoparticles [46,47] have also been previously reported to enhance the thermal stability of PHA, supporting the current results for PCL.

**Figure 6.** Thermogravimetric analysis (TGA) curves of the electrospun PCL (**left**) and PCL-containing palladium nanoparticles (PdNPs) fibers (**right**).

**Table 2.** Values of thermal stability obtained from the thermogravimetric analysis (TGA) curves of the electrospun PCL and PCL/PdNP electrospun fibers in terms of degradation temperature at 5% of mass loss (T5%), maximum degradation temperature of the two degradation peaks (Td1, Td2), and residual mass at 900 ◦C (R900).


#### *3.4. PHB Electrospun Fiber Morphology*

Similarly to the FIB-SEM investigation of PCL fibers, the sectioning of PHB mats was also carried out, and it is displayed in Figure 7. The SEM images in Figure 7 indicate that the PHB fibers morphology is similar to the one reported in earlier work, with the fiber cross-section ranging between 200 and 400 nm, and showing a smooth and beads-free fiber morphology [20–22].

**Figure 7.** SEM images of the PHB sample. (**a**) Overview of PHB nanofibers, (**b**) PHB fibers after FIB sectioning.

The cross-section of the individual PHB fiber shown in Figure 8 revealed that the PdNP agglomerated even more strongly than seen in PCL at the core of fiber, even with a surfactant, which was also visualized with the 3D reconstructions present in Figure 9. The 3D tomography allowed us to verify the presence of PdNP in the individual PHB fiber section scanned, similarly to the PCL data shown in Figure 3. The estimation of the Pd nanoparticle concentration of the fiber section analyzed yielded a concentration of 3%, suggesting that Pd is not as well-distributed as in PCL. This is a relevant finding that suggests that even when a surfactant was added to the PHB to improve filler distribution, as suggested by our earlier work [20], the distribution was still seen lower than for PCL without a surfactant. The reason for this could be explained by the different chemistry, but also by the fact that PHB is known to be a more rigid polymer that crystallizes to a greater extent into thick spherulites than PCL, which has more available free volume in the amorphous phase for dispersion.

Figure 10 shows the TEM analysis of the fibers and the corresponding film, which are in agreemen<sup>t</sup> with the FIB-SEM results, in which some Pd agglomerations could be observed but also the presence of some Pd nanoparticles dispersed and distributed across the material matrix. These results are also in agreemen<sup>t</sup> with previous findings for PHB [20].

**Figure 8.** The cross-sectional SEM image of two PHB fibers with Pd nanoparticles after FIB sectioning.

**Figure 9.** (**a**) The side view of the 3D reconstruction of the PHB fiber (in semi-transparent yellow) and Pd nanoparticles (in white), including the SEM image (obtained from FIB sectioning) in the middle of the reconstruction, (**b**) 3D reconstruction of the PHB fiber with Pd nanoparticles inside (**c**) 3D reconstruction of particles only from the inside of PHB fiber 2. The binding box size for this reconstruction had the following dimensions: 1.996 × 2.464 × 0.612 nm.

**Figure 10.** TEM images of (**a**) the PHB/PdNP fibers and (**b**) the PHB/PdNP film.

#### *3.5. Morphology of the Multilayers*

Figure 11 shows the SEM images of the multilayer structures obtained, which contain a paper substrate, and electrospun PHB/PCL fibers containing PdNP and their annealed films. In all cases, the amount of PCL deposited (1 hr coating time) was lower than the amount of PHB (2 hr coating time) because PHB is a better water barrier than PCL. In a previous paper, PHB containing Pd nanoparticles showed a significant decrease in oxygen absorption after annealing; hence, in this work we intended to explore the feasibility of using some PCL in the coating, in the hope of enhancing the oxygen scavenging effect of Pd. As a result, the PCL layer has a lower thickness, and it was set as the top layer. After annealing, there is an additional packing of the fibers, resulting in an even lower layer thickness, as can be seen in Figure 11.

The thickness of the paper layer was 117 μm and, of course, the fiber mean cross section of the paper was found to be higher, 18.37 ± 2.45 μm, compared to the electrospun fibers. Figure 11b shows that the surface of the electrospun neat PCL fibers exhibit significant fiber interconnections suggesting either remnant solvent induced coalescence and/or coarser fibers due to a drop in electric field as a result of deposition over other insulating materials as compared to direct deposition over the metallic collector (compare with Figure 1a).

The multilayers presented in Figure 11c–f indicate the coalescence of fibers during the annealing step as expected, leading to a much less porous continuous film. Even though the samples had similar morphologies, it seems that sample (c) showed somewhat greater porosity, and sample (**f**) showed the least porosity. This was surely the effect of annealing in both coatings, leading to a higher level of packing structure.

In agreemen<sup>t</sup> with previous studies carried out on coatings with electrospun fibers over paper or polymer substrates, it is seen that the adhesion achieved after annealing between layers was very strong, due to the high surface-to-volume ratio of the electrospun fibers [22,48].

**Figure 11.** SEM images of the top view, and the cross section of (a) paper; (**b**) paper/PHB fibers/PCL fibers; (**c**) paper/PHB fibers/PCL-PdNP film; (**d**) Paper/PHB-PdNP fibers/PCL film; (**e**) Paper/PHB-PdNP fiber/PCL-PdNP film; (**f**) paper/PHB film/PCL-PdNP film.

#### *3.6. Passive and Active Barrier Properties of the Multilayers*

#### 3.6.1. Water Vapor Passive Permeance

In general, the barrier properties of materials depend on the solubility and diffusion of the permeants, and hence, they depend on the permeant size, shape, and polarity; but also on the crystallinity, degree of cross-linking, and polymer chain segmental motion of the polymer matrix, among other factors [49,50].

The diffusion coefficient of water in an amorphous or semi-crystalline polymer is related to the particular molecular dynamics or segmental motions within the amorphous regions of the polymer. In addition, in semi-crystalline materials, a low crystallinity index and the formation of crystals of inferior quality confer a high degree of mobility to the macromolecular chains, resulting in lower barrier performance [51]. It is known that PHA is a better barrier material than PCL [52]. However, for fiber-based materials such as the paper/PHB fibers/PCL fibers multilayer generated here, the barrier performance was expected to be as low as paper, due to the existing porosity between the adjacent fibers.

The water permeance data of the multilayer samples are gathered in Figure 12. From this figure, it can be clearly seen that the samples that contained fibers and that had porosity at the surface did not significantly enhance the barrier of the multilayers compared to neat paper. However, it can also be seen that when the two electrospun coatings underwent annealing, this led to a porosity reduction and stronger adhesion, and the water barrier performance was significantly increased.

**Figure 12.** Water vapor permeance (WVP) of paper and Paper/PHB/PCL multilayers with and without palladium nanoparticles (PdNPs). Different letters indicate significant differences among samples (*p* < 0.05).

The higher water barrier of the sample paper/PHB film/PCLPdNP film is then explained by a reduction in sample porosity, and also by the expected improvement in the crystalline morphology that occurs in the sample after annealing, which is known be both impermeable to the diffusion of sorbed water molecules, and impose restraints to the mobility of the amorphous phase [53]. The barrier data gathered for this sample (4.3 × 10–11 kg/m<sup>2</sup>·s<sup>−</sup>1·Pa−1) is in the same order of magnitude as the results obtained in a previous work for PHB (9.6 × 10–11 kg/m<sup>2</sup>·s<sup>−</sup>1·Pa−1) [20].

#### 3.6.2. Active Oxygen Scavenging Performance

Figure 13 shows the oxygen scavenging rate (OSR) of the PCL and PHB fibers, and the prepared paper based multilayers. The oxygen absorption of the fibers and resulting multilayers was investigated at 23 ◦C, with an initial oxygen concentration of 1.0% in the headspace of the measuring flasks, and at an RH of 50%. As reported earlier, due to the higher barrier of the material against water and oxygen compared to PCL, the PHB-PdNP fibers are not very efficient as oxygen scavengers at low or intermediate relative humidity, and even at high relative humidity as a film. However, PCL fibers and films are extremely quick at removing oxygen from the head space, suggesting that PCL is more adequate for hosting PdNP for oxygen scavenging purposes. The higher fractional free volume of the PCL polymer allows for moisture and oxygen to reach the catalyst quickly, and hence oxygen removal is more efficient. The greater reduction in the OSR of the fibers is, of course, related to the high surface-to-volume ratio of the electrospun fibers as compared to the annealed films. The dissociation rate of the hydrogen molecules into hydrogen atoms over the Pd surface in the films depends on the available surface area that is presented by the PdNP within the film.

As mentioned above, another important factor to be considered is that oxygen depletion is dependent on the RH conditions applied, and oxygen scavenging decreases when the RH increases [15,54]. In this work, we intended to conduct testing at medium relative humidities to simulate an intermedium case study. Figure 13 clearly indicates that, as expected, neither the paper nor the neat PCL polymer had any oxygen scavenging capacity.

Figure 13 also indicates that the best performing multilayer materials, in terms of OSR were, as expected, paper/PHB fibers/PCL-PdNP film and paper/PHB-PdNP fibers/PCL-PdNP film. In the cases where the intermediate layer is a PHB film or where the PdNP are not at the surface, the performance is reduced, since oxygen and moisture will possess slower kinetics of diffusion.

**Figure 13.** Oxygen depletion of PCL and PHB fibers, and paper/PHB/PCL multilayers and films with and without palladium nanoparticles (PdNPs). Values were measured at 50% relative humidity (RH).
