**Cutin from** *Solanum Myriacanthum* **Dunal and** *Solanum Aculeatissimum* **Jacq. as a Potential Raw Material for Biopolymers**

**Mayra Beatriz Gómez-Patiño 1, Rosa Estrada-Reyes 2, María Elena Vargas-Diaz <sup>3</sup> and Daniel Arrieta-Baez 1,\***


Received: 4 August 2020; Accepted: 26 August 2020; Published: 28 August 2020

**Abstract:** Plant cuticles have attracted attention because they can be used to produce hydrophobic films as models for novel biopolymers. Usually, cuticles are obtained from agroresidual waste. To find new renewable natural sources to design green and commercially available bioplastics, fruits of *S. aculeatissimum* and *S. myriacanthum* were analyzed. These fruits are not used for human or animal consumption, mainly because the fruit is composed of seeds. Fruit peels were object of enzymatic and chemical methods to get thick cutins in good yields (approximately 77% from dry weight), and they were studied by solid-state resonance techniques (CPMAS 13C NMR), attenuated total reflection-Fourier transform infrared spectroscopy (ATR-FTIR), atomic force microscopy (AFM) and direct injection electrospray ionization mass spectrometry (DIESI-MS) analytical methods. The main component of *S. aculeatissimum* cutin is 10,16-dihydroxypalmitic acid (10,16-DHPA, 69.84%), while *S. myriacanthum* cutin besides of 10,16-DHPA (44.02%); another two C18 monomers: 9,10,18-trihydroxy-octadecanoic acid (24.03%) and 18-hydroxy-9S,10R-epoxy-octadecanoic acid (9.36%) are present. The hydrolyzed cutins were used to produce films demonstrating that both cutins could be a potential raw material for different biopolymers.

**Keywords:** cutin; cuticles; bioplastics; biopolymers; solanum: CPMAS 13C NMR

### **1. Introduction**

The cuticle is the outer membrane that covers the aerial parts of plants, such as the stem, leaves, flowers, and fruit. In the evolution of plants, the cuticle plays a critical role against the loss of water from internal tissues [1–3]. In the same way, this biopolymer plays an essential physiological role as it is considered a first barrier that prevents the entry of pathogens and pesticides [4,5]. The cuticle consists primarily of cutin (a C16 and C18 long-chain hydroxy acid polyester), cell wall polysaccharides (cellulose, hemicellulose, and pectin), as well as epicuticular fatty acids [6–8]. Cutin, and other important natural biopolymers such as lignin, cellulose, and chitin, has been shown to have important bioplastic properties [9–13]. For this reason, they have been considered as models for plastic materials with biodegradable characteristics that eventually could replace conventional plastics derived from petroleum in specific industrial uses [14].

Raw renewable materials have a high impact on the cost of bio-based plastic production and, in this regard, different efforts have been directed to use biomass to get or produce biopolymers. Biopolymers such as starch, cellulose, lignocellulosic materials, and proteins; bio-derived monomers like polylactic acid (PLA), polyglycolic acid (PGA), and biodegradable polymers from petrochemicals (aliphatic polyesters, aromatic co-polyesters and polyvinyl alcohols) have been investigated as sources for bioplastics [15,16].

In this sense, the cutins of some fruits for human consumption, such as tomatoes, citrus have shown good physicochemical characteristics as biopolymers [9,17–20]. However, the ethical problem that could be generated has led researchers to use industrial waste products. In fact, the vegetable food processing industry generates a significant amount of waste worldwide [21]. Thus, from agro-industrial residues, biomaterials have been generated, and some of them have been used in the food packaging industry, and other bioplastic applications [22,23].

In the present work, we have searched for fruits that are not for human consumption, which present the same chemical characteristics of the cuticles used for biopolymers applications in order to be considered promising candidates as a raw material to produce bioplastics. In this sense, the cuticles of the fruits of two species of the *Solaneum* genus were studied. *S. aculeatissimum* and *S. myriacanthum* are shrubs that grow in the wild, covered with thin spines up to 18 mm long, that produce small fruits of approximately 2–3 cm, which are mainly filled with seeds. *S. aculeatissimum* is native to Brazil, but it could be found in tropical Africa and Asia. In Mexico, it is distributed in the states of Jalisco, Oaxaca, Chiapas, Veracruz, and Puebla [24]. It is considered a toxic plant due to the alkaloids present in the seeds and leaves [25,26]. Its fruit is green when it is immature and red when it is ripe. *S. myriacanthum* is native to central and south America, although it is also distributed in Asia, mainly in India. In Mexico, it is distributed in the states of Chiapas, Veracruz, Oaxaca, and Puebla [24]. Anthelmintic properties are attributed to the extracts of the fruit [27]. Its fruit is green when immature and yellow when ripe.

*S. aculeatissimum* and *S. myriacanthum* cutins were extracted and analyzed by means of CPMAS 13C NMR, ATR-FTIR, AFM and DIESI-MS, and their components and physicochemical characteristics were determined and compared with other cutin components. From their hydrolyzed components, films were obtained and characterized and, from these results, both Solanum species could be considered as a raw material for biopolymers used in different fields of the plastic industry.

### **2. Materials and Methods**

### *2.1. Chemicals*

Trifluoroacetic acid (TFA), KOH, and other reagents were purchased from Sigma-Aldrich (St. Louis, MO, USA). The enzymes *Aspergillus niger* pectinase (EC 3.2.1.15) (specific activity ≥ 5 unit/mg protein), *A. niger* cellulase (EC 3.2.1.4) (specific activity ≥ 0.3 units/mg protein) and *A. niger* hemicellulose (EC 3.2.1.4) (specific activity 2.3 units/mg protein) were purchased from Sigma Chemicals (St Louis, MO, USA).

### *2.2. Isolation of Cutin*

*S. aculeatissimum* and *S. myriacanthum* fruits were collected in Cuetzalan, Puebla (20◦01 48.7" N 97◦29 04.3" W) in September 2018. Fruits were washed with tap water, cut, and the seeds were removed to obtain the cuticle. The cutin was obtained using a previously reported protocol [9]. Briefly, the cuticle was treated with *A. niger* pectinase (EC 3.2.1.15, St Louis, MO, USA) (10 mg/mL) for 1 week. After this, cell wall polysaccharides were removed with an enzymatic digestion using *A. niger* cellulose (EC 3.2.1.4, St Louis, MO, USA) (80 mg/mL) for 1 week and *A. niger* hemicellulose (EC 3.2.1.4, St Louis, MO, USA) (80 mg/mL) for 1 week. To complete the extraction, a Soxhlet procedure was done with methylene chloride:methanol (1:1 *v*/*v*, 48 h, St Louis, MO, USA) to remove residual compounds of cutin such as monosaccharides and waxes. Five hundred grams of *S. aculeatissimum* dried peel yielded 386 g (77.2%) of cutin and 500 g of *S. myriacanthum* dried peel yielded 394 g (78.8%) of cutin. The resulting

cutins were analyzed by Cross Polarization Magic-Angle Spinning (CPMAS 13C NMR), attenuated total reflection-Fourier transform infrared spectroscopy (ATR-FTIR) and atomic force microscopy (AFM).

### *2.3. Treatment of Cutin with Trifluoroacetic Acid (TFA)*

One hundred and fifty milligrams of the obtained cutin from the *S. aculeatissimum* or *S. myriacanthum* fruits was added to an aqueous TFA solution 2.0 mol·L−<sup>1</sup> and stirred at 115 <sup>±</sup> 5 ◦C for 2 h in separated experiments. Each reaction was filtered, and the insoluble material was washed using chloroform-methanol (1:1, *v*/*v*, St Louis, MO, USA) for 1 h to obtain the TFA-hydrolyzed cutin (TFA-HC). The TFA-HC was separated by filtration, dried, and analyzed by CPMAS 13C NMR. The TFA solution was co-evaporated with methanol and the resulting solids were redissolved in methanol to give a clear brown solution, which was later analyzed by DIESI-MS (Bruker Daltonics, Biellerica, MA, USA) and solution-state NMR (Billerica, MA, USA) [28].

### *2.4. Alkaline Hydrolysis of the Cutin with KOH*/*MeOH*

Fifty milligrams of *S. aculeatissimum* or *S. myriacanthum* cutin was added to 50 mL of 1.5 mol·L−<sup>1</sup> methanolic KOH solution, and the mixture was stirred at room temperature for 24 h. After this time, the reaction was filtered, neutralized, and monomers were extracted with CHCl3-MeOH. The dried extract was weighed, dissolved in CHCl3-MeOH and analyzed by DIESI-MS (Bruker Daltonics, Biellerica, MA, USA).

### *2.5. Preparation of Cutin Films*

Twenty-five milligrams of hydrolyzed *S. aculeatissimum* or *S. myriacanthum* cutin were added to 5 mL of ultrapure methanol:chloroform (1:1, *v*/*v*, St Louis, MO, USA) solution. The solution was sonicated for 30 s and it was deposited in plastic Petri dishes for making films using the casting method. On the other hand, 5 μL of the solution were deposited on a watch glass to study the structures of self-assembled layers. After this, films were kept in a chemical hood to remove residual solvents from the films [29].

### *2.6. NMR Spectroscopy*

*S. aculeatissimum* and *S. myriacanthum* cutin and TFA-HC were analyzed using standard CPMAS 13C NMR experiments carried out on a Varian Instruments Unityplus 300 widebore spectrometer (Palo Alto, CA, USA) equipped for solid-state NMR. The resonance frequency was 74.443 MHz, with a customary acquisition time of 30 ms, a delay time of 2 s between successive acquisitions and a CP contact time of 1.5 ms. Typically, each 30 mg sample was packed into a 5 mm rotor and supersonic MAS probe from Doty Scientific (Columbia, SC, USA), then spun at 6.00 (±0.1 kHz) at room temperature for approximately 10 h. No spinning sidebands were observed upon downfield from the major carbonyl, aromatic, or aliphatic carbon peaks, presumably due to motional averaging and/or excessive broadening of such features.

Soluble products derived from the TFA hydrolysis were examined using 1H NMR. Experiments were conducted on a Bruker Instruments ASCEND 750 spectrometer (Billerica, MA, USA). The resonance frequency was 750.12 MHz, with a typical acquisition time of 2.1845 s and a delay time of 1.0 s between successive acquisitions. The 1H and 13C chemical shifts are given in units of δ (ppm), using tetramethylsilane (TMS) as internal standard.

### *2.7. ATR-FTIR Spectroscopy*

Attenuated Total Reflectance Fourier transform infrared spectroscopy (ATR-FTIR) spectra were recorded with a BOMEM 157 FTIR spectrometer (Bomem Inc., Quebec, Canada) equipped with a deuterated triglycinesulfate (DTGS) detector. The instrument was under a continuous dry air purge to eliminate atmospheric water vapor. The spectra were recorded in the region of 4000 to 400 cm<sup>−</sup>1.

### *2.8. Atomic Force Microscopy*

The samples for the Atomic Force Microscopy analysis (AFM) were prepared by fixing to a metallic disk with double-sided tape. The images themselves were taken using a MultiMode AFMV (Bruker, SantaBarbara, CA, USA) in air with an RTESP cantilever, and operating the AFM in tapping mode. The size of each image was 5 <sup>×</sup> 5 <sup>μ</sup>m2. The roughness parameters *R*<sup>q</sup> and *R*<sup>a</sup> were determined using the expressions *<sup>R</sup>*<sup>q</sup> <sup>=</sup> <sup>√</sup>ΣZ21 <sup>=</sup> N and *<sup>R</sup>*<sup>a</sup> <sup>=</sup> <sup>1</sup> <sup>=</sup> <sup>N</sup>ΣNj <sup>=</sup> 1 jZjj, where *<sup>R</sup>*<sup>q</sup> is the root mean square average of the height deviations, Ra is the arithmetic average of the absolute values of the surface height deviations, Z is the height value, and N is the number of data points. These parameters were obtain using the NanoScope Analysis image software.

### *2.9. Mass Spectrometry*

Direct Ionization analysis (DIESI-MS) was done on a Bruker MicrOTOF-QII system, using an electrospray ionization (ESI) interface (Bruker Daltonics, Biellerica, MA, USA) operated in the negative ion mode. A solution of 10 μL of the sample resuspended in 1 mL of methanol was filtered with a 0.25 μm polytetrafluoroethylene (PTFE) filter and diluted 1:100 with methanol. Diluted samples were directly infused into the ESI source and analyzed in negative mode. Nitrogen was used with a flow rate of 4 L/min (0.4 Bar) as a drying and nebulizer gas, with a gas temperature of 180 ◦C and a capillary voltage set to 4500 V. The spectrometer was calibrated with an ESI-TOF tuning mix calibrant (Sigma-Aldrich, Toluca, Estado de México, México).

MS/MS analysis was performed using negative electrospray ionization (ESI−), and the obtained fragments were analyzed by a Bruker Compass Data Analysis 4.0 (Bruker Daltonics, Technical Note 008, 2004, Bruker Daltonics, Biellerica, MA, USA). An accuracy threshold of 5 ppm was established to confirm the elemental compositions.

### **3. Results**

*S. aculeatissimum* and *S. myriacanthum* fruits were collected in the Cuetzalan, Puebla (México) region (Figure 1). Usually, the mature fruit is a globose berry 2–3 cm in diameter, and it is composed of the peel (≈0.2 mm thick) that represent a 25–30% percent of the fruit, 10% of a polysaccharides layer, and 50–60% of seeds. Once the peels were washed and dried, cutins were obtained by previously published methods. Cutin obtained from the dry peels were in very high yield (≈77%) in relation to other fruit cutins, such as tomato or citrus fruits (≈0.5%) [9,30]. Most of the compounds hydrolyzed with the enzymatic treatment were identified as monosaccharides (Glu and Fru, data not shown), and cutins were characterized by solid-state resonance techniques (CPMAS 13C NMR), ATR-FTIR, and AFM.

**Figure 1.** Photographs of the plants and fruits of (**A**) *S. aculeatissimum* Jacq, and (**B**) *S. myriacanthum* Dunal.

### *3.1. CPMAS 13C NMR Analysis*

Cutins obtained from *S. aculeatissimum* and *S. myriacanthum* were analyzed by CPMAS 13C NMR, and their spectra are shown in Figure 2. According to our previous studies in fruit cuticles, typical resonances of aliphatic-aromatic polyesters are exhibited: bulk methylenes (20–35 ppm), oxygenated aliphatic carbons (55–85 ppm), aromatics and olefins (105–155 ppm), and carbonyl groups (172 ppm) signals.

**Figure 2.** CPMAS 13C NMR spectra of cutins from *S. aculeatissimum* and *S. myriacanthum*.

Some of these signals belong to the carbohydrate moieties (C6 at 60 ppm, C2,3,5 at 70–75 ppm, C4 at 83 ppm, and C1 at 101–105 ppm), some of these peaks could overlap with oxygenated aliphatic signals. However, every cutin showed unique NMR characteristics: more aromatic peaks are evident in *S. aculeatissimum* (Figure 2, upper spectrum), while in *S. myriacanthum* (Figure 2, lower spectrum) peaks at 52 and 56 ppm are present.

To garner more information about these materials, cutins from both fruits were the object of a TFA hydrolysis. It has been demonstrated that TFA hydrolysis could be used to remove non-cellulosic polysaccharides with the advantage that TFA is easy to remove by evaporation rather than a loss-prone neutralization step [28].

*S. aculeatissimum* cutin was found to be resistant to TFA hydrolysis. There was a minimal weight loss, and as seen in Figure 3, and most of the peaks remain as in the spectra without TFA treatment. However, *S. myriacanthum* shows ≈ 8% of weight loss, and this can be attributed to the disappearance of compounds with peaks at 50 ppm. The soluble part obtained from the TFA hydrolysis was studied, and it was found that these peaks belong to an epoxidated C18 long-chain aliphatic acid (see Supplementary Material).

**Figure 3.** CPMAS 13C NMR spectra of cutins from *S. aculeatissimum* and *S. myriacanthum* after trifluoracetic acid (TFA) hydrolysis (TFA-HC).

### *3.2. Infrared Spectroscopy Analysis of the S. Aculeatissimum and S. Myriacanthum Cutins*

Isolated cutins have been characterized in situ at their functional chemical groups as well as their interactions at the cuticular levels with exogenous chemicals [8,31]. The ATR FT-IR analysis of *S. aculeatissimum* and *S. myriacanthum* cutins (Figure 4) were characterized as follows: hydroxyl groups of the polysaccharide domain and residual carboxylic acids showed its absorption maxima as broadband at 3860 cm−1, characteristic intense bands corresponding to the asymmetrical and symmetrical stretching vibrations of the methylene CH2 region at 2905 and 2850 cm<sup>−</sup>1, with the bending vibrations at 1462 and 1350 cm−1, which came from the aliphatic components present in the cutin. Another group of signals associated with the cutin matrix is that from 1600 to 1750 cm−<sup>1</sup> attributed to the carbonyl C=O stretching band in ester groups, and their asymmetric stretching vibrations of C–CO–O at 1100 cm−1. These assignations agree with those reported and used in the study of non-isolated plant cutins [8].

Figure 4 shows that IR spectra for both cutins are very similar, and the most intense bands correspond to the main domains of this polyester: aliphatic and polysaccharides groups. However, two groups of signals are making the difference between them. For *S. aculeatissimum* cutin, a group of bands at 1500 to 1650 cm−<sup>1</sup> related to aromatic and C=C functional groups are less intense in *S. myriacanthum* cutin, due to the low presence of aromatics. On the other hand, the group of signals at 1100 cm−<sup>1</sup> is broader and more intense in *S. myriacanthum*, possibly because of a poliesterification with at least two different long-chain acids. These observations agreed with the CPMAS 13C NMR analysis.

**Figure 4.** Attenuated Total Reflectance Fourier transform infrared spectroscopy (ATR-FTIR) spectra of *S. aculeatissimum* and *S. myriacanthum*.

#### *3.3. Atomic Force Microscopy Analysis*

The cuticles obtained from the fruits of *S. aculeatissimum* and *S. myriacanthum* were analyzed through AFM. Figure 5 shows that these cutins are thicker than other fruit cutins, such as tomatoes, lemon, orange. The AFM amplitude error images showed that cutin surfaces are composed mainly of fibers that give the characteristic roughness (Figure 5C,D). The fibers are more homogeneous in the *S. aculeatissimum* cutin with an average thickness of 34 nm, while in the *S. myriacanthum* cutin they are irregularly present, with fibers ranged from 125 to 23 nm that were observed. The roughness study showed that *S. aculeatissimum* has a *R*q of 1.8 nm and a *R*a of 1.3 nm, while the cutin of *S. myriacanthum* showed a lower roughness with a *R*q of 3.4 nm and a *R*a of 2.6 nm.

**Figure 5.** Atomic force microscopy tapping mode topographical images from (**A**) *S. aculeatissimum* (**B**,**C**): 5.0 and 1.0 μm, respectively) and (**D**) *S. myriacanthum* (**E**,**F**): 5.0 and 1.0 μm, respectively).

### *3.4. Alkaline Hydrolysis (KOH*/*MeOH)*

To study the main aliphatic components present in the *S. aculeatissimum* and *S. myriacanthum* cutins, a complementary analysis was done with alkaline hydrolysis. In both cases, around ≈93% of the cuticular material was hydrolyzed. Soluble products from the alkaline hydrolysis were analyzed by means of direct-injection electrospray ionization mass spectrometry in negative mode (DIESI-MS, see Supplementary Material) and the compounds identified by the *ms*/*ms* analysis, are reported in Table 1.



[M − H]−exact: Molecular Weight exact, [M − H]−obs: Molecular Weight observed, % RA: % Relative Area. Error [ppm]: Absolute value of the deviation between measured mass and theoretical mass of the selected peak in [ppm].

Even when most of the compounds are present in both cutins, some differences can be observed. The main constituent identified in *S. aculeatissimum* cutin was 10,16-dihydroxyhexadecanoic acid (10,16-DHPA), an important monomer present in different cutins such as tomato, citrus cuticles and green pepper [32], in a 69.84% of the relative abundance. Aromatic and some derivatives compounds were detected in agreement with the CPMAS 13C NMR data. 10,16-DHPA was found in *S. myriacanthum* cutin. However, two other significant monomers are present: 9,10,18-trihydroxy-octadecanoic acid and 18-hydroxy-9S,10R-epoxy-octadecanoic acid in 24.03 and 9.36%, respectively. According to the TFA-hydrolysis analysis, the epoxilated long-chain aliphatic acid was hydrolyzed and obtained almost pure, according to the NMR analysis (see Supplementary Material). This observation could suggest that it is present in a different domain from the other components. The predominance of C16 long-chain acids in cutins is very common and corroborates previous cutin reports. However, it is important to highlight that most of the 25% of the main monomers in *S. myriacanthum* cutin are C18 acids. The presence of these C16 and C18 monomers could be the reason for the broadband esterification detected at 100 cm−<sup>1</sup> in the ATR-FTIR spectrum. Aromatic compounds are not present as in *S. aculeatissimum* cutin, which agrees with the NMR analysis.

### *3.5. Analysis of the Films Prepared from Hydrolyzed Cutins*

To demonstrate that *S. aculeatissimum* and *S. myriacanthum* cutins could be a good material for biopolymer, films were prepared by simple blending in solvents. Representative photographs of the films prepared from the hydrolyzed cutins are shown in Figure 6. Both samples have a waxy consistency, but their surface was quite homogeneous. Films were characterized through ATR-FTIR and AFM.

**Figure 6.** Photographs of the films prepared from hydrolyzed cutins.

### *3.6. ATR-FTIR Analysis of the Films*

Analysis using Fourier transform infrared (FTIR) spectroscopy indicated that films from hydrolyzed cutins keep the spectral features of the original cutins. However, most of the signals demonstrate that bands associated with ester groups disappeared, especially the absorption at 1630 cm−<sup>1</sup> that belongs to the stretching of C=O of ester groups. However, the presence of the band at 1127 cm<sup>−</sup>1, ascribed to the asymmetric stretching vibrations of C–CO–O, demonstrates that part of this polyester network remains, or monomers were partially polymerized (Figure 7).

**Figure 7.** ATR-FTIR spectra of the films from hydrolyzed cutins of *S. aculeatissimum* and *S. myriacanthum.*

### *3.7. AFM Analysis of the Films*

Atomic Force Microscopy (AFM) analysis shows a different topography from that observed in the original cutins. There is no occurrence of fibers that could be attributed to the cellulose or pectin presence [33]. According to DIESI-MS analysis, there is not a presence of sugars or some oligoor polysaccharides in the soluble hydrolyzed cutins. The roughness study showed that film from *S. aculeatissimum* cutin has a value of a *R*<sup>q</sup> 0.527 nm and a *R*<sup>a</sup> 0.406 nm, while that obtained from *S. myriacanthum* cutin showed values of *R*q 0.973 nm and Ra 0.584 nm (Figure 8).

**Figure 8.** Atomic force microscopy tapping mode topographical images from (**A**) *S. aculeatissimum* (**B**,**C**): 5.0 and 1.0 μm, respectively) and (**D**) *S. myriacanthum* (**E**,**F**): 5.0 and 1.0 μm, respectively).

The homogeneity could be attributed to a good organization and a high degree of order of the monomers, oligomers or polymers present in the hydrolyzed cutin. This characteristic is highly important to get films with small porous or cavities distributed along the surface.

#### **4. Conclusions**

In this work, we have demonstrated that *S. aculeatissimum* and *S. myriacanthum* cuticles have a good percentage of cutin—around 70%, from the dry weight, more than other studied fruits such as tomato or citrus fruits (≈0.5%). These cutins have the same monomers composition reported in other fruits cuticles such as tomato, citric fruits, or pepper, were the main component was 10,16-DHPA. Films obtained from the hydrolyzed cutins showed a good homogeneity. Furthermore, the fact that these fruits are not for human or animal consumption makes it feasible for them to be considered as a potential raw material to produce sustainable composite materials as an alternative to traditional plastics.

### **Supplementary Materials:** The Supplementary Materials are available online at http://www.mdpi.com/2073-4360/ 12/9/1945/s1.

**Author Contributions:** M.B.G.-P. and D.A.-B. conceived and designed the main ideas of this paper, carried out the NMR and DIESI-MS experiments, analyzed the experimental results, and wrote the paper. R.E.-R., and M.E.V.-D. carried out the cuticle and compounds extraction experiments and help to discuss the results. The authors read and approved the final manuscript. Investigation, D.A.B, R.E.-R., M.E.V.-D. and M.B.G.-P.; Project administration, D.A.B and M.B.G.-P.; Supervision, D.A.B and M.B.G.-P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Consejo Nacional de Ciencia y Tecnologia (CONACyT) for funding Project No. 253570 and SIP-IPN grants No. 20201066 and 20201968.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## **Controlled Release, Disintegration, Antioxidant, and Antimicrobial Properties of Poly (Lactic Acid)** /**Thymol**/**Nanoclay Composites**

**Marina Ramos 1,\*, Elena Fortunati 2, Ana Beltrán 1, Mercedes Peltzer 3,4, Francesco Cristofaro 5, Livia Visai 5,6, Artur J.M. Valente 7, Alfonso Jiménez 1, José María Kenny <sup>2</sup> and María Carmen Garrigós <sup>1</sup>**


Received: 13 July 2020; Accepted: 17 August 2020; Published: 20 August 2020

**Abstract:** Nano-biocomposite films based on poly (lactic acid) (PLA) were prepared by adding thymol (8 wt.%) and a commercial montmorillonite (D43B) at different concentrations (2.5 and 5 wt.%). The antioxidant, antimicrobial, and disintegration properties of all films were determined. A kinetic study was carried out to evaluate the thymol release from the polymer matrix into ethanol 10% (*v*/*v*) as food simulant. The nanostructured networks formed in binary and ternary systems were of interest in controlling the release of thymol into the food simulant. The results indicated that the diffusion of thymol through the PLA matrix was influenced by the presence of the nanoclay. Disintegration tests demonstrated that the incorporation of both additives promoted the breakdown of the polymer matrix due to the presence of the reactive hydroxyl group in the thymol structure and ammonium groups in D43B. Active films containing thymol and D43B efficiently enhanced the antioxidant activity (inhibition values higher than 77%) of the nano-biocomposites. Finally, the addition of 8 wt.% thymol and 2.5 wt.% D43B significantly increased the antibacterial activity against *Escherichia coli* and *Staphylococcus aureus 8325-4,* resulting in a clear advantage to improve the shelf-life of perishable packaged food.

**Keywords:** PLA; nanocomposites; functional properties; thymol; migration; films

### **1. Introduction**

The use of biopolymers in packaging applications is considered a suitable alternative to the petrochemical counterparts contributing to the limitation of the environmental problems caused by the accumulation of plastic waste [1]. Among the different bio-based polymers that can be used for such purpose, poly (lactic acid) or PLA is a biodegradable polyester obtained from 100% renewable

resources. It is also a highly versatile material with many commercial applications in the textile, medical, automotive and packaging sectors. PLA shows some desirable properties such as its inherent biodegradability, biocompatibility, stiffness, high strength, thermo-plasticity, and high transparency [2]. However, PLA-based films show some significant drawbacks which limit their performance in processing and the final use, such as poor barrier properties to gases, low thermal and mechanical resistance, slow crystallization and brittleness [3].

These shortcomings can be overcome by the development of nano-biocomposites based on PLA reinforced with nanoparticles, which has been introduced as a promising option [4]. Souza et al. [5] assumed that the incorporation of nanoclays into PLA-based formulations is a promising alternative to improve their barrier and mechanical properties without altering the transparency and compostability of the final material. Among the different nanoparticles, an interesting approach to improve PLA properties is the reinforcement with nanoclays such as montmorillonites, resulting in the development of PLA nanocomposites [6]. The use of this type of nanomaterials in food packaging applications can also modify the internal atmosphere due to the modification of barrier properties, thereby delaying ripening and extending the shelf-life in packaged foods [7].

The development of new nanocomposites with unique characteristics is desirable, especially for the food industry due to their full range of applications [8]. The satisfaction of consumers' requirements for healthier and highly nutritional foods, as well as the increase in the need for long shelf-life for fresh food, have resulted in a high interest in the use of functional packaging systems with a controlled release of active substances embedded in eco-friendly plastic materials [9]. Active packaging systems incorporate agents with specific functionalities into the polymer matrix [10,11], reacting with or releasing specific components to increase food quality, taking advantage of these interactions [2]. Low density polyethylene (LDPE) was combined with thymol, eugenol, and carvacrol embedded in montmorillonite or halloysite to produce active nanocomposite films [12]. These formulations improved the main properties of LDPE as well as food quality by the increase in antimicrobial and antioxidant performance, and enhanced barrier properties. Montmorillonite was also used in alginate-based films with lemon essential oil to obtain antibacterial and antifungal films [13]. Consequently, the combination of natural compounds extracted from plant species with antimicrobial and antioxidant properties with bio-based polymers to enhance their functional properties and to extend food shelf-life is a promising strategy to be applied in the food sector [14–20]. In particular, thymol which is the main phenolic compound present in thyme and oregano essential oils has been used due to its antimicrobial and antioxidant properties that make it useful to be incorporated as an active compound in packaging formulations [21–24]. To the best of our knowledge, the combination of thymol and montmorillonite in PLA-based films to obtain active nanocomposites with both antimicrobial and antioxidant properties has not been extensively studied. This combination can be considered an interesting approach in food science to reduce the oxidative and microbial deterioration of food products to increase food quality, safety and shelf-life while maintaining the material properties. Several works on PLA nanocomposites have been already published [25–29], but the present work could contribute to fill a gap in the development of new sustainable nanocomposite films with combined antioxidant and antibacterial performance by controlling the release of the active agent. In addition, it is important to validate the potential use of these materials in active packaging applications.

In a previous work, these novel nano-biocomposite films based on PLA with D43B and thymol for active packaging were successfully characterized in their physicochemical properties [30]. However, the evaluation of the controlled release of the active agent into a specific food-grade simulant (ethanol 10% *v*/*v*), disintegration, antioxidant and antimicrobial properties of the obtained nano-biocomposite films has not been published yet. Therefore, the aim of this study is the evaluation of the functional and compostability properties of these films for their potential use as active food packaging materials, demonstrating also the effective release of thymol from the active nano-biocomposite PLA films.

### **2. Materials and Methods**

### *2.1. Materials*

Commercial PLA-4060D (Tg = 58 ◦C, 11–13 wt.% D-isomer) was supplied in pellets by Natureworks Co., (Minnetonka, MN, USA). Thymol (99.5%), 2,2-Diphenyl-1-picrylhydrazyl (DPPH, 95%), methanol and ethanol (High-performance liquid chromatography, HPLC grade) were supplied by Sigma-Aldrich (Madrid, Spain). The commercial nanoclay used was Dellite®43B (D43B) (Laviosa Chimica Mineraria S.p.A. Livorno, Italy), a dimethyl-benzyldihydrogenated tallow ammonium modified montmorillonite with a cation exchange capacity (CEC) of 95 meq/100 g clay, a bulk density of 0.40 g cm−<sup>3</sup> and a particle size distribution between 7–9 μm.

### *2.2. Nano-Biocomposite Films Preparation*

PLA-based nano-biocomposite films incorporated with D43B (as nanofiller) and thymol (as active agent) were developed by melt blending in a Haake Polylab QC mixer (ThermoFischer Scientific, Walham, MA, USA) at 160 ◦C as already described in a previous work [30]. Five formulations were obtained by combining thymol (8 wt.%) and D43B at two different loadings (2.5 and 5 wt.%) in PLA matrices. Neat PLA was used as control. Homogenous and transparent films were obtained by compression-molding as described in [29]. The final mean film thickness was 190 ± 5 μm measured with a Digimatic Micrometer Series 293 MDC-Lite (Mitutoyo, Japan) at five random positions.

In our previous work [29], it was concluded that the addition of thymol did not significantly affect the thermal stability of PLA, but some decrease (around 15%) in elastic modulus was observed due to the slight plasticizing effect induced by the active additive. The incorporation of D43B and thymol to PLA did not result in an apparent enhancement in oxygen barrier properties, but the tensile behavior was improved due to the intercalation and partial exfoliation of nanoparticles through the polymer matrix, as observed by X-Ray diffraction (XRD). Moreover, the intrinsic transparency of PLA was not affected by the addition of both components and most of the thymol initially added to PLA (around 70–75%) remained in the nanocomposites after processing, ensuring the potential applicability of these films as active systems.

### *2.3. Migration and Mathematical Di*ff*usion Analysis*

Migration tests for thymol released from the different nano-biocomposite films were performed, in triplicate, into ethanol 10% (*v*/*v*) as food simulant following the legislation for food contact materials EU/10/2011 [31] and the European Standard EN 13130-2005 [32]. Double-sided, total immersion migration tests were performed with films (12 cm2). 20 mL of food simulant were put in contact with the nano-biocomposite films (area-to-volume ratio around 6 dm<sup>2</sup> L<sup>−</sup>1) at 40 ◦C in an oven (J.P. Selecta, Barcelona, Spain) for 10 days. A blank sample (pure simulant) was also studied.

On the other hand, the release mechanism of thymol from PLA nano-biocomposite films was also studied over time by modelling the obtained results for 15 days in ethanol 10% (*v*/*v*). Different approaches have been applied to assess the migration of additives and contaminants from food packaging films: (i) a quantitative assessment of *MF,*<sup>∞</sup> to analyze the diffusion process by using Equation (1) to fit the experimental data,

$$\frac{M\_{F,t}}{M\_{P,\rho}} = \left(\frac{M\_{F,\infty}}{M\_{P,0}}\right) \cdot \left(1 - e^{-k't}\right) \tag{1}$$

where *M*P,0 is the initial amount of thymol inside the polymeric matrix, determined by using the HPLC-UV method optimized in our previous study [30]; *k'* is a constant; and *MF,t* is the mean of thymol released mass to the food simulant at a defined time; (ii) a Fick's approach (diffusion-controlled systems) to assess the migration of thymol from films by using mathematical modelling (Equation (2)). This differential equation provides a general description of the migration of an additive or contaminant from an amorphous polymer packaging film:

$$\frac{\partial \mathbf{C}}{\partial t} = \frac{\partial}{\partial \mathbf{x}} \Big( D \frac{\partial \mathbf{C}}{\partial \mathbf{x}} \Big) \tag{2}$$

where *D* is the diffusion coefficient; *c* is the concentration of the released species; *t* is time (s); and *x* is the space coordinate.

Apparent partition coefficients (α) can be calculated by Equation (3) from the values obtained for *MF,*∞/*MP,0* by fitting Equation (1),

$$\frac{M\_{F,\infty}}{M\_{P,0}} = \frac{1}{(1+\alpha)}\tag{3}$$

where α is defined as:

$$a = \frac{V\_F}{K\_{P\_rF} V\_P} \tag{4}$$

where *VP* and *VF* are the volumes of the polymer sample (*P*) and food simulant (*F*), respectively; and *KP,F* is the partition coefficient of thymol in the system between the samples and the solution, which can be assumed as constant at low concentrations and also for its relative solubility at the equilibrium between PLA and the food simulant [33].

Crank solved Equation (2) and formulated initial and boundary conditions, from which Equation (5) was obtained as a solution. This equation was applied for a plane sheet of thickness *l*, and the initial condition for testing −*l*/2 < *x* < *l*/2, considering constant the thymol's concentration released, a boundary condition of a partition coefficient between both phases and for one-dimensional diffusion of thymol in a limited volume solution [34],

$$\frac{M\_{\rm F,f}}{M\_{\rm F,\infty}} = 1 - \sum\_{n=1}^{\infty} \frac{2\alpha(1+\alpha)}{1+\alpha+\alpha^2 q\_n^2} \exp\left[\frac{-Dq\_n^2 t}{l^2}\right] \tag{5}$$

where *qn* is the non-zero positive roots of *tanqn* = −α *qn* and *l* is the polymer matrix half-thickness.

The root mean square error (RMSE) was used to estimate the quality of the model fitting and it was calculated by following Equation (6),

$$\text{RMSE} = \left[ \sum\_{i=1}^{n} \right]^{\frac{\left(y\_i - y\_i\right)^2}{n}} \right]^{\frac{1}{2}} \tag{6}$$

where *yi* and *yˆi* are, respectively, the experimental and predicted residual value; and *n* is the number of experimental points per migration curve.

High performance liquid chromatography coupled to an ultraviolet spectrophotometry detector (HPLC-UV) was used to determine the amount of thymol released from films at different migration times. An Agilent 1260 Infinity-HPLC Diode Array Detector (DAD) (Agilent, Santa Clara, CA, USA) and an Agilent Eclipse Plus C18 (100 mm × 4.6 mm × 3.5 μm) column were used. The mobile phase was composed of acetonitrile/water (40:60) at 1 mL min−<sup>1</sup> flow rate. 20 μL of the extracts were injected and thymol was detected at λ = 274 nm. Analyses were performed in triplicate. Calibration standards were run at different concentrations between 12.5 and 780 mg kg−<sup>1</sup> directly obtained from a stock solution (1000 mg kg<sup>−</sup>1) using appropriately diluted standards of thymol in ethanol 10% (*v*/*v*). The method was validated by the calculation of the main analytical parameters affecting the determination of thymol in the studied food simulant. Limit of detection (LOD) and limit of quantification (LOQ) values were determined by using regression parameters from the calibration curve (3 Sy/x/<sup>a</sup> and 10 *Sy*/*x*/*a*, respectively; where *Sy*/*<sup>x</sup>* is the standard deviation of the residues and a is the slope of the calibration curve) obtaining

values of 0.29 mgThymol kg−<sup>1</sup> and 0.96 mgThymol kg<sup>−</sup>1, respectively. An excellent linearity was obtained with a determination coefficient R2 of 0.9994.

### *2.4. Determination of Antioxidant Activity*

The antioxidant activity of the obtained migration extracts was evaluated to study the effect of thymol released from the nano-biocomposite films into ethanol (10%, *v*/*v*) after 10 days by using the spectrophotometric method based on the formation of the stable radical 2,2-diphenyl-1-picrylhydrazyl (DPPH), as described elsewhere [30].

### *2.5. Bacterial Strains, Culture Conditions and Antibacterial Activity*

*Escherichia coli* (*E. coli*) was provided by the Zooprofilattico Institute of Pavia (Italy) as an isolate strain whereas *Staphylococcus aureus 8325-4* (*S. aureus 8325-4*) was obtained from Timothy J. Foster (Department of Microbiology, Dublin, Ireland). *E. coli* and *S. aureus 8325-4* were grown overnight under aerobic conditions at 37 ◦C in Luria Bertani Broth (LB) and Brian Heart Infusion (BHI) (Difco Laboratories Inc., Detroit, MI, USA), respectively. The final density of these cultures was established at 1 <sup>×</sup> <sup>10</sup><sup>10</sup> cells mL<sup>−</sup>1, determined by comparison of the optical density at 600 nm.

The evaluation of the antibacterial activity of the films was performed in 100 μL of a diluted cell suspension (1 <sup>×</sup> 103 cells mL<sup>−</sup>1) of *E. coli* and *S. aureus 8325-4* maintained overnight. Bacterial strains were added to 3 <sup>×</sup> 3 mm2 samples, seeded at the bottom of a 96-well tissue culture plate and incubated at three different temperatures: 4 ◦C, 24 ◦C and 37 ◦C for 3 h and 24 h, respectively. These temperatures were selected to simulate refrigeration conditions (4 ◦C), ambient temperature (24 ◦C) and the usual incubation temperature of microbiological tests (37 ◦C). Regarding time, 3 h simulated a fast contact between the polymer sample and the bacteria whereas 24 h is the usual time used to evaluate the microbiological growth in antimicrobial tests.

Furthermore, 96-well flat-bottom sterile polystyrene culture plates were used as control under the same experimental conditions. At the end of each incubation period, bacterial suspensions were serially diluted and, after the incubation for 24/48 h at 37 ◦C, cell survival was expressed as the percentage of colony-forming unit, CFU, of bacterial growth on active nano-biocomposite films compared to those obtained for the neat PLA film.

### *2.6. Study of Disintegrability Under Composting Conditions*

Disintegration tests were performed by following the ISO 20,200 Standard [35]. Film samples for disintegration tests were cut in pieces (20 <sup>×</sup> 20 mm2) and they were buried at 5 cm depth in perforated boxes with a solid synthetic bio-waste and incubated at 58 ◦C for 35 days. Different time intervals were selected to recover samples from their burial, and they were further tested after 0, 2, 4, 7, 10, 14, 21, 28 and 35 days. The degree of disintegration (%) was calculated by normalizing the sample weight at different stages of incubation to the initial weight, following Equation (7):

$$\text{Disintegrability} \left( \% \right) = \frac{\mathcal{W}\_{\bar{i}} - \mathcal{W}\_{\bar{t}}}{\mathcal{W}\_{\bar{i}}} \cdot 100 \tag{7}$$

where *Wi* is the initial dry plastic weight, and *Wt* is the dry plastic weight at the end of the test.

The degradation of the chemical structure of the nano-biocomposite films during the disintegration tests was estimated by comparing Fourier transform infrared spectroscopy (FTIR) spectra and differential scanning calorimetry (DSC) thermograms at different test stages. Furthermore, the changes in the visual appearance of the samples were also evaluated.

DSC analysis of samples at different disintegration times was carried out from −25 to 180 ◦C at a heating rate of 10 ◦C min−<sup>1</sup> by using a DSC Mettler Toledo 822/e equipment (Schwarzenbach, Switzerland) working under nitrogen atmosphere (50 mL min<sup>−</sup>1). FTIR spectra of degraded samples were recorded by using a Jasco FT-IR 615 spectrometer (Jasco Inc., Easton, MD, USA) in attenuated total reflection (ATR) mode at 400–4000 cm−<sup>1</sup> range.

### *2.7. Statistical Analysis*

Statistical analysis of results was performed by using the SPSS commercial software (Version 15.0, Chicago, IL, USA). A one-way analysis of variance (ANOVA) was carried out. Differences between means were assessed based on confidence intervals using the Tukey test at a *p* < 0.05 significance level.

### **3. Results**

### *3.1. Migration Test and Antioxidant Activity of the Active PLA Nano-Biocomposite Films*

The use of nanofillers in active packaging systems has revealed their ability in controlling the release of active additives from polymer matrices, at suitable rates, improving their action [7]. In this work, the effect of the nanoclay in controlling the release kinetics of thymol from the PLA nano-biocomposite films to ethanol 10% (*v*/*v*) at 40 ◦C was evaluated. The amounts of thymol migrated in the simulant after 10 days were 285.0 <sup>±</sup> 3.3, 275.5 <sup>±</sup> 13.8 and 235.3 <sup>±</sup> 19.4 mgthymol kg−<sup>1</sup> food simulant for PLA/T, PLA/T/D43B2.5 and PLA/T/D43B5, respectively. These results indicate that the formulation with the highest amount of D43B (PLA/T/D43B5) showed the lowest migration rate, retaining a higher amount of thymol in the polymer structure after 10 days. This behavior is in agreement with previously reported studies [36] indicating that the tortuosity effect imposed by the presence of D43B in the diffusion of the active compound through the matrix plays an important role in the release of thymol from PLA-based nano-biocomposite films. Campos-Requena et al. [37] observed that the intercalated morphology of organic modified montmorillonite/LDPE nanocomposite films had some influence on the release rate of thymol resulting in a decrease by approximately 15%, providing a controlled release material.

The antioxidant capacity of thymol released from the developed nano-biocomposite films was measured by analyzing the extracts obtained after 10 days by using the DPPH method. The inhibition values obtained were 77.8 ± 0.1, 77.0 ± 0.4, and 77.8 ± 0.8% for PLA/T, PLA/T/D43B2.5, and PLA/T/D43B5, respectively, showing the efficient antioxidant performance of thymol added in PLA-based nano-biocomposite films. This high antioxidant activity can be attributed to the capability of the thymol phenolic hydroxyl groups to convert the phenolic oxygen anion in an alkaline environment [38]. Moreover, thymol has been reported to be a better antioxidant in lipids than its isomer carvacrol due to the more significant steric hindrance of the phenolic group [39].

#### Mathematical Modelling of Thymol Released from PLA Nano-Biocomposite Films

Release studies are necessary and highly relevant when active substances with antimicrobial or antioxidant properties are incorporated into packaging materials to enhance the safety and quality of food during long storage. The kinetics of these processes is indicative of the transport phenomena inside the matrix and gives an accurate estimation of the effective release rate of the active compounds. In this process, the migrant starts to diffuse through the amorphous portion of the polymer matrix toward the interface with the driving force of the concentration gradient. The migrant's concentration is partitioned between the two media until its potential chemical values, in both the polymer and the food, reach equilibrium. Thus, the release is the result of diffusion, dissolution and equilibrium processes, which are often described by the Fick's second law [40].

Consequently, the release mechanism of thymol was evaluated by modelling the results obtained at different times (up to 15 days) in ethanol 10% (*v*/*v*). Figure 1a–c depicts the normalized plots obtained for the average amount of thymol released to the food simulant at a defined time, *MF,t*, by the amount of thymol released at the equilibrium at time t→∞, *M*F,∞, vs. time *t* (hours).

The quantitative assessment of *MF,*<sup>∞</sup> allowed the analysis of the diffusion process. For such purpose, Equation (1) was used to fit the experimental data. The corresponding data for α and *KP,F* (Table 1) were computed considering that *VF* was 20 cm3 of ethanol 10% (*v*/*v*) and the area of the PLA-based films used in these tests was 12 cm2.

**Figure 1.** Normalized release of thymol from different polymer matrices: (**a**) PLA/T, (**b**) PLA/T/D43B2.5, and (**c**) PLA/T/D43B5; (**d**) plots of <sup>1</sup> <sup>π</sup> <sup>−</sup> <sup>1</sup> <sup>α</sup> <sup>×</sup> *MF*,*<sup>t</sup> MF*,0 0.5 versus *t* 0.5 for the migration of thymol from: PLA/T (-), PLA/T/D43B2.5 (•), and PLA/T/D43B5 (), into ethanol 10% (*v*/*v*).

**Table 1.** Parameters obtained for the release of thymol from PLA-based films into ethanol 10% (*v*/*v*).


According to the results shown in Table 1, two main conclusions can be obtained: (i) the cumulative amount of thymol released into ethanol 10% (*v*/*v*) decreased from 38% (without nanoclay) to 35 and 31% for 2.5 and 5 wt.% of D43B, respectively; (ii) the analysis of the partition coefficients (α and *KP,F*) showed their influence in the thymol diffusion mechanism. Therefore, it can be assumed that the release of thymol is governed by the Fick's second law (Equation (2)) and the diffusion coefficients (*D*, cm<sup>2</sup> s−1) were calculated (Table 1) from the least-square fit of Equation (5) to the experimental data (solid lines in Figure 1a–c). The RMSE (Equation (6)) of the experimental and estimated values between the calculated (*yi*) and observed (*yˆi*) results for *MF,t*/*MF,*<sup>∞</sup> was minimized, providing a reliable indication of their fit, and it could be considered promising if the experimental error is taking into account the *MF,t*/*MF,*<sup>∞</sup> ratio. A good fit between calculated and experimental data was obtained for PLA/T and PLA/T/D34B2.5, with RMSE values of 0.0773 and 0.0698, respectively. However, a higher value was obtained for PLA/T/D43B5 (RMSE: 0.114), particularly for long-range times.

A more in-depth analysis of the fitting between experimental and calculated values showed that these results allow estimating the kinetics of thymol release at short times. Positive deviations of the fitting line for short times (i.e., *MF,t*/*MF,*<sup>∞</sup> < 0.60) and negative deviations for *MF,t*/*MF,*<sup>∞</sup> > 0.60 were observed in all cases. For such purpose, Equation (8) can be used for a linear regression analysis and as a simplified release model derived from Equation (5) [41]:

$$\left[\frac{1}{\pi} - \frac{1}{\alpha} \cdot \frac{M\_{F,l}}{M\_{P,0}}\right]^{0.5} = -\frac{D^{\prime 0.5}}{\alpha \cdot l} \cdot t^{0.5} + \frac{1}{\pi^{0.5}}\tag{8}$$

Diffusion coefficients for short times, (*D'*, cm<sup>2</sup> s<sup>−</sup>1), were computed by using the linear fitting of Equation (8) to the experimental data (Figure 1d). Results obtained showed a very good fit between computed and experimental values for the first term in Equation (8) as a function of*t0,5*, with coefficients of determination (*R*2) higher than 0.999, suggesting that the experimental release data are well described by the proposed diffusion model for short-range times.

However, it was observed that the total data range cannot be fully characterized by a Fickian diffusion process, probably due to the lack of fitting caused by the last points in the plot resulting in poor results in the fitting to the first data, called short-range time. However, it was also observed that better fitting values were obtained after application of Equation (8). Therefore, the discrepancy in *D* values obtained (*D* and *D'*) from Equation (5) and Equation (8) was a good indication that a non-Fickian release model was observed in this system. The failure of the Fickian solution process to predict the release kinetics may be due to the polymer matrix's structural modifications caused by the progressive deterioration of the PLA matrix due to the direct contact with ethanol, which can act as a plasticizer in this system. This could result in the erosion of the polymer by opening the PLA's internal structure, creating interstitial spaces that could favor the release of thymol over time due to the concentration gradient [42,43]. Moreover, as the diffusion rate increased (*D* > *D'*), the intermolecular interactions between ethanol molecules and PLA chains were enhanced at long times [44]. No reports on the diffusion coefficient for thymol in nanocomposite films based on PLA have been published, but higher values than those obtained in this study were reported in polypropylene (PP) films using ethanol 10% (*v*/*v*) as food simulant (1.75 <sup>×</sup> 10−<sup>10</sup> cm<sup>2</sup> s<sup>−</sup>1) [45]. Torres et al. [46] evaluated the thymol release from LDPE films in ethanol 10% (*v*/*v*). They indicated that the diffusion coefficient values ranged from 7.5 <sup>×</sup> <sup>10</sup>−<sup>8</sup> to 1.8 <sup>×</sup>10−<sup>8</sup> cm<sup>2</sup> s−1, which were higher than those obtained in this work. These differences could be mainly due to the lower density and linear structure of LDPE compared to PLA, resulting in higher mass transport properties.

*D* values calculated in this work are consistent with the thymol release profiles shown in Figure 1a–c. The presence of D43B led to a decrease in the thymol release during the study at 40 ◦C, which was consistent with our previous study showing a shift in the XRD peak of D43B, suggesting the intercalated morphology of the PLA-D43B nanocomposites. In this case, a shift of the clay diffraction peak to lower angles of 2θ = 4.6◦ was observed, corresponding to an interlayer distance of 35.6 Å [30]. Moreover, Campos-Requena et al. [37] reported that the influence of the chemical modification of clays on the active compound profile release could be a factor in the modelling of an active packaging system with controlled release of volatile compounds.

### *3.2. Antibacterial Activity*

The antibacterial activity of the developed active nano-biocomposite films was evaluated against Gram-positive (*S. aureus* 8325-4) and Gram-negative bacteria (*E. coli*) by using the direct contact method. Table 2 shows the cell viability related to neat PLA, expressed as the percentage of microorganisms proliferated onto the PLA-based films after 3 and 24 h incubated at 4, 24 (room temperature), and 37 ◦C, respectively. Significant differences in cell viability for both bacterial strains were observed for almost all formulations compared to neat PLA at the tested experimental conditions. Exciting features could be observed when comparing the results obtained for active formulations containing thymol (PLA/T, PLA/T/D43B2.5, and PLA/T/D43B5) and their non-active counterparts (PLA/D43B2.5 and PLA/D43B5). These results are in accordance with those reported by Liu et al. [47] who found that the antibacterial

activity of phenolic monoterpenes, including thymol, is related to their ability to be released through the polymer matrix over time allowing their continuous availability and diffusion through the bacterial cell membrane. Thymol can attach to the cell surface, and thereafter, penetrate the phospholipid bilayer of the cell membrane. The relative position of the hydroxyl group is crucial for the bioactivity of thymol, which explains its superior antimicrobial action as compared to other plant phenolics, such as 2-amino-p-cymene, which has a similar structure than thymol except for the hydroxyl group [48].

**Table 2.** Antibacterial activity of PLA-based nano-biocomposite films obtained at 2 incubation times (3 h and 24 h) and 3 different temperatures (4, 24 and 37 ◦C) against *E. coli* and *S. aureus* 8325-4, expressed as cell viability (%).


Data are expressed as cell viability (%), which corresponds to the percentage of the CFU of bacteria growth in nano-biocomposite films compared to that obtained in PLA (set as 100%). Results are expressed as mean ± standard deviation, *n* = 3. Different superscripts (a, b, c, d, and e) within the same column at a specific temperature indicate statistically significant different values compared at 3 and 24 h for *S. aureus* 8325-4 and *E. coli* at different temperatures.

PLA formulations with only D43B, without the presence of thymol, also showed antibacterial activity against *E.* coli and S. aureus 8325-4 strains at both incubation times and all tested temperatures (4, 24 and 37 ◦C). PLA/D43B2.5 and PLA/D43B5 showed similar cell viability in most of the tested conditions. The only set of experimental conditions where PLA/D43B2.5 showed significantly higher antimicrobial activity than PLA/D43B5 was at 4 ◦C and 24 h for S. aureus 8325-4, with values 88.6 ± 0.9 and 94.4 ± 1.7%, respectively. These results are in agreement with those reported by De Azeredo et al. [49], who concluded that organo-modified montmorillonites could also produce the rupture of cell membranes by themselves, resulting in the inactivation of both Gram-positive and Gram-negative bacteria. This effect was attributed to the presence of quaternary ammonium groups able to react with lipids and proteins in the microorganism cell wall. Hong and Rhim [50] proved that organically modified clay powders with a quaternary ammonium salt, such as D43B, possess strong antimicrobial activity against *S. aureus*, *Listeria monocytogenes*, *Salmonella typhimurium,* and *E. coli* O157:H7.

The obtained antibacterial activity in the ternary systems resulted much more efficient and statistically significant with reference to the PLA control film. The percentage of bacteria viability was lower for PLA/T/D43B2.5 and PLA/T/D43B5 incubated with *S. aureus* 8325-4 strains (44.3 ± 1.4% and 48.5 ± 1.1%, respectively) and *E. coli* strains (47.2 ± 1.2% and 47.3 ± 1.4%, respectively) for 3 h at 37 ◦C (Table 2). Regarding incubation time, the percentage of the surviving fraction of bacteria did not show significant differences (*p* > 0.05) between 3 and 24 h for both bacterial strains, confirming the bacteriostatic action of the PLA-based nanocomposites. Concerning incubation temperature, the percentage of cell viability for both bacterial strains was lower when incubating at 37 ◦C whereas it was moderately increased at 4 ◦C. These results are in line with the thymol release data, thus confirming that the temperature may influence the diffusion and swelling properties of the PLA matrix, promoting the diffusion of thymol through the biopolymer structure.

#### *3.3. Disintegrability Under Composting Conditions*

Table 3 shows the values of weight loss of each sample and Table 4 shows the visual appearance of nano-biocomposite films at different times under composting conditions. It was observed that after 4 days, the disintegration rate of PLA-based materials increased significantly for binary and ternary systems showing an evident fragmentation, reaching all materials a degree of disintegration exceeding 90% after 35 days. In fact, after 4 days, all samples changed their visual appearance with a general whitening effect, loss of transparency, and evident deformation and size reduction. These results are indicative of the beginning of the hydrolytic degradation process caused by simultaneous changes in the refractive index due to water absorption, with the formation of low molecular weight by-products and the resulting increase in the PLA crystallinity [51].

According to Su et al. [52] this first step, covering the first 5 days of this study where the weight loss was small, corresponds to slow bulk degradation processes with a surface-erosion mechanism that can be mainly caused by hydrolysis, resulting in small molecules (mostly water) that can diffuse through the polymer matrix. The diffusion rate is influenced by several factors, such as crystallinity, cross-linking degree, and other morphological properties. After 5 days, the weight loss increased dramatically reaching values higher than 40% after 7 days and a continuous increase with time up to 35 days when more than 90% of the initial weight was lost (Table 3). In that period, the hydrolysis reaction was still important, and the average molecular weight of PLA decreased continuously forming small fragments easier to disintegrate since the internal chains were increasingly exposed. In that period, water, compost, and the microbiota generated in the reactor can penetrate into the gaps in the PLA structure formed by the hydrolysis reactions, contributing to a clear acceleration of the disintegration process which causes visual modifications by the formation of small particles, fragmentation and irreversible changes in mechanical properties.


**Table 3.** Disintegrability values (mean ± standard deviation, *n* = 3, %) of PLA and nano-biocomposite films at different times under composting conditions. Different superscripts (a, b, c, and d) within the same row indicate statistically significant different values.


**4.**VisualofPLAandnano-biocompositefilmsatdifferenttimesundercompostingconditions.Differentsuperscriptswithinthe

The use of nanoclays, such as D43B, and thymol as active additive influenced the disintegration rate in compost of PLA since this process is strongly dependent on the hydrophilic/hydrophobic character of the nanocomposite and this parameter changes due to the presence of hydroxyl groups from thymol and the organic modifier of D43B [53]. Hydroxyl groups can contribute to the heterogeneous hydrolysis of PLA by absorbing water from the medium resulting in a noticeable formation of labile bonds in the PLA structure with the consequence of a significantly higher disintegration rate [54]. It was also observed that binary and ternary systems suffered physical breakage with a considerable increase in weight loss (Table 4) after 7 days, showing significant differences in the disintegration profile when comparing neat PLA and nano-biocomposite films. Results at longer times showed that physical degradation progressed with burial time, resulting in the complete disintegration of all the initial samples after 35 days, when the disintegration degree exceeded 90% covering the ISO 20,200 requirements (Table 3).

FTIR analysis of neat PLA and nano-biocomposite films at different times was carried out to evaluate the structural changes produced by the disintegration process in all formulations. Figure 2 shows the FTIR spectra obtained for neat PLA, PLA/T and PLA/T/D43B5 after 0, 7 and 21 days under composting conditions. PLA showed characteristic bands at 1750 cm−<sup>1</sup> (C = O), 1440 cm−<sup>1</sup> (CH–CH3), and 1267 cm<sup>−</sup>1(C–O–C) as well as three peaks at 1123, 1082 and 1055 cm−<sup>1</sup> related to the C–C–O groups. After 7 days, the intensity of the three peaks related to the C–C–O groups decreased and after 21 days, these peaks disappeared for all formulations. Similar results were obtained by Fortunati et al. [53] who proposed that the modification of the intensity of peaks related to the C–C–O groups can be associated with the scission of the PLA interchain bonds produced by hydrolysis reactions occurring during disintegration tests. Moreover, the C–O–C stretching vibration at 1267 cm−<sup>1</sup> was also affected by the depletion of the lactic acid and oligomer molecules caused by microorganisms, leaving highly reactive carboxylate end groups [55]. FTIR results agreed with those achieved for the disintegration weight loss and with the progressive disintegration of all samples with increasing testing time.

Figure 3a shows an example of the DSC thermograms obtained from the first heating scan for PLA, PLA/T and PLA/T/D43B5 films at different composting times. The endothermic peak observed immediately after the Tg at day 0 for all the tested materials corresponds to the enthalpic relaxation process. This effect was related to the aging process of PLA, which was previously observed by other authors [56]. However, the initially amorphous PLA-based materials developed multiple endothermic peaks just after 7 days under composting conditions. In fact, the enthalpic relaxation peak gradually disappeared due to the hydrolysis process at short incubation times. Yang et al. [57] related this behavior to the moisture absorption happening under composting conditions, since water could serve as a plasticizer agent in PLA matrices. These effects could also be related to the well-dispersed nanofiller inside the polymer matrix. Olewnik-Kruszkowska et al. [58] associated the hydrolytic degradation to the decrease in all thermal properties and they confirmed that the highest changes in the Tg value could be related to the dispersion of the nanofiller inside the polymer matrix.

The gradual disintegration suffered when increasing the testing time resulted in the observation of new melting peaks related to the formation of crystalline structures with different perfection degrees in the PLA matrix (Figure 3a). These results were found for all samples and they can be correlated with the observed visual changes, since hydrolysis promotes crystallization in the polymer matrix, resulting in significantly essential changes in the visual appearance of the testing samples and their disintegrability behavior. Similar results were reported by other authors, who suggested that the appearance of multiple melting peaks could be related to the formation of different crystal structures due to the polymer chain scission produced during degradation [59,60].

Figure 3b shows an example of the DSC thermograms recorded during the second heating scan for PLA, PLA/T and PLA/T/D43B5 samples submitted to the disintegration test. It was observed that after 2 days, all PLA-based films showed a significant decrease in Tg. Previous work by our research group showed that this decrease in Tg values was due to the increase in the mobility of the polymer chains as a consequence of the hydrolytic process [55] and the formation of lactic acid oligomers and low molecular weight by-products with a plasticizing effect in the polymer structure and the consequent changes in their visual appearance.

**Figure 2.** FTIR spectra of PLA, PLA/T, and PLA/T/D43B5 before (0 days) and after different incubation times (7 and 21 days) under composting conditions.

**Figure 3.** DSC thermograms, 1st heating scan (**a**) and 2nd heating scan (**b**), of PLA-based nanobiocomposite films after different composting times.

#### **4. Conclusions**

The incorporation of thymol, as active additive, and D43B, as nano-reinforcing agent, into PLA has shown as an accessible and useful route for the preparation and modification of PLA nano-biocomposite films properties. An improvement in functional properties of PLA-based films was obtained due to the addition of the active additive and the nanoclay resulting in enhanced antimicrobial and antioxidant properties, demonstrating the high potential of the developed formulations for food packaging applications without compromising the inherent biodegradation properties of the PLA matrix. The obtained results suggest the possibility of controlling the release of active additives in the design of active nano-biocomposite films through the incorporation of laminar nanoclays by the decrease in the diffusion of thymol through the polymer matrix by the formation of tortuous paths. The antibacterial activity of these active nanocomposites was proved against two different bacterial strains, showing the PLA/T/D43B2.5 formulation the best results against both *S. aureus 8325-4* (44.3 ± 1.4%) and *E. coli* (47.2 ± 1.2%) at 37 ◦C and 3 h of incubation time. From a practical point of view, the combination of 8 wt.% of thymol and 2.5 wt.% of D43B added into a commercial PLA matrix showed high potential for the development of new bio-based and biodegradable active packaging films with application in prolonging the shelf-life of fresh food.

**Author Contributions:** Conceptualization, M.R., M.P., A.J. and M.C.G.; methodology M.R., F.C., L.V. and A.J.M.V.; validation, M.R., E.F., M.C.G. and A.J.; formal analysis, M.R., E.F., A.B., L.V., M.P., A.J.M.V., A.J., J.M.K. and M.C.G.; investigation, M.R., F.C. and A.J.M.V.; resources, M.R., J.M.K., A.J. and M.C.G.; data curation, M.R., A.B., E.F., L.V., A.J.M.V. A.J., and M.C.G.; writing—original draft preparation, M.R.; writing—review and editing, M.R., E.F., M.P., A.B., J.M.K., F.C., L.V., A.J.M.V., A.J. and M.C.G.; supervision, M.R., A.J. and M.C.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** Authors would like to thank Spanish Ministry of Economy and Competitiveness (MAT2017-84909-C2-1-R) and Generalitat Valenciana (IDIFEDER/2018/007) for their support of this research. Marina Ramos would like to thank University of Alicante (Spain) for the UAFPU2011-48539721S predoctoral research grant. AJMV thanks ´Coimbra Chemistry Center and Fundação para a Ciência e Tecnologia (FCT) for the financial support through the programmes UID/QUI/UI0313/2020 and COMPETE.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Design and Preparation of Polysulfide Flexible Polymers Based on Cottonseed Oil and Its Derivatives**

**Yurong Chen 1,2,3, Yanxia Liu 1,2,3, Yidan Chen 2,3, Yagang Zhang 1,2,3,\* and Xingjie Zan <sup>3</sup>**


Received: 21 July 2020; Accepted: 17 August 2020; Published: 19 August 2020

**Abstract:** Polysulfide-derived polymers with a controllable density and mechanical strength were designed and prepared successfully using bio-based cottonseed oil (CO) and its derivatives, including fatty acid of cottonseed oil (COF) and sodium soap of cottonseed oil (COS). The reaction features of CO, COF and COS for polysulfide polymers were investigated and compared. Based on the free radical addition mechanism, COF reacts with sulfur to generate serials of polysulfide-derived polymers. COF strongly influences the density and tensile strength of these polymer composites. Whereas COS was not involved in the reaction with sulfur, as a filler, it could increase the density and tensile strength of polysulfide-derived polymers. Moreover, the results showed that these samples had an excellent reprocessability and recyclability. These polysulfide-based polymers, with an adjustable density and mechanical strength based on CO and derivatives, could have potential applications as bio-based functional supplementary additives.

**Keywords:** polysulfide-derived polymers; cottonseed oil; fatty acid of cottonseed oil; sodium soap of cottonseed oil

### **1. Introduction**

Elemental sulfur derives mainly from industrial process such as petroleum refining. It is currently recognized as a vital basic chemical stock for rubber production, chemical fertilizer, antimicrobial agents and chemical dyes [1–5]. Although elemental sulfur has been applied in many fields, it is extensively stockpiled every year because production far surpasses demand [6]. Since 1,3-diisopropenyl benzene was found to stabilize polysulfide chains in inverse vulcanization and the resulting thermoplastic copolymers exhibited great processability and promising potential for applications in the electrical industry as cathodes [7], various functional sulfur-containing polymers and sulfur–organic copolymers synthesized by using different cross-linkers have been reported [8–10]. These sulfur-containing polymers and sulfur–organic copolymers have drawn considerable attention due to their application potentials for solid electrodes [11,12], camera lenses and medium infrared ranges [13], as well as heavy metal remediation [14]. In order to make the best of mass-produced sulfur and meet the demands of sustainable development, it would be desirable to seek renewable and sustainable bio-based materials to prepare value-added materials with sulfur. Bio-based materials feature many advantages such as their inexpensiveness, accessibility and availability in large quantities. The design and synthesis of bio-based functional materials also fits in with and benefits green chemistry and sustainable engineering, aimed at making the best of natural resources with more environmentally benign and eco-friendly approaches [15–19].

Vegetable oils, such as canola oil, sunflower oil and linseed oil [20], have been explored for the preparation of functional sulfur–organic copolymers with the method of inverse vulcanization. Cotton, as a vital kind of commercial crop, has a wide range of planting areas worldwide and a higher yield compared with other commercial crops. Therefore, cottonseed oil has more advantages in view of yield and price than other vegetable oils. Currently, the explorations for cottonseed oil are concentrated on the development of biodiesel with lower environmental pollution, lower production costs and greater safety [21–24], whereas polysulfide-derived polymers based on cottonseed oil and derivatives are rarely reported on. It is highly desirable to design and prepare value-added functional polymer composites with cottonseed oil and its derivatives.

In our previous work [25], cottonseed oil was used as a renewable cross-linker to react with industrial byproduct sulfur and the resulting sulfur-containing plant rubber polymers were prepared successfully. These plant rubber polymers could remove mercury ions from aqueous solution as bio-based absorbents. Traditional rubber materials could serve as supplementary additives for improving the mechanical properties or densities of polymers [26–29]. It would be interesting and ideal if the mechanical property and rigidity of these sulfur-containing plant rubber polymers were tunable. In the work reported here, we explored the possibility of preparing value-added sulfur-containing functional polymer with cottonseed oil and its derivatives with goal of achieving controllable densities and adjustable mechanical strength.

In this work, though the inverse vulcanization process, mass-produced sulfur, renewable cottonseed oil (CO) and its derivatives were taken advantage of, and a series of novel polysulfide-based polymers with controllable densities and adjustable mechanical strength were successfully prepared. Cottonseed oil derivatives involved in this work included fatty acid of cottonseed oil (COF) and sodium soap of cottonseed oil (COS). These polysulfide-derived polymers could have potential applications as bio-based functional supplementary additives.

### **2. Materials and Methods**

### *2.1. Materials*

Cottonseed oil (CO, food grade), fatty acid of cottonseed oil (COF) and sodium soap of cottonseed oil (COS) from cottonseed oil were obtained from Shihezi Kanglong Oil Industry and Trade Company (Shihezi, China). Sulfur (S, powder, ≥99.5%,) was purchased from Tianjin Baishi Chemical Industry Co. Ltd. (Tianjin, China). CO, COF, COS and their mixture, including CO/COF, CO/COS, as well as CO/COF/COS, is referred to as COX collectively for ease of subsequent discussion. Moreover, except for CO, residual raw materials were also called COY in order to differentiate them from COX.

### *2.2. Preparation of Polysulfide-Derived Polymers*

About 10.0 g of elemental sulfur powder (S) was added into a 100 mL vial equipped with a magnetic stir bar and then melted while stirring at 150 ◦C. A certain amount of COX was then added to the above molten liquid, while stirring and heating were continued to ensure efficient mixing and reactions between reactants. After a certain period, the mixture was cooled to room temperature and then the resulting polysulfide-derived polymers were obtained. The related details can be found in Table 1.

### *2.3. Characterization*

The morphological analysis of the prepared samples was performed by a field emission scanning electron microscopy (FE-SEM, SUPRA 55vp, ZEISS, Oberkochen, Germany) with an Oxford detector, operating with 2.00 kV electron beams. Infrared (IR) spectra were recorded on a Fourier Transform Infrared Spectrometer (VERTEX-70, Bruker, Karlsruhe, Germany) using the ATR (Attenuated Total Reflectance) method with a wave number ranging from 400 to 4000 cm<sup>−</sup>1. Gas Chromatography–Mass Spectrometry (GC–MS) was acquired on a Headspace injection gas chromatograph mass spectrometer

(7890B GC/7697A/5977B MSD, KEYSIGHT, Santa Rosa, CA, America) using a Rtx-5MS column (30 m long × 0.25 mm thickness × 0.25 μm ID), with an injection temperature of 130 ◦C, column temperature of 270 ◦C, gas flow rate of 24.0 mL·min−1, and electron ionization used to obtained nominal masses. Density analysis was carried out on an electronic densitometer (XF-120MD, Xiongfa, Xiamen, China) with testing samples tailored in the cubic dimension of 10 mm × 10 mm × 10 mm. Mechanical testing was carried out on a microcomputer-controlled electronic tensile testing machine (C43-104, MTS, Rochester, MN, America) according to the national standard GB/T 528-2009 with dumbbell-shaped splines at an elongation rate of 500 mm·min−<sup>1</sup> and the test length and thickness of splines were 20.0 ± 0.5 mm and 2.0 ± 0.2 mm, respectively.



Time <sup>1</sup> refers to the period from the moment that all reactants were fully mixed together to the moment that the reactions were finished, which does not include the period of cooling samples to room temperature.

### *2.4. Self-Healing Experiments*

To study the reprocessability and recyclability of the related samples, self-healing experiments were performed by cutting samples into pieces and putting the fragments together for remolding under 10 MPa at 160 ◦C for 10 min using a thermo-compressor (R3212, Qien, Zhengzhou, China). The ratio of the tensile strength of the healed samples to those of the original one was determined as the recovery ratio to measure the reprocessability and recyclability of samples.

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

### *3.1. Synthesis and Characterization*

To investigate whether elemental sulfur could react effectively with cottonseed oil derivatives including fatty acid of cottonseed oil (COF) and sodium soap of cottonseed oil (COS), the resulting polysulfide-derived polymers were analyzed. As depicted in Figure 1a–e, when elemental sulfur reacted with CO, COF and COS, respectively, the resulting polysulfide-derived polymers SCO and SCOS appeared as brown elastic bulk and tanned plastic bulk, respectively, while SCOF appeared as a black viscous fluid. Moreover, the tanned plastic SCOS bulk could also be kneaded into a ring-shaped or U-shaped object at room temperature, which implies that SCOS is quite flexible and features excellent processability compared with SCO and SCOF. Figure 1f,g displays the surface morphology of SCO and SCOS under a scanning electron microscope. There were large numbers of fragments on the surface of SCO, whereas the surface of SCOS was occupied by a sticky object conjoined with many small particles. Furthermore, an FT-IR spectrogram (Figure 2) was also used to analyze the structural changes in SCO and SCOF, as well as SCOS. Compared with raw materials CO, COF and COS, the prepared SCO and SCOF both showed that the moderate-strength peak at 3009 cm−<sup>1</sup> disappeared (with the exception of SCOS). The absence of the peak was attributed to the vanishing of =C–H bond stretching vibration in cottonseed oil and fatty acid of cottonseed oil [4]. Meanwhile, according to Table 2, the relative intensity of the peak at 3009 cm−<sup>1</sup> in SCO and SCOF samples decreased by about 90% compared with

that of raw materials CO and COF, whereas relative intensity of the peak at 3009 cm−<sup>1</sup> in SCOS samples had no obvious changes compared with that of the raw material COS. The quantitative analysis of the FT-IR spectrogram further indicated that there existed an obvious reaction between raw materials CO and COF and products SCO and SCOF. The macroscopic and microscopic differences between samples as well the as differences in the FT-IR spectrogram implied differences in chemical reactivity between sulfur and different COX. According to the previous work [20], under a high temperature, elemental sulfur could initiate ring-opening polymerization and further react with the unsaturated double bonds in plant oils and their derivatives. The essence of the reaction was that, above the melting point temperature, 120 ◦C, sulfur was first melted and then heated up further to a higher temperature to generate sulfur free radicals to further react with C=C bonds in plant oils by the free radical addition mechanism.

**Figure 1.** The surface morphology and plasticity of real samples. (**a**,**b**) Digital photos of sulfur cottonseed oil (SCO) and sulfur fatty acid of cottonseed oil (SCOF), respectively; (**c**–**e**) digital images of sulfur–sodium soap of cottonseed oil (SCOS): (**c**) cubic SCOS, (**d**) ring-shaped SCOS and (**e**) U-shaped SCOS; (**f**,**g**) SEM images of SCO and SCOS, respectively (mS: mCOX = 1:1).

**Figure 2.** FT-IR spectrogram of the related samples (mS: mCOX = 1:1).


**Table 2.** Quantitative data for FT-IR spectrogram of the related samples (mS: mCOX = 1:1).

GC–MS analysis data (Table 3) show that cottonseed oil contains different fatty acid components, including both saturated and unsaturated fatty acids. There are a large number of C=C bonds derived from unsaturated fatty acid, including linoleic acid and oleic acid. These C=C bonds are the key motifs and act as active cross-linking sites with sulfur to generate cross-linking elastic SCO (Figure 3). This also explains why the =C–H bond stretching vibration in the FT-IR spectra of SCO disappeared compared with raw material CO. The results show that the content of unsaturated fatty acid was about 75% in COF. Theoretically, COF could react with sulfur to generate elastic plant rubber in a similar manner to SCO. However, the resulting SCOF was a viscous fluid instead of presenting solid-state elasticity. This could be due to the fact that the vanishing of glycerinum decreased the cross-linking effect of unsaturated fatty acid in COF, and the obtained SCOF had a low cross-linking degree (Figure 3). This could also explain why the FT-IR spectra of SCO and SCOF were similar but the appearance and character of them were different. Sodium soap based on unsaturated fatty acid was the dominant component of COS, which mainly included sodium linoleate and sodium oleate. In theory, COS could react with sulfur to generate an elasticity similar to SCO. However, the resulting SCOS appeared as a plastic bulk. Moreover, compared with the FT-IR spectrogram of raw material COS, the peak at 3009 cm−<sup>1</sup> still existed in SCOS, which may imply that there was no reaction between sulfur and COS. This was due to the high melting point of sodium linoleate and sodium oleate, which exceeded 190 ◦C. Therefore, these sodium soaps were not effectively involved in the reaction with sulfur at 150 ◦C. Instead, they were mingled with polysulfide fragments as granular padding (Figure 3). The observation of the chemical reactivity with different cottonseed oils and their derivatives was consistent with the above analysis of the SEM images of SCOS (Figure 1f,g).

**Table 3.** Representative methyl esterification products of cottonseed oil by Gas Chromatography–Mass Spectrometry (GC–MS) analysis.


### *3.2. Density Analysis*

As depicted in Figure 4a, when the mass ratio of sulfur to COX was 1.0, the density of SCO/COF as well as SCO/COF/COS serial samples (while the mass ratio of COF to COS was 2.0) decreased with the increase in the mass ratio of COF or COF/COS to CO and was lower than that of SCO samples. These results imply that the introduction of COF could decrease the density of the prepared polymer composites. On the other hand, the density of SCO/COS and SCO/COF/COS serial samples (while the mass ratio of COF to COS was lower than 2.0) increased significantly with the rising mass ratio of COS or COF/COS to CO and, noticeably, the density of these samples were higher than that of SCO serial samples, which indicated that the introduction of COS could increase the density of the resulting polymer composites. Moreover, Figure 4b reveals that the density of samples gradually increased with

the increase in the mass ratio of sulfur to COX, which implies that increasing the sulfur content is helpful for elevating the density of samples to some extent.

**Figure 3.** Procedures for the preparation of SCO, SCOF and SCOS, respectively.

**Figure 4.** Relative density of the prepared samples. (**a**) Relative density variation in samples versus the mass ratio of COY to CO (mS/mCOX = 1); (**b**) relative density variation in samples versus the mass ratio of sulfur to CO, COF, COS and their mixture, including CO/COF, CO/COS, as well as CO/COF/COS (COX).

### *3.3. Mechanical Strength Analysis*

Figure 5a shows that when the mass ratio of sulfur to COX was 1.0, the tensile strength of SCO/COF serial samples decreased with the increasing mass ratio of COF to CO. Similarly, the tensile strength of SCO/COF/COS serial samples decreased with the increasing mass ratio of COF/COS to CO, while the mass ratio of COF to COS was 2.0. Meanwhile, the tensile strength of both SCO/COF and SCO/COF/COS serial samples were lower than that of SCO serial samples, which implies that the introduction of COF could decrease the mechanical strength of samples. However, when the mass ratio of COY to CO was lower than 1.0, the tensile strength of SCO/COS serial samples increased significantly with the increase in the mass ratio of COS to CO. The tensile strength of SCO/COF/COS serial samples also increased with the increasing mass ratio of COF/COS to CO, while the mass ratio of COF to COS was lower than 2.0. Most importantly, the tensile strength of these samples was higher than that of SCO serial samples, which reveals that the introduction of COS could enhance the mechanical strength of samples effectively. However, when the mass ratio of COY to CO was higher than 1.0, the tensile strength of SCO/COS and SCO/COF/COS serial samples (the mass ratio of COF to COS was lower than 2.0) started to decline, which was due to the decreasing cross-linking density resulting from the lessening of the content of CO. Figure 5b demonstrates that the tensile strength of samples initially increased with the increase in the mass ratio of sulfur to COX and gradually reached the maximum when the mass ratio of sulfur to COX was 1.0 and then decreased dramatically. This could be due to local stress concentration on the surface and inside the samples [30] because of the increase in the content of sulfur.

**Figure 5.** Tensile strength of the prepared samples. (**a**) Tensile strength variation in samples versus the mass ratio of COY to CO (mS/mCOX = 1); (**b**) tensile strength variation in samples versus the mass ratio of sulfur to COX.

### *3.4. Reprocessability and Recyclability*

To further study the reprocessability and recyclability of the prepared samples, SCO/COS serial samples with a higher density and tensile strength were chosen as representative samples. As shown in Figure 6a, SCO/COS serial samples can be remolded into coherent and smooth dumbbell-shaped splines when they are cut into small pieces and after hot pressing. Figure 6b,c show the tensile strength variation in SCO/COS serial samples versus the mass ratio of reactants after multiple reprocesses. The tensile strength of SCO/COS serial samples exhibited a recovery ratio above 90% after first reprocessing and the secondary recovery rate was above 85%, which was due to a large number of reversible disulfide bonds in the samples. These results demonstrate that the reversible cross-linking made the samples capable of reprocessing and recycling [31]. However, the results show that the tensile strength of these materials decreased significantly after more than two cycles. Our hypothesis was that the cross-linked sulfur–sulfur bonds would degrade after more than two cycles and simple hot pressing would not be efficient enough to aid in recovering and rebuilding those damaged bonds. These are actually the disadvantages of this type of material and it is important to find a way to solve these problems.

**Figure 6.** Reprocessability and recyclability of SCO/COS serial samples. (**a**) Digital photos of thermal reprocessing ability of SCO/COS serial samples; (**b**) tensile strength variation in SCO/COS serial samples versus the mass ratio of COY to CO after multiple reprocesses; (**c**) tensile strength variation in SCO/COS serial samples versus the mass ratio of sulfur to COX after multiple reprocesses.

### **4. Conclusions**

A series of polysulfide-derived polymers with a controllable density and mechanical strength were prepared successfully based on cottonseed oil (CO) and its derivatives, including fatty acid of cottonseed oil (COF) and sodium soap of cottonseed oil (COS). Based on the free radical addition mechanism, which is similar to the reaction mechanism of SCO, COF reacted with sulfur generates serial samples containing COF. COF can decrease the density and tensile strength of polysulfide-based polymers, whereas COS was not effectively involved in the reaction with sulfur due to the high melting point of sodium linoleate and sodium oleate. Even so, COS could act as a padding component, which could increase the density and tensile strength of polysulfide-derived polymers. The results demonstrated that the prepared polymer composites had an excellent reprocessability and recyclability, attributed to the large number of reversible disulfide bonds formed in the formation of plant rubber. These polysulfide-derived polymers with a controllable density and mechanical strength, based on CO and derivatives, could have potential applications as bio-based functional supplementary additives.

**Author Contributions:** Conceptualization, Y.Z. and Y.C. (Yurong Chen); methodology, Y.C. (Yurong Chen) and Y.L.; software, Y.L. and Y.C. (Yurong Chen); validation, Y.L and Y.C. (Yurong Chen); formal analysis, Y.Z. and X.Z.; investigation, Y.C. (Yurong Chen), L.Y. and Y.C. (Yidan Chen); resources, Y.Z. and X.Z.; data curation, Y.C. (Yurong Chen) and Y.C. (Yidan Chen); writing—original draft preparation, Y.C. (Yurong Chen); writing—review and editing, Y.Z.; visualization, Y.C. (Yurong Chen) and Y.L.; supervision, Y.Z. and X.Z; project admini-stration, Y.Z.; funding acquisition, Y.Z. All authors contributed substantially to the work reported. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was financially supported by the National Natural Science Foundation of China (21464015, 21472235), the Xinjiang Tianshan Talents Program (2018xgytsyc 2-3) and UESTC Talents Startup Funds (A1098 5310 2360 1208).

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **E**ff**ect of Iignocellulosic Nanoparticles Extracted from Yerba Mate (***Ilex paraguariensis***) on the Structural, Thermal, Optical and Barrier Properties of Mechanically Recycled Poly(lactic acid)**

**Freddys R. Beltrán 1,2,\*, Marina P. Arrieta 1,2,\*, Gerald Gaspar 1, María U. de la Orden 2,3 and Joaquín Martínez Urreaga 1,2**


Received: 29 June 2020; Accepted: 27 July 2020; Published: 29 July 2020

**Abstract:** In this work, yerba mate nanoparticles (YMNs) were extracted from *Ilex paraguairiencis* yerba mate wastes and further used to improve the overall performance of mechanically recycled PLA (PLAR). Recycled PLA was obtained by melt reprocessing PLA subjected to an accelerated ageing process, which involved photochemical, thermal and hydrothermal ageing steps, as well as a final demanding washing step. YMNs (1 and 3 wt.%) were added to the PLAR during the melt reprocessing step and further processed into films. The main goal of the development of PLAR-YMNs bionanocomposites was to increase the barrier properties of recycled PLA, while showing good overall performance for food packaging applications. Thus, optical, structural, thermal, mechanical and barrier properties were evaluated. The incorporation of YMNs led to transparent greenish PLAR-based films with an effective blockage of harmful UV radiation. From the backbone FTIR stretching region (bands at 955 and 920 cm<sup>−</sup>1), it seems that YMNs favor the formation of crystalline domains acting as nucleating agents for PLAR. The morphological investigations revealed the good dispersion of YMNs in PLAR when they are used in the lowest amount of 1 wt.%, leading to bionanocomposites with improved mechanical performance. Although the addition of high hydrophilic YMNs increased the water vapor transmission, the addition of 1 wt.% of YMNs enhanced the oxygen barrier performance of the produced bionanocomposite films. These results show that the synergistic revalorization of post-consumer PLA and nanoparticles obtained from agri-food waste is a potential way for the production of promising packaging materials that meet with the principles of the circular economy.

**Keywords:** poly(lactic acid); mechanical recycling; yerba mate; bionanocomposites

### **1. Introduction**

The development of bioplastics has raised a fair amount of interest in recent years. This is due to the constant growth of the consumption of fossil-fuel based plastics, which is leading to important environmental and raw materials availability problems. Among the most important bioplastics is poly(lactic acid) (PLA), which is an aliphatic polyester produced from renewable resources. PLA is obtained, on an industrial scale, via the ring-opening polymerization of lactide, the cyclic dimer of lactic acid, which is in turn produced by the fermentation of carbohydrates present in renewable feedstock such as corn, sugar beet or potato [1,2]. Due to its intrinsic biocompatibility and biodegradability, PLA was initially developed with a focus on biomedical applications. However, the development of new grades with improved thermal, mechanical and optical properties has turned PLA into one of the most important bioplastics on the market, with applications on several industrial sectors, such as the textile, automotive and especially in short-term applications, such as those coming from the food packaging sector [3,4]. This wide variety of applications is leading to a continuous growth on the production of PLA, reaching a global production capacity of 270 kt in 2019 [5].

The use of PLA in applications commonly dominated by fossil-fuel-based plastics, such as packaging, could lead to important advantages from the sustainability point of view. However, it is worth noting that a massive use of PLA might result in some environmental problems. The newer grades of PLA, designed with demanding applications in mind, are very resistant and are only biodegradable at specific industrial conditions (i.e., 58 ◦C, RH% - 65, pH - 7.5, *C*/*N* relationship between 20:1 and 40:1) [6,7], which are not available in the environment (i.e., landfill) [8,9]. Hence, an inadequate management of the generated residues could lead to the accumulation of PLA wastes. Furthermore, the transition to a circular economy model has to be considered. In this circular economy model, plastics play a prominent role, as it can be seen from the strategies and directives proposed by the European Commission, including the need to replace single-use plastics by the end of 2020 [10–12]. These policies promote not only the reduction of plastic waste, but also the recovery of such wastes in order to reuse them, retaining their value. Therefore, the incorporation of PLA into the circular economy model constitutes a major challenge, which could be achieved through the mechanical recycling of PLA-based plastic wastes [13].

Although mechanical recycling allows for reducing the consumption of raw materials and the emissions related to the manufacture of PLA, previous studies [14–16] point out that mechanical recycling promotes chain scission reactions in PLA, resulting in a decrease of the molecular weight and of some important properties such as thermal stability and Vickers hardness. Therefore, the development of cost-effective and environmentally friendly methods to recover the properties of mechanically recycled PLA, and thus improve its recyclability, is a key challenge. In this regard, several alternatives have been proposed as valid approaches to increase the overall performance of recycled PLA, such as the use of thermal treatments [17], reactive extrusion with cross-linking agents and chain extender additives [18,19] and the use of inorganic fillers [3]. Furthermore, to guarantee the packaging green nature, another potential alternative is the utilization of reinforcements derived from renewable resources. For instance, in a recent work [20], the addition of small amounts of silk fibroin nanoparticles led to the improvement of thermal, mechanical and gas barrier properties of recycled PLA. It is widely known that nanocomposites show excellent mechanical, thermal, and gas barrier properties compared with the conventional polymeric materials or composites, [21,22]. Thus, the use of natural reinforcements for the development of PLA-based nanocomposites, intended for food packaging applications, represents a good option to improve the overall performance of PLA and recycled PLA (PLAR), without influencing the transparency which is very important for consumers acceptance [3,22,23].

From a circular economy point of view, it would be interesting to evaluate upgrading methods that also allow to valorize other wastes, such as those coming from agri-food or textile industries. In this regard, lignocellulosic residues from agri-food products are mainly considered as waste or low-value by-product [24–26]. Nevertheless, other lignocellulosic biomass derivatives have been recognized as optimal reinforcing fillers for the bioplastic industry due to the fact that they are biobased, light, stiff as well as non-abrasive for the plastic processing machinery [27–29]. In this context, several lignocellulosic nanoparticles have shown their ability to enhance the PLA overall performance, in terms of thermal, mechanical and barrier properties while also providing some anti-oxidant properties, thus increasing its interest in food packaging applications [27,29–31]. Polymer nanocomposites refer to multiphase polymeric systems where at least one of the constituent phases, commonly the nanofiller, has at least one dimension in the nanoscale range (<100 nm) [22]. The nanoparticles dimensions and properties depend on the raw material utilized for the extraction and the chemical process selected for their production [28]. A simple and aqueous extraction procedure to obtain lignocellulosic nanoparticles

from yerba mate waste was recently proposed [32]. Yerba mate (*Ilex paraguairiensis*, Saint Hilaire) tree originates from the subtropical region of South America, and naturally grows in a limited zone within Argentina, Brazil and Paraguay. It is generally consumed as infusion due to its good taste and well-known antioxidant properties [32,33]. Yerba mate is composed from about 35% α-cellulose [34], 25% hemicellulose [34] and 25–30% lignin [34,35]. The presence of lignin results in yerba mate containing different amounts of polyphenols (i.e., caffeic and chlorogenic acids) [33], xanthines (i.e., caffeine and theobromine), flavonoids (i.e., catechin, quercetin, kaempferol and rutin) [36,37], amino acids, saponin and tannins as well as some vitamins (i.e., C, B1, and B2) [36,38]. Nowadays, Brazil is the largest producer of Mate (around 350 kt annually) [39], followed by Argentina, which produced 270 kt in 2019 [40], and Paraguay (around 100 kt annually). Its high consumption leads to the generation of a high amount of yerba mate wastes, since, after being used as infusion, it is wasted without any kind of revalorization [32,41]. Thus, their use for lignocellulosic nanoreinforcements production could not only provide a sustainable revalorization to such waste, as it was already demonstrated for virgin PLA [32], but it could also potentially help to recover the properties of mechanically recycled PLA by developing bionanocomposites with interest in the food packaging field. In fact, as yerba mate is a rich source of polyphenols, which display an antiradical activity similar to pure gallic acid (20 mg/mL) [38], it has gained interest as sustainable additive that could be used to improve and modulate the properties of biopolymers. For instance, yerba mate extract provided a significant improvement of a starchy polymeric matrix stability in acidic and alkaline media [37]. Moreover, yerba mate extract has been added to starch treated by hydrostatic pressure to increase the loading capacity, obtaining interesting carriers for antioxidants, in which the antioxidant activity was maintained after the high pressure treatment without changing the yerba mate polyphenols profile [33]. Similarly, yerba mate has been encapsulated into electrospun zein fibers, improving the thermal stability and proving antioxidant activity, and thus showing interest as antioxidant releasers for food packaging applications [42]. Lignocellulosic yerba mate nanoparticles (YMNs) has also been added to PLA (in 5 wt.%), showing that the high amount of polyphenols protects the polymeric matrix from the thermal degradation during processing, and yielding bionanocomposites with significantly improved mechanical performance, although they showed a somewhat green tonality [32].

The main objective of the present research is to study the effects of lignocellulosic nanoparticles extracted from yerba mate wastes on the properties of mechanically recycled PLA, aiming to revalorize both yerba mate and PLA wastes by developing high-performance bionanocomposites intended for food packaging applications. Yerba mate nanoparticles (YMNs) were obtained by means of an aqueous extraction procedure, followed by two filtration steps, following a previously optimized recipe [32]. Recycled PLA (PLAR) was obtained by subjecting PLA to an accelerated ageing process previously optimized [14], which involved photochemical, thermal and hydrothermal ageing steps, as well as a final demanding washing step to simulate the washing conditions used on an industrial recycling level. The bionanocomposites were prepared by extrusion followed by a compression molding process. The YMNs were previously freeze dried to obtain a powder. Considering the high amount of –OH on the surface of lignocellulosic nanoparticles, which induces high attraction between them, particularly during the freeze-drying process, the nanoparticles were characterized by means of dynamic light scattering (DLS) and Transmission Electron Microscopy (TEM), before and after the freeze-drying process. The structure of the recycled PLA reinforced with yerba mate nanoparticles was characterized using infrared (FTIR) and UV-visible spectroscopic techniques, Differential Sacanning Calorimetry (DSC), Scanning Electron Microscopy (SEM) and intrinsic viscosity (IV) measurements. The effect of the nanoparticles on the thermal stability was measured using Thermogravimetric analysis (TGA), while the mechanical performance was evaluated by nanoindentation measurements. Finally, regarding the potential application in the packaging field, special attention was given to the gas barrier performance, which is of critical importance in food packaging applications. Thus, the permeability to oxygen gas and water vapor of the obtained materials was measured and compared. The results show that the yerba mate nanoparticles can significantly enhance the barrier to oxygen in the recycled material.

### **2. Materials and Methods**

### *2.1. Materials*

PLA, under the commercial name IngeoTM 2003D, was purchased from NatureWorks (Minnetonka, MN, USA). This grade presents a melt mass-flow rate of 6 g/10 min (2.16 kg at 210 ◦C). The yerba mate (*Ilex paraguariensis*) residue was collected after yerba mate infusion (Taragüi, Argentina) consumption in our own laboratory.

### *2.2. Nanoparticle Extraction from Yerba Mate Residues*

The lignocellulosic-based nanoparticles, named yerba mate nanoparticles (YMNs), were obtained from yerba mate infusion wastes following an already developed recipe [32]. In brief, the residue of yerba mate infusion after its consumption was dried in an oven at 60 ◦C for 24 h. Then, 6 g of yerba mate infusion residue were mixed with 200 mL of distilled water and heated up to 100 ◦C under reflux during 60 min with vigorous magnetic stirring (1000 rpm). Next, the solid residue was eliminated by simple filtration, while the obtained mate extract solution was filtered off again (filter paper Whatman Grade 41:20–25 μm particle retention), frozen and further subjected to a freeze-drying process using a Flexi-Dry Freeze Dryer (FTS Systems, Stone Ridge, NY, USA) to obtain a powder as it is schematically represented in Figure 1. The obtained powder of YMNs was stored at 40 ◦C under vacuum during 24 h to remove any moisture before melt compounding process.

**Figure 1.** Schematic representation of mate nanoparticles' extraction procedure.

### *2.3. Preparation of the Samples*

The procedure followed for the ageing and subsequent obtainment of recycled PLA based materials is presented on Figure 2. Firstly, Ingeo 2003D pellets were processed by melt extrusion in a Rondol Microlab counter-rotating twin-screw extruder (Microlab, Rondol, France) with an *L*/*D* ratio of 20. The extrusion process was carried out at 60 rpm, using the following temperature profile (from hopper to die): 125, 160, 190, 190, 180 ◦C. The obtained material was transformed into films (thickness = 200 ± 10 μm) using an IQAP-LAP hot plate press (IQAP Masterbatch Group S.L., Barcelona, Spain) at 190 ◦C. Secondly, the films (PLAV) were subjected to an accelerated ageing process, consisting of the following stages: (i) 40 h of photochemical degradation using an ATLAS UVCON chamber (Chicago, IL, USA), equipped with eight F40UVB lamps; (ii) 468 h of thermal degradation in

an oven at 50 ◦C and (iii) 240 h of hydrolytic degradation in distilled water at 25 ◦C. Thirdly, the aged samples were subjected to a demanding washing process, which was used in previous studies [14], using an aqueous solution of NaOH (1.0 wt.%) and Triton X (0.3 wt.%).

**Figure 2.** Procedure followed for the obtainment of the PLAR-YMN bionanocomposites.

Lastly, the washed material was ground, and melt compounded together with yerba mate nanoparticles, in different proportions, at the same conditions used for the obtainment of PLAV films. Table 1 summarizes the different materials obtained in this study.


**Table 1.** Materials obtained after the recycling process.

### *2.4. Characterization Techniques*

The hydrodynamic size of YMNs were measured by means of a dynamic light scattering (DLS) analyzer. The obtained YMNs, in powder form, were dispersed in water (1 mg mL<sup>−</sup>1) by sonication and further measured at 20 ◦C in a Zetasizer Nano series ZS DLS equipment (Malvern Instrument Ltd., Malvern, UK).

YMNs were also observed by Transmission Electron Microscopy (TEM) in a JEOL JEM-1010 operating (JEOL Ltd., Tokyo, Japan) at 100kV. One droplet of YMNs aqueous suspension (1 mg mL<sup>−</sup>1) was deposited on carbon-coated copper grids and dried at room temperature during 20 min before TEM observation. The nanoparticles' length and width were measured from the TEM images with ImageJ software; the mean and standard deviation of 15 nanoparticle measurements are reported.

Intrinsic viscosity measurements were performed, at 4 different concentrations in chloroform, at 25 ◦C in an Ubbelohde viscometer. All the solutions were filtered prior to the intrinsic viscosity measurement.

UV-Visible spectroscopy tests were conducted in a Varian Cary 1E UV-Vis spectrophotometer (Varian, Palo Alto, CA, USA) equipped with an integrating sphere and using a scanning speed of 400 nm/min. The overall transmittance in the visible region was then calculated according to the ISO 13468 standard.

Fourier transform infrared (FTIR) spectra of the different materials were recorded in Nicolet iS10 spectrometer (Thermo Fisher Scientific, Waltham, MA, USA), equipped with a diamond Attenuated Total Reflectance (ATR) accessory, using 16 scans and a resolution of 4 cm<sup>−</sup>1. The surface crystallinity degree (*X*c) of each nanocomposite was calculated from the absorbance of the band at 955 cm−1, measured in both the amorphous PLAR (*I*0) and the nanocomposite (*If*), using Equation (1) [43]:

$$X\_{\mathcal{C}} = \left(\frac{I\_0 - I\_f}{I\_0}\right) \times 100\% \tag{1}$$

The cryo-fractured surface microstructure of the cross section of each bionanocomposite film was observed by field emission scanning electron microscopy (FE-SEM) in a JEOL JSM 7600F microscope (JEOL Ltd., Tokyo, Japan). Films were previously sputtered with a gold layer to make them conductive.

Differential scanning calorimetry (DSC) scans were performed in a TA Instruments Q20 calorimeter (New Castle, DE, USA). Samples of 5 mg were placed in aluminum pans and subjected to the following protocol (under nitrogen atmosphere): (i) heating from 30 to 180 ◦C at 5 ◦C/min; (ii) isothermal step at 180 ◦C for 3 minutes; (iii) cooling from 180 to 0 ◦C at 5 ◦C/min; (iv) isothermal step at 0 ◦C for 1 min and (v) heating from 0 to 180 ◦C at 5 ◦C/min.

Thermogravimetric analysis (TGA) was conducted on 10 mg samples using a TA Instruments TGA2050 thermobalance (New Castle, DE, USA). The samples were heated from 40 to 800 ◦C at 10 ◦C/min under nitrogen atmosphere. The onset degradation temperature (*T*10) was calculated at 10% of mass loss, and the maximum degradation temperature (*T*max) was determined from the peak of the first derivative of the TGA curve (DTG).

The water vapor transmission rate (WVTR) of the materials was measured, three times, by gravimetry according to the ISO 2525 standard. Thin films (9 ± 2 μm) of the samples were prepared by solvent casting from 0.01 g/mL chloroform solutions. The permeability cups were filled with 2 g of dry silica gel, sealed with the sample film and then placed in a desiccator with a saturated KNO3 solution at 23 ◦C (approximately 90% RH). The cups were weighed each hour for 6 h. WVTR (g/day cm2) was determined using Equation (2):

$$WVTR = \frac{240\*(m\_t - m\_0)}{A\*t} \tag{2}$$

where *mt* is the mass of the cup at time *t*, *m*<sup>0</sup> is the initial mass of the cup and *A* is the exposed area of the film.

Oxygen permeability tests were conducted at 30 ◦C in a homemade permeation cell, using a gas pressure of 1 kPa.

Nanoindentation tests were carried out using a Shimadzu DUH-211S dynamic Ultra-Microhardness Tester (Shimadzu Corporation, Kyoto, Japan), equipped with a Berkovich indenter. The measurements were conducted at room temperature (24.5 ± 0.5 ◦C), using a maximum load of 10 mN and a loading rate of 1.4632 mN/s. Maximum load was held for 5 s, and then it was retired. Each measurement was replicated 6 times.

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

### *3.1. Yerba Mate Nanoparticles' Characterization*

The obtained mate extract solution after filtration was analyzed by DLS (Figure 3a), showing a monomodal size distribution from 85 to 103 nm, with an average size of 94 nm. There is a shoulder at higher sizes, around 500 nm, which has been related with the formation of agglomerates [32]. Considering that the DLS technique is designed to calculate the hydrodynamic diameter of spherical particles, the nanosize as well as the morphological aspect of the YMNs were further examined by TEM (Figure 3b). Individualized lignocellulosic 2D YMNs were observed. The yerba mate solution was then freeze-dried to obtain a powder, obtaining a yield of 19% ± 5%, which is in good agreement with previously reported work [32]. The size analysis of the obtained powder of yerba mate nanoparticles

was also carried out by DLS (Figure 3c) and it revealed a monomodal size distribution with a dimension ranging from 450 to 545 nm, with an average size of 495 nm. From the TEM image of YMNs powder (Figure 3d), it can be seen that the YMNs tend to agglomerate due to the natural tendency of both, lignin and cellulose, to re-agglomerate and form strong hydrogen bonds as the water sublimate during freeze-drying process [44,45]. Nevertheless, they still showed sub-micron size with dimensions of 525 ± 136 in length and 302 ± 96 nm in width (see zoom in Figure 3d).

**Figure 3.** YMNs solution: (**a**) DLS measurements and (**b**) TEM images. YMNs powder: (**c**) DLS measurements and (**d**) TEM images.

### *3.2. Structure and Morphology of the PLA-YMN Bionanocomposites*

The FTIR spectra of YMNs, PLAR and PLAR-3YMN are reported in Figure 4. The broad absorption band in the range of 3000–3700 cm−<sup>1</sup> present in YMNs can be ascribed to the stretching vibration of the −OH groups in lignin as well as in cellulose molecules. The successful hemicellulose removal from yerba mate residue was confirmed by the absence of the band around 1730 cm−<sup>1</sup> in YMNs [32], which corresponds to the acetyl and ester groups in hemicelluloses [46]. The spectrum that corresponds to PLAR-3YMM bionanocomposite shows a broad band at 3320 cm−<sup>1</sup> (stretching vibration of the –OH groups) and a shoulder at 2920 cm−<sup>1</sup> (C–H stretching vibration) (Figure 4a) that confirm the successful incorporation of YMNs in the recycled PLA [32]. The stretching vibration of the carbonyl group (–C=O) of PLA appears at 1750 cm−<sup>1</sup> (Figure 4b) [47]. Moreover, the FTIR-ATR spectra of the bionanocomposites show very slight changes in the intensity of the bands at 920 and 956 cm−<sup>1</sup> (Figure 4b). These absorptions have been assigned to skeletal C–C stretching mode coupled with CH3 rocking one [48–50]; while the band centered at 920 cm−<sup>1</sup> corresponds to the 103 helix chain conformation, characteristic of the crystalline forms, the band at 956 cm−<sup>1</sup> is assigned to the amorphous phase. In this work, the crystallinity degrees in the surface of the nanocomposites were calculated from the absorbances of the band at 955 cm−<sup>1</sup> in the different materials, using Equation (1). The values

obtained, 16.7% for PLAR-3YMN and 12.5% for PLAR-1YMN indicate that the YMNs act as nucleating agents for recycled PLA.

**Figure 4.** FTIR spectra of YMNs, recycled PLA (PLAR) and PLAR-3YMN bionanocomposite: (**a**) in the 4000–2500 cm−<sup>1</sup> region and (**b**) in the 1900–600 cm−<sup>1</sup> region.

The effect of the addition of YMNs on the microstructure of mechanically recycled PLA was studied by means of SEM analysis. Neat virgin PLA (Figure 5a) shows the typical regular and smooth surface of an amorphous polymer. PLAR (Figure 5b) shows a very similar behavior than that of neat PLA (Figure 5a) with a rather more ductile pattern. This more plastic behavior could be ascribed to the already commented chain scission reactions that take place during the accelerated ageing and mechanical recycling, because the shorter polymer chains formed in these degradation processes plasticize the polymeric matrix. Meanwhile, PLAR-YMN bionanocomposites (Figure 5c,d) exhibited a rougher surface due to the YMNs' reinforcing effect, as it has been already observed in virgin PLA blended with lignocellulosic nanoparticles [23,32].

**Figure 5.** SEM observations of: (**a**) PLAV, (**b**) PLAR, (**c**) PLAR-1YMN and (**d**) PLAR-3YMN. (10,000× magnifications).

The fracture surface depends on the concentration of YMNs (Figure 5c,d). In fact, in PLAR-1YMN bionanocomposite (Figure 5c), YMNs appear uniformly dispersed, with no phase separation between the nanoparticle and the polymeric matrix. However, in PLAR-3YMN (Figure 5d) some micro-holes can be seen, thus suggesting that YMNs in bionanocomposite containing 3 wt.% of YMNs show poor interfacial adhesion with PLAR matrix. Micro-holes have been already observed in virgin PLA reinforced with lignin nanoparticles and have been related to the formation of YMNs' aggregates during bionanocomposite processing [29].

### *3.3. Properties of the PLA-YMN Bionanocomposites*

### 3.3.1. Effect of the Addition of YMNs on the Intrinsic Viscosity

Intrinsic viscosity is related to the molecular weight of PLA, which plays a very important role in the final thermal and barrier properties of the materials. Furthermore, intrinsic viscosity is important from a processing point of view since industrial forming processes are frequently designed to operate at specific IV values. Thus, in order to get information regarding the effect of YMNs on the processing of PLAR-based bionanocomposites, the values of the intrinsic viscosity (IV) of PLAR in all the samples was determined by dissolving each sample in chloroform, followed by a filtration step to eliminate the YMNs. In accordance with previous works, Figure 6 shows that PLAR has an intrinsic viscosity around 14% lower than PLAV due to the degradation experimented [12,16] during the accelerated ageing, washing and reprocessing steps.

**Figure 6.** Intrinsic viscosity values of the different samples.

Regarding the effect of the addition of yerba mate nanoparticles, Figure 6 shows that the material with only 1 wt.% of YMNs presents an intrinsic viscosity value 12% lower than PLAR. However, the sample with 3 wt.% of YMNs shows an intrinsic viscosity higher than that of the unfilled recycled material. This behavior might suggest that the addition of the nanoparticles produces two counteracting effects on the intrinsic viscosity of recycled PLA. On the one hand, the high hydrophilicity of the yerba mate nanoparticles might cause the absorption of small amounts of water during processing, which could result in a significant hydrolytic degradation of PLA during melt compounding. A similar behavior was observed in other PLA/lignocellulosic filler composites. For instance, Arrieta et al. [51] observed the reduction in the molecular weight of PLA bionanocomposites in virgin PLA reinforced with cellulose nanocrystals. Similarly, Way et al. [52] reported that PLA filled with lignocellulosic fibers showed a more severe degradation during melt processing than its unfilled counterpart. On the other hand, the antioxidant nature of yerba mate nanoparticles (due to the presence of phenolic

compounds) could contribute to reduce the degradation of the polymer during extrusion, as it has been pointed out by Arrieta et al. [32] in a previous work. The results shown on Figure 6 indicate that at lower concentrations of YMNs, the negative effect of the hydrolysis prevails; however, at higher concentrations, the effect of the higher amounts of antioxidant compounds present in YMNs is more important, resulting in materials with higher IV values.

### 3.3.2. Thermal Properties

The effect of the addition of YMNs on the thermal properties of mechanically recycled PLA was studied by means of DSC and TGA. Figure 7 and Table 2 summarize the DSC results obtained for the different materials. As can be seen in Figure 7, PLAV show the characteristic thermal transitions of PLA: (i) a glass transition (*T*g) around 60 ◦C; (ii) a broad cold crystallization exothermic peak (*T*cc) above 100 ◦C and (iii) a melting endotherm (*T*m) centered at 150 ◦C. As for the behavior of PLAR, Figure 7 and Table 2 show that it has, overall, the same thermal transitions as PLAV. However, there are some noteworthy differences. Firstly, PLAR shows a narrower cold crystallization peak, which is also located at temperatures 15 ◦C lower than the virgin material. This difference could be attributed to the degradation of PLA during mechanical recycling, since the shorter polymer chains have increased mobility and crystallize more easily [14]. This behavior is also reflected in the higher values of the cold crystallization and melting enthalpies (Δ*H*cc and Δ*H*<sup>m</sup> respectively) of PLAR. Secondly, Figure 7 shows that there are differences in the melting endotherm of the recycled material, since PLAR shows two well-defined melting peaks. This behavior has been reported in previous studies [53], and it has been attributed to the occurrence of a melt recrystallization phenomenon. Such a phenomenon consists of the melting of less ordered crystalline domains at lower temperatures, their rearrangement into crystalline structure as the temperature increases and a final melting of the more ordered crystals at a higher temperature. The fact that PLAR shows this behavior could also be explained by the degradation of the polymer during the recycling process. The shorter polymer chains present in PLAR could rearrange themselves during heating more easily due to their increased mobility and hence form more stable crystalline structure, which melt at higher temperatures.

**Figure 7.** DSC scans corresponding to the second heating of the materials.


**Table 2.** DSC (second heating) results as well as TGA parameters of the different materials.

Regarding the effect of the presence of the yerba mate nanoparticles, both Figure 7 and Table 2 show that the thermal behavior of PLAR-1YMN and PLAR-3YMN are very similar to that of mechanically recycled PLA. However, some differences can be seen in the cold crystallization temperature. Both Figure 7 and Table 2 show that the addition of the nanoparticles leads to a slight decrease in *T*cc values. This behavior suggests that yerba mate nanoparticles act as nucleating agents, promoting the cold crystallization of PLA at lower temperatures, as it was seen by means of FTIR-ATR. The nucleating effect of different organic-based fillers has been previously reported by different authors, such as Fortunati et al. [44], Arrieta et al. [23] and Lizundia et al. [54] in PLA/cellulose nanocrystals bionanocomposites; as well as by Beltrán et al. [3] who studied recycled PLA/silk fibroin nanoparticles nanocomposites. It is also worth noting that PLAR-3YMN presents a melting behavior closer to PLAR than to PLAV, despite its higher IV value. This could also be explained by the nucleating effect of YMNs, since it allows for the occurrence of the melt recrystallization mechanism, despite the limited mobility of the longer polymer chains present in PLAR-3YMN. Nevertheless, the observed differences are rather small, thus indicating that the effect to the yerba mate nanoparticles on the thermal transitions of recycled PLA is limited.

The effect of the recycling process as well as the addition of YMNs on the thermal properties of PLAR was also investigated by dynamic TGA measurements. The weight loss (TGA) and derivative (DTG) curves of virgin PLA (PLAV), recycled PLA (PLAR) and PLAR-YMNs bionanocomposites are reported in Figure 8, while the thermal parameters obtained from these curves are summarized in Table 2. All samples show a one-step degradation processes. While virgin PLA (PLAV) shows the highest maximum onset degradation temperature (*T*<sup>10</sup> = 334.2 ◦C), PLAR-based samples presented a decrease in the thermal stability, as shown by the decrease in the onset degradation temperature, which has been ascribed to the presence of shorter polymer/oligomeric chains with lower thermal stability [14]. These results are in good agreement with the already commented reduction in the molecular weight when discussing the intrinsic viscosity measurements. In this context, Burgos et al. developed different PLA formulations plasticized with oligomeric lactic acid (OLA) and also observed a reduction of the onset degradation temperature, which decrease with increasing amounts of OLA [55]. The incorporation of 1 wt.% of YMNs did not promote significant changes in the thermal behavior of PLAR. While the *T*<sup>10</sup> values slightly increased, the *T*max remained almost invariable. However, with the incorporation of 3 wt.% of YMNs, both the *T*<sup>10</sup> and the *T*max decreased. These results may seem rather surprising considering that the material with 3 wt.% YMNs has a higher IV than PLAR. However, similar findings were observed by Fortunati et al. in virgin PLA reinforced with 1 and 3 wt.% of cellulose nanocrystals. They reported that the thermal stability of PLA decreased as the nanocellulose content increased and ascribed this behavior to the lower thermal stability of the cellulose nanocrystals (maximum degradation rate at about around 291 ◦C) [30]. The DTG curve of the YMNs used in this work, which has been previously reported and analyzed [32], shows two maxima at 215 and 315 ◦ C, below the main PLA degradation temperature, so that the presence of 3 wt.% YMNs could explain the decrease in the thermal stability of the nanocomposite.

**Figure 8.** Dynamic (**a**) TGA and (**b**) DTGA curves of binary PLA nanocomposite films.

### 3.3.3. Optical Properties

The visual appearances of virgin PLA, recycled PLA and YMN-reinforced bionanocomposites are shown in Figure 9a. From the visual appearance of the films, it is possible to observe that the recycled PLA remains transparent, with no apparent differences with virgin PLA. Meanwhile, bionanocomposites presented a somewhat green tonality, which was more evident in the case of the bionanocomposite with a higher amount of YMNs (PLAR-3YMN). In a previous work, virgin PLA has been reinforced with 5 wt.% of similar yerba-mate-based lignocellulosic nanoparticles and the developed films presented a brown tonality [32]. The transmission values in the visible (400–800 nm) and UV region of the spectra were determined by using UV-Vis spectroscopy (Figure 9b,c).

**Figure 9.** PLAV, PLAR and bionanocomposites (PLAR-1YMN and PLAR-3YMN): (**a**) visual appearance, (**b**) UV-vis spectra and (**c**) visible light transmission rates.

The spectra show that films obtained from PLAV and PLAR are highly transparent in the visible region. In good agreement with the visual appearance of the films, the spectra show that the presence of YMNs leads to significant decreases in the visible light transmission (Figure 9b). The overall transmission rate in this spectral region falls from values higher than 90% in PLAV and PLAR to values clearly below 80% in biocomposites (78.6% and 76.1% of light transmission rates in PLAR-1YMN and PLAR-3YMN, respectively), although these materials remain transparent (Figure 9c). The presence of lignocellulosic aggregates in PLAR-3YMN decreased the visible light transmittance of the PLAR-based film, in good agreement with SEM analysis. Similar results have been observed in PLA/lignin nanoparticles bionanocomposites [31].

It is worth noting that PLAR shows lower UV light transmission than PLAV, with the appearance of a small absorption peak centered at 277 nm. This band is related with the formation of chain-end carboxyl groups, as a consequence of the degradation of the polymer that take place during the recycling process [20,56]. In the case of YMN-reinforced recycled plastics, this region is overlapped with different absorptions due to the polyphenols (i.e., chlorogenic acid, caffeic acid and rutin [33]) present in YMNs.

The above spectra reveal that YMNs produce a strong UV blocking effect in the recycled PLA matrix. Other authors have already reported the UV blocking effect in virgin PLA reinforced with different lignocellulosic nanoparticles, such as in PLA/lignin nanoparticles bionanocomposites [31,57]; PLA/cellulose nanocrystals nanocomposites [51] and also in virgin PLA reinforced with 5 wt.% of similar yerba-mate-based lignocellulosic nanoparticles [32]. In the case of PLAR-1YMN film, the presence of only 1 wt.% of YMNs was able to block around 90% of UV-B and C, and this UV blocking effect was more marked in PLAR-3YMN, as could be expected.

In summary, it can be said that the addition of YMNs nanoparticles to the recycled PLA, in a proportion less than or equal to 3 wt.%, has an overall positive effect on the optical properties of the material. On the one hand, the transparency in the visible region is reduced, but the sheets of these bionanocomposites remain transparent, which is important in many cases of food packaging, because seeing the packed food through the packaging film is one of the most important requirements for consumers' acceptance. On the other hand, the presence of YMNs greatly reduces UV transmission, thus slowing down the degradation of the contents of the container.

### 3.3.4. Barrier Properties

The barrier properties against different gases are very important in food packaging applications, which is the most important market for PLA. Therefore, the effect of the addition of the YMNs on the gas barrier properties of mechanically recycled PLA was measured; the main results are reported in Figures 10 and 11.

Figure 10 shows the WVTR of the different samples. The obtained values are similar to those reported in previous studies for PLA based materials [58]. It can be seen that mechanical recycling led to a slight increase in the WVTR of PLA. To explain this behavior, one should consider that the gas permeability, and hence the WVTR, of semicrystalline polymers depends on two factors: the diffusion coefficient and the solubility of the gas. These factors are affected by the molecular weight, structure and free volume of the polymer and by the temperature and nature of the gas molecules [12,29]. The observed increase in the WVTR of the mechanically recycled PLA could be related to the generation of terminal carboxyl and hydroxyl groups during the ageing and mechanical recycling, which decreases the hydrophobic character of the polymer, thus facilitating the passage of water vapor through the films. Regarding the effect of the nanoparticles, it can be observed that the nanocomposites show higher WVTR values than both unfilled PLAV and PLAR samples, which could be explained by the hydrophilic nature of the YMNs due to the high amount of –OH groups. In this context, Kim et al. [59] studied the WVTR of PLA reinforced with pristine lignin and acetylated lignin, reporting higher WVTR values for the PLA-lignin composites in comparison with neat PLA. This behavior was ascribed to the hydrophilic nature of pristine lignin. Meanwhile, acetylated lignin-based composites were able

to decrease the WVTR values of neat PLA. Similarly, Espino-Perez et al. [60] developed PLA loaded with cellulose nanowhiskers (5, 14 and 30 wt. %), reporting that WVTR increased with the cellulose nanowhiskers content. This behavior was related to the hydrophilic nature of cellulose structures. In this work, PLAR-1YMN, the material with the lower amount of hydrophilic YMNs, shows higher WVTR than PLAR-3YMN, which can be related to the lower viscosity observed in PLAR-1YMN. This low viscosity, due to a stronger degradation, indicates the presence of more hydrophilic terminal groups in the polymer chains, which can explain the higher value of WVTR.

**Figure 10.** Water vapor transmission rate of the different materials.

**Figure 11.** Oxygen permeability of the different materials.

Figure 11 presents the oxygen permeability coefficient, measured in Barrer (1 Barrer = 3.35·10−<sup>16</sup> mol m/m2 s Pa), of the different samples. It can be observed that the ageing and the mechanical recycling cause only a slight decrease in the oxygen permeability of PLA, despite the degradation observed by means of IV measurements. Similar results have been reported in a previous study [14] and have been attributed to the presence of two counteracting effects of the mechanical recycling on the permeability of PLA. On the one hand, the presence of shorter polymer chains might reduce the free volume inside the polymer, due to their better ability to rearrange themselves, reducing the diffusion coefficient. On the other hand, the generation of terminal –COOH and –OH groups during the ageing and recycling lead to an increase in the affinity between the polymer and the gas molecules, increasing the solubility of the gas into the polymer. The concurrence of these counteracting effects leads to the overall small changes observed in the oxygen permeability.

As for the behavior of the bionanocomposites, Figure 11 shows that the oxygen permeability is significantly reduced with the addition of 1 wt.% of YMNs. The reduction of the oxygen permeability due to the incorporation of cellulose nanocrystals have been already observed in virgin PLA/cellulose nanocrystals based bionanocomposites [51,61]. This behavior could be explained by the barrier effect caused by the dispersion of the nanoparticles in the polymer matrix, which leads to an increase in the tortuosity of the diffusion path traveled by the gas going through the polymer film. However, the oxygen permeability increased, reaching values close to those of unfilled PLA, when the amount of YMNs was 3 wt.%. It is well known that the tortuosity of the diffusion path depends on several factors (i.e., shape and aspect ratio of the filler, degree of dispersion or exfoliation, filler loading and orientation, adhesion to the matrix, moisture activity, filler-induced crystallinity, polymer chain immobilization, filler-induced solvent retention and porosity) [61]. Thus, this result could be due to the poor dispersion of the nanoparticles in the PLAR-3YMN sample, as was observed in SEM photographs. The poor dispersion of the nanoparticles might result in the formation of micro-pores in the polymer matrix, which act as low-resistance paths for the gas diffusion through the polymer. Therefore, this result underlines the success of the dispersion of low amounts of YMNs (1 wt.%) during melt-compounding process and its reinforcement effect produced in the final formulation.

### 3.3.5. Mechanical Properties

Mechanical properties play a very important role in food packaging applications; consequently, nanoindentation tests were conducted to determine the effect of the addition of YMNs on mechanically recycled PLA. Figure 12 shows the indentation hardness and the Young modulus of the different materials. It can be seen that both hardness and modulus values are in good agreement with those found in the literature for PLA samples [62,63]. It could also be seen that mechanical recycling led to a slight decrease in the mechanical properties of PLA, due to the degradation of the polymer during the ageing, washing and reprocessing steps. Similar results have been reported in previous works [14,62], who found that mechanical recycling led to small decrease in the mechanical properties of PLA.

**Figure 12.** Indentation hardness (**a**) and Young modulus (**b**) of the different samples.

Regarding the effect of the addition of the YMNs, it can be seen that samples with 1 wt.% YMN and 3 wt.% YMN show slightly higher values for hardness and modulus than unfilled PLAR. This result suggests that the presence of the YMNs nanoparticles has a reinforcing effect on the recycled PLA matrix. Similar trends have been reported in other PLA nanocomposites, for instance, Zaidi et al. [63,64] reported increases in both indentation hardness and the Young modulus with the addition of low amounts of organically modified montmorillonite. It is worth noting that, despite the overall improvement of the mechanical properties of recycled PLA with the addition of YMNs, better results are observed in the material with only 1 wt.% YMN. This behavior agrees with that observed in the oxygen permeability measurements and highlights the relevance of the better dispersion of lower amounts of YMNs.

### **4. Conclusions**

The effect of the addition of lignocellulosic nanoparticles extracted from food waste, specifically yerba mate waste, on the structure, mechanical and barrier properties of mechanically recycled PLA (PLAR) was studied. PLAR was obtained by subjecting a commercial grade of PLA to accelerated ageing followed by mechanical recycling. Lignocellulosic yerba mate nanoparticles (YMNs) were extracted from yerba mate waste in an aqueous extraction process and added to PLAR in the reprocessing step at two levels (1 and 3 wt.%). FTIR and SEM analysis confirmed the successful incorporation of YMNs into the PLAR matrix.

Ageing and mechanical recycling cause the degradation of PLA, leading to a decrease in the molecular weight, thermal stability and barrier performance. The addition of small amounts of YMNs significantly modifies some properties of the material, depending on the YMNs content. The nanoparticles act as nucleating agents, thus facilitating the crystallization of PLAR, without significantly reducing the average molecular weight. Although the nanoparticles slightly reduce the thermal stability of the material, due to their lower thermal stability, the material remains stable under processing conditions. Bionanocomposites with 1 wt.% of YMNs show a good dispersion of the nanoparticles; however, when the YMNs' content rises up to 3 wt.%, although no phase separation was detected, YMNs tend to aggregate, inducing the formation of micro-voids. Thus, the addition of only 1 wt.% YMNs improved the mechanical performance and reduces oxygen permeability, a key property in food packaging materials. However, if the YMNs content rises to 3%, the effect on the oxygen barrier is negative, due to dispersion problems and the formation of micro-voids. In general, the incorporation of YMNs increases the water vapor transmission rate, due to the hydrophilic character of the nanoparticles. As for light transmission, another key property in food packaging, the addition of YMNs slightly reduces transmission in the visible region, but the recycled material remains transparent. However, nanoparticles dramatically reduce transmission in the UV areas of the spectrum, which can help slow down the degradation of the container's content.

Overall, the results obtained indicate that the addition of yerba mate nanoparticles could lead to obtaining recycled PLA with good properties for the intended use and with significant improvements in some key properties, such as the barrier to UV light and oxygen. Considering that these nanoparticles are also obtained from a food residue and using an environmentally friendly extraction process, the use of YMNs could be the basis of a useful and potentially competitive method to improve the recyclability of PLA and other similar polymers.

**Author Contributions:** Conceptualization, F.R.B., M.P.A. and J.M.U.; methodology, F.R.B., M.P.A., G.G. and M.U.d.l.O.; formal analysis, F.R.B., M.P.A., G.G., M.U.d.l.O. and J.M.U.; investigation, F.R.B., M.P.A. and G.G.; writing—original draft preparation, F.R.B. and M.P.A.; writing—review and editing, F.R.B., M.P.A., M.U.d.l.O. and J.M.U.; supervision, M.U.d.l.O. and J.M.U.; project administration, M.U.d.l.O. and J.M.U.; funding acquisition, M.U.d.l.O. and J.M.U. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was founded by European Union's Horizon 2020 research and innovation program under grant agreement No. 860407 BIO-PLASTICS EUROPE; MINECO-SPAIN under project CTM2017-88989-P and Universidad Politécnica de Madrid under project UPM RP 160543006.

**Acknowledgments:** The authors thank the staff of the ICTS National Center for Electron Microscopy (CNME), UCM, Madrid (Spain) for their assistance with transmission and electron scanning microscopy.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article Cucumis metuliferus* **Fruit Extract Loaded Acetate Cellulose Coatings for Antioxidant Active Packaging**

### **Marina Patricia Arrieta 1,\*, Luan Garrido 2, Simón Faba 2, Abel Guarda 2, María José Galotto <sup>2</sup> and Carol López de Dicastillo 2,\***


Received: 12 May 2020; Accepted: 28 May 2020; Published: 29 May 2020

**Abstract:** A new active coating was developed by using *Cucumis metuliferus* fruit extract as antioxidant additive with the aim of obtaining an easy way to functionalize low-density polyethylene (LDPE) films for food packaging applications. Thus, an extraction protocol was first optimized to determine the total phenolic compounds and the antioxidant activity of CM. The aqueous CM antioxidant extract was then incorporated into cellulose acetate (CA) film-forming solution in different concentrations (1, 3 and 5 wt.%) to be further coated in corona-treated LDPE to obtain LDPE/CA-CM bilayer systems. CA and CA-CM film-forming solutions were successfully coated onto the surface of LDPE, showing good adhesion in the final bilayer structure. The optical, microstructural, thermal, mechanical and oxygen barrier performance, as well as the antioxidant activity, were evaluated. The active coating casted onto the LDPE film did not affect the high transparency of LDPE and improved the oxygen barrier performance. The antioxidant effectiveness of bilayer packaging was confirmed by release studies of *Cucumis metuliferus* from the cellulose acetate layer to a fatty food simulant. Finally, the LDPE/CA-CM active materials were also tested for their application in minimally processed fruits, and they demonstrated their ability to reduce the oxidation process of fresh cut apples. Thus, the obtained results suggest that CA-CM-based coating can be used to easily introduce active functionality to typically used LDPE at industrial level and enhance its oxygen barrier, without affecting the high transparency, revealing their potential application in the active food packaging sector to extend the shelf-life of packaged food by prevention of lipid oxidation of fatty food or by prevention fruit browning.

**Keywords:** *Cucumis metuliferus*; extraction; antioxidant activity; coating; cellulose acetate; LDPE; bilayer packaging; active packaging

### **1. Introduction**

Increasing ecological concern aimed towards a reduction of the environmental impact of short-term plastics (i.e., packaging, disposable cutlery, agricultural mulch films, etc.) is contributing to a move towards a circular economy model, in which a more sustainable plastic industry is continuously developing. The deliberate introduction of bio-based and biodegradable plastics in the field of packaging will make it possible to reduce the consumption of non-renewable petrochemical-based resources, as well as to prevent the accumulation of plastics waste in the environment (i.e., landfills, oceans, etc.) within the frame of the circular economy [1,2].

Cellulose acetate (CA) is a thermoplastic biodegradable polymer extensively studied for packaging applications owing to its excellent optical transparency and high toughness, and because it has the advantage of being produced primarily through the esterification of the renewable and most abundant polymer in nature: cellulose [3–5]. In fact, among cellulosic derivates, CA is extremely attractive for the packaging field mainly because of its low price, good biodegradability and non-toxicity, as well as due to its having a better processability than cellulose, as it can be processed either by solvent-casting, melt-blending or electrospinning approaches [6–8]. Moreover, CA has been widely used for the development of active packaging materials by means of the incorporation of active compounds (i.e., antioxidant and antimicrobial substances) into the CA polymeric matrix [9–11]. Active food packaging has advantages over direct addition of active compounds into the foodstuff, such as the use of additives in lower concentrations, as well as extension of shelf life due to the controlled release of active compounds during storage [12,13]. Moreover, it has an important effect on the reduction of deterioration reactions that begin at the food surface due to a more significant interaction with the surface of the packed food [11,14,15]. Additionally, there is a growing trend in the food packaging industry to replace synthetic additives with natural antioxidants in both petrochemical-based [16–19] and bio-based and biodegradable polymers [10,20–22], particularly with tocopherol, catechols, essential oils, and plant extracts [16,20,23].

In this regard, coatings are progressively becoming more widely recognized as a powerful tool for extending the shelf life of food by improving many properties of plastic materials as well as a simple method for providing specific active functions (e.g., antioxidant, antimicrobial, etc.) to the final food packaging [24]. In fact, the addition of active compounds into food coatings or packaging coatings has some advantages, as they act only at the surface level, and can be applied at any stage of the food supply chain [13,21]. In the food packaging sector, coating technology represents the most efficient and affordable solution for attaining high barrier properties against oxygen for light packaging, particularly in the case of polyolefins (i.e., polypropylene (PP) and polyethylene (PE)) [25]. In this regard, the application of CA-based active coatings on a typical film packaging material (e.g., low-density polyethylene (LDPE)) is an advantageous alternative for easily providing the final material with specific active performance. In fact, LDPE is one of the most widely consumed polymers in the food packaging field; it is extensively used in film to cover foodstuff due to its low cost, high resistance to tearing, low heat seal temperature, and high water barrier, as well as its high production efficiency [2,19,26]. However, it presents high oxygen permeability, which is a crucial property for plastic food packaging films [27]. Several strategies have been explored to improve the LDPE barrier performance for food packaging applications through blending [28,29], development of nanocomposites [30,31], or by using multi- or bi-layer approaches (i.e., surface coatings, sandwich structures, electrospun deposition, etc.) [25,32–35]. Coating approaches are of high interest since they make it possible to obtain a bi-layer structure using an easy, scalable and cost-effective method at an industrial level. Thus, applying a CA-CM layer to commercially available LDPE films by a simple coating process has the potential to reduce the oxidation process of packed food, providing the CA with better intrinsic oxygen barrier performance than LDPE, as well as offering the additional advantage of giving active packaging technology to the final formulation by simply incorporating antioxidant compounds into the CA-based film-forming solution.

*Cucumis metuliferus* (Cucurbitaceae) is an annual climber plant, native to Africa, that grows specifically from South Africa to tropical Africa [36]. It is known for its potential benefits to human health, and it has been suggested that it possesses antifungal, antimicrobial, antiviral and antioxidant effects, as well as chelating power [37,38]. It is called African homed melon, jelly melon and kiwano. The commercial culture of *C. metuliferus* began in New Zealand, where it was commercialized as an exotic fruit during the eighties. The main commercial advantages of *C. metuliferus* are that it grows rapidly and remains in good condition for around 6 months without cold storage [39,40]. For this reason, the commercialization was extended, and nowadays it grows in New Zealand, Australia, Chile, Argentina, Venezuela, Spain, Portugal, Germany, Italy, Israel and California [37,40]. Young *C. metuliferus* fruit is dark green with mottled light green spots, while as it ripens it becomes bright orange with

very sharp spines [36]. In the interior is a mass of green, translucent, slightly mucilaginous juice-sacs enclosing many tightly packed, flat seeds [39]. Although the antioxidant ability of *C. metuliferus* has been determined [36,38], to the best of our knowledge its use in the development of antioxidant active packaging coating has not yet been proposed.

The main goal of the present research work was to assess the potential production of antioxidant active coatings for food packaging proposes based on cellulose acetate loaded with *C. metuliferus* fruit extract (CM). Initially, the extraction of antioxidant agents from *C. metuliferus* fruit was optimized by evaluating extraction procedures using different solvents: water, ethanol and ethanol 50%. Then, the obtained extract was incorporated into a cellulose acetate solution in different proportions (1, 3 and 5 wt.%). While coatings require substrates with high surface energy, LDPE is known to possess low surface energy as a consequence of its non-polar nature. Hence, LDPE is frequently surface treated to promote good adhesion between the polyolefin and the coating [33,41]. Thus, the obtained CM-functionalized CA film-forming solutions were coated onto commercial LDPE films. The obtained bi-layer structures were fully characterized considering the intended use in the active food packaging field. Thus, the correct adhesion of CA coating into corona-treated LDPE film was corroborated by scanning electron microscopy (SEM). The effect of the CA-based coating on the optical properties of LDPE was investigated by UV-visible measurements and the determination of colorimetric properties in the CIELab space. The mechanical and barrier performances were also evaluated with the aim of assessing their suitability for the food packaging sector. Finally, since these materials are intended for active food packaging applications, the release ability of the antioxidant compounds of *C. metuliferus* fruit from bilayer materials was analyzed in a fatty food simulant, as well as in direct contact with fresh-cut apples in order to get information of the possible application of these sustainable materials at an industrial level intended for both fatty food and fresh fruit protection.

### **2. Materials and Methods**

### *2.1. Materials*

Cellulose acetate with Mn = 30,000 and 39.8 wt.% acetyl content (CA degree of substitution = 2.5 [42]) was supplied by Sigma-Aldrich (Santiago, Chile). A commercial corona-treated low-density polyethylene (LDPE) film was kindly supplied by EDELPA (Santiago, Chile). The *Cucumis metuliferus* fruits were obtained at a local market in Santiago de Chile. 2,2-azinobis(3-ethylbenzothiazoline-6 sulphonate) (ABTS), Folin Ciocalteu phenol reagent, anhydrous sodium carbonate, gallic acid and 6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid (Trolox) were purchased from Sigma Aldrich. Acetone (99.9% HPLC grade) and absolute ethanol (99.9% HPLC grade) were supplied by Merck (Santiago, Chile).

### *2.2. Methods*

### 2.2.1. *Cucumis metuliferus* Extraction Optimization

The *C. metuliferus* fruits were cut into slices and dried at 40 ◦C for 48 h. They were mechanically grinded to obtain powder by means of a cutter grinder. With the aim of obtaining the CM extract with highest antioxidant performance, it was extracted from *C. metuliferus* fruit powder using three different solvent systems: water, ethanol and 50% aqueous ethanol (v/v, EtOH 50%). Approximately 500 mg of *C. metuliferus* fruit powder was dispersed in 30 mL of each solvent and vigorously stirred for 180 min at 40 ◦C. The obtained viscous extracts were then filtered twice and used for the determination of radical scavenging activity and the measurement of total phenolic content (PC). Figure 1 show the schematic representation of *C. metuliferus* extract (CM) extraction procedure from *Cucumis metuliferus* fruit.

The total phenolic content (PC) of the *C. metuliferus* extract was colorimetrically determined by means of the Folin-Ciocalteu method according to the methodology adapted in previous work by Lopez de Dicastillo et al. [10]. In brief, 0.2 mL of Folin-Ciocalteu reagent and 3.1 mL of distilled water were mixed with 0.1 mL of each CM extract and kept in darkness for 5 min. Then, 0.6 mL of anhydrous Na2CO3 20% (w/v) was added, shaken and then kept in the dark for 2 h. The PC was

determined by means of the absorbance at 765 nm in a UV-vis spectrophotometer. The measurements were performed in triplicate and the results were expressed as mg gallic acid equivalent (GAE) per 100 g of dried sample.

**Figure 1.** Schematic representation of antioxidants extraction procedure from *Cucumis metuliferus* fruit.

The evaluation of the antioxidant ability of *C. metuliferus* fruit extract in the three different solvents assayed was determined by means of Ferric Reducing Antioxidant Power (FRAP), as well as radical 2,2 -azinobis(3-ethylbenzothiazoline-6-sulphonate (ABTS·+) methods, since they are simple, inexpensive and robust techniques [10,43]. FRAP follows a Single Electron Transfer (SET) method, and thus detects the ability of the antioxidant to transfer one electron to deactivate the reactive functional group of ferric 2,4,6-tripyridyl-s-triazine (TPTZ) [12]. In fact, FRAP measures the reduction to a blue ferrous product, which is absorbed at 593 nm [12,44]. Meanwhile, the ABTS method monitors the inhibition of oxidation of a suitable substrate, which may be neutralized either by direct reduction via SET or by radical quenching via Hydrogen Atom Transfer (HAT) [12,44]. HAT-based methods measure the antioxidant ability to quench free radicals by hydrogen donation [44]. The inhibition of the cationic radical ABTS·<sup>+</sup> due to the presence of antioxidant compounds from *C. metuliferus* fruit extract was followed by the reduction of the characteristic wavelength absorption spectrum at 715 nm [16,45]. FRAP and ABTS methods were performed in triplicate, and results were expressed as Trolox equivalents per 100 g of fruit sample.

### 2.2.2. Preparation of *Cucumis metuliferus*-Loaded Cellulose Acetate-Coated LDPE Films

Cellulose acetate (CA) coating was prepared by dissolving 6 g of CA in 40 mL of acetone (0.15 g/mL) at 50 ◦C under stirring. The antioxidant cellulose acetate coating (CA-CM) was prepared by adding 1, 3 and 5 wt.% of the *C. metuliferus* antioxidant extract (CM) respect to CA weight (Figure 2a).

**Figure 2.** Schematic representation of the bi-layer film-forming process: (**a**) film coating solution, (**b**) coating application over LDPE film followed by drying process, and (**c**) visual appearance of the CA-CM coated LDPE films.

Each acetate film-forming solution was poured onto the corona-treated side of LDPE films with a lab-scale automatic applicator (Multicoater RK Printcoat, model K303, Royston, UK) equipped with a steel horizontal rod to obtain a homogeneous wet coating material of 80 μm (speed of 5 m/min) and at room temperature (Figure 2b). The coated LDPE films were immediately dried at 50 ◦C for 3 min and transparent films were obtained (Figure 2c). CA-CM films were also prepared for comparison. Thus, the film-forming solutions were casted on a 50 mm-diameter Petri dish and dried at 50 ◦C for 3 h.

Figure 2 schematically summarizes the bi-layer structure film preparation procedure, starting from the film-forming solution (Figure 2a), its application onto the LDPE film surface with the automatic applicator (Figure 2b), and the resulting bilayer LDPE film coated with CA-CM (Figure 2c).

#### 2.2.3. Film Characterization

### Characterization of *Cucumis metuliferus*-Loaded Cellulose Acetate-Coated LDPE Films

The thickness of the obtained bilayer films was measured with a Digimatic Micrometer Series 293 MDC-Lite (Mitutoyo, Tokyo, Japan) at ten random positions over the film surface.

The absorption spectra in the 700–250 nm region of bilayer films were obtained using a Perkin Elmer (Lambda 35, Waltham, MA, USA) UV-VIS spectrophotometer.

The color properties of the films were measured in the CIELab space in a Minolta colorimeter CR-410 Chroma Meter (Minolta Series, Tokyo, Japan). The colorimeter was calibrated with a white standard tile. Measurements were carried out in quintuplicate at random positions over the CA-based coating surface layer of the LDPE films and average values for these five tests were used to calculate the total color differences (ΔE) induced by the presence of CA and CA-CM coatings into LDPE by means of Equation (1):

$$
\Delta \mathbf{E} = \sqrt{(\Delta \mathbf{a}^\*)^2 + (\Delta \mathbf{b}^\*)^2 + (\Delta L)^2} \tag{1}
$$

where a\*, b\* and *L* are the color coordinates *L* (lightness), a\* (red-green) and b\* (yellow-blue).

The cross cryo-fractured surface microstructure of the cross-section of the bi-layer structures was observed by scanning electron microscopy (SEM) using a JEOL F-6335 microscope. Samples were previously sputtered with a gold layer to make them conductive.

Thermogravimetric measurements were carried out in a Mettler Toledo Gas Controller GC20 Stare System TGA/DCS thermal analyzer (Schwerzenbach, Switzerland). The experiments were conducted under dynamic mode and under nitrogen atmosphere (flow rate of 50 mL/min). Film samples were heated from room temperature to 700 ◦C at 10 ◦C/min. The initial degradation temperatures (T0) were determined at 5% mass loss. Meanwhile, the temperatures at the maximum degradation rate (Tmax) were calculated from the first derivative of the TGA curves (DTG) for CA (TmaxCA), as well as for LDPE (TmaxLDPE).

The mechanical properties of the LDPE and LDPE/CA bilayer films were determined by tensile test measurements at room temperature with an IBERTEST ELIB 30 (S.A.E. Ibertest, Madrid, Spain) machine equipped with a 100 N load cell. Tests were performed in rectangular strips (dimensions: 100 <sup>×</sup> 10 mm2), initial grip separation of 50 mm and crosshead speed of 2 mm/min. Five different samples were tested, and average values of tensile strength and elongation at break were reported.

The oxygen permeation rates of the LDPE and LDPE/CA films were determined at 23 ◦C and 0% relative humidity (RH) by means of an OXTRAN MOCON model 2/21 ML (Lippke, Neuwied, Germany). Films were previously purged with nitrogen for a minimum of 16 h prior to exposure to an oxygen flow of 10 mL/min. The oxygen permeability coefficient (OP) is proportional to oxygen transmission rate per thickness, OTR\*e (e = thickness, mm), and thus, the OTR\*e values were used to compare the oxygen barrier properties of the films.

Release studies of the active compounds from the CA-CM coated LDPE films were conducted by immersion of the films into a fatty food simulant (simulant D1 = solution of 50% ethanol) at 40 ◦C for 10 days [46]. Double sided, total immersion migration tests were carried out by total immersion of

3 cm2 piece of each film in 5 mL of food simulant (area-to-volume ratio = 6 dm2/L) contained in a glass vial. Since ABTS·<sup>+</sup> is an indicator radical that can be neutralized either by direct reduction via SET or by radical quenching via HAT [12], the antioxidant performance of the developed film formulations was measured by means of ABTS method. Therefore, the antioxidant activity of the *C. metuliferus* fruit extract released in the fatty food simulant was regularly analyzed by the scavenging activity of stable free ABTS·<sup>+</sup> radicals, expressed as Trolox equivalents per film area.

The obtained CA-CM coated LDPE films were also tested as fresh fruit browning prevention systems. Thus, LDPE/CA and LDPE/CA-CM films were used to pack fresh-cut apples. Apples were previously washed with tap water, peeled and sliced with a clean knife and packed with the developed bilayer materials in direct contact with the CA-CM layer. The browning of apples was indirectly measured by colorimetrical measurements in the CIELab space at 30 ◦C for 92 h. The packed sliced apples color changes were measured in a Minolta colorimeter CR-410 Chroma Meter (Minolta Series, Tokyo, Japan). The colorimeter was calibrated with a white standard tile. Measurements were carried out in quintuplicate at random positions over the packed apple surfaces and average values for these five tests were used to calculate the total color differences (ΔE) by Equation (1).

Significant differences in the determination of PC as well as in the assessment of antioxidant activity of *C. metuliferus* fruit extract (FRAP and ABTS methods) were statistically calculated by one-way analysis of variance (ANOVA) with OriginPro 8 software using Tukey's test with a 95% confidence level. Similarly, for bilayer films the colorimetric coordinates determinations, tensile test measurements, the release studies of CM-CA-coated LDPE films, as well as the color changes measurements in packed sliced apples, were also statistically calculated by one-way analysis of variance (ANOVA) with OriginPro 8 software using Tukey's test with a 95% confidence level.

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

### *3.1. Antioxidant Activity of Fruit Extracts*

The antioxidant ability of natural extracts is highly dependent on the chemical structure of the active compounds, as well as on the mechanisms used (SET and/or HAT) [12]. Therefore, there is no general standardized method for the extraction of antioxidant agents from heterogeneous systems, such as foods and crops [10]. With the aim of evaluating the most effective extraction process, various solvent systems were assayed based on previous works [43,47]. Table 1 reports the polyphenolic content (PC), as well as the antioxidant activities measured by two methods: FRAP (which operates by the SET mechanism) and ABTS (which operates by both the HAT and SET mechanisms) of the resulting fruit extracts. The lowest polyphenolic extraction efficiency was obtained for pure ethanol, while PC and antioxidant power values of aqueous and aqueous/ethanol extractive solutions did not present significant differences. Matsusaka et al. studied the PC of edible (pulp) and non-edible (seed and peel) parts of *C. metuliferus* from Japan, extracted in EtOH 50%, and similar values were obtained [38]. The peel and seeds showed higher phenolic content than the edible pulp [38].

**Table 1.** Total phenolic content (PC) and antioxidant measurements (ABTS and FRAP) of *C. metuliferus* extract in different solvents.


a–b Different superscripts within the same column indicate significant differences between formulations (*p* < 0.05).

With respect to antioxidant ability, it is known that the FRAP method is more specific for hydrophilic antioxidants, while ABTS is a good method for evaluating both lipophilic as well as hydrophilic antioxidants [48]. The FRAP and ABTS methods (Table 1) indicated that FRAP values were higher for aqueous and 50% ethanolic extractive solutions (without significant differences, *p* > 0.05) than for ethanol (*p* < 0.05). Matsusaka et al. also determined the radical scavenging activity using the ABTS method, obtaining around 200 μmol Trolox/g of whole fruit (edible and non-edible parts), which is approximately 5 mg GAE/100g fruit, which is lower than the results obtained here. Motlhanka studied the antioxidant performance of aqueous, methanolic and under chloroform *C. metuliferus* extracts (pulp and skin) using the DPPH method, and the results indicated that aqueous extract exhibited the strongest antioxidant response, while methanolic extract possessed moderate antioxidant response and low activity in chloroform [36]. Both works were in accordance in confirming that principal phenolic compounds were mainly extracted by using distilled water. Although the chemical composition of *C. metuliferus* has aroused little scientific interest, it is known that the pulp contains beta carotene and vitamins A (retinol), B (B1–B3, B5, B6 and B9) and C [37]. Meanwhile, the seeds are rich in linoleic acid [49], α-tocopherol and γ-tocopherol [37,49], lipases, lipoxygenases enzymes [49] and inorganic ions, such as potassium, calcium, iron, magnesium, phosphorus and zinc [37,49]. It has been reported that the fruit also comprises alkaloids, carbohydrates, cardiac glycosides, flavonoids (i.e., rutin, miricetin and quercetin), saponins, tannins, steroids and terpenoids [37,40]. Although PC values manifested clear differences between extracts, the results concluded that extracts were rich in both hydrophilic antioxidants (as determined by FRAP and ABTS), and lipophilic antioxidants (as determined by ABTS). In fact, the antioxidant activity determined by FRAP showed the lowest values for pure ethanol (*p* > 0.05) and higher values for water and aqueous/ethanol extractive solutions, without significant differences between the water and aqueous/ethanol extractive solutions (*p* > 0.05), in accordance with the PC results. This fact was probably because flavonoids, which are generally more soluble in ethanol, can be bonded with saccharide groups, which are more water soluble, as has already been reported in previous work [43]. The correlation between PC and FRAP values between extracts occurred principally because the PC method is based on the oxidation of phenols by a molybdotungstophosphoric reagent through single electron transfer [43]. On the other hand, the ABTS values of extracts did not present significant differences. Non-glycoside phenolic compounds such as flavanol and flavones generally present better solubility in alcoholic extractive solutions. These PCs were probably molecules with higher chemical composition where a simple molecule is able to scavenge several radical molecules and whose antioxidant activities through the HAT mechanism were also taken into account [43,44,50–52]. Due to the obtained results (Table 1), and considering environmental aspects in evaluating the extractive effectiveness, water was selected for the extraction procedure, which is schematically represented in Figure 1. In brief, dried *C. metuliferus* fruit was mechanically grinded to obtain a powder and dispersed in distilled water, which was then heated (40 ◦C, 180 min at 500 rpm). The resulting solution was filtered twice (double ring qualitative filter paper GE, Grade fast 101), frozen and then the extract was concentrated to dryness by means of a freeze drying process. The obtained viscous extract was then used for the preparation of antioxidant coatings.

### *3.2. Coating Process for Bilayer Film Forming*

*C. metuliferus* extract (CM) was incorporated at three different concentrations (1, 3 and 5 wt.%) into cellulose acetate solution for the development of active coating film-forming solutions. It is widely known that a polymer should be soluble in a solvent with a similar solubility parameter (δ), and thus δ represents an important parameter when working with polymeric solutions [8,14,42]. Thus, good solubility of CA into acetone is ascribed to their close solubility parameters, which are between 19.6 and 25.1 MPa1/<sup>2</sup> [8,42,53] for CA and between 19.9 and 20.1 MPa1/<sup>2</sup> [8,53] for acetone. Concerning the solubility parameters of the main components of CM which are 18.9 MPa1/<sup>2</sup> for betacarotene [54], 18.7 MPa1/<sup>2</sup> for retinol [55], 20.2 MPa1/<sup>2</sup> for α-tocopherol and 20.3 MPa1/<sup>2</sup> for γ-tocopherol [54], good miscibility between CA and CM should be expected.

It is known that the viscosity of a polymeric solution greatly depends on the polymer concentration [56], thus it can be regulated by simple varying the polymer concentration in the film-forming solution. Therefore, 6 g of CA powder was firstly dispersed in 40 mL of acetone under continuous stirring at 50 ◦C, until complete dissolution [42]. Since in this work CM was obtained by means of an aqueous extraction procedure, it should be taken into account that low amounts of residual water can act as a non-solvent, which can potentially compete against the interactions between CA and acetone [42]. In fact, the solubility parameter of water is 47.9 MPa1/<sup>2</sup> [8,42], and thus the solubility parameter of solvent will increase as the presence of water increases in the acetone:water system. From semi-dilute to concentrated polymeric solution, the polymer dimensions decrease until critical concentration (c+) is reached, at which point they shrink to their unperturbed dimensions and remain constant [57]. Necula et al. studied several polymeric solutions of CA in acetone 95% *v*/*v* at different concentrations up to 0.4 g/mL, at different temperatures (from 20 to 50 ◦C). CA critical concentration at which the polymer coils begin to overlap with each other (0.013 > c\* > 0.018 g/mL), as well as the critical concentration for reaching the unperturbed state, c<sup>+</sup> (0.098 > c<sup>+</sup> > 0.142 g/mL) (c<sup>+</sup> - 8 c\*), were determined [58]. Thus, in the present work, in order to ensure that the cellulose acetate coils in acetone (or acetone with low amounts of water as non-solvent, i.e., acetone > 90%) were able to contract toward the unperturbed size state, the selected concentration of CA and/or CA-CM film-forming solution was 0.15 g/mL.

Due to the non-polar nature of LDPE for coating applications and for effectively formation of adhesion joints, it needs a previous surface treatment [41]. Thus, a commercial corona-treated LDPE was selected as substrate in order to increase the poor adhesion properties of LDPE. It should also be taken into account that mass and/or heat transfer takes place, and the polymeric systems become thermodynamically unstable during the solvent coating process, and therefore, phase separation can take place [42]. The molecules of CA in acetone (boiling point 56 ◦C) are characterized by high chain rigidity, but the chain stiffness decreases as temperature increases and, as a result, their flexibility increase [59]. Thus, in order to select the coating drying conditions, the temperature was increasingly varied within the temperature range from 45 ◦C to 50 ◦C and the time was varied between 2 and 3 min through trial-and-error practice until bilayer films with good-quality visual appearance were obtained. That is, the coating parameters were adjusted until a homogeneous solution coating completely covered the LDPE film without apparent phase separation. The processing drying temperature and time of CA and/or CA-CM coated onto the LDPE films were 50 ◦C and 3 min, since these processing parameters made it possible to obtain transparent films without visual defects (Figure 2c). The obtained film thicknesses ranged from 50 ± 2 μm to 63 ± 3 μm, confirming the low thickness of the CA coating in the final bilayer formulation. All LDPE/CA-coated film formulations were transparent without affecting the high transparency of the LDPE (see upper image in Figure 2c), even at the highest *C. metuliferus* fruit extract concentration of 5 wt.% (see lower image of LDPE/CA-CM5 in Figure 2c).

#### *3.3. Optical and Morphological Properties*

The processing conditions used here made it possible to obtain transparent and thin bilayer films (see thickness in Table 2). It should be highlighted that transparency is one of the most important characteristics of the polymeric films for food packaging. Thus, these results were confirmed by means of the determination of their optical properties (Figure 3). No significant differences were observed on the light transmission along the visible region of the spectra (400–700 nm), suggesting that CM was homogeneously dispersed over the CA matrix. The transparency of the LDPE/CA films was measured in the range 540–560 nm (see zoom in Figure 3). The addition of CA had practically no effect on the high transparency of the LDPE. Similarly, the incorporation of CM into the CA matrix had practically no effect on the high transparency of the LDPE/CA, particularly when it was added at low amounts such as 1 wt.% and 3 wt.% (LDPE/CA-CM1 and LDPE/CA-CM3). Meanwhile, the incorporation of the highest amount of CM (5 wt.%) produced a slight reduction in the transparency of the LDPE/CA (see zoom Figure 3), but high transparency was still observed, as can be seen in the visual appearance of this bilayer film (see as example the lower image of LDPE/CA-CM5 in Figure 2c). With respect to the UV spectra region (250–400 nm), the LDPE/CA film showed a reduction of the transmittance of

LDPE due to the fact that CA absorb light in the region below 250 nm. This absorption was slightly reduced with increasing amounts of CM due to the decreasing CA content in the formulation.

The color parameters of films were measured in the CIELab space (Table 2). All materials showed high lightness values. The CA and CA-CM coating application did not produce significant changes in *L* values, which is in good accordance with the high transparency observed for the visual appearance of the films (Figure 2c). The negative values obtained for the a\* coordinate are indicative of deviation towards green color, but these values were very close to zero. This coordinate decreased particularly in the LDPE/CA-CM5 film with the highest amount of CM, showing significant (*p* > 0.05) differences with respect to LDPE/CA-CM materials with lower amounts of CM (LDPE/CA-CM1 and LDPE/CA-CM3 films). Meanwhile, no significant differences in the b\* coordinate were observed between LDPE/CA-based films with respect to the LDPE film, with the exception of LDPE/CA-CM5 film, which showed significant differences (*p* > 0.05) towards positive values, which are indicative of deviation towards yellow color. Similarly, the highest color differences with respect to uncoated LDPE film were observed for LDPE/CA-CM5. Nevertheless, it should be highlighted that all formulations showed lower ΔE values than 0.3, being considerably lower than ΔE of ± 2.0, which is the value typically considered to be the threshold of perceptible color difference for the human eye [13], and even lower than ΔE of ± 0.5, which is the total color difference able to be recognized by a sensorial panel [60].

**Figure 3.** UV-vis spectra of LDPE and LDPE/CA-coated films.


**Table 2.** Color properties of CA-CM-coated LDPE films.

a–b Different superscripts within the same column indicate significant differences between formulations (*p* < 0.05).

The morphological structure of polymeric films is an essential characteristic, since it directly affects the mechanical and barrier performance of the final materials, particularly important in the packaging sector, where it can ultimately influence the commercial success. The adhesion between polymeric layers in multilayer systems is frequently evaluated by observing the microstructure of the materials using microscopic techniques [21,61]. Figure 4 shows the micrograph of the cross-section surfaces of LDPE-, LDPE/CA- and LDPE/CA-CM-based films analyzed by SEM. The SEM analysis was carried out to evaluate the morphological investigation of the bilayer structures, as well as to evaluate the effect of active films on the microstructure at the different concentrations of CM (1, 3, and 5 wt.%) with respect to the CA polymeric matrix used to produce the different LDPE/CA-CM-based formulations. In the SEM micrographs, both polymeric layers can be clearly distinguished (see arrows), showing very good adhesion, with no detachment being observed, revealing that cellulose acetate had been successfully coated onto the surface of corona-treated LDPE (Figure 4a). The LDPE layer presented the typical smooth surface of LDPE in all bilayer formulations [60]. In LDPE/CA film, the CA layer presented a homogenous structure without the presence of pores on the coating structure, suggesting that no pores were formed during the process as a consequence of the acetone evaporation, as can occur in CA-based film processed by solvent casting [62]. This result confirmed the success of the coating process developed here in which the CA/acetone ratio used, as well as the drying conditions (50 ◦C during 3 min), are crucial. The addition of CM into CA coating film-forming solutions did not affect the adhesion of either polymeric layer (Figure 4b–d). However, some compact rougher structures were observed with increasing amounts of CM in the CA layer, which was particularly evident in the LDPE/CA-CM5 film (Figure 4d). This behavior can be ascribed to interaction among active components that tend to agglomerate at high concentrations.

**Figure 4.** SEM of cross-fracture surface: (**a**) LDPE/CA, (**b**) LDPE/CA-CM1, (**c**) LDPE/CA-CM3, and (**d**) LDPE/CA-CM5. 2000× (inset figures 5000×).

Although further studies should be performed to ensure the good adhesion between CA-CM coating and corona-treated LDPE substrate (e.g., sealability, as well as friction and scratch resistance), optical and morphological SEM analysis of bilayer systems revealed that good adhesion had been achieved through the applied drying process parameters (time and temperature) immediately after coating CA-CM film forming solution onto LDPE film. In fact, on one hand, CA-CM coating layer had practically no effect on the LDPE substrate transparency, while on the other hand the absence of porous structures and/or phase separation in SEM images suggests good adhesion between both layers.

### *3.4. Thermal Properties*

With the aim of studying the thermal degradation of each layer, CA and CA-CM-based formulations were prepared by solvent casting, and the TGA parameters, as well as the residue at 700 ◦C, are summarized in Table 3. Meanwhile, the TGA and DTG curves are shown in Figure 5a,b, respectively.

**Table 3.** TGA thermal properties of CA-CM-based films and LDPE/CA-CM-based films.

**Figure 5.** Thermogravimetric analysis: (**a**) TGA of CA-based layer, (**b**) DTG of CA-based layer, (**c**) TGA of LDPE/CA-based bilayer, and (**d**) DTG of LDPE/CA-based bilayer.

A small mass loss below 130 ◦C belonging to the volatilization of the volatile matter and/or to the evaporation of absorbed and bound water was seen in all CA-based films [6,60]. Subsequently, there was a thermal degradation (from around 180 to 300 ◦C) related to the loss of acetyl groups, followed by acetic acid volatilization, which could further catalyze the decomposition of cellulose [63]. Next, the two typical thermal degradation processes of cellulose acetate were also observed, corresponding to the fragmentation of macromolecular structure of the cellulosic chain (TmaxCA = 364 ◦C), followed by the last thermal degradation step, which starts at around 450 ◦C, belonging to the carbonization of

products (≈550 ◦C) to ash [6,60]. Neat CA film still yielded small residual ashes after degradation at 700 ◦C (less than 10%), since CA requires higher temperatures in order to achieve practically no residue (790 ◦C) [63]. The incorporation of *C. metuliferus* extract reduced the thermal stability of cellulose acetate matrix by reducing the onset degradation and maximum degradation temperatures to lower values. After 700 ◦C, the residual ashes for the CA-CM samples were somewhat higher, probably consisting of: (i) positive interaction between CM components and cellulosic structures formed during degradation (i.e., hydrogen-bonding interaction) that delayed the end of the main degradation step of the cellulose structure, (ii) lignocellulosic structures extracted from CM (lignin degradation takes place in a wide range of temperature, from 100 to 900 ◦C [64]), and/or (iii) inorganic components of CM.

The effect of the addition of the CA coating onto LDPE film was also investigated by TGA (Figure 5c,d). The addition of the cellulose acetate coating layer reduced the high thermal stability of LDPE, since CA presented lower thermal stability than LDPE. Thus, the onset degradation temperature was shifted approximately 80 ◦C toward lower values, from 429 ◦C in LDPE to 347 ◦C in the LDPE/CA film, while the Tmax of LDPE was slightly reduced or largely maintained.

The effect of the addition of CA-CM coating produced a similar behavior, and both the T0 and Tmax of CA shifted to lower values, following the same tendency as that of the CA films (Table 3). Meanwhile, the Tmax of LDPE was not affected by the presence of the CA-CM-based coatings. After 700 ◦C, the bilayer films presented practically no residue (less than 1%). Nevertheless, it should be highlighted that no significant degradation took place in the temperature region from room temperature to 200 ◦C, which is a considerably higher temperature than that at which the films are intended to be used during the food packing process, as well as during storage.

### *3.5. Mechanical and Oxygen Barrier Properties*

Films for food packaging are required to maintain their integrity with the aim of withstand the stress that occurs during shipping, handling and storage [1,2]. Thus, the mechanical properties of LDPE-, LDPE/CA- and LDPE/CA-CM-based films were studied by mean of tensile test measurements. Based on the tensile test results (Table 4), it seems that the mechanical properties of the coated LDPE films (LDPE/CA- and LDPE/CA-CM-based films) were controlled by the polyethylene layer. Nevertheless, it should be mentioned that multilayer films of plastic combined with biopolymers generally possess poor mechanical properties due to the poor mechanical strength of the biopolymers [65]. For instance, Shin et al. studied corn zein-coated LDPE, and their mechanical strength could not be measured due to the high brittleness of the corn zein layer, since it broke before the LDPE in the bilayer system [65]. In the present work, the LDPE/CA-based bilayer films exhibited a somewhat higher tensile strength, probably due to the composite structure and the higher tensile strength of CA polymeric matrix [5] with respect to that of LDPE [32], although without significance differences (*p* < 0.05). However, CA possessed very low elongation at break [5] and thus, it is probable that the very thin CA-based coating broke before the LDPE in the bilayer structure during the tensile test measurements. However, this was undetectable from the stress-strain curve (not shown) due to the very thin character of the CA layer. In fact, as Table 4 shows, it seems that the high flexibility of LDPE was not affected by the presence of the CA-CM coating (*p* > 0.05) in bilayer formulations. Moreover, comparing the LDPE/CA-CM-based films with respect to the LDPE/CA formulation, it seems that CM did not affect the mechanical performance of the LDPE/CA film, confirming the well dispersion of the *C. metuliferus* extract in the CA polymeric matrix, as was noted in SEM analysis (see Figure 4). Similar findings on the mechanical performance of LDPE coated with methylcellulose containing murta leaf (*Ugni molinae* Turcz) extract were observed in a previous work by Hauser et al. (2016). In that case, although the elongation at break of neat methylcellulose did not exceed 15%, high elongation at break (higher than 160%) in the bilayer structures was observed [32]. They ascribed this behavior to homogenous methylcellulose coating formation with good adhesion to the corona-treated LDPE [32].

One of the major challenges for coatings intended for LDPE is to increase the low oxygen barrier performance of this polymer. CA is recognized to have a higher barrier performance (OTR values around 650 cm3/m<sup>2</sup> day [6]) than LDPE (LDPE film = 4750 650 cm3/m<sup>2</sup> day, thickness = 0.05 mm). Thus, the application of a CA coating onto LDPE film drastically reduces the oxygen permeability, reducing the OTR\*e values by between 19% and 31% (Table 4). The LDPE/CA-CM5 film presented slightly higher oxygen transmission values, probably due to its having the lowest homogeneity as a result of its high CM extract concentration. Although the oxygen barrier performance obtained here did not provide the final packaging material with a strong oxygen barrier performance, such as those provided by other polymeric matrices with well-known oxygen barrier performance (i.e., poly(ethylene terphthalate) (PET) with OTR\*e < 3 cm3 mm/m<sup>2</sup> day [4,66], EVOH which exhibits low OTR values under dry conditions with OTR\*e < 4 cm<sup>3</sup> mm/m<sup>2</sup> day [67], or calcium and sodium caseinates with OTR\*e < 7 cm<sup>3</sup> mm/m<sup>2</sup> day [13]), it showed the effectiveness of CA and CA-CM coating to improve these properties due to the good adhesion onto the corona-treated LDPE substrate. The improvement of the LDPE films' barrier performance by coating it with different biopolymers such as whey protein [35], gelatin [66], chitosan or corn zein [65] has been already observed.

**Table 4.** Tensile test and oxygen barrier properties of LDPE/CA-CM-based films.


<sup>a</sup> Different superscripts within the same column indicate significant differences between formulations (*p* < 0.05).

### *3.6. Antioxidant Activity of Active Bilayer Systems*

Considering that lipids are one of the main targets of oxidative reactions and, thus, lipid oxidation process responsible of a major problem in both natural and processed foodstuff [68], the antioxidant developed bilayer films were studied in direct contact with a fatty food simulant (simulant D1 = solution of 50% ethanol) [46]. Meanwhile, considering the complexity of different compounds in *C. metuliferus* fruit extract, the release studies of the active agents from the CA-CM coating layer of the LDPE films were indirectly measured through the determination of the total antioxidant activity into the food simulant, due to the fact that it is proportional to antioxidant release kinetics [10]. To evaluate both, lipophilic as well as hydrophilic antioxidants released, ABTS method was used and the measurements were performed after 1, 3 and 10 days in contact with. Non containing *C. metuliferus* fruit extract LDPE/CA film was also analyzed as control material and, as expected, did not show any ABTS radical scavenging activity (not shown). The antioxidant release kinetics indicated that more than 50% of active compounds were released during the first day (Figure 6). Subsequently, the release capacity moved on towards an equilibrium value on the third day in contact with the food simulant. The release kinetic of the active compounds of CM followed the second Ficks' law of diffusion with an exponential growth to a maximum, in accordance with already reported works of active cellulose acetate films (i.e., CA loaded with ascorbic acid [69], L-tyrosine [69], thymol [70] and red onion extract [10]). As it was expected, the antioxidant activity increased with increasing amount of CM in the formulations. Thus, the higher antioxidant effectiveness was for the film with the higher amount of CM (LDPE/CA-CM5 film). However, higher antioxidant effect was observed in other cellulose acetate films such as CA loaded with 5 wt% or red onion extract which showed around 1 mg Trolox/dm<sup>2</sup> film [10]. This result can be related with the lower PC and antioxidant performance of the *C. metuliferus* extract with respect to that of red onion. Nevertheless, it should be taken into account that the use of an active internal CA-based layer coated in an external LDPE layer may contribute to the effectiveness of antioxidant performance by slowing down the release rates and extending their action due to the interaction between both polymeric layers and the less exposition to the food simulant [61]. The major antioxidant ability of CM has been attributed to non-edible parts of the fruit (seed and peal) [38]. Thus, CM can

result interesting not only for the development of antioxidant packaging materials, but also towards the use of this fruit waste from agri-food industry as a valorization resource of bioactive compounds giving an added value to the non-edible waste.

**Figure 6.** The *C. metuliferus* fruit extract antioxidant activity measured by ABTS method. a–b Different superscripts within the same day indicate significant differences between formulations (*p* < 0.05).

### *3.7. Anti-Browning E*ff*ect on Packaged Fresh-Cut Apple*

Another promising field of coating technology application is in the fresh and minimally processed fruit sector, which are highly perishable products [21]. Thus, the obtained active-coated LDPE films were also tested in the prevention of browning in fresh-cut apples (Figure 7).

**Figure 7.** (**a**) Visual appearance of apples packed with LDPE/CA film: (**a**)/(**A**) immediately packed, (**a**)/(**B**) after 48 h and (**a**)/(**C**) after 72 h; and LDPE/CA-CM3 film: (**a**)/(**D**) immediately packed, (**a**)/(**E**) after 48 h, and (**a**)/(**F**) after 72 h packed; (**b**) Total color change evolution of fresh-cut apples packed with LDPE/CA and LDPE/CA-CM bilayer films over 96 h. a–c Different superscripts within the same day indicate significant differences between formulations (*p* < 0.05).

It is well known that fresh fruit browning is caused by enzymatic oxidation of phenolic compounds mediated by polyphenol oxidase activity, and two strategies for inhibiting this process are through the reduction of oxygen and addition of antioxidants [71]. Figure 7 shows the visual appearance of apple slices packed with non-active bilayer film (LDPE/CA, upper images in Figure 7b), as well as with active LDPE/CA-CM films, using LDPE/CA-CM3 as example (down images in Figure 7a) stored at 30 ◦C to simulate the worst foreseeable conditions. As can be seen, apples packed in LDPE/CA without *C. metuliferus* extract clearly exhibited a browning effect after 48 h (Figure 7a(B)) and 72 h (Figure 7a(C)). Meanwhile, this effect was less pronounced in LDPE/CA-CM-based formulations (Figure 7a(D–F)). These findings were corroborated by the determination of the evolution of color differences (ΔE) of packed apples with all LDPE/CA-based bilayer formulations over 96 h of storage (Figure 7b). As expected, the highest color differences were observed in packed apples with un-functionalized LDPE/CA film Figure 7b, which was in good accordance with the visual browning observed in Figure 7a(B,C). Meanwhile, those packed fresh-cut apples containing CM in the coating layer showed less color difference, which decreased with increasing amounts of CM in the formulations (Figure 7b). In fact, LDPE/CA-CM5 formulation was able to reduce the ΔE value by around 50% with respect to the LDPE/CA film, reaching values around ΔE = 2.

### **4. Conclusions**

Antioxidant compounds of *Cucumis metuliferus* (CM) fruit were successfully extracted by means of an aqueous extraction process and further incorporated into cellulose acetate (CA) matrix to develop antioxidant active coatings. CA was dissolved in acetone and CM was further added in concentrations of 1, 3 and 5 wt.%. The good miscibility of the film-forming solution was directly related to the fact that the main components of *C. metuliferus* show solubility parameters close to those of CA, as well as to acetone. The CA-CM-based film-forming solutions were successfully coated onto corona-treated LDPE films through a simple process. Very thin CA-CM layers were obtained, since the films' thickness varied from 50 μm in the case of LDPE to thicknesses between 60 and 65 μm in the case of the bilayer LDPE/CA-CM films. SEM observations confirmed the proper adhesion of the CA coating onto the LDPE film for intended use as bilayer packaging materials. CA and CA-CM-based coatings induced a decrease in the thermal stability of LDPE, but exhibited enough thermal stability (T0 > 300 ◦C) for the intended use (i.e., during food packing or storage). The CA-CM-based layer provided improved oxygen barrier to LDPE film and did not affect its high transparency or colorlessness. Meanwhile, the CA layer containing different amounts of CM extract (1, 3 and 5 wt.%) showed its effectiveness as an antioxidant carrier, since CM either underwent a sustained release into a fatty food simulant, exerting free radical scavenging activity, or reduced the browning of fresh-cut apples in direct contact. Since the coating process proposed here is simple, and extremely flexible and low-cost, it is expected that the transfer of these active coatings from laboratory scale to industrial production will be easily feasible.

**Author Contributions:** Conceptualization, M.P.A., C.L.d.D., A.G. and M.J.G.; methodology, M.P.A., C.L.d.D., L.G., S.F., A.G. and M.J.G.; formal analysis, M.P.A., C.L.d.D., L.G. and S.F.; investigation, M.P.A., C.L.d.D., L.G., S.F., A.G. and M.J.G.; resources, A.G. and M.J.G.; data curation, M.P.A., C.L.d.D., L.G. and S.F.; writing—original draft preparation, M.P.A., C.L.d.D.; writing—review and editing, M.P.A., C.L.d.D., L.G., S.F., A.G. and M.J.G.; visualization, M.P.A., C.L.d.D., L.G., S.F., A.G. and M.J.G.; supervision, A.G. and M.J.G.; project administration, M.P.A. and C.L.d.D.; funding acquisition, M.P.A., C.L.d.D., A.G. and M.J.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Santander-UCM (PR87/19-22628) and FONDECYT 1200766 projects. Marina Patricia Arrieta thanks MINECO for her postdoctoral contract: Juan de la Cierva-Incorporación (FJCI-2017-33536).

**Acknowledgments:** The authors thank the staff of the ICTS National Center for Electron Microscopy (CNME), UCM, Madrid (Spain) for their assistance with transmission and electron scanning microscopy. The authors thank EDELPA who kindly provide corona-treated low-density polyethylene (LDPE) film.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
