**Characterization Study of Empty Fruit Bunch (EFB) Fibers Reinforcement in Poly(Butylene) Succinate (PBS)**/**Starch**/**Glycerol Composite Sheet**

### **Rafiqah S. Ayu 1, Abdan Khalina 2,\*, Ahmad Sa**ffi**an Harmaen 1, Khairul Zaman 3, Tawakkal Isma 2, Qiuyun Liu 4, R. A. Ilyas <sup>5</sup> and Ching Hao Lee 1,\***


Received: 24 May 2020; Accepted: 22 June 2020; Published: 15 July 2020

**Abstract:** In this study, a mixture of thermoplastic polybutylene succinate (PBS), tapioca starch, glycerol and empty fruit bunch fiber was prepared by a melt compounding method using an industrial extruder. Generally, insertion of starch/glycerol has provided better strength performance, but worse thermal and water uptake to all specimens. The effect of fiber loading on mechanical, morphological, thermal and physical properties was studied in focus. Low interfacial bonding between fiber and matrix revealed a poor mechanical performance. However, higher fiber loadings have improved the strength values. This is because fibers regulate good load transfer mechanisms, as confirmed from SEM micrographs. Tensile and flexural strengths have increased 6.0% and 12.2%, respectively, for 20 wt% empty fruit bunch (EFB) fiber reinforcements. There was a slightly higher mass loss for early stage thermal decomposition, whereas regardless of EFB contents, insignificant changes on decomposition temperature were recorded. A higher lignin constituent in the composite (for high natural fiber volume) resulted in a higher mass residue, which would turn into char at high temperature. This observation indirectly proves the dimensional integrity of the composite. However, as expected, with higher EFB fiber contents in the composite, higher values in both the moisture uptake and moisture loss analyses were found. The hydroxyl groups in the EFB absorbed water moisture through formation of hydrogen bonding.

**Keywords:** empty fruit bunch fiber (EFB); polybutylene succinate (PBS); starch; glycerol; characterizations; biocomposite; polymer Blends

### **1. Introduction**

The development of biodegradable materials has attracted much research interest by scientists on worldwide. Aliphatic polyesters are among the most promising materials for the production of high-performance biodegradable plastics. One of the polyesters, polybutylene succinate (PBS) which is commercially available in the market, has very high fame as a high-performed bioplastic [1]. Many recent studies have selected PBS as the composite matrix for various applications and purposes [2–4].

PBS is synthesized from succinic acid and 1,4-butanediol (BDO) via a polycondensation process, and exhibits balanced performance in thermal and mechanical properties as well as processability [5]. It is more thermally stable than PLA polymer [6]. PBS is able to undergo biodegradation and even disposal in compost, moist soil, fresh water (by activated sludge), or sea water. It also can be composted by microorganism activities to convert it into CO2, H2O, and inorganic products under aerobic conditions, or CH4, CO2, and inorganic products under anaerobic conditions. The biodegradability of PBS depends mainly on its chemical structure and especially on its hydrolysable ester bond in the main chain, which is susceptible to microbial attack [7,8]. One study prepared a reactive-PBS polymer (RPBS) with insertion of toluene-2,4,diisocyanate (TDI) chemical in different ratios and blends with starch. The properties of the blended specimens were found to be significantly improved, even with only 10 wt% of RPBS. The TDI chemical insertion smoothened the PBS/starch polymer blend's surface, showing better miscibility of the two phases [9]. However, PBS has some negative properties such as slow crystallization rate, low melt viscosity, and softness. These have restricted its processing condition and potential applications. Polymer mixing with other materials is commonly used, to develop new blend materials that are suitable for specific working environments or specific purposes. However, most of the polymers are not miscible with each other and tend to phase-separate in a melt state [10]. Besides, although a fast crystallization reaction can happen when mixing with other materials, this may cause deterioration of PBS composite's strength [11]. Therefore, plasticizers such as glycerol were added to overcome and improve the flexibility of PBS polymer [12]. The council of the IUPAC (International Union of Pure and Applied Chemistry) has defined a plasticizer as "a substance or material incorporated in a material (usually a plastic or elastomer) to increase its flexibility, and workability by lowering glass transition temperature (Tg)" [13]. Glycerol is a pure anhydrous structure and has a specific gravity of 1.261 g·mL−1, melting point of 18.2 ◦C and boiling point of 290 ◦C under normal atmospheric pressure [14]. On the other hand, grafting is another method to improve a compatibilizer between two materials. Suchao-in et al., 2013, have grafted PBS on tapioca starch blends. Results revealed a strong interfacial adhesion of the blend and enhanced modulus properties, as evidenced from SEM micrographs [15].

Starch is one of the materials that is readily available, low cost and one of the important bioresources used in the food industry, e.g., as a thickener and gelling agent. It also possesses good physical, mechanical and oxygen barrier properties, that give it potential to become active film [6,16]. It is much more reliable and chemically stable than other spacers [17]. Starch is a natural polymeric product and is found in almost every plant. Usually the main sources of starch come from tapioca, potato, maize, rice and wheat [18]. Starch contains two different molecular structures, linear (1,4)-linked α-d-glucan amylose and highly (1,6)-branched α-d-glucan amylopectin. The starch molecules are tied by van der Waals bonds and strong intermolecular hydrogen bonds. Common native starch granules have a semi-crystalline, radially oriented spherulitic structure. They contain water on different structural levels [19]. Amylopectin consist of a branching chain that forms double helices and produce crystalline structure of the granules, whereas amylose is amorphous and interspersed among amylopectin molecules [20]. Some starch polymers form helical structures due to the existence of α linkages, which contribute to its extraordinary properties and enzyme digestibility [21]. The relative amounts of amylose and amylopectin depend upon the plant source. Corn starch granules typically contain approximately 70% amylopectin and 30% amylose [22]. However, native starch itself cannot be satisfactorily used due to its hydrophilicity and brittleness which lead to the poor mechanical properties, so it requires some chemical modification to overcome this drawback [23]. Blending thermoplastic starch with PBS is one of the frequently selected options by researchers. Higher water resistance, good processability, fully biodegradable, and superior mechanical properties were being claimed for PBS/corn starch blend with glycerol plasticizers [24].

On the other hand, extensive investigation has been carried out to study the effects of natural fiber reinforcement on polymer composites [25–27]. The majority of outcomes have agreed that reinforced natural fiber has a better performing load transfer mechanism, and results in higher mechanical properties [28,29]. Empty fruit bunch (EFB) fibers have shown comparable quality to high strength kenaf bast fibers [30]. However, the hydrophilic nature of the EFB fiber is found to be incompatible with the hydrophobic polymer matrix. This caused poor interfacial adhesion between the fiber and matrix, leading to lower performances. Chemically treated EFB fibers had greater thermal and morphologies properties [31]. Moreover, it consists of wood-like constituents (cellulose, hemicellulose and lignin), showing lower thermal stability towards high heat environments, yet producing high residue at

high temperature [32]. Furthermore, the hydrophilic behavior is expected to have higher moisture absorption, leading to swelling of the EFB fiber. Nevertheless, the extremely low cost of EFB fiber as a byproduct and its 100% biodegradable properties have created a high interest in it [33].

This study is a continuation of previous study, which investigated the characterization of high volume contents of EFB fiber reinforced in PBS/tapioca starch composite [34]. The high volume of fiber reinforcement found deterioration of mechanical properties due to poor interfacial bonding, evidenced from SEM micrograph and this is not accepted by the market, and similar findings were reported that show a lower tensile strength when alkaline treated-sugarcane fibers were inserted without any plasticizers [35]. Hence, in the present study, a lower volume of EFB fiber was added into the PBS/starch composite sheet with glycerol plasticizers to improve compatibility. This study has filled the knowledge-of-gap on low EFB fiber reinforcement in PBS/starch composite sheet with plasticizer fillers. The outcomes of this investigation (mechanical, morphological and thermal characterization) could serve as valuable knowledge for future developments on EFB fiber reinforcement in polymer composite.

### **2. Experimental**

### *2.1. Materials*

PBS in the form of pallets were bought from PTT Public Company Limited in Thailand. Density of PBS is 1.26 g/cm3. Tapioca starch in form of powder was obtained from PT Starch solution in Indonesia. Empty fruit bunch fiber (EFB) was used and obtained from Polycomposite Sdn Bhd in Negeri Sembilan. The EFB were chopped using a grinder machine and sieved to get an average 300–600 μ in size. Meanwhile, glycerol was purchased from Duro Kimia Sdn Bhd in Selangor. The properties of materials as tabulated in Table 1.


**Table 1.** Properties of polybutylene succinate (PBS), starch and empty fruit bunch (EFB) fiber.

### *2.2. PBS Composite Preparation*

The PBS pallets and EFB fiber was first dried in an oven at 80 ◦C to prevent excessive hydrolysis which can compromise physical properties of the polymer. Starch, glycerol and EFB were dry mixed in an industrial mixer machine and sieved to remove excessive lumps during the mixing process. Then, PBS and the mixed compound of starch/EFB/glycerol were added into an industrial counter rotating extruder feeder for a total of 300 kg per processing. After that, the compound was melted in an industrial extruder machine comprising 10 heat zones, which were set temperatures in between 115–145 ◦C with rotation speed of 80 RPM. As a result of the shear stress imposed on fibers during compounding, homogenization of PBS/starch/fiber/glycerol was carried out by cycling the mixture in the extruder for 15 min and then extruded through a 2 mm gauge strand die at a rate of 10 mm/s. The melted compound was then passed through a calendaring machine before producing a sheet. Then, the sheets were cut into shapes according to specific characterization testing. The image of the extruded compound is shown in Figure 1.

**Figure 1.** Sheet extrusion process.

### 2.2.1. Mechanical Properties (Tensile Properties)

The tensile testing of the composite was conducted using a 5 kN Bluehill INSTRON Universal Testing Machine. The test was carried out according to ASTM standard D-638. The specimens were cut into dog bone shape by a plastic molder machine with the specifications of 120 <sup>×</sup> <sup>120</sup> <sup>×</sup> 2 mm3 of length, width and thickness respectively. The composites were gripped at a 30 mm gauge length and the crosshead speed was set at 2.0 mm/min. All specimens were kept in a conditioning room and the test was run at 22 ◦C and relative humidity (RH) at 55%. Seven specimens were tested per test condition.

### 2.2.2. Mechanical Properties (Flexural Properties)

Flexural test of the composite was performed using 5 kN Bluehill INSTRON Universal Testing Machine. Test samples were cut to the dimension of 70 <sup>×</sup> 15 <sup>×</sup> 2 mm<sup>3</sup> and three-point bending tests were performed according to ASTM D790 standard. The crosshead speed was set at 2 mm/min with a support span-to-depth ratio of 16:1. All specimens were kept in a conditioning room and the test was run at 22 ◦C with the relative humidity (RH) at 55%. Seven specimens were tested per test condition.

### 2.2.3. Morphological Analysis

Morphology of the samples was observed using Hitachi S-3400N scanning electron microscope (SEM) equipped with energy dispersive X-ray (EDX) under an accelerating voltage of 15 kV and at an emission current of 58 μA. The tensile-tested-samples were gold sputtered before observation to avoid the charging effect during sample examination. SEM helps to analyze the microscopic structure and characterization of the compound on the basis morphology and structural changes.

### 2.2.4. Thermal Analysis

The thermal stability of the samples was characterized using a TA Instruments Q500 thermogravimetric analyzer, TGA. About 6 mg of the sample was scanned from 30 to 700 ◦C at a heating rate of 20 ◦C min−<sup>1</sup> under a nitrogen gas atmosphere.

### 2.2.5. Moisture Absorption and Moisture Loss Analysis

Sample sheets of rectangular shape with dimensions of 15 <sup>×</sup> <sup>15</sup> <sup>×</sup> 0.5 mm3 were dried in a vacuum oven at 60 ◦C for 24 h and weighed prior to testing. The vacuum dried rectangular sheets were immersed in distilled water at 20 ◦C to determine the water absorption and soluble ratio. The sample was taken out to measure the water absorption and soluble ratio in a certain time, and then the same sample was vacuum dried to measure the weight loss of the sample. The weights of the original sample and the sample after water absorption were designated as *W*<sup>0</sup> and *W*1, and the dry weight of the water extracted sample was designated as *W*2. The value of moisture absorption was obtained by Equation (1):

$$\text{Moisture uptake} = \frac{W\_1 - W\_2}{W\_2} \times 100\% \tag{1}$$

with the value of the soluble ratio derived from Equation (2):

$$\text{Soluble ratio} = \frac{W\_0 - W\_2}{W\_0} \times 100\% \tag{2}$$

Three measurements were performed for each sample, and the result was reported as the average value. This procedure followed the short-term immersion standard method ASTM D570-98.

On the other hand, seven samples were prepared for the moisture content evaluation. The samples were placed in normal climatic conditions at room temperature (27 ± 2 ◦C) with 65% relative humidity of air for 24 h before being weighed. Percentages of moisture content were determined by using Equation (3). The samples were heated in the oven for 24 h at 105 ◦C. Before heating the samples were measured as *M*0. After 24 h in the oven, the fiber was weighed again as *M*1. Therefore:

$$\text{Moisture content} \left(\% \right) \frac{M\_1 - M\_0}{M\_0} \times 100\% \tag{3}$$

### **3. Result and Discussion**

### *3.1. Mechanical Testing*

Filler reinforcement is an important factor in determining mechanical properties of the composite. The most crucial factor that affects the mechanical properties of the fiber reinforced materials is its fiber/matrix interfacial adhesion. The strength of the interfacial bonding was determined by several factors, such as the nature of the fiber and polymer components, fiber aspect ratio and processing procedure [36,37]. The mechanical properties of the PBS composite are presented and illustrated in Table 2 and Figures 2 and 3, respectively. It was clearly shown that the tensile and flexural strength of specimens were decreased for fiber reinforcement up to 8 wt%. This is due to poor dispersion and incompatibility between fillers and the PBS matrix according to previous studies [38,39]. Fibers are unable to disperse evenly in the PBS matrix, creating high stress concentration spots, resulting in a dramatic reduction in tensile strength [40]. However, increments in tensile and flexural strength were observed, indicating that the reinforcing ability of the natural fibers has overcome the shortage from the interfacial adhesion factor. A previous study reported the same trend, that higher fiber contents led to an improvement in the tensile strength of the matrix due to the interaction related to the fiber contents [41].

On the contrary, there were relatively higher mechanical properties for a 0% EFB specimen (which contained 30 wt% of starch/glycerol with a 2:1 ratio) in a current study, when compared to a previous study, which only gave 16.12 and 21.78 MPa for tensile and flexural strength, respectively, for pure PBS polymer [34]. The insertion of starch supposedly reduces the composite's strength performance due to low compatibility [6]. However, the addition of glycerol has the adverse effect of strength deterioration by localization of a compatibilizer at the interface for a stable morphology from a SEM micrographic [42].

Accoding to Thirmizir et al., the flexural strength of PBS composites was higher than neat PBS polymer. [8]. Higher fiber loadings have improved the flexural strength due to mechanical interlocks found between fiber and matrix. The fiber/matrix mechanical interlocking was expected to act as a mechanism to withstand the bending force in flexural testing. On the other hand, flexural strength was reduced by 6% for 8 wt% EFB fiber reinforcement composites. This may be attributed to interruption of the continuous long polymer chain by the presence of hydrophilic lignocellulose. Similarly, higher flexural strength values were recorded for higher EFB fiber reinforcement specimens. EFB fibers work as a carrier of loads in the matrix, synchronized with the tensile performance.


**Table 2.** Mechanical properties of PBS/starch/glycerol and EFB blends.

<sup>a</sup> Starch/Glycerol in 2:1 ratio.

**Figure 2.** Tensile strength and tensile modulus of PBS composites.

**Figure 3.** Flexural strength and flexural modulus of PBS composites.

### *3.2. Morphological Analysis*

Figure 4 shows SEM images for specimens' surface morphology, under 500× magnification. The strength performances of the composites are directly affected by morphology status. Figure 4a shows a smooth and regular PBS surface, while Figure 4b shows the image of modified tapioca starch granules on the surface. Figure 4c,d shows the presence of EFB fiber, which consists of long fibers surrounded on the PBS/starch matrix. The poor adhesion of fibers on the matrix shows correlation with the reduction of mechanical properties for "8 wt% EFB" specimens. The poor impregnation makes it easier for the fiber to be pulled out, and causes a lower strength performance for the composite. This trend was also reported by a previous researcher [36]. For Figure 4e,f, it can be observed that the fibers are adhered to the matrix. The longitudinal fibrous shapes of the fibers were evenly mixed and evenly distributed on the matrix surface. The fibers mix homogenously with the matrix and are not clearly seen on the surface morphology analysis. This indicates the good fiber/matrix adhesion. On the other hand, the void between the EFB fiber and matrix is less, which gives a better fiber/matrix bonding and increased mechanical strength to the composite.

**Figure 4.** SEM micrographs for: (**a**) raw PBS, (**b**) 0% EFB, (**c**) 8% EFB, (**d**) 12% EFB, (**e**) 16% EFB and (**f**) 20% EFB.

### *3.3. Thermal Analysis*

Thermogravimetric analysis (TGA) is a useful method for quantitative determination of the degradation behavior, thermal stability and mass change in a composite. The appearance of starch and glycerol in the PBS polymer composite has reduced the thermal stability of the specimen in generally. Excess amounts of glycerol have taken part in the reaction with hydroxyl groups of the PBS polymer, which promoted a lower thermal stability [43]. However, with more starch/glycerol contents replaced by EFB fibers, the effects of glycerol are lesser and gradually dominated by EFB fibers.

Figure 5 shows the TGA profiles of the EFB composites, while Table 3 lists the mass loss in every stage with peak temperature until sample reach 600 ◦C. There is a small but noticeable step between 75–95 ◦C, which was due to the presence of free water in the composite. Other researchers also have reported that this is due to water removal, as starch has a higher tendency to absorb moisture [6]. It also was reported that at the initial stage weight loss may be ascribed to the evaporation of water in the fiber [32,44,45]. Sharp transitions at peak 2 and 3 between 200–265 ◦C is due to decomposition of polysaccharide components in the starches. At higher temperatures, hemicellulose degradation occurs, followed by cellulose degradation [46]. Both degradation processes involve complex reactions (dehydration, decarboxylation, among others) as well as breakage of C-H, CO and C-C bonds [47]. Apart from this, lignin starts to degrade at a temperature range between 250–450 ◦C. Lignin degradation generates water, methanol, carbon monoxide and carbon dioxide [48,49]. PBS matrix is a thermally stable biopolymer and it begins to degrade near 300 ◦C with high degradation rates, as similarly found by Lee et al. [50]. In this analysis, there was slightly higher mass loss for early stage thermal decomposition whereas insignificant changes on decomposition temperature, regardless of EFB contents. However, with the higher lignin constituent in the composite there was a higher mass residue, which would turn into char at high temperature. This observation indirectly proves the dimensional integrity of the composite. Besides, the better mechanical performance for the high natural fiber reinforcement could offer wider the applications for this composite material.

**Figure 5.** TGA profiles of EFB composites.


**Table 3.** A summary of peak temperatures for EFB composites.

### *3.4. Moisture Uptake and Average Loss of Moisture Contents*

The amount of water absorbed in the composite was calculated by weight difference between before and after samples exposed to water. Figure 6 shows moisture uptake over the time and average loss of moisture contents for EFB composites. The moisture uptake test was conducted to identify the amount of water absorbed by the composites while the average loss of moisture content is to measure the mass loss after being subjected to heat. Generally, the moisture uptake was depending on several factors such as volume fraction of fiber, voids, viscosity of matrix, humidity and temperature [51].

**Figure 6.** Moisture uptake analysis and average loss of moisture contents for EFB composites.

Water absorption is one of the disadvantages of applying lignocellulosic materials. Insertion of starch components into PBS polymer (0% EFB), comes with expected higher water absorption [52,53], as the starch component may take up to 300% of water absorption, as reported previously [9]. However, when a portion of the starch/glycerol is replaced by EFB fiber, higher values are found in both moisture uptake and loss analyses. This is because of the hydrophilic properties of the natural fibers in the poor interfacial bonding, leading to higher increments of moisture uptake, due to the presence of hydroxyl groups. Hence, it was observed that 20% EFB composite has the highest moisture uptake. The hydroxyl groups absorbed water moisture through formation of hydrogen bonding. The higher moisture content of the natural fiber may result in a weak interfacial bonding between the fiber and matrix [54]. The water molecules were absorbed in the inter-fibrillar space of the cellulosic structure that exists in the fiber and causes cracks and micro voids in the composite surface [55]. During immersion of the samples in water, capillarity action conducts water molecules to fill the voids, causing cracks and dimensional change. Swelling of fiber also leads to interfacial debonding and thereby reduction of mechanical strength [56,57]. In this study, at 6 to 8 h immersion, samples reached stable moisture contents, showing a saturation point,

where no more water was absorbed. Similarly, when subjected to heat, the high EFB loadings composite loses a higher amount of water content. This shows that fiber reinforcement improves strength profiles, yet may cause higher susceptibility to moisture attack, thereby reducing overall composite properties.

### **4. Conclusions**

In this study, the effect of fiber content on the mechanical and thermal properties of polybutylene succinate (PBS) composites were mainly evaluated. The control specimen (0% EFB) was compared with PBS polymer to discuss the changes affected by the appearance of starch/glycerol components. Generally, insertion of starch/glycerol provided better strength performance, but worse thermal and water uptake to all specimens.

On the other hand, it was found that there was poor interfacial adhesion between the EFB and PBS matrix, leading to lower mechanical properties. Fortunately, this was overcome and improved by higher fiber reinforcement, that regulated a better load transfer mechanism. Higher fiber loadings have improved the flexural strength due to mechanical interlocks found between the fiber and matrix. As a result, the tensile and flexural strength had increases of 6.0% and 12.2%, respectively, for 20 wt% EFB reinforcements.

In the SEM micrographic, it shows a smooth surface for PBS, while appearances of the EFB fiber show poor adhesion on the matrix, and was found to correlate with the mechanical properties analysis. On the other hand, the void between the EFB fiber and matrix was less and gave better fiber/matrix for a high fiber volume content composite.

A total of four thermal degradation peaks were recorded in the TGA analysis. The first peak was observed at 75–95 ◦C, due to the presence of free water in the composite. Sharp transitions at peak 2 and 3 between 200–265 ◦C were due to decomposition of the polysaccharide components in the starches and natural fibers. The last thermal decomposition peak was recorded at around 350 ◦C, which was responsible for the degradation of the PBS matrix. In this analysis, there was a slightly higher mass loss for early stage thermal decomposition, whereas insignificant changes on decomposition temperature were recorded, regardless of EFB contents. However, the higher lignin constituent in the composite had a higher mass residue, which would turn into char at high temperature. This observation indirectly proves the dimensional integrity of the composite. Moreover, the better mechanical performance of the high natural fiber reinforcement could offer wider applications for this composite material.

The moisture uptake over time and average loss of moisture contents for EFB composites were analyzed in this study. The higher the EFB fiber content in the composite, the higher values in both moisture uptake and loss data were found. This is expected due to the hydrophilic properties of the natural fibers that lead to higher increments of moisture uptake, due to the presence of hydroxyl groups. Hence, it was observed that 20% EFB composite has the highest moisture uptake. In this study, at 6 to 8 h immersion, samples reached a stable moisture content, showing a saturation point, where no more water was absorbed. Similarly, when subjected to heat, the high EFB loadings composite loses a higher amount of water content. This shows that fiber reinforcement improves the strength profile yet may cause higher susceptibility to moisture attack, thereby reducing overall composite properties.

As concluding remarks, the present results suggest that the use of 20% EFB fiber contents in the composite may be a potential candidate for effectively improving the properties and performances of the composite for future application. Nevertheless, the content of starch/glycerol may need to strategically planned to obtain a balance between performance and costing.

**Author Contributions:** Conceptualization, A.K. and R.A.I.; methodology, R.S.A.; validation, C.H.L.; formal analysis, Q.L.; investigation, R.A.I.; resources, A.S.H.; data curation, T.I.; writing—original draft preparation, R.S.A.; writing—review and editing, C.H.L.; visualization, K.Z.; supervision, A.K.; project administration, A.K.; funding acquisition, A.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Malaysian Industry Government Group of High Technology (MIGHT) under Newton Engku Omar Fund Grant no 6300873 (Safebiopack food packaging).

**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* **Mechanical and Rheological Behaviour of Composites Reinforced with Natural Fibres**

### **Mariana D. Stanciu 1,\*, Horatiu Teodorescu Draghicescu 1, Florin Tamas <sup>2</sup> and Ovidiu Mihai Terciu <sup>1</sup>**


Received: 23 May 2020; Accepted: 17 June 2020; Published: 22 June 2020

**Abstract:** The paper deals with the mechanical behaviour of natural fibre composites subjected to tensile test and dynamic mechanical analysis (DMA). Three types of natural fibre composites were prepared and tested: wood particle reinforced composites with six different sizes of grains (WPC); hemp mat reinforced composites (HMP) and flax reinforced composite with mixed wood particles (FWPC). The tensile test performed on universal testing machine LS100 Lloyd's Instrument highlights the elastic properties of the samples, as longitudinal elasticity modulus; tensile rupture; strain at break; and stiffness. The large dispersion of stress–strain curves was noticed in the case of HMP and FWPC by comparison to WPC samples which present high homogeneity of elastic–plastic behaviour. The DMA test emphasized the rheological behaviour of natural fibre composites in terms of energy dissipation of a material under cyclic load. Cole–Cole plots revealed the connection between stored and loss heat energy for studied samples. The mixture of wood particles with a polyester matrix leads to relative homogeneity of composite in comparison with FWPC and HMP samples which is visible from the shape of Cole–Cole curves. The random fibres from the hemp mat structure lead to a heterogeneous nature of composite structure. The elastic and viscous responses of samples depend on the interface between fibres and matrix.

**Keywords:** natural fibre composites; mechanical properties; elastic behaviour; viscous response

### **1. Introduction**

The natural fibres, vegetal, animal or mineral, consist of sustainable resources for composite materials used both in industrial applications and building structures. The source of vegetal fibres is different parts of plants such as bust for jute, hemp, flax, ramie, kenaf, leaf for sisal, banana and manila hemp (abaca), seeds in the case of cotton, coir and oil palm, wood and grass stem [1–4]. All vegetal fibres contain cellulose, hemicellulose and lignin in various proportions. For composite materials, reinforcement can be done with continuous fibres or with short fibres, in yarn or mat form. The ratio of mechanical properties and low weight, the possibilities to design different volumetric composition with effects on the mechanical, thermal, optical and electrical properties, and the environment protection and biodegradable properties of some natural composites are demonstrated by numerous pieces of research [5–7].

The main criteria for assessing new natural fibre composites were usually [8–10]: the degree of capitalization of the vegetal raw material and of other materials; the efficiency in use of raw material sources; volume or surface density; the limit values of the resistances to different mechanical stresses (traction parallel and perpendicular to/on the surface of the plate, bending, shearing in different planes and directions, detachment, fatigue, etc.); the rigidity and elasticity of the products, expressed by the values of the longitudinal and transversal modulus of elasticity; the physical properties of lignocellulose composite materials (profile density, swelling coefficient and water absorption in different environments and periods, degree of penetration of different substances, humidity, content in crystalline substances, etc.); the ecological properties (volatile emissions, etc.); and biodegradation and recyclability. The natural fibre reinforced polymers are characterized by two stress–strain states: an elastic behaviour as a response to a fast applied force and a viscous behaviour as response to a slowly applied force. The fibre reinforcement plays an important role in elastic behaviour of polymeric composite, as the matrix contributes to the viscous behaviour. For instance, the most efficient volume fractions of fiber content is maximum 50–65 m% according to [11]. The tests to determine the flow and relaxation phenomena provide important information about the dimensional stability of the polymer [12,13]. Some studies have shown that the natural fibres used for reinforcement are thermally unstable above 200 ◦C, for which reason the used matrices are based on polyethylene (PE), PP, polyolefin, polyvinylchloride and polystyrene and thermosets (which can be cured below this temperature) [11–16].

Most studies on wood reinforced composites (plastic wood composites) have analysed either the influence of the particle volume fraction/resin, or the influence of the size of wood particles, or the influence of the fibre type used as reinforcing elements on the mechanical properties of composites [14,15]. Mechanical properties of wood plastic composites (WPC), determined by tensile and bending tests, show an increase of the elasticity modulus E and yield modulus at the same time with the increasing of the particle size between 0.1 mm and 0.2 mm. These mechanical parameters are lower for composites with larger particles (1.00–4.00 mm) [8,12–16]. The mechanical characteristics of plastic composites with wood sawdust as filler material were studied in previous research by [14–18]. They carried out a comparative analysis of the effects of water/seawater absorption on the degradation of mechanical properties obtained at bending, in the case of hybrid composites made of polyester resin reinforced with fiberglass E and wood flour as filler. Thus, the maximum flexural stress decreases by almost 33% after immersion in sea water of wood sawdust mixture and epoxy resin during 6572 h [15,16]. Other studies are addressed to degradation of lignocellulosic composites both by weathering, ageing or humidity, because, from a mechanical point of view, the exposure of lignocellulosic composites to different aggressive factors can produce modifications of the elastic characteristics, and can induce stress concentrations at the surface of the polymeric material (which can lead to its premature failure) [19–24]. The modifications of the elastic-dynamic properties of a composite made of small waste oak particles of various sizes and a polyester resin subjected to photo degradation by UV radiation and thermal degradations were studied by [16,25–31]. From the point of view of the change of the energy loss capacity (internal friction) after the exposure to UV radiations, the most stable are the specimens reinforced with 0.4 mm particles, whereas the most unstable are those reinforced with 1 mm particles [16]. Ref. [32] studied new lignocellulose composites with carbon nanotube, having improved the mechanical properties, stability and fire retardment. The mechanical behaviour of fibre-reinforced composites depends on the size of the fibres (diameter and length), their distribution in the composite structure, the strength and elasticity of the fibres, the chemical stability and thermal resistance of the matrix and the nature of the fibre-matrix interface. Depending on the main directions of stress in the final product, the layers can be oriented differently so as to ensure a multidirectional distribution of stress during applications [33,34]. Numerous studies have shown that the method of reinforcement, length and nature of the fibres influence the mechanical properties of the composite. There is a linear relationship between the increase of the fibre content and the increase of the elastic modulus of the composite [35–37]. The complexity of the composite behaviour consists of the differences between the rigidity of the components that lead to the development of shear stress at the interface between the matrix and the fibres [38–40]. The determination of the mechanical and dynamic properties of composites reinforced with natural fibres plays a key role in the different structures and applications for both the proper design of the structures and the prediction of their

lifetime cycle. The finished products from natural fibre reinforced composites obtained from wood waste or hemp or flax and polyester resin are exposed to environmental and technological risks such as high relative humidity, temperature, UV radiation, vibrations, etc., reducing their resistance to different loads. The mechanical and dynamical characterisation of lignocelluloses composites based on wood particles, flax woven and polyester resin have not been investigated yet.

The aim of this paper is to examine the mechanical and dynamical properties of three different types of three categories of composites (wood particle reinforced composites, hemp mat reinforced composites and flax reinforced composite with mixed wood particles). The mechanical characterization of natural fibre composites is very important from the design and analysis as well as from the life prediction point of view, and it was obtained by a uni-axial tension test according to SR EN ISO 527-4, where the longitudinal Young's modulus, tensile of rupture and strain at break are obtained. The polyester resin as a matrix in the studied composites is used for outdoor applications due to its high resistance to environment factors. The natural fibre composites proposed and studied in this piece of research have a good potential to be used in different parts of automotive structures as door panels, ornaments or for the indoor parts of boats. The effects of thermal degradation in terms of glass transition and heat deflection temperature were detecting by thermal dynamical mechanical analysis. Dynamic mechanical analysis (DMA) was carried out in order to provide quantitative information about the performance of material to cyclic stress and variation of temperature. Based on dynamic properties determined by DMA (stiffness, energy dissipation), the viscous-elastic properties of manufactured natural composites can be improved in order to increase the quality of composites. For the producers of composite materials and for the users of the products made from them, the present research offers information about the possibilities of use versus limitations depending on the determined mechanical properties.

### **2. Materials**

Because the aim of this study is to compare the mechanical characteristics of composites reinforced with natural fibres, the sample preparation is presented for each type of specimens: wood particle reinforced composites abbreviated as WPC, hemp mat and polyurethane resin abbreviated as HMP (in longitudinal and transversal direction of mat fibres), flax and wood particle reinforced composites (abbreviated as FWPC). The specimens have the specific shape and dimensions of tensile test composite materials, according to ASRO SR EN ISO 527 and were made by a hand lay-up process.

### *2.1. Natural Fibre Composites*

### 2.1.1. Wood Particle Reinforced Composite (WPC)

The WPC specimens were obtained by mixing wood particles with polyester resin. The oak wood particles resulted from the mechanical processing of the wood logs from a Romanian factory. Oak particles were conditioned at 5–6% moisture content and their specific gravity was established by a pycnometer [41]. For the production of the lignocellulosic composites, in the first stage, the wood particles were sorted according to their size using granulometric sieves. Five classes of oak wood particles were obtained: less than 0.04 mm (coded WPC 0.04); between 0.04 ÷ 0.1 mm (coded WPC 0.10); between 0.1 ÷ 0.2 mm (WPC 0.20); between 0.2 ÷ 0.4 mm ((WPC 0.40); between 0.4 ÷ 1 mm ((WPC 1.00) and from 1 mm to 2 mm (WPC 2.00) (Table 1). From each type of grain sizes used as reinforcement, five specimens for the tensile test were prepared using the same volume fraction of 25% in a mixture with 440-M888 POLYLITE type polyester resin (Table 2), obtaining a total of 25 WPC type samples) (Figure 1a). For the DMA test, 2 specimens for the test with constant temperature and 2 specimens for the test with temperature variation (Figure 1b) were prepared from each type of wood particle reinforced composites The physical features of WPC samples for the tensile test are shown in Table 1 and for the DMA test in Table 3.


**Table 1.** The physical characteristics of samples for the tensile test.

**Table 2.** The characteristics of the polyester resin type 440-M888 Polylite, at 23 ◦C.


**Table 3.** The physical characteristics of the samples for the DMA test (Legend: DMA operated under constant temperature T = const.; DMA operated under temperature variation T).


**Figure 1.** The specimens made from wood particles and polyester resin: (**a**) specimens for the tensile test; (**b**) samples for DMA.

### 2.1.2. Hemp Mat Reinforced Composites

The analysed composite material contains hemp mat and polyurethane resin (RAIGITHANE 8274/RAIGIDUR CREM), with 50% percent of reinforcing natural fibres. This type of natural fibre composite is used in the automotive industry for the interior panel of the car door. To evaluate the mechanical behaviour of composite reinforced with hemp fibres subjected to the tensile test, 5 samples on the longitudinal direction of the mat, coded HMPL were cut from the plate and 5 samples on the transversal direction of the mat, coded HMPT (Figure 2a). For the DMA test, 2 specimens for the test with constant temperature and 2 specimens for the test with temperature variation (Figure 2b) were prepared from each type of hemp mat reinforced composites. The physical features of hemp mat samples for tensile testing in Table 1 and for the DMA test in Table 3.

**Figure 2.** The specimens made from hemp mat reinforced and polyurethane resin: (**a**) geometry for tensile test: HMPL—sample cut in longitudinal direction; HMPT—sample cut in transversal direction; (**b**) geometry for the DMA test: HMPL—sample cut in longitudinal direction; HMPT—sample cut in transversal direction.

### 2.1.3. Flax Reinforced Composites

For these tests, the specimens were made of polyester resin reinforced with 6 layers of flax fabric and oak wood particles with dimensions between 0.1 ÷ 0.2 mm, arranged between layers as it can be seen in Figure 3. The 6 layers of fabric have the same orientation of the warp and weft threads, respectively. The total volume percentage of reinforcement with natural fibres, in this case is approx. 30%. Samples for the tensile test were cut from the composite plate on the two main directions of the fabrics, respectively the warp direction (named FWPC\_L) and the weft direction (named FWPC\_T) (Figure 3a). For the DMA test, from each type of FWPC 2 specimens for the test with constant temperature and 2 specimens for the test with temperature variation were prepared (Figure 3b). The geometrical characteristics of FWPC for the tensile test are presented in Table 1 and for the DMA test in Table 3.

**Figure 3.** The specimens made from flax and wood particles reinforced and polyester resin: (**a**) samples for tensile test: FWPC\_L—samples cut in longitudinal direction; FWPC\_T—samples cut in transversal direction; (**b**) samples for DMA test.

### *2.2. Experimental Set-Up*

### 2.2.1. Tensile Test

To determine the elastic characteristics of a material, the samples were subjected to a static tensile test. In this study, for the analysis of the mechanical behaviour of the composites, the specimens were tested on the universal testing machine LS100 Lloyd's Instrument belonging to the Mechanical Engineering Department of Transilvania University of Bras,ov. The specimens were loading with a constant speed of 1 mm/min until breaking. The elongation was measured simultaneously with loading using extension device (Figure 4a).

For data acquisition, the Nexygen Plus software was used. After the tensile tests (according to SR EN ISO 527-4), the characteristic curve, the specific deformation, the longitudinal elastic modulus, the rupture tension of each reinforced composite were determined, and on the basis of the load curves, the average deformation energy for each type of sample was calculated. The fracture of samples was analysed with optical devices.

**Figure 4.** The experimental devices: (**a**) sample WPC 2.00 during the tensile test (Legend: 1—sample; 2—extension device for the elongation measurement; 3—tensile test machine jaws); (**b**) experimental set-up for the flexural test (3 points bending); (**c**) the DMA equipment (Legend: 1—sample; 2—sample supports; 3—loading device; 4—conditioner chamber; 5—temperature sensor).

### 2.2.2. Dynamic Mechanical Analysis

The rheological characteristics of the natural fibre composites were measured with the dynamic mechanical analysis by using the Dynamic Mechanical Analyzer DMA 242C Netzsch Germany at the Institute of Research and Development for Technical Physics in Ia¸si. The method is based on ASTM D7028-07 which covers the procedure for the determination the glass transition temperature of polymer composites under the flexural oscillation mode. Thus, the complex modulus E \*, with its two components (the conservation modulus E' and the loss modulus E") and the damping factor tan δ were determined in two cases: under isothermal conditions (T = 30 ◦C) for 30 min and with temperature variation between 30 to 120 ◦C, for 45 min. The specimens having the shape and geometry as it can be seen in Figure 4b were subjected to a flexural test. The input data were set up to 6 N for the applied force with frequency of 1 Hz. In Table 3, the specimen features for the DMA test are presented. For the DMA test, the samples have the same width and thickness as in the case of the tensile test and the length between supports is standard being set-up at 45 mm (Figure 4c).

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

The results in terms of qualitative and quantitative values of mechanical properties of natural fibre composites will be presented successively and then as a comparison. In the case of wood particle reinforced composites, it was noticed that, the smaller the reinforcing particles, the better the mechanical properties to traction. Nevertheless, the highest values of the elasticity modulus and of the tensile strength were obtained in the case of the composites reinforced with 0.2 mm particles. This is highlighted in the literature as well in the case of the polypropylene matrix composites (PP) [42]. The mechanical properties of the composites made of polypropylene and wood particles (WPC) in the case of tensile and bending tests show an increase in the values of the elasticity and resistance modulus simultaneous with the increase of the particle size between 0.25 and 2 mm, which is then followed by a slight decline of these values for larger particles (2–4 mm) [42]. The more elongated the particles, the more the mechanical properties increase because the contact between the reinforcing elements and the matrix occur over a larger surface. The first debonding of the matrix and the dispersed fibres occur near the breaking point. The characteristic curves for the same category of composites reinforced with wood particles did not display a large dispersion of values, the mixture between matrix and fibres being homogeneous (Figure 5).

**Figure 5.** Characteristic curve stress–strain of wood particle reinforced composites: (**a**) stress–strain characteristic curves ofWPC0.04 specimens; (**b**) stress–strain characteristic curves ofWPC0.10 specimens; (**c**) stress–strain characteristic curves of WPC0.20 specimens; (**d**) stress–strain characteristic curves of WPC0.40 specimens; (**e**) stress–strain characteristic curves of WPC1.00 specimens; (**f**) stress–strain characteristic curves of WPC2.00 specimens.

Tensile tests performed by other researchers have shown that natural fibre fabric reinforced composite materials have major differences in the mechanical properties of traction in the direction of the warp and weft [42–45] as it can be seen in the case of HMP specimens (Figure 6a,b) and FWPC specimens (Figure 6c,d). It is known that the fibre mat is used as reinforcement to assure a quasi-isotropy of composite plates due to the random orientation of the fibres. Despite this assumption, the tests performed on hemp mat composites showed that there are significant differences between the tensile properties in the two directions of the mat (longitudinal and transversal direction). In the case of flax wood particle reinforced composites FWPC, the trend is similar to HMP regarding the mechanical properties in the weft and warp direction. Equally, it can be observed that HMP show a great spread of stress–strain curves in the longitudinal direction (Figure 6a) by comparison to FWPC which indicated a great dispersion of the curves in the transversal direction (Figure 6d).

*Polymers* **2020**, *12*, 1402

**Figure 6.** Characteristic curves stress–strain of natural fibres reinforced composites: (**a**) stress–strain curves for HMP\_L samples; (**b**) stress–strain curves for HMP\_T samples; (**c**) stress–strain curves for FWPC\_L samples); (**d**) stress–strain curves for FWPC\_T samples.

In Figure 7, it can be noticed that the FWPC samples present a rigid behaviour compared to the HMP\_L samples which behave viscously. The WPC0.20 samples are also rigid by comparison to the other wood particle reinforced composite.

**Figure 7.** Comparison of stress–strain curves for tested samples.

The mechanical properties of tested samples in terms of the average values of the longitudinal elasticity modulus, stress at break, the strain and stiffness in percent are summarized in Table 4. In the case of WPC specimens, it can be observed that the modulus of elasticity varies from 2877 MPa, for the specimens reinforced with particles about 1 ÷ 2 mm, up to the maximum value of 4012 MPa, for the specimens reinforced with particles about 0.1 ÷ 0, 2 mm. The minimum values of the tensile strength were noted for specimens reinforced with particles about 1 ÷ 2 mm (WPC 2.0), being 19.5 MPa, and the maximum being 26 MPa, for specimens reinforced with particles from 0.1 to 0.2 mm (WPC 0.2). For the HMP specimens cut in the longitudinal direction, an average tensile strength value of 26 MPa was obtained, and in the weft direction 32 MPa, approximately 23% higher than in the longitudinal direction. In the case of the longitudinal modulus of elasticity, it is observed that, in the transverse direction, its value is approximately 63% higher than in the longitudinal direction (Table 3). For the FWPC specimens cut in the warp direction, tensile strength values between 23.9 and 27.3 MPa were obtained, and in the weft direction between 31.4 and 42 MPa. The longitudinal elasticity modulus is higher in the transversal direction (weft) by almost 95% than in the warp direction (longitudinal). Although the mechanical properties of lignocelluloses samples are relatively low, they can be used in some applications as exterior and interior products, being valuable for the possibility of integrating wood residues from processing operations or from recycled wood in the form of chips and fibres.


**Table 4.** Average values of elastic characteristics obtained after the tensile test. Legend: E—longitudinal elasticity modulus; STDV—standard deviation; σr—tensile of rupture; εr—percentage strain at break; k—stiffness.

### *3.1. Dynamic Mechanical Analysis*

### 3.1.1. Isothermal Conditions

The rheological characteristics of the natural fibre composites in terms of complex modulus E \*, with its two components (the storage modulus E' and the loss modulus E") and the damping tan δ were determined with the DMA. The storage modulus E' represents the capacity of materials to withstand the applied loading, being the expression of the elastic constant of the composite. The energy dissipation due to the internal friction of the material is called loss modulus E" and it represents the viscous modulus. The ratio between the loss and storage modulus represents damping (tan δ). tan δ is an indicator of how efficiently the material loses energy to molecular rearrangements and internal friction [46]. In Figures 8–10, the elastic and viscous responses of samples to dynamic loading with frequency of 1 Hz can be noticed. The capacity of WPC samples to store the deformation energy decreases slowly while increasing the time of loading. Similar behaviour is noticed in the case of the HMP samples with the mention that there are clear differences between the samples cut in the longitudinal direction compared to those cut in the transverse direction (Figures 8, 9 and 10b).

**Figure 8.** Variation of the complex modulus in time: (**a**)WPC samples; (**b**) HMP samples; (**c**) FWCP samples.

**Figure 9.** Variation of the storagemodulus E'in time: (**a**)WPC samples; (**b**)HMP samples; (**c**) FWCP samples.

**Figure 10.** Variation of the loss modulus E" in time: (**a**) WPC samples; (**b**) HMP samples; (**c**) FWCP samples.

Both types of samples have a similar viscous-elastic behaviour at the beginning, so that, after 10 min of stress, the samples cut in the longitudinal direction show a stiffening phenomenon compared to the samples cut in the transverse direction whose elastic behaviour decreases suddenly after 18 min of cyclic loading (Figures 8 and 9c). As far as the values of complex modulus and storage modulus for each type of the tested samples are concerned, it can be noticed that the lower values are noticed in the case of the HMP sample by comparison to WPC and FWPC. Initially, there are 1120–1260 MPa and, after 5 min, they decreased by 12.5% in the case of HMP\_L maintaining constant value for remaining time exposure to cyclic loading. The HMP\_T indicated a higher decreasing by almost 28% after 10 min and then a stabilization of the values at around 790 MPa. In the case of the WPC samples, the complex modulus ranges between 3250 MPa (WPC0.40) and 4270 MPa (WPC0.20). It can be noticed that the values for the flexural test in a dynamic regime are similar to Young's modulus values obtained in the tensile test. FWPC samples obtained the highest values for the dynamic modulus (5900–6060 MPa) in comparison to the other types of specimens. The loss modulus increases by increasing the time exposure to cyclic loading in the case of WPC and FWPC samples while the HMP samples indicated a decrease of this viscous component (Figure 10). This behaviour is due to the type of matrix: the WPC and FWPC, which contain as matrix polyester resin, indicated a slightly higher network density and it is slightly more cross-linked than the HMP sample based on polyurethane resin. The minimum values of energy dissipation due to internal frictions are indicated in the case of HMP samples. In the longitudinal direction (HMP\_L), the viscous modulus decreases by 28% during the cyclic loading and, in the transversal direction (HMP\_T), the decrease is 50%. For FWPC, the internal frictions increase by increasing the exposure time to loading; the overall value varied from minimum 465 MPa to maximum 560 MPa (in the case of FWPC\_T) and from minimum 490 MPa to maximum 565 MPa (in the case of FWPC\_L). For the WPC samples, the loss modulus varied in accordance with the wood particle sizes: the highest value of energy dissipation is recorded for smaller wood particles (WPC0.04). Regarding the variation of loss modulus, two groups can be noticed: WPC0.04, WPC0.10, and WPC0.20 indicated

an increase in internal friction and WPC1.00, WPC2.00, WPC0.40 indicated a slight decrease of the viscous modulus.

The damping tan δ as a ratio between the loss and the storage modulus is an expression of the energy dissipation of a material under cyclic load, and it depends on the interface and adhesion between fibres and matrix. Any rigid material is characterized by a high damping value, whereas any ductile material indicates a low damping value [47]. In this sense, the damping tan δ for the HMP samples tends to decrease by increasing the loading time, since, for the WPC and FWPC samples, the damping increases (Figure 11).

**Figure 11.** Variation of the damping tan δ in time: (**a**) WPC samples; (**b**) HMP samples; (**c**) FWCP samples.

### 3.1.2. Temperature Variation

The viscous-elastic behavior of composites is emphasized by the temperature variation during cyclic loading. The stiffness and rigidity stability of composites at a certain temperature can be observed on the storage modulus curves. Thus, similar behavior between the WPC and FWPC samples can be noticed in Figure 12a,c: the glassy region presented between 30 ◦C and 40 ◦C characterized the rigidity of composites due to polymeric chains; in the second region (40–80 ◦C), the storage modulus decreases drastically because by increasing the temperature, the internal friction in polymeric chains is accelerated leading to a rubbery region. The HMP composite behavior differs, their stiffness being affected right from the start of the test and decreasing by temperature increase (Figure 12b). Despite this behavior, the reinforcement with hemp fibers and also the reinforcement with flax fabric lead to higher values of storage modulus by comparison to wood particle reinforced composites.

**Figure 12.** Variation of the storage modulus E' with temperature: (**a**) WPC samples; (**b**) HMP samples; (**c**) FWCP samples.

Figure 13 presents the effect of reinforcement on the damping as a function of temperature at 1 Hz frequency. From this type of chart, the glass transition temperature (*T*g) can be extracted from the peak of damping variation curves.

**Figure 13.** Variation of the damping tan δ with temperature: (**a**) WPC samples; (**b**) HMP samples; (**c**) FWCP samples.

According to [48–50], lower damping values represent the improved interfacial interaction as it can be noticed in the case of FWPC and HMP (Figure 13b,c), while a higher damping value is recorded in the case of poor interfacial adhesion as it can be observed in Figure 13a—the WPC samples. Thus, the *T*g for the WPC samples is around 78 ◦C (for 2.00 mm and 1.00 mm wood particle sizes) and 82 ◦C for smaller wood particles. The *T*g for FWPC is lower than the one for WPC, being 70 ◦C. The HMP samples indicated a different behaviour. In Figure 14, the Cole–Cole charts are shown in order to analyse the connection between stored and loss heat energy for the studied samples. As [49–51] highlights, this kind of plot is useful to interpret the modification of viscous-elastic material with different reinforcement, as it is illustrated in Figure 14a—the WPC samples. The mixture of wood particles with polyester matrix leads to relative homogeneity of composite by comparison to the FWPC and HMP samples which is visible from the shape of Cole–Cole curves. The random fibres from the hemp mat structure lead to a heterogeneous nature of the composite structure.

**Figure 14.** Cole–Cole plot: (**a**) WPC samples; (**b**) FWCP samples; (**c**) HMP\_L sample; (**d**) HMP\_T sample.

### *3.2. Fracture Analysing*

In Figure 15, the ways of breaking of the tested composites are shown. Thus, it is noticeable that the breakage of the composite reinforced with wood particles (WPC) is produced through the simultaneous destruction of the matrix and of the dispersed fibres (Figure 15). In the case of the HMP samples, the matrix is fractured first; then, the dispersed fibres break. The plane of fracture is obtained on the area with the minimum resistance of the interface between the matrix and the fibres or in the area where the fibres are missing or do not have a good adhesion with the matrix. For the FWPC composites which contain both dispersed fibres (wood particles) and flax fabric, the fracture is produced first in the matrix mixed with wood particles and then the layers of flax fabric fail. It is appreciated that the use of a fabric structure (flax) doubles the tensile strength compared to mat reinforcement. The mechanism of failure differs between short fibres/particles reinforcement and long fibres [52,53]. The tension stress causes interface debonding. [53–56] considered that the final interface between the short fibres and the matrix is easy to debond in the loading process. At the end of the final interface, the stress transfer from the matrix to the fibres depends on the shear stress only on the axial interface (Figure 16a). In the case of long fibre composites, the interface stress (shear stress) is higher due to the adhesion of resin to fibres (Figure 16b).

**Figure 16.** Mechanism of failure modes: (**a**) the case of short fibres reinforced composites; (**b**) the case of long fibres reinforced composites.

### **4. Conclusions**

The mechanical behaviour of composite materials from three types of natural fibres was studied. The results demonstrated that both tensile and rheological behaviour depends on the size of fibres, disposure of fibres (randomly or fabric) and the type of matrix. The mechanical properties of natural fibres composites differ within the same type of composite. For instance, WPC0.20 recorded the higher values of Young's Modulus and tensile strength in comparison with WPC2.00 which have the lower values. It can be concluded that the best wood particles size is 0.20 mm from mechanical point of view. In case of HMP and FWPC, there is strong relation between direction of loading and weft/yarn direction. The longitudinal elasticity modulus is higher in the transversal direction (weft) by almost 80–95% than in the warp direction (longitudinal). The flax woven reinforcement in case of FWPC leads to the best mechanical properties from all types of tested composites.

Regarding DMA results, HMP and also FWPC present higher values of storage modulus by comparison to wood particle reinforced composites although the increasing temperature produces a decreases in viscous-elastic behaviour for all types of samples. The *T*g for all tested natural fibres composite range between 65 and 85 ◦C, the matrix having the main role in modification of polymers stiffness.

The HMP presents a good capacity to absorb the energy of deformation, partially due to polyurethane resin and to the type of dispersed fibres. In the case of a structure made of lignocelluloses composite materials, consisting of layers reinforced with natural fibres as fabrics and layers reinforced with wood particles, by adding an additional layer reinforced with wood particles on the visible surface of the panels. Thus, structures with superior aesthetic properties can be made which no longer require coating with other materials. Although the structure is no longer symmetrical, it has visible surfaces with natural textures whose colour can be changed by using wood particles of different species.

The further work will focus on simulation of mechanical behaviour of complex structures made from natural fibres composites using the elastic properties determined in experimental tests and predicting the stress and strain states of them.

**Author Contributions:** Conceptualization, M.D.S. and F.T.; methodology, H.T.D. and O.M.T.; software, H.T.D. and O.M.T.; validation, M.D.S. and F.T.; formal analysis, H.T.D. and O.M.T.; investigation, M.D.S. and O.M.T.; resources, F.T.; data curation, H.T.D.; writing—original draft preparation, O.M.T. and M.D.S.; writing—review and editing, H.T.D. and F.T.; visualization, M.D.S.; supervision, O.M.T.; funding acquisition, M.D.S. and F.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Sectorial Operational Program Human Resources Development (SOP HRD), financed from the European Social Fund and by the Romanian Government under the contract number POSDRU/88/1.5/S/59321 and Program partnership in priority domains PNII under the aegis of MECS UEFISCDI, project No. PN-II-PT-PCCA-2013-4-0656.

**Acknowledgments:** We are grateful to the manager Dorin Rosu, PhD and to the technical staff of S.C. Compozite S.A., Brasov, Romania, manufacturer of composite products for supplying the specimens for the experimental research of the present article. We are grateful to Savin Adriana, head of laboratory "Non-destructive tests" of the National Institute of Research and Development for Technical Physics in Iasi, for the DMA tests.

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

### **Abbreviations and Notations**


### **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* **Sustainable Micro and Nano Additives for Controlling the Migration of a Biobased Plasticizer from PLA-Based Flexible Films**

**Laura Aliotta 1,2,\*, Alessandro Vannozzi 1,2, Luca Panariello 1,2, Vito Gigante 1,2, Maria-Beatrice Coltelli 1,2,\* and Andrea Lazzeri 1,2**


Received: 20 May 2020; Accepted: 13 June 2020; Published: 17 June 2020

**Abstract:** Plasticized poly(lactic acid) (PLA)/poly(butylene succinate) (PBS) blend-based films containing chitin nanofibrils (CN) and calcium carbonate were prepared by extrusion and compression molding. On the basis of previous studies, processability was controlled by the use of a few percent of a commercial acrylic copolymer acting as melt strength enhancer and calcium carbonate. Furthermore, acetyl n-tributyl citrate (ATBC), a renewable and biodegradable plasticizer (notoriously adopted in PLA based products) was added to facilitate not only the processability but also to increase the mechanical flexibility and toughness. However, during the storage of these films, a partial loss of plasticizer was observed. The consequence of this is not only correlated to the change of the mechanical properties making the films more rigid but also to the crystallization and development of surficial oiliness. The effect of the addition of calcium carbonate (nanometric and micrometric) and natural nanofibers (chitin nanofibrils) to reduce/control the plasticizer migration was investigated. The prediction of plasticizer migration from the films' core to the external surface was carried out and the diffusion coefficients, obtained by regression of the experimental migration data plotted as the square root of time, were evaluated for different blends compositions. The results of the diffusion coefficients, obtained thanks to migration tests, showed that the CN can slow the plasticizer migration. However, the best result was achieved with micrometric calcium carbonate while nanometric calcium carbonate results were less effective due to favoring of some bio polyesters' chain scission. The use of both micrometric calcium carbonate and CN was counterproductive due to the agglomeration phenomena that were observed.

**Keywords:** poly(lactic acid); poly(butylene succinate); plasticizer migration; diffusion

### **1. Introduction**

The environmental impact of plastic wastes, also due to the limited disposal methods, is a continuously growing public concern worldwide. This problem has encouraged research and industrial interest on biobased and biodegradable polymers that could overcome sustainability issues and environmental challenges posed by the production and disposal of oil-derived plastics [1,2]. In packaging and agriculture applications, the use of bio-based biodegradable polymers is a strong advantage both for environment and customers [3]. However, it is expected that the demand for these biopolymers will increase and, over the coming few decades, bioplastic materials can complement and gradually substitute the oil-based plastics in different sectors [2,4]. Nowadays different biopolymers

can be found not only in food and agriculture applications, but also in the medicine and cosmetic sectors. The development of renewable polymeric modulated materials tailored for specific applications is a subject of active research interest worldwide [5].

Poly(lactic acid) (PLA) is particularly interesting because it exhibits mechanical properties (Young's modulus of about 3 GPa, tensile strength between 50 and 70 MPa with an elongation at break of around 4%, and an impact strength close to 2.5 kJ/m2) that make it useful for a wide range of applications [6,7]. It can be obtained from renewable resources (e.g., corn, wheat, or rice) and it is not only biodegradable and compostable, but also recyclable. However, in particular, its biocompatibility makes it appealing for biomedical and cosmetic applications. If compared to other commercialized biopolymers (such as poly(hydroxylalkanoates) (PHAs), poly(caprolactone) (PCL), and starch), PLA is easily processable [8]. However, PLA's brittleness and its poor heat resistance limit strongly the application of PLA. The improvement of PLA's toughness can be reached in different ways (that can also be adopted contemporary): plasticization, copolymerization, and melt blending with flexible polymers [9,10]. However, the blending technique is the common method adopted to overcome the PLA brittleness making it useful in those applications (like film production) where high flexibility and toughness are requested. In literature different successful studies can be found where PLA was blended, in different quantities, with other biodegradable polymers such as: polycaprolactone [11], poly(vinyl alcohol) (PVOH) [12,13], polybutylene(succinate) (PBS) [14–16], poly(butylene succinate-co-adipate) [17–19], and poly(butylene adipate-co-terephthalate) (PBAT) [20–22].

Furthermore, the combination of additives such as plasticizers into biopolymers and their blends, is a common practice to further improve the mechanical flexibility, and processability limitations. Generally, plasticizers are a class of low-to-medium molecular weight compounds up to a few thousand, it is expected that their demand will increase reaching approximately the 9.75 million of tons in 2024 [23]. In various applications, like film formation or coating dispersion, the adequate selection of plasticizer greatly improves the processing; however, the choice of these plasticizers, especially in biobased applications, is limited by the required safety, environmental favorability, and chemical and physical properties that dictate their miscibility [2].

It is therefore evident that, unlike classic plastic commodities (e.g., HDPE) which possess good starting physical-chemical properties, biopolyesters need to be improved and in this context biopolymers can release a major amount of additives that are not covalently bound to the polymer, at all stages of the plastic's lifecycle via migration of liquids or solids or via volatilization [24]. The result of this release is the transfer of chemicals (such as plasticizers) that can affect human health and can contaminate soils and water [25,26]. In fact, especially in natural environments, liquid additives can quickly migrate out of plastic and can be absorbed by roots affecting the plant development [27]. On the other hand, a controlled release of beneficial substances contained in the blends are very useful in the cosmetics and biomedical sectors [28,29].

For all of these aspects, the study of migration is very important especially for films containing plasticizers that are not chemically attached to polymeric chains (not reactive plasticizers). Depending on certain conditions, liquid plasticizers can come out from the polymer matrix. During service and storage, this loss is problematic because it leads to unwanted changes of the mechanical properties (loss of flexibility and toughness with an increment of stiffness). Clearly, this decrement of the mechanical properties will be more evident and dangerous in those polymeric matrices that are brittle in the not plasticized state (like PLA). This plasticizer release, as well as creating problems of stiffening, can lead to variations in the crystallization of the samples due to the reorganization over time (aging) of the crystalline structures [30]. Even the plasticizer molecular weight and linearity of the plasticizer influence the migration and so an average plasticizer molecular weight is required to ensure long-term plasticizer retention in the polymeric matrix [31].

According to the type of plasticizer, different strategies can be adopted to prevent the plasticizer migration (such as internal plasticization [32], polymer surface modification [33,34], or addition of nanofillers or ionic liquid [35,36]). The methodologies and the advantages/disadvantages of these strategies are extensively reported in literature [37]. Among the strategies mentioned, the role of micro and nano filler addition on the plasticizer release is particularly interesting.

In fact, the use of fillers has a double effect: they can reduce plasticizer migration thanks to the creation of tortuosity that forces the plasticizer molecules to follow a longer path to leave the polymeric structure [38] and the adsorption mechanism. Fillers, especially nanoparticles or nanofibrils, are extremely small and they possess a large number of groups capable of interactions on their surfaces that make the absorption of other substances easy. On the basis of these characteristics, the plasticizer migration can be reduced both by the absorption on the surface of plasticizer molecules and by the steric resistance that makes the passage of the plasticizer difficult [37]. On the other hand, filler addition can improve the mechanical properties of the materials (such as toughness) according to the rigid filler toughening mechanism [39]. Various nano-scale and micro-scale fillers with different geometries (such as: montmorillonite, silica, calcium carbonate, and aluminum oxide) are reported to improve not only the properties of polymers such as toughness, stiffness, and heat resistance [40,41] but they also can limit the plasticizer migration [35,38]. For example, it has been observed that organic montmorillonite (OMMT), nano-SiO2, and nano-CaCO3 have a strong adsorption force and diffusion inhibition [35,37]. Therefore, the presence on the fillers surface of functional groups which are active and can easily bond with plasticizer molecules is another important aspect that influences the migration of the plasticizer [42]. It was observed that the migration rates decrease with the increasing of filler content [43]. However, it is essential to reach a homogeneous dispersion of the particles in the polymeric matrix to hinder the plasticizer migration. In fact, the increase of fillers content does not lead always to better barrier properties if there is a poor dispersion of nano particles in the matrix [35]. Clearly, great attention must be paid to the combination of polymeric plasticizer and micro/nano additives to control the eventual plasticizer migration and at the same time to reach an optimal combination of mechanical properties (good balance between stiffness and toughness).

Understanding the mechanism and the kinetics of plasticizer loss is fundamental also to evaluating the short- and long-term performance of the final plasticized material. For instance, if the correlation between mechanical properties and plasticizer concentration is known, it will be possible to predict the change of mechanical properties and, consequently, the lifetime of products. Furthermore, it will be also possible to develop new methods to eliminate/control the plasticizer migration.

In this paper, the attention is focused on the results of previous studies where suitable biobased skin compatible films were successfully prepared and investigated. Different additives were added: calcium carbonate, chitin nanofibrils as functional filler (dispersed by poly(ethylene glycol) (PEG)), melt strength enhancer (Plastistrength), and acetyl tributyl citrate (ATBC) plasticizer. The addition of all these additives has led to an improvement of the melt processability, of the mechanical properties (the films resulted flexible and high resistant) and of the antimicrobial and anti-inflammatory properties [44–47]. However, troubles were encountered during the storage of films where 4 wt % of micrometric calcium carbonate was present [44] due to the plasticizer migration. Interestingly, any data were reported about the effect on plasticizer migration of natural nanofibers obtained by sea food waste, chitin nanofibrils used in these bionanocomposites for their indirect anti-microbial activity [47].

In the present paper, the identification of the diffusion coefficients for the examined blends was carried out. Furthermore, in order to limit and/or to control the plasticizer migration, different types of commercial calcium carbonate particles (micro and nanometric) and chitin nanofibrils were used. The effect of their addition on the films processability and on the mechanical properties was carried out to ensure that no significant variations occurs in the already optimized rheological and mechanical properties. Finally, the effect of the introduction of CaCO3 and chitin nanofibrils on the ATBC migration was evaluated through the diffusion coefficient calculation. The diffusion coefficients were obtained, applying analytical correlations based on the Fick's second law, by regression of the experimental migration data plotted as the square root of time.

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

### *2.1. Materials*

The polymeric granules and additives used in this work for the blend's preparation are:


### *2.2. Characterization of Fillers*

A Brunauer-Emmett-Teller (BET) analysis of the chitin nano-fibrils and of the three calcium carbonate fillers was carried out, with a micromeritics instrument-Gemini V analyzer (Micromeritics, Atlanta, GA, USA)—in order to determine the surface area, the total porosity and the number of particles per gram of the fillers. The measurement was performed with the same procedure used in the previous works of Coltelli et al. [47].

### *2.3. Blends Preparation and Torque Characterization*

PLA/PBS blends preparation was carried out with a Haake Minilab II (Thermo Scientific Haake GmbH, Karlsruhe, Germany) co-rotating conical twin-screw extruder. Before processing the materials were dried in air circulated oven at 60 ◦C for at least 24 h. The molten materials were recovered in filaments for the subsequent tests. For each extrusion cycle, 6 g of PLA/PBS pellets, manually mixed with the other additives were fed through a little hopper into the mini-extruder. After the feeding, the molten material flowed in a closed circuit for 1 min at the end of which it was recovered. During this time, the torque was measured as a function of time, at least ten experimental measurements were performed for each blend compositions to guarantee the reliability and the consistency of the test. The final torque value is taken once that the melt stabilizes. The extrusion was carried out at 190 ◦C with a screw rotating speed of 110 rpm.

The compositions of the PLA/PBS blends investigated in this work, chosen considering the results of previous studies [44,47], are reported in Table 1.

To this purpose, the PLA/PBS ratio was maintained equal to 0.8 for all the formulations and also the ratio between ATBC and the polymeric matrix of PLA/PBS was kept constant and equal to 0.2. Furthermore, a fixed quantity (2 wt %) of Plastistrength (PST) was used for all blend compositions (except for F1 blend). The F3 formulation, according to previous studies [44,47] contained also a few percent (4 wt %) of micro-calcium carbonate (2AV). However, as mentioned, plasticizer migration was observed in these samples. Consequently, it has been decided to add a greater amount of calcium carbonate (7 wt %) and two other types of calcium carbonate with different particle size distributions to the blends to evaluate their effect on ATBC migration. F5 and F6 blends Smartfill and CCR were respectively added. The addition of chitin nanofibrils alone (F7 formulation) and coupled with the calcium carbonate particles, that showed better capability in hindering the plasticizer diffusion, was also investigated (F8 formulation).

For the preparation of the blends containing chitin nanofibrils (F7 and F8) PEG 6000 was used to achieve a better dispersion of the fibrils. The procedure adopted and the quantity of chitin nanofibrils chosen are described in the previous works of Coltelli et al. [47,48].


**Table 1.** Blends name and composition.

### *2.4. Melt Flow Rate*

A CEAST Melt Flow Tester M20 (Instron, Canton, MA, USA) equipped with an encoder was used to investigate the melt flow behavior of the blends. The encoder, following the movement of the piston, acquires the melt volume rate (MVR) data. For each blend, three tests were carried out following the standard ISO 1133D [49]. According to the ISO procedure, the sample is simply preheated for 40 s at 190 ◦C after that, a weight of 2.160 kg is released on the piston and then, after 5 s, a blade cuts the spindle starting the real test. Every 3 s, MVR value is recorded by the encoder. The molten material, flowing through the capillary of specific diameter and length, is recovered and the MFR value is obtained. Before the test, the pelletized filaments of each polymer blend were dried in an air oven at 60 ◦C for one day.

### *2.5. Thermal Characterization by Di*ff*erential Scanning Calorimetry (DSC)*

Differential scanning calorimetry (DSC) measurements were performed with a DSC TA Instruments Q200 (TA Instruments, New Castle, UK), equipped with a RSC cooling system. Indium was used as the standard for calibration with aluminum hermetic pans. About 10 g of material was analyzed for each blend. Nitrogen was used as purge gas at a rate of 50 mL/min. The samples were heated at 10 ◦C/min from −40 ◦C to 220 ◦C. Only the first scan was considered to take into account the samples thermal history. Glass transition temperature (*Tg*), melting temperature (*Tm*), cold crystallization temperature (*Tcc*), melting and cold crystallization enthalpies (Δ*Hm* and Δ*Hcc*) were determined by using the TA Universal Analysis software. In particular, the enthalpies of melting (Δ*Hm*) and cold crystallization (Δ*Hcc*) were determined from the corresponding peak areas in the heating thermograms; while the melting temperature (*Tm*) and the cold crystallization temperature (*Tcc*) were recorded at the maximum of the melting peak and at the minimum of the cold crystallization peak, respectively.

The crystallinity percentage (Xcc) of PLA in the blends was calculated as [50]

$$X\_{\mathcal{L}} = \frac{\Delta H\_{m,PLA} - \Delta H\_{\text{cc},PLA}}{\Delta H^{\text{o}}\_{m,PLA} \cdot X\_{PLA}} \tag{1}$$

where Δ*Hm,PLA* and Δ*Hc,PLA* are the melting enthalpy and the enthalpy of cold crystallization of PLA, *XPLA* is the weight fraction of PLA in the sample and Δ*H*◦ *m,PLA* is the melting enthalpy of the 100% crystalline PLA, equal to 93 J/g [51].

### *2.6. Tensile Test*

The mechanical properties of the blends were evaluated by tensile tests carried out at room temperature. An INSTON 5500R universal testing machine (Canton, MA, USA), equipped with a 100 N load cell, was used. The machine, interfaced with a computer running a MERLIN software (INSTRON version 4.42 S/N-014733H), was assembled with compressed air grips (initial grip separation: 25 mm). The crosshead speed was set at 100 mm/min. The preparation of tensile specimens was carried out using the pelletized strains come out from the micro-compounder. The pellets were dried in an oven at 60 ◦C for 24 h to avoid the water uptake; subsequently, they were used for the film preparation by compression molding. The pelletized materials were pressed between two Teflon sheets at 180 ◦C for 1 min with a pressure of 3 tons, using a NOSELAB ATS manual laboratory heat press.

The mechanical tests were performed on an ISO 527-2 type A [52] dumbbell specimens obtained from the films cut with a Manual Cutting Press EP 08 (Elastocon, Brahmult, Sweden). At least 10 specimens were tested for each sample and the average values were reported.

### *2.7. Scanning Electron Microscopy Analysis (SEM)*

Samples morphologies were investigated by scanning electron microscopy (SEM) with a FEI Quanta 450 FEG instrument (Thermo Fisher Scientific, Waltham, MA, USA). The micrographs of samples fractured with liquid nitrogen and sputtered with a layer of gold were collected. The metallic layer makes the surface electrically conductive, allowing the backscattered electrons to generate the images.

### *2.8. Migration Tests*

Films prepared by compression molding (adopting the same procedure of film preparation described in Section 2.5) were used for the migration tests. To evaluate the weight loss of films due to the ATBC migration, three pieces of film for each formulation were put between two paper sheets. In this way, thanks to the capillarity forces related to the ATBC absorption from the paper sheets, the plasticizer is removed from the surface of the film ensuring a migration kinetics controlled by diffusion. The samples were kept in an oven at 60 ◦C (above Tg) to make the test severe and to accelerate the migration process. Periodically, the films were weighed to estimate the weight loss as a function of

time. The migration tests were stopped after 1500 h. For each formulation, the percentage weight loss of the film as a function of time was determined with the following relationship.

$$\%wt\,\text{loss} = \frac{wt\_{t,fill} - wt\_{t0,film}}{wt\_{t0,film}} \cdot 100 \tag{2}$$

where *wtt,film* is the film weight at the time *t* and *wtt*0, film is the film weight at the beginning of the test (*t* = 0). To separate the effect of calcium carbonate (used as an additive for hinders the plasticizer migration), from that of the loss of ATBC, was calculated the percentage of weight loss normalized respect to the amount of ATBC initially present in the film

$$\% \text{wt lost } ATBC = \frac{\text{lost } ATBC \text{ mass}}{\text{initial } ATBC \text{ mass}} \cdot 100 \tag{3}$$

In this way, it is possible to evaluate the films' oiliness, thanks to the total weight loss of the film and appreciate the percentage by weight of ATBC that migrates from the film independently from the amount of plasticizer in the film.

### **3. Theoretical Analysis**

For the prediction of the migration of monomers or additives, mathematical models are often used. In particular, the second Fick's law is generally applied.

For a 3D system, the second Fick's law is expressed as

$$\frac{\partial \mathbf{c}}{\partial t} = D \left[ \frac{\partial^2 \mathbf{c}}{\partial \mathbf{x}^2} + \frac{\partial^2 \mathbf{c}}{\partial y^2} + \frac{\partial^2 \mathbf{c}}{\partial z} \right] \tag{4}$$

If the diffusion is one-dimensional, that means that there is a gradient concentration only along one axis (for example x-axis), Equation (4) can be written in a simpler form (Equation (5)).

$$\frac{\partial \mathbf{c}}{\partial t} = D \frac{\partial^2 \mathbf{c}}{\partial \mathbf{x}^2} \tag{5}$$

Commonly, the prediction of a substance migration through a polymer matrix above its glass transition temperature is described by the Fick's second law based on one directional transfer (Equation (5)) [53]. General solutions of the diffusion equation can be adopted for a variety of initial and boundary conditions, provided the diffusion coefficient is constant. Crank and Vergnaud [54,55] proposed and classified different types of solutions for different geometries and boundary conditions. Taking into account the sample geometry adopted for the migration experiments, it is possible to apply one of the Crank's solutions of the Fick's second law. The equation used is reported below (Equation (6)) and it has been adopted successfully for plasticized polymeric systems [56,57]

$$\mathbf{C}\_{\mathbf{x}} = \frac{1}{2} \mathbf{C}\_{0} \left[ \text{erf}\left(\frac{h-\mathbf{x}}{2\sqrt{D\cdot t}}\right) + \text{erf}\left(\frac{h+\mathbf{x}}{2\sqrt{D\cdot t}}\right) \right] \tag{6}$$

The parameters involved in Equation (3) are:


It can be observed that Equation (6) is symmetrical about *x* = 0 this means that the system can be cut in half by a plane at *x* = 0 without affecting the concentration distribution. It must be pointed out that the equation was obtained assuming that the migration occurred from the plasticized matrix to the same pure matrix, whereas our experiments were performed using paper, thus by removing the migrated plasticizer on the surface. Furthermore, if it is assumed that the plasticizer is initially distributed with a known concentration in the film and, if it is also supposed that the diffusion coefficient of the plasticizer in the polymer can be treated as a constant, a simple form of the Crank's equation can be obtained by Equation (7) [54,58]

$$M\_{\rm l} = 2 \text{C}\_{\rm P0} \sqrt{\frac{Dt}{\pi}} \iff \frac{M\_{\rm l}}{\mathcal{C}\_{\rm P0}} = \sqrt{D} \cdot 2 \sqrt{\frac{t}{\pi}} \tag{7}$$

where *Mt* (mg/cm2) is the total plasticizer lost from the film at time *t* (s), *Cp*<sup>0</sup> (mg/cm3) is the initial migrant concentration in the polymer and D (cm2/s) is the diffusion coefficient. Equation (7) also assumes that the film is sufficiently thick that the concentration of plasticizer at the mid-plane remains at its original value (*Cpo*) and this can be physically achieved if less than about of 15–20% of the plasticizer is lost [58]. If experimental migration data are available, the diffusion coefficient can be thus obtained by linear regression of the migration data as function of the square root of time (according to Equation (7)). Clearly, to adopt Equation (7), there must be sufficient data points available in the early part of the migration graph. Nevertheless, the principal criticism of this simplified equation is that the diffusion coefficient of plasticizer varies with the plasticizer concentration and thus this equation can be applied only for small amounts of plasticizer (low concentration). In fact, it is known in [59,60] that the diffusivity increases with the plasticizer concentration due to the increment of free volume and mobility of the polymer caused by the plasticizer addition. This dependence of diffusion with plasticizer concentration is well described by the exponential equation [54,61]

$$D(\mathbb{C}) = D\_{\mathbb{c}0} \mathbf{e}^{\mathbf{a}\mathbb{C}} \tag{8}$$

where *Dc*<sup>0</sup> is the zero-concentration diffusivity, and a is the plasticization coefficient related to the plasticizer efficiency. When the diffusivity is a function of the concentration, the differential equation (Equation (5)) is not linear. Normally for migration modeling the diffusion coefficient (*D*) is seen as concentration independent, which in most cases (if the system is not highly plasticized) can be acceptable [62].

However, it must be pointed out that two kinetics migration modes dominate the plasticizer loss: the diffusion mode (above mentioned) and the evaporation mode [23]. Considering only the one-dimensional problem in the x-direction (similarly to Equation (5)), the evaporation condition can be described by the mass balance [54,63]

$$-D(\mathbb{C})\left(\frac{\partial \mathbb{C}}{\partial \mathbf{x}}\right) = F(\mathbb{C} - \mathbb{C}\_{\mathbb{C}}) \tag{9}$$

In Equation (9), the mass transfer related to the diffusion process (left term) is equated to the mass transfer of the plasticizer from the surface (right term). Obviously, evaporation can occur if the plasticizer concentration at the surface is greater than the concentration corresponding to the environment saturated with plasticizer (*Ce*) [23].

Thus, the plasticizer migration will be affected by the coexistence of two phenomena: diffusion and evaporation; the overall rate of plasticizer loss will be determined by the slower process (diffusionor evaporation-controlled). In the case of diffusion-controlled system, the evaporation rate will be faster than diffusion rate. On the other-hand, in the case of evaporation-controlled system the evaporation rate will be slower than diffusion rate and often this leads to a formation of a plasticizer film on the surface of the analyzed sample [64]. Generally, it is possible to understand if the system is diffusion- or

evaporation-controlled, by the shape of the concentration profile obtained plotting the curves of mass loss versus the square root of the time (Figure 1).

**Figure 1.** Examples of diffusion-controlled (blue line) and evaporation-controlled (black line) shape concentration profile.

The shape of concentration profile for an evaporation-controlled system, is flatter (S-shaped) if compared to the diffusion-controlled system [23].

It is evident that, for cases where the chosen plasticizer has a very high boiling point if compared to the temperature in which the material is currently used, the plasticizer will be accumulated on the external surfaces and it will form a thin film. The presence of this thin film of plasticizer that cannot easy evaporate due to the low temperature, limits the kinetic of plasticizer migration (the system will be evaporation-controlled). It is obvious that the concentration of plasticizer and the operative temperature will strongly influence the migration kinetics. Zhang et al. [65] demonstrated for a PLA system plasticized with acetyl triethyl citrate (ATC) that the ATC migration increases with the increasing of ambient temperature. The diffusion-controlled mechanism is activated for this system around 100–135 ◦C. Consequently, due to the similarity of the Zhang system with the polymeric system studied in this paper and also considering that the temperature adopted for the migration test (60 ◦C) is well below from the evaporation temperature of ATBC, it will be expected that the system will be evaporation-controlled.

The different film blends used for this work showed a plasticizer release phenomenon during the storage. From a practical point of view, it is difficult that the external film layer of plasticizer on the sample surface is not altered by external factors (presence of paper packaging which absorbs the plasticizer for example). Furthermore, there are some sectors (like cosmetics) in which a controlled release of these films is foreseen (for example in beauty masks). For these cases, it is evident how the diffusion process is the dominating mechanism.

In this work, the migration studies were carried out with the goal of investigating the diffusion-controlled mechanism in order to evaluate the effect of micro and nano calcium carbonate addition on the plasticizer diffusion coefficient. Consequently, the experimental migration tests were also carried out in order to make the diffusion the kinetically controlling mechanism. The shape of concentration profile obtained for all the compositions examined (Figure 2), being diffusion shaped, confirmed the diffusion-controlling mechanism.

**Figure 2.** Concentration profile for all the compositions examined in this paper.

### **4. Results and Discussion**

### *4.1. Migration Results and Determination of the Di*ff*usion Coe*ffi*cients*

The trends of the weight loss percentage as function of the time are reported for all the blends in Figure 3. In Figure 3a, the loss percentages are reported for the reference formulations (F1, F2, and F3) compared to F4 where the quantity of micro-calcium carbonate (2AV) was incremented from 4 to 7 wt %. It can be noticed that F1 formulation has the higher weight loss if compared to F2. Both the formulations do not have micro-calcium carbonate; nevertheless, their plasticizer kinetic release is markedly different. This difference can be attributed to the addition of PST that, favoring the formation of intermolecular interaction [44], decreases the ATBC migration making more difficult the plasticizer diffusion. It can be observed that the addition of 4 wt % of micro-calcium carbonate (F3 blend), do not alter significantly the ATBC weight loss and only a slight slowing is observed if compered to F2. On the other hand, an increment from 4 wt % to 7 wt % significantly decreases the migration of ATBC and the weight loss percentage is reduced under the 0.5%. The major quantity of filler significantly slowed the plasticizer kinetic release. This can be ascribed to a major tortuosity path (generated by the increment of micro-calcium carbonate particles number) that the ATBC molecules must encounter. The plasticizer molecules have to follow a longer path in order to leave the polymeric structure [38].

The effect of the three different types of calcium carbonate particles (micro- and nano-, surface coated and not) are reported in Figure 3b. The best barrier properties to the ATBC migration are obtained with 2AV (F4). Moving from micro- to nano- calcium carbonate (F6 formulation), it would be expected a significant reduction of the plasticizer migration [35]. However, the data show a very similar trend with a slight decrease in the loss of ATBC. This result shows a poor efficiency of CCR (which also has a cost higher than Omycarb 2-AV). The explanation of this behavior can be ascribed to the presence of zones rich in agglomerates, as highlighted by the SEM analysis shown in Figure 4. It is likely that these surface coated nano-particles' agglomeration is related to the manufacturing process of CCR. CCR, in fact, is a precipitated calcium carbonate where a water coating process was used. Generally the 'wet' processes have a minor surface coated area if compared to the 'dry' processes [66]. The coating in aqueous medium is different from solvent or dry coating, the process is controlled by micelle absorption followed by the micelle collapse into double or multiple layers during the drying stage. It has been demonstrated that for this process the monolayer coating is incomplete [67]. This incomplete coverage is responsible of the partially particles agglomeration. However, the presence

of some nano-agglomerates cannot be the only reason for which the efficiency of the nano-metric fillers is not good enough. In fact, it must be considered that these nano-fillers are surface coated with fatty acids to improve their dispersion, reducing the surface tension between a hydrophobic and non-polar polymer and inorganic polar hydrophilic particles [68,69]. These surface agents, containing a polar group and a long aliphatic chain, can alter the ATBC absorption on the particles' surface, worsening the hindering of ATBC migration. It is known in fact that the coating with fatty acids reduces the wettability of the particles to solvents such as water and n-decane [70]. Hence it is reasonable, on the basis of the results obtained, to hypothesize a negative chemical affinity (that reduces the ATBC absorption) between the ATBC and the aliphatic chains of fatty acids used for the surface coating. This conjecture can be reflected in the F5 formulation that contains surface-coated micro-calcium carbonate particles (Smartfill). The results achieved in the ATBC loss are worsened if compared to the not-surface-covered calcium carbonate particles, confirming the probable low absorption capacity of the calcium carbonate particles' surface treated with fatty acids.

The results are worse where the coating with fatty acids is better; in fact, Smartfill is a ground calcium carbonate surface coated with a 'dry' process, hence it has a better surface coating if compared to CCR. Consequently, the ATBC absorption capacity of the calcium carbonate appears to be greatly worsened by this coating. The best results on the other hand are obtained with a calcium carbonate having no surface coverage.

Finally, in Figure 3c the effect of chitin nanofibrils alone and coupled to 2AV can be observed. It can be observed that chitin nanofibrils alone are capable of limiting the plasticizer release, and the migration level of ATBC is comparable to the F4 formulation. The coupling between chitin nanofibrils and 2AV does not show any type of synergy. The presence of agglomerates (that can be observed in the SEM of Figure 4) leads to a worsening of the final material barrier properties.

The quantitative results of the maximum weight loss, calculated according to Equations (2) and (3), are reported in Table 2 and confirms the trends of Figure 3.


**Table 2.** Film weight loss and ATBC lost, calculated according to Equations (2) and (3), for all film formulations.

The surficial area of the different fillers used in the present paper were determined by BET analysis and the results are reported in Table 3.

From Table 3, it can be noticed that CCR presents the higher Langmuir area among the calcium carbonates and a number of particles per gram two order of magnitude greater as a consequence of nanometric dimension. Moreover, chitin nanofibrils show a very high surficial area. The fillers with a higher surficial area should develop more interactions with the polymeric matrix.

However, it must be pointed out that a high surface area makes the aggregation of particles easier, confirming what has been observed by the SEM (Figure 4), where micrometric particles or agglomerates are revealed in all the samples. Chitin nanofibrils show the highest surficial area among the fillers, making them capable of agglomerating easily, resulting in micrometric bundles (Figure 4, F7 sample). This justifies the use of PEG to separate the fibrils as much as possible [47].


**Table 3.** Surficial areas, porosity, and number of particles per gram of the four fillers used in this paper.

(**b**)

**Figure 3.** *Cont*.

**Figure 3.** Weight loss percentage as function of the time for (**a**) F1, F2, F3 and F4 formulations, (**b**) F4 compared to F5, F6, (**c**) F4 compared to F7 and F8.

(**F6**) (**F7**)

**Figure 4.** *Cont*.

(**F8**)

**Figure 4.** SEM micrographs of **F4**, **F5**, **F6**, **F7**, and **F8** formulations. The **F8** micrographs was obtained by backscattered electrons to better evidence CaCO3 particles (white).

From the weight loss trends reported in Figure 3, two migration regimes can be observed: a first linear regime where a high quantity of plasticizer is lost in a short time and a second regime where the mass loss is lower and almost constant. The diffusion mechanism, as reported previously, is clearly influenced by the addition of the fillers (by their typology and quantity). However, another factor must be considered. The films, in fact, were maintained at 60 ◦C, slightly above the PLA glass transition temperature. At this temperature, the presence of compatible plasticizer (like ATBC) enhances the free volume of chains which induces molecular mobility. This reduces the Tg, but it can also favor the crystallization process [71]. The plasticizer goes only in the amorphous regions and if these amorphous regions decrease due to the crystallinity increment, the plasticizer migration will be accelerated. Considering that fillers can also act as nucleating agents (especially CaCO3) [40], film crystallization can become the main cause of ATBC loss. To better understand the role of the crystallization phenomena to the ATBC release, DSC analysis was performed considering only the first heating scan to take into account the real condition of the films produced by compression molding (i.e., considering their thermal history).

In Table 4 for all the formulations the values of glass transition temperature (*Tg*), cold crystallization temperature (*Tcc*), cold crystallization enthalpy (Δ*Hcc*), melting temperature (*Tm*), and melting enthalpy (Δ*Hm*) are reported. Furthermore, the PLA percentage of crystallinity (*Xcc*), calculated according to Equation (1), was reported. It can be observed that all the blends, which were held at 60 ◦ C, are well above their *Tg*. Since these are plasticized films, their *Tg*, compared to pure PLA, are shifted towards lower values. A marked decrement of PLA crystallinity from F1 to F2 formulation containing the PST can be observed. In literature, other authors confirmed that the PST addition decreases the number of crystals [72], this capability combined to the intermolecular interactions created by PST, limits considerably the plasticizer migration. The addition of 4 wt % of micrometric calcium carbonate (2AV) does not alter the crystallinity percentage initially present in the film. By adding larger quantities of filler (4–7 wt %), the starting crystalline content is significantly reduced independently of the type of calcium carbonate used. A greater quantity of filler not only hinders the diffusion process by increasing the tortuosity diffusion paths, but it also worsens the mobility of the polymeric chains and slows down the crystallization kinetics. It is known, in fact, that rigid fillers (such as calcium carbonate) can act as heterogeneous nucleation sites (nucleating agents) only if added in small quantities and if enough time is provided to the system for crystallizing [6]. Initially, the crystallinity degree does not influence the plasticizer migration that it is only affected by the type of filler used. At this purpose, the data confirm the negative role of the surface coating on the migration of ATBC making the micrometric 2AV the best choice in limiting the migration process.

On the other hand, F8 formulation shows a different crystallization behavior due to the interactions occurring between calcium carbonate particles and chitin nanofibrils. The not uniform distribution related to the formation of agglomerates facilitates the plasticizer migration and, at the same time, the PLA crystallization. The combination of these factors explains the major loss of ATBC that was encountered.


**Table 4.** Results of differential scanning calorimetry analysis (first heating).

To better understand the migration mechanism of the plasticizer and if this is influenced during the time by an eventual crystallinity increment (due to the permanence of the sample at 60 ◦C for a long period), DSC analysis were performed at the end of the migration tests. The comparison was made only between the reference formulation F1, the formulation F2 containing the PST, the best filler calcium carbonate added formulation (F4) and for the chitin nanofibrils added formulation (F7).

In Table 5, the DSC results at the end of the migration test for F1, F2, and F4 formulations are summarized. In this way, it is possible to verify how the crystallinity varies during the heat treatment at 60 ◦C. In Figure 5 the DSC thermograms before and after the migration test are also reported.

**Figure 5.** Comparison between thermograms before and after the migration test (first heating).


**Table 5.** DSC results for F1, F2, F4, and F7 formulations at the beginning and at the end of the migration test (first heating).

It can be observed that, over time, the crystallinity of the samples increases independently if there is the presence of PST or filler. This means that the migration of plasticizer over the time is influenced by the crystallinity increase that reduces the amorphous regions in where the plasticizer is situated and favors its migration. Anyway, it seems that independently from the additives, the system reaches the same value of PLA crystallinity (about 40%) that can be assumed as the final value reached over the time.

At this point, the diffusion coefficients for all formulations were evaluated adopting the linear diffusion coefficient obtained by regression of the migration data as a function of the square root of the time according to the simple form of the Crank's equation (Equation (7)). According to Equation (7), by plotting the values of *Mt*/*Cp*<sup>0</sup> (where Mt is the weight variation of the samples at time *t* per film area (mg/cm2) and *Cp*<sup>0</sup> is the initial density of the film (mg/cm3)) as a function of the time square root, the diffusion coefficient corresponds to the slope of the linear part of the curve. Clearly, this simplified equation, considering only the early stages of the migration where the trend is linear, does not take into account the crystallization phenomenon that may occur in in long periods and can induce a further plasticizer migration over the time. Consequently, the diffusion coefficients were also calculated by applying the not-simplified Crank's solution of the second Fick's law (Equation (6)). Considering that the films are very thin (0.025 cm), the concentration variation of the ATBC along the entire film thickness can be neglected thus x can be imposed equal to zero in Equation (6). The results obtained with Equation (6) make the calculation of the diffusion coefficients at various instants possible until the end of the migration tests (after 1500 h), considering in this way also the crystallization processes. An average diffusion coefficient weighted as a function of the plasticizer concentration was thus considered and compared with the diffusion coefficient obtained from Equation (7); the results are reported in Table 6.


**Table 6.** Diffusion coefficient calculated according to Equation (7) and to Equation (6).

The values of the diffusion coefficients calculated in the first stages by Equation (7) highlight the better efficiency of 2AV in limiting the plasticizer migration; in fact, for F4 formulation, the diffusion coefficient is lower of an order of magnitude. On the other hand, F8 shows the highest value of diffusion coefficients; the coexistence of the chitin nanofibrils and 2AV that forms agglomerates helps the ATBC diffusion leaving more routes for the plasticizer migration. Another good result is obtained by the diffusion coefficient of F6, containing the nano-metric CCR calcium carbonate. This result is coherent to what is observed from the ATBC mass loss. However, these diffusion coefficients consider only the initial diffusion mechanism where a very low value of crystallinity content was present for many formulations. As it was observed, the crystallinity increases up to 40% after 1500 h, consequently to consider the effects at longer times, the diffusion coefficients calculated by Equation (6) are more realistic.

First of all, it can be observed that considering the entire process, there is an increase of all diffusion coefficients and this confirms that the calculation only at the first stages lead to a D underestimation. A great worsening of the diffusion coefficients is registered for F1, F2, and F8 formulations coherently to what it was observed experimentally. In fact, these formulations lost a higher quantity of ATBC. The results confirm also that 4 wt % of 2AV is not a sufficient quantity to efficiently limit plasticizer migration; on the other hand, passing from 4 to 7 wt %, very good results are achieved. For longer times, the different behavior to the plasticizer migration for the three types of calcium carbonate emerges. The diffusion coefficients obtained confirm what was registered from the migration tests. In fact, it was observed that Smartfill was the worst filler in hindering the ATBC migration and, over long periods of time, its diffusion coefficient dramatically decreases. On the other hand, CCR seems to quite effectively limit plasticizer migration for long times with a slight decrement of its diffusion coefficient that passes from 1 <sup>×</sup> 10−<sup>12</sup> cm2/s to 8.3 <sup>×</sup> 10−<sup>11</sup> cm2/s. However, if we also compare these results with 2AV it can be observed that the improvement in the migration coefficients, and consequently in the ATBC loss, are not so evident to justify the use of this more expensive filler. Omycarb-2AV in fact, possess a *D* value of 8.2 <sup>×</sup> <sup>10</sup>−<sup>11</sup> cm2/s perfectly comparable with F6 diffusion coefficient. Noteworthy is the greatest difference in the diffusion coefficient calculated with the Equations (6) and (7), registered for the F6 formulation. Additionally, considering the crystallinity values over time (reported in Table 5), it can be observed that the crystallization kinetics for this type of formulation is slower therefore the higher ATBC release will occur in longer times. Equation (7) can therefore give a fairly truthful estimation of the diffusion coefficients only for those systems that quickly lose the plasticizer and for which there are no great differences of crystallinity over time.

As far as concern the formulation containing only the chitin nanofibrils (F7), a decrease in plasticizer migration should be observed. In fact, it is known in [73] that their addition improves the barrier properties. Effectively, the diffusion coefficient obtained is lower and confirms the mass loss results obtained. However, to contrast the ATBC migration, 2AV is still the most efficient.

$$D = \alpha \text{s} \mathbf{t} \cdot \mathbf{t}^{-1} \tag{10}$$

It can be observed from Figure 6 that the most significant variations of *D*, and therefore of the migration of the plasticizer, occur in the first 100 h—after the curves tend to flatten and the values of *D* tend to zero—meaning that the diffusion process finishes. The most flattened curve is that of the F4 formulation which will therefore have the lowest diffusion coefficient value over time and will therefore be able to release less plasticized content, further confirming the good efficiency of 2AV.

The rheological properties (torque, MFI, and MVR) properties were therefore evaluated and compared with those of the previous works [44,47].

The melt properties (MFR, MVR, and torque values) of each formulation are summarized in Table 7.

The addition of PST melt strength enhancer leads to an increment of torque (and in parallel to a decrement of MFR value) as it can be observed passing from F1 to F2 compositions. Clearly, the addition of a rigid filler (in this case micro calcium carbonate 2AV) leads to a marked decrement of the MFR value (F3 blend). As it can be expected, a further addition of micro-calcium carbonate (from 4 wt % to 7 wt %) further decreased the MFR from 9.4 to 7.6 (g/10 min) for F3 and F4 blends respectively. The MFR (similarly to MVR), affected by the molecular weight and by the interactions occurring in the melt material, can be correlated to the degree of disentanglement of the fluid blend [74]. The more the melt material is disentangled, the more the diffusion coefficient increases as calculated by Equation (6) (so considering only the starting part of the mass loss versus time trend) The material above its glass transition consists of a net of entangled macromolecules. A high molecular weight as well as efficient interactions contribute to increasing the degree of entanglement making more rigid the net and more difficult the diffusion of ATBC molecules. On the contrary, a high degree of disentanglement, due to lack of interactions with additives or fillers, as well as a decrease of molecular weight (cutting macromolecules), make the net more suitable for diffusion increasing D. The data of MFR, measured in the melt, and D decrease from F1 to F4. In fact, the degree of disentanglement decreases and D decreases accordingly (Figure 7).

**Figure 6.** Trend of the diffusion coefficient over the time for all the formulation examined.


**Table 7.** Torque, MVR, and MFR values of the prepared blends.

However, in F5 using the same weight percent (7 wt %) of Smartfill with respect to F4 and CCR in F6, an increase in MFR was registered. It is likely that the surface coating of the calcium carbonate particles creates less friction in the molten matrix during the extrusion process. Moreover, as the removal of water during the drying of filler before the extrusion is more difficult for nanometric fillers, some chain scission due to water transesterification with polyester macromolecules can occur [75] resulting in shorter macromolecules. The latter effect influences the MFR confirming, this being the measurement done in the melt in dynamic conditions, so it is more dependent on the capability of macromolecules to flow. Both of these effects generate disentanglements that increase the value of D with respect to F4. In general, the observed variations in the Torque and MVR values are not so significant to alter the final processability independently of the filler used and also by the greater

quantity of calcium carbonate added. Interestingly, F6 shows a very low value of D despite having the highest value of MFR. This is likely due to chain scission.

**Figure 7.** Comparison between MFR data and D calculated by Equation (6).

The occurring of chain scission was demonstrated by preparing a blend with the same composition of F6 but by previously drying better the nano-calcium carbonate, for two days at 60 ◦C at reduced pressure (50 KPa), and it was observed that the torque value was 77.0 ± 5.0 Ncm and the MFR value for this blend was 7.7 <sup>±</sup> 0.6 cm3/10 min, in agreement with a higher molecular weight of the biopolyesters with respect to F6.

It can be observed that, for F6, the correspondent low value of *D* evidenced very good interactions, counterbalancing the decrease in molecular weight and reasonably achieved thanks to the nano-dimension of the filler (extended surficial area) significantly enhancing interactions despite a sub-optimal chemical affinity due to fatty acids and despite partial agglomeration. Chitin nanofibrils (F7) and micro-nanometric uncoated carbonate (F5), having an intermediate dimension between that of micro-carbonate and nano-carbonate, showed a similar behavior in agreement with the similar dimensions shown in phase morphology analysis of blends. In general, the analysis of the melt fluidity of the blends was fundamental to show that molecular weight can be another important parameter affecting the diffusion behaviors of polymeric materials.

### *4.2. Mechanical Characterization of Blends*

After the evaluation of the different types of filler added to the ATBC migration, a general screening concerning the mechanical properties was carried out in order to evaluate if the mechanical properties (that were optimized in previous works [44,47]) were significantly altered by the greater amount of calcium carbonate added and by the different types of fillers used.

The variations of the tensile properties (yield stress (σy), stress, and strain at break (σ<sup>b</sup> and εb)) are reported in Table 8.

The increment up to 7 wt % of the different calcium carbonate fillers weakens the material. The greater quantity of rigid calcium carbonates added makes the final material more rigid, leading to a decrease of stress and elongation at break and to an increment of the yielding stress. However, comparing the final stress and strains values of the F3 formulation with those of F4, F5, and F6 it can be concluded that nevertheless this flexibility reduction the tensile properties of the plasticized PLA/PBS blends remains still good. The fillers introduced in these formulations, leads to a modification of the first part of the curve which assumes a typical trend of a more resistant system. In fact, it can be observed the capability of PST and fillers to increase the yield stress compared to the F1 formulation, which exhibits a behavior similar to an elastomer (Figure 8). The addition of an uncoated micrometric

carbonate such as 2AV, capable of dispersing adequately during extrusion, ensures a yield value higher than other types of carbonates.


**Table 8.** Tensile properties of the films obtained from each blend.

**Figure 8.** First part of the stress-strain curves for the formulations examined.

The combination of 2AV with chitin nanofibrils does not lead to an improvement of the tensile properties, as can be observed in F8 compared to F4, especially as regards the elongation at break. For F8 formulation, the yield stress is very similar to the result obtained with chitin nanofibrils alone. On the other hand, the comparison with the F4 formulation shows a marked decrease in the yield stress. The nanofibrils seem to not allow calcium carbonate to reinforce the formulation. Standing between the carbonate particles and the polymer matrix, they cause a reduction of the filler-matrix adhesion as it can be observed from the SEM image (Figure 4) where the micrometric calcium carbonate was found both in the matrix and near the areas containing nanofibrils.

### **5. Conclusions**

On the basis of previous studies, the plasticized poly(lactic acid) (PLA)/poly(butylene succinate) (PBS)-based films containing chitin nanofibrils (CNs), calcium carbonate, and a small percentage of a commercial melt strength enhancer, have been investigated from the point of view of the plasticizer migration. It was observed that, during storage, these films lost a significant amount of plasticizer. The effect of the addition of different types of calcium carbonate (nanometric and micrometric, surface coated and not) and chitin nanofibrils was investigated with the purpose of reducing or controlling the plasticizer migration. The Crank's solution of the Fick's second law was adopted to obtain the

quantative values of diffusion coefficients. In addition, the evolution of the crystallinity that can induce the plasticizer migration was considered.

The results showed that the not surface coated calcium carbonate (2AV) is the more effective in hindering the plasticizer migration. Micrometric calcium carbonate reduced significantly the ATBC migration but at the same time did not lead to a significant change in the processability and mechanical properties of the already optimized formulations.

Regarding the surface coated calcium carbonate, the surface coating (generally made with fatty acids) probably limits the ATBC absorption on the calcium carbonate particles surfaces, leading to worse results in ATBC migration. Moreover, the occurrence of chain scission due to difficulties in removing humidity from nanofiller before blending determined a decrease in molecular weight, resulting in macromolecular net disentanglement that allowed a better diffusion of ATBC, as concluded by studying the fluidity in the melt of blends. For these reasons, the migration control is not significantly improved passing from micro- to nano- calcium carbonate particles, despite the increased surficial area of fillers. From a point of view of cost, reduction of the plasticizer migration, and quantity added, the nanometric calcium carbonate particles are less efficient and this is ascribable to their agglomeration tendency and high hygroscopicity.

Chitin nanofibrils alone can also influence the plasticizer migration thanks to their capability of hindering the plasticizer diffusion, slowing but not efficiently limiting the migration, and thus suggesting new potentialities of such nano additives in films or products with a controlled release.

The coupling of chitin nanofibrils and 2AV did not lead to significant results. Negative effects have been encountered probably related to the synergistic agglomeration tendency of chitin nanofibrils and calcium carbonate.

The crystallinity reached from samples after migration test seems not affected by the type of filler added and about 40% of PLA crystallinity was obtained. Clearly, this marked crystallinity increment over time affects the plasticizer migration due to the decreasing portion of mobile amorphous fraction in which the plasticizer is located.

In general, the analysis of all the parameters affecting migration of plasticizers suggests promising strategies to better exploit the properties of inorganic micro and nano particles, as well as natural nanofibrils in biocomposites.

**Author Contributions:** Conceptualization, M.-B.C. and L.A.; Experiments, A.V., L.P., and L.A.; Calculations, V.G. and L.A.; Data curation, L.P. and A.V.; Data interpretation L.A. and M.-B.C.; Writing—original draft preparation, L.A. and A.V.; Writing—review and editing, M.-B.C.; Visualization, L.A.; Supervision, A.L.; Project administration, M.-B.C. and A.L.; Funding acquisition, M.-B.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partially funded by the Bio-Based Industries Joint Undertaking under the European Union Horizon 2020 research program (BBI-H2020), PolyBioSkin project, grant number G.A. 745839.

**Acknowledgments:** We thank the Centre for Instrumentation Sharing-University of Pisa (CISUP) for their support in the use of FEI Quanta 450 FEG scanning electron microscope. We would also like to thank Pierfrancesco Morganti for interesting discussion.

**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* **Dog Wool Microparticles**/**Polyurethane Composite for Thermal Insulation**

**Francisco Claudivan da Silva 1,2, Helena P. Felgueiras 3, Rasiah Ladchumananandasivam 1,4, José Ubiragi L. Mendes 1, Késia Karina de O. Souto Silva <sup>4</sup> and Andrea Zille 3,\***


Received: 10 April 2020; Accepted: 9 May 2020; Published: 11 May 2020

**Abstract:** A polyurethane (PU)-based eco-composite foam was prepared using dog wool fibers as a filler. Fibers were acquired from pet shops and alkaline treated prior to use. The influence of their incorporation on the PU foams' morphological, thermal, and mechanical properties was investigated. The random and disorganized presence of the microfibers along the foam influence their mechanical performance. Tensile and compression strengths were improved with the increased amount of dog wool microparticles on the eco-composites. The same occurred with the foams' hydration capacity. The thermal capacity was also slightly enhanced with the incorporation of the fillers. The fillers also increased the thermal stability of the foams, reducing their dilatation with heating. The best structural stability was obtained using up to 120 ◦C with a maximum of 15% of filler. In the end, the dog wool waste was rationally valorized as a filler in PU foams, demonstrating its potential for insulation applications, with a low cost and minimal environmental impact.

**Keywords:** dog wool fibers; fillers; polyurethane; eco-composites; renewable resources

### **1. Introduction**

Global energy consumption is estimated to increase by 53% within the next 10 years [1]. One of the simplest and most cost-effective ways to reduce the energy demands and the greenhouse gas emissions is through building insulation. If properly selected, an effective insulation may save energy by requiring less for space cooling in the summer and heating in the winter, and thus reducing the use of natural resources (e.g., petroleum and gas) [2]. Thermal insulation is achieved by means of a material or composite materials endowed with high thermal resistance. Over the years, many options have been proposed and tested, including fiberglass, mineral wool, and foams (e.g., polyurethane, PU, and polyvinyl chloride, PVC) [3].

PU is formed of stiff and flexible segments, endowing PU foams with versatile properties and light weight and making them particularly desirable for insulation. They are obtained by a reaction between polyfunctional alcohols (polyol polyether or polyol polyester) and polyisocyanate [4]. Their foaming appearance is possible due to the production of a blowing agent (e.g., carbon dioxide) during exothermal polymerization, which remains enclosed within the material and ensures the foam insulating performance [5,6]. Depending on the amount, proportions, and characteristics

of the components, three categories of PU foams can be defined: Flexible, semi-rigid, and rigid, the last being the preferred for insulation purposes due to its highly cross-linked and closed-cell structure, good mechanical and chemical resistance, low density, and low water adsorption [7]. Further, the R-value (measure of how well a two-dimensional barrier resists the conductive flow of heat) of rigid PU foams is among the highest of any insulating material, thus ensuring efficient heat retention and/or consistent temperature control of refrigerated environments [8].

In recent years, the development of PU-based composite foams with interdisciplinary functions has expanded considerably with the goal of increasing their mechanical performance, broadening their application, and preserving the environment by using lower amounts of PU [9,10]. Natural fibers have attracted much attention as potential reinforcements for composites due to their availability, biodegradability, and low cost [11]. These fibers have been involved in a growing type of polymer composites, the eco-composites, which describe combinations of materials with environmental and ecological potential and/or produced using materials from renewable resources [12–14]. So far, vegetable fibers such as flax, hemp, jute, and kenaf have been the most explored due to their low density, variable mechanical properties, and intrinsic biodegradability [15–17]. However, animal-based fibers are starting to demonstrate their potential as well. Feather keratin fibers have been shown to possess a hollow structure, filled with air, responsible for their low density and low dielectric constant, properties highly desirable in composites for electronic or automotive applications [18,19]. Silk fibers have been investigated to produce composites for tissue engineering due to their increased oxidation resistance and improved antibacterial and UV-light protection properties [20]. Animal-derived wastes, such as wool fibers, have also been successfully embedded in a polymeric film-forming matrix of cellulose acetate, with potential applications in the packaging and agricultural industries [21].

Even though this is an environmentally friendly solution to animal waste disposal, very few reports have been published on the subject.

In the present work, we explore the use of discarded dog wool fibers as a reinforcement agent in the production of PU-based eco-composites for thermal insulation. According to the Statistical Institute of Brazil (IBGE), there are in the country 52 million dog pets. The goal was to determine the efficiency of this mixture and the potentialities of animal wastes for industrial applications. To the authors' knowledge, this is the first report on the use of dog wool fibers as reinforcement in PU-based eco-composites. Various fiber percentages were combined with PU castor oil. The resulting eco-composite foams were characterized in terms of their physical, thermal, and mechanical properties in light of the desirable application.

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

### *2.1. Materials*

Respan, a semi-flexible and biodegradable polyurethane from castor oil (PU) resin, acquired from Resichem Chemicals LTDA (São Paulo, Brazil), was used as matrix. Castor oil is a vegetable oil pressed from castor beans. Dog wool fibers were used as reinforcement and were collected in pet shops in the metropolitan area of the city of Natal (Natal, Brazil). PU was used as control. All remainder chemicals were acquired from VWR International and used without further purification.

### *2.2. Treatment of Dog Wool Fibers*

Dog wool fibers were initially washed in 0.05 M sodium hydroxide (NaOH) solution, to remove impurities present along the surface, and dried at 50 ◦C for 24 h. After, they were ground in a micro-slicer (Urschel, Chesterton, IN, USA) to obtain microparticles of ≈30 mesh screen.

### *2.3. Preparation of Eco-Composites*

Dog wool microparticles were thoroughly mixed with the semi-flexible PU resin using a commercial mixer, to guarantee the homogeneity of the composite structure. The mixture was then poured onto a steel mold, which was tightly closed, and submitted to a controlled expansion process to induce strong interactions between matrix and reinforcement (Figure 1). Eco-composite plates were produced with dimensions of 30 <sup>×</sup> 30 <sup>×</sup> 1 cm3 and different ratios of fiber in their composition (Table 1). Then, 100% PU plates were also produced and used as control.

**Figure 1.** (**a**) Fiber microparticles, (**b**) mixture of PU resin and the fiber microparticles, and (**c**) the eco-composite.


**Table 1.** Eco-composites' composition.

### *2.4. Scanning Electron Microscopy (SEM)*

Morphological analyses of the fibers and the eco-composites were carried out using a SEM TM 3000 HITACHI (Hitachi, Chiyoda, Tokyo, Japan). Backscattering electron images were realized with an acceleration voltage of 15 kV that enabled the visualization of the distribution of the fiber reinforcement along the polymeric matrix.

### *2.5. Particle Size Distribution*

The particle size distribution was performed in a laser diffraction particle size analyzer model CILAS 1180 (Cilas, Orléans, France) at the laser light wavelength of 635 nm. The equipment is able to measure particles ranging from 0.04 to 2500 μm. The size distributions of the samples were determined based on Fraunhofer diffraction theory and expressed as frequency (%) vs. particle diameter (μm). The measurement was carried out with samples of 0.2 g in accordance to the standard BS ISO 13320:2009.

### *2.6. Fourier-Transformed Infrared (FTIR)*

FTIR spectra of the eco-composites with various reinforcement percentages were collected using a Shimadzu spectrometer, model FTIR-8400S, IRAffinity-1 (Shimadzu, Kyoto, Japan), coupled with an attenuated total reflectance (ATR) accessory, the PIKE MIRacle™ single reflection with a ZnSe crystal (PIKE Technologies, Madison, WI, USA). Spectra were obtained in the range of 4000–500 cm<sup>−</sup>1, from 30 scans at a resolution of 4 cm<sup>−</sup>1. All measurements were performed in triplicate.

### *2.7. Thermal Properties*

The thermal properties in the castor polyurethane samples were determined using the KD2 Pro (Decagon Devices, Pullman, WA, USA) equipment coupled with a thermal sensor twin needle SH1, which uses the transient line heat source method to measure thermal diffusivity, specific heat (heat capacity), thermal conductivity, and thermal resistivity. All analyses were performed at room temperature following the standards ISO EN 31092-1994. An average of 10 readings was taken for each sample and the data were reported as mean ± standard deviation. TGA was performed on a DTG-60H model (Shimadzu, Kyoto, Japan) using a platinum pan. The TGA trace was obtained in the range of 30–300 ◦C under nitrogen atmosphere, flow rate of 50 mL/min, and temperature rise of 10 ◦C/min. Results were plotted as percentage of mass loss vs. temperature. DSC was carried on a Power Compensation Diamond DSC (Perkin Elmer, Waltham, MA, USA) with an Intracooler ILP, based on the standards ISO 11357-1:2016, ISO 11357-2:1999, and ISO 11357-3:1999. Samples were dried at 60 ◦C for 1 h and placed in an aluminum sample pan before testing. The analysis was carried out in nitrogen atmosphere with a flow rate of 50 mL/min. The DSC analysis was carried out at three stages: The first heating, cooling, and second heating, all at the heating rate of 10 ◦C/min, in order to eliminate the thermal history of the samples. The thermogram was obtained in the range of 20 to 500 ◦C.

### *2.8. Mechanical Properties*

The eco-composites' tensile strength and compression capacities were examined using an X 300KN Universal Testing Machine (Shimadzu, Kyoto, Japan). The tensile strength of the eco-composites was determined following the ASTM D3039 with a specimen of 3 mm of thickness and 25 mm of width (75 mm<sup>2</sup> of cross-section) and the compression test according to NBR 8082. In the compression test, deformation was measured when the machine was activated to reduce the thickness of the specimen in 10% at speed of 0.25 cm/min. It was calculated by the formula *R*<sup>c</sup> = F/A, where *R*<sup>c</sup> is the compression strength at 10% deformation (Pa), F is the force (N), and A is the test area of the sample (m2).

### *2.9. Hydration Capacity*

The eco-composites' water absorption capacity was measured following the ASTM D2842. Three replicates were used of each eco-composite. Samples were initially dried at 50 ◦C for 24 h and then transferred to a desiccator and left for 15 min until they reached room temperature. Samples were weighed in their dry state (mdry). After, they were immersed in distilled water (*d*H2O) and measured continuously (24 random intervals) until saturation was reached. The saturation point was determined when the sample weight reached a constant value (mwet). The samples' hydration capacity was determined using the Equation (1):

$$\% \text{ Water Absorption} = \frac{m\_{\text{wet}} - m\_{\text{dry}}}{m\_{\text{dry}}} \times 100\tag{1}$$

### *2.10. Dilatometry*

The thermal expansion coefficient of the samples was determined on the device NETZSCH model DIL 402 PC (Netzsch, Selb, Germany). The samples were made with dimensions of 25 mm in length and 8 mm in diameter. The tests were carried out under an argon gas flow of 5 mL/min at the heating gradient from room temperature to 170 ◦C. The heating rate was 5 ◦C/min.

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

### *3.1. Particle Size and Eco-Composites' Morphology*

SEM micrographs of the dog wool fibers, in their natural state (untreated), treated with NaOH, and combined with the PU resin as reinforcement to form eco-composites were taken (Figure 2). As expected, there were substantial differences between the fibers before and after treatment with NaOH. The impurities present along the fibers (Figure 2a) were eliminated after NaOH washing, revealing the efficiency of this alkali treatment and leaving the surface clean and unspoiled (no evidences of degradation, Figure 2b), with a desirable open structure capable of interacting with the polymeric matrix. The incorporation of the fibers within the PU matrix was evidenced in Figure 2d. As can be observed, the porosity and morphology of the composite up to 15% of fiber content (Figure 2e) was not significantly different from the pure PU resin (Figure 2c). A very porous structure is characteristic of the PU foam. The average pore size (mean of 50 measures) of the composite was 84 ± 50 μm and the pore size of the pure PU was 83 ± 40 μm. At the fiber content of 20%, the PU structure became instable with large (~0.5 mm) collapsed structures and holes of ~0.2 mm around the fibers. Even though fibers' distribution and orientation were random, they were preferentially detected in compact areas, both in the interior and the borders of the foam cells. Similar outcomes were obtained with composites of rigid PU foam reinforced with cellulose fiber residues [5].

(**a**) (**b**)

**Figure 2.** *Cont.*

**Figure 2.** SEM micrographs of the dog wool fibers (**a**) in their natural state and (**b**) after treatment with 0.05 M of NaOH. Micrographs of the pure PU (**c**), PU + 10% of fibers (**d**), PU + 15% of fibers (**e**), and PU + 20% of fibers (**f**) at 100× magnification.

Regarding their size, the dog wool fibers were considered microparticles. According to data from Figure 3, they were very heterogeneous in size, varying from 1 to 700 μm, with the largest amount measuring between 30 and 40 μm. This heterogeneity is to be expected since the dog wool wastes were collected from various pet shops that use different wool treatments and cutting tools. These factors can then condition the micro-slicer precision and, consequently, the grounding process.

**Figure 3.** Dog wool microparticle size distribution.

### *3.2. ATR-FTIR Spectra*

The spectra profiles of the eco-composites formed of PU and different amounts of dog wool microparticles are shown in Figure 4. Between 3200–3450 cm−<sup>1</sup> was located one of the most important PU bands. This was attributed to the symmetric and asymmetric stretching vibrations of the N-H groups from the urethane and urea, which result from the reaction between water and isocyanate [22]. However, as observed by the spectrum of the dog wool fibers, a large peak at 3300 cm−<sup>1</sup> is typical of the stretching vibrations of –NH groups in keratin [23]. As the amount of dog wool fibers increased in the composite, this region became broader, which indicates a larger number of intermolecular hydrogen being promoted by these microparticles [24]. A very small peak corresponding to C–H stretching of the aliphatic CH=CH was identified at 3008 cm−1, while at 2950 and 2850 cm−<sup>1</sup> the asymmetric and symmetric stretching vibrations of C–H were observed, respectively. A peak around 2270 cm<sup>−</sup>1, associated with the stretching vibrations of the NCO group of the isocyanate, was detected in all

formulations. However, it was more important in those eco-composites containing higher amounts of dog wool microparticles. This is indicative of the presence of unreacted isocyanate [25]. The peaks at 1710, 1240, and 1070 cm−<sup>1</sup> relate to the stretching vibrations of the C=O and C–O of the ester groups, while the overlapping bands between 1540 and 1517 cm−<sup>1</sup> can be attributed to the stretching and bending vibrations of the C–N and N–H of the urethane moieties, respectively. These two peaks could also be assigned to the C–N stretching and N-H bending vibrations of amide II in wool fibers. This explains the increasing definition and clarity of these two peaks as the percentage of dog wool fibers rose in the eco-composite [26]. A very small increase in the composite of the peak at 1650 cm−<sup>1</sup> can be observed. This peak in the dog wool spectrum is attributed to the α-helix of the keratin structure [27]. This peak can be considered as a direct measure of the presence of the fiber in the composite.

**Figure 4.** ATR-FTIR spectra of the pristine PU and Dog Wool (DW) and of the eco-composites containing 5% to 20% of dog wool microparticles.

### *3.3. Thermal Properties*

Degradation steps associated with temperature rising were identified on PU and PU-based eco-composites via TGA (Figure 5). In the pure dog wool, a weight loss between 25 and 100 ◦C was observed due to the evaporation of the incorporated water. The second decomposition starting at around 200 ◦C could be attributed to the denaturation and degradation of the keratin molecules. According to literature, the disulfide bonds are cleaved between 230 and 250 ◦C [28]. In the composite, but not in the pure PU, a first very small step of degradation was detected between 25 and 100 ◦C (Figure 5 inset) and refers to the initial volatilization of moisture from the foams due to the evaporation or dehydration of hydrated cations [29,30]. This step was more important on the fiber-reinforced composites because of the wool fibers' affinity towards water molecules, which tends to increase moisture retention [24,31]. The first degradation step for the pristine PU was detected at ≈260 ◦C and was attributed to the cleavage of the PU polymeric backbone, initiating with the polyol component degradation (urethane chains) and, then, progressing to the isocyanate component degradation (ester bonds) [32]. At 300 ◦C, 12% of the original mass was already lost with the remaining 88% being further decomposed into amines, small transition components, and CO2 [33]. Because of the wool fibers' incorporation, the eco-composites were more quickly prone to degradation. This occurred because keratin wool fibers, such as dog hair fibers, start decomposing at temperatures superior to 200 ◦C. In

fact, from this point, denaturation of the helix structure and the destruction of chain linkages, peptide bridges, and the skeletal degradation occurs. At temperatures closer to 300 ◦C, several chemical reactions take place with the fibers being decomposed into lighter products and volatile compounds such as CO2, H2S, H2O, and HCN [34]. From all formulations, the eco-composites containing 5% of dog wool microparticles were capable of retaining more of their original mass, ≈ 91%, at 300 ◦C.

**Figure 5.** TGA of pristine PU and dog wool (DW) and the eco-composites containing 5, 10, 15, and 20% of dog wool microparticles measured between 25 and 300 ◦C, performed at a heating rate of 10 ◦C/min in a nitrogen atmosphere. The inset represents the initial part of the PU and PU composites between 25 and 130 ◦C.

DSC thermograms of the PU and PU-based composites prepared with different percentages of dog wool microparticles were acquired between 20 and 500 ◦C (Figure 6). The first heating cycle between 20 and 120 ◦C (Figure 6a) and the cooling cycle between 120 and 20 ◦C (Figure 6b) did not shown any significant event. In the second heating cycle (Figure 6c), the first endothermic peak for PU was detected at ≈300 ◦C, which, as seen earlier, is associated with the cleavage of the PU polymeric backbone, initiating with urethane chains and continuing to the ester bonds. In Figure 6d, it can be observed a detail of the second heat cycle between 100 and 180 ◦C. In this region a *T*<sup>g</sup> is observed in all the thermograms relative to the hard urethane segments [35]. However, the *T*<sup>g</sup> of the composites starting from 10% of dog wool content were lower (~150 ◦C) than corresponding PU control (~160 ◦C) and the 5% composite. It seems that the presence of the fibers affected the state of crystallinity in the PU matrix by reducing the *T*g towards lower temperatures. These *T*g are very small since polyurethane is mostly amorphous and suggest that the fibers improve the mobility of soft segment in PU, reducing the hydrogen bonding interactions [36]. For the eco-composite foams, the first endothermic peak occurred earlier, at ≈220 ◦C, with the initial denaturation of the wool fiber helix structure and the destruction of chain linkages. At temperatures ranging from 300 to 340 ◦C, the main polymeric chains in the eco-composite started degrading together with the remaining components of the wool fibers. These data are consistent with the TGA observations. The last endotherm peak registered for all foams was detected around 460 ◦C and can be attributed to the final degradation of the remaining residual polymeric chains and dog wool fibers into carbon char, small transition components, and volatile species [33,34].

**Figure 6.** *Cont.*

**Figure 6.** DSC thermogram of the pristine PU and the eco-composites containing 5, 10, 15, and 20% of dog wool microparticles collected at a heating rate of 10 ◦C/min in a nitrogen atmosphere. (**a**) The first heating cycle between 20 and 120 ◦C, (**b**) the first cooling cycle between 120 and 20 ◦C, (**c**) the second heating cycle between 20 and 500 ◦C. (**d**) Detail of the second heating cycle between 100 and 180 ◦C showing the *T*g.

In foamed systems, the dominant heat transfer modes are thermal radiation and gas-gas and solid-solid conduction. In PU foams, the total conductivity ranges about two-thirds of the conductivity of stagnant air because there is low conductivity gas, or foaming agent, inside the foam [33]. Here, the addition of the wool fibers to the eco-composites had little influence on the foams' thermal conductivity (Table 2), maintaining the values within the expected ranges, desirable for thermal insulation, and approximated to those of polystyrene (one of the most common materials applied in thermal insulation) [37–39]. The thermal or heat capacity measures the amount of energy required to raise the temperature of a material one degree. Data from Table 2 demonstrates, again, that the PU and the eco-composites presented very similar values. However, the addition of 20% dog wool

microparticles increased the composite thermal capacity above the pristine PU. Hence, this formulation requires more heat for temperature variations to occur, thus maintaining insulation more effectively. Thermal diffusivity describes the rate of temperature spread through a material and is a function of the thermal conductivity and the heat thermal capacity. As such, since thermal conductivity was the lowest in the eco-composites containing 20% of dog wool fibers, the same happened with the thermal diffusivity. It has been shown that thermal diffusivity is dependent on the organization of the foaming cells, their dimension, and the type of blowing agent applied [40]. Here, it is likely that the random disposition of the microparticles along the polymeric matrix may have compromised these specific thermal properties. Finally, in order to be classified as an insulating material, the foam must be endowed with a high thermal resistance. Data shows that thermal resistance decreased slightly with the addition of dog wool fibers. Even though these values are acceptable for thermal insulation, it seems that by increasing the percentage of fibers within the eco-composite this property is also enhanced. Thus, future studies will be conducted to confirm this premise.


**Table 2.** Main thermal properties of pristine PU and the eco-composites (*n* = 3, S.D. ± 3).

### *3.4. Mechanical Properties*

The foams' tensile stress and compression performance were measured with and without the addition of the dog wool microparticles (Figures 7 and 8, respectively). PU achieved the highest percentage of elongation from the tested formulations (≈50%), although requiring less stress (≈1.25 MPa) to break than the reminding eco-composites. In fact, with the addition of only 5% dog wool microparticles, the stress necessary to reach a similar elongation state (≈46%) was almost double, ≈2 MPa (Figure 7). This behavior is explained by the interactions established between the polymer matrix and the fiber arrays, which led to the disorganization of the PU original structure. At this percentage, small changes were induced in the foam's morphology. It is possible the microparticles migrated and filled existing defects, thus increasing the force necessary to break the material. A stiffness enhancement was registered with superior percentages of dog wool fibers. This is related to the higher rigidity of the foam solid phase in consequence of the fiber's contribution [5]. Because of the heterogeneous and disorganized orientation and distribution of the fibers along the composite, there was no proportion between the force applied-elongation capacity and the percentage of fiber reinforcement.

Even though the 5% eco-composites registered the most balanced performance between stress applied and elongation capacity (Figure 7), their resistance to compression was the lowest from the group (Figure 8). It is likely the rearrangements the polymeric foam underwent, to accommodate the fibers, promoted the development of an anisotropic-like material in which the mechanical resistance was more important in one direction than in the other [41]. Also, the presence of gaps along the foam in response to the addition of the microparticles and the alterations in the PU original structure may have contributed to this phenomenon. Irregularities in the foams' organization are more likely to occur in composites containing smaller amounts of reinforcement fibers than higher [5]. In turn, the increased percentage of filler can induce a decrease in the reactivity of the components in the system, affecting the foam expansion and increasing its density and rigidity, consequently, improving the compression strength [42]. As such, it was expected the maximum compressive stress to strain to be endured by the composite with the largest percentage of filler (20%).

**Figure 7.** Stress (MPa) versus elongation at break (%) of the pristine PU and the dog wool-reinforced eco-composites.

**Figure 8.** Compressive stress versus strain of the pristine PU and the dog wool-reinforced eco-composites.

### *3.5. Hydration Capacity*

The water adsorption capacity of the pristine PU and the dog wool-reinforced eco-composites was followed up to six days in *d*H2O, with samples being weighed every 24 h until water saturation was reached. Data from Figure 9 revealed pristine PU as the foaming material with the least hydration capacity, reaching a saturation state with only 4% of water in its composition. The eco-composites were found more attractive to water molecules with their hygroscopic capacity augmenting as the percentage of microparticles increased, that is, from 5% water content in the 5% dog wool eco-composites to 11% water content registered for the 20% dog wool-reinforced eco-composite. These results are explained by the ability of wool fibers to bind and absorb large amounts of water [31]. Water permeability in wool fibers is dictated mainly by cell membrane lipids. However, much remains to be understood on this front. The interaction between fibers and water is quite complicated; at low relative humidity, a water molecule monolayer can be formed by the interaction with specific fiber polar side chains, while at high relative humidity water associates with the peptide backbone of the fiber, generating a multilayer absorption. Fiber swelling also occurs, as a result of the breaking of hydrogen bonds between and within protein chains, due to water molecules rising over the surface and within the intercellular spaces; thus, generating even more interaction sites for water molecules [43,44].

**Figure 9.** Water adsorption capacity of the pristine PU and the dog wool-reinforced eco-composites over time.

### *3.6. Dilatometry*

Thermal expansion is defined as the increase in a material's volume in response to temperature rising. As the temperature rises, molecular agitation increases thereby growing the distance between the molecules. Figure 10 shows evidences of PU dilation with the increase in temperature from 50 to 160 ◦C. The addition of the dog wool fibers reduced significantly the polymer expansion with temperature variations. In fact, with a 5% addition of microparticles, the foam's volume did not even alter. By incorporating a higher content of wool fibers, the foam's dilatation reached negative values, indicative of the loss in material stability and capacity to maintain its structural integrity with heating. The best results showing structural stability up to 120 ◦C were obtained using a maximum of 15% of dog wool fibers.

**Figure 10.** PU and eco-composites' dilatation with increasing temperature, from 30 to 160 ◦C.

### **4. Conclusions**

PU eco-composites reinforced with dog wool fibers were successfully produced at different percentages. Alkaline treatment was effective in removing impurities from the fibers without compromising their integrity. Fibers were incorporated along the polymeric matrix in a random and disorganized manner. Still, they were effective in increasing the foams' mechanical resistance, namely tensile and compression strengths. The thermodynamic behavior suffered little changes with the incorporation of the dog wool fillers, being the most important the improvement in thermal capacity. Additionally, the hydration capacity was significantly improved in response to the wool fibers' water permeability and increased capacity to bind with water molecules. Dilatometry studies revealed the capacity of the fillers to increase the thermal stability of the foam, reducing their expansion with heating. Data demonstrated the potential of this combination to produce new alternative solutions for insulation using low-cost, sustainable resources and with minimal environmental impact.

**Author Contributions:** F.C.d.S., K.K.d.O.S.S., and J.U.L.M. performed the main analysis and data collection, finalized the paper, and performed data interpretation. R.L. supervised and performed data interpretation. A.Z. and H.P.F. performed data analysis and finalized the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** Authors acknowledge the *Fundação de Apoio a Pesquisa do Estado do Rio Grande do Norte* (FAPERN) for financing this work. They thank the pet shops from Natal city for donating the dog wool fibers used in the experiments. H.P. Felgueiras and A. Zille also acknowledge project UID/CTM/00264/2019 of Centre for Textile Science and Technology (2C2T), funded by national funds through FCT/MCTES.

**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* **Polypropylene**/**Basalt Fabric Laminates: Flexural Properties and Impact Damage Behavior**

### **Pietro Russo 1, Ilaria Papa 2, Vito Pagliarulo <sup>3</sup> and Valentina Lopresto 2,\***


Received: 27 April 2020; Accepted: 7 May 2020; Published: 8 May 2020

**Abstract:** Recently, the growing interests into the environmental matter are driving the research interest to the development of new eco-sustainable composite materials toward the replacement of synthetic reinforcing fibers with natural ones and exploiting the intrinsic recyclability of thermoplastic resins even for uses in which thermosetting matrices are well consolidated (e.g., naval and aeronautical fields). In this work, polypropylene/basalt fabric composite samples were prepared by film stacking and compression molding procedures. They have been studied in terms of flexural and low-velocity impact behavior. The influence related to the matrix modification with a pre-optimized amount of maleic anhydride grafted PP as coupling agent was studied. The mechanical performances of the composite systems were compared with those of laminates consisting of the pure matrix and obtained by hot-pressing of PP pellets and PP films used in the stacking procedure. Results, on one side, demonstrated a slight reduction of both static and dynamic parameters at the break for specimens from superimposed films to ones prepared from PP pellets. Moreover, an outstanding improvement of mechanical performances was shown in the presence of basalt layers, especially for compatibilized samples.

**Keywords:** polypropylene; basalt fibers; composite laminate; flexural; impact damage

### **1. Introduction**

The growing diffusion of plastics in all industrial fields and concerns related to their environmental impact and secure disposal at the end of the useful life has been the main reason motivating academic and industrial research towards the study and development of new eco-sustainable materials. So, an increasing interest in natural fibers as polymers reinforcement (conventional and bio-sources derived) has been detected. In this context, basalt fibers have gained outstanding attention as reinforcing fibers concerning traditional glass and carbon-based ones [1,2]. Specifically, the interest of the research toward the use of basalt fibers was driven by their remarkable properties such as relatively low cost, sustainability, enhanced mechanical properties, non-flammability, high chemical stability, excellent sound and thermal insulation [3]. Basalt fibers, derived from natural rocks in volcanic regions with technologies similar to that of glass fibers' forming, are mainly composed of silica (SiO2) and alumina (Al2O3) [4] and have a melting temperature ranging from 1500 ◦C and 1700 ◦C with an average diameter between 9 and 13 micrometers. In the last decades, basalt fibers, included in thermosetting and thermoplastic resins for uses in several fields as transportation, defense and constructions, have largely demonstrated their potentiality as eco-sustainable substitutes of glass fibers, especially if an adequate adhesion at the interface with the surrounding polymer phase is ensured.

Concerning thermosetting matrices, it is possible to obtain great benefits, primarily in terms of mechanical and thermal properties, including basalt fibers in epoxy, polyester and vinyl ester resins [5–9]. However—with the awareness that the sustainability of new products can be encouraged by the recyclability of the involved matrices—recently, the focus of the research has been increasingly devoted to thermoplastic systems mainly based on polypropylene [10–12], polyethene [13,14], polyamides [15–18], polyesters [19,20] as well as polymer blends [21] and hybrids [22,23].

In particular, polypropylene resins, primarily used for mass productions due to their low cost and full versatility in terms of processability and properties, still receive an extraordinary interest with research efforts mainly aimed to improve their performances further and extend their field of application more and more.

In this regard, Szabo et al. [11] demonstrated that the mechanical properties of polypropylene composites packed with different short fibers are strongly affected by the content and direction of the same. As witnessed by scanning electron micrographs, samples showed damage mechanisms as fiber pull out in the perpendicular direction and debonding along the longitudinal direction depending on the manufacturing process used.

Guo et al. [12] considered basalt fabric modified with the aid of a silane coupling agent KH550 to reinforce a commercial polypropylene resin. Hydrophilicity and lipophilicity tests of modified basalt fibers allowed a preliminary optimization of the pre-treatment parameters. Time and coupling agent content at values of 1.5 h and 6% by weight have been varied, respectively. The surface modification of basalt fibers significantly improved tensile, impact resistance, bending properties and thermal stability of the PP matrix giving rise to composites with better dynamic viscoelasticity compared to ones containing the same content of unmodified basalt fibers.

Greco et al. [24] highlighted the crucial role of structural features as the microcrystal size and the amorphous content of basalt fibers in determining their mechanical properties as well as the expected enhancement of mechanical performances of polypropylene composites by increasing the fiber-matrix interface goodness.

In the present paper, polypropylene/basalt fabric composite laminates, prepared by film stacking and hot-pressing procedures, were analyzed. Flexural and low-velocity impact properties were studied, taking pure PP specimens as reference materials and considering influences related to the matrix modification with a pre-optimized amount of a maleic anhydride grafted PP as a coupling agent. Mechanical results, also interpreted in light of indentation measurements, demonstrated that, although filming of the matrix does not significantly compromise mechanical performances of the PP matrix, the presence of basalt layers induces a marked improvement of composite performances. Moreover, the preliminary modification of the matrix to enhance the interfacial adhesion further improves the flexural performances, especially in terms of strength allowing to withstand higher impact loading concerning neat PP based composite laminates. Premised that for both investigated composites, no penetration seems to occur and the more significant matrix-reinforcement interaction for compatibilized composite laminamakesmake them less prone to plastic deformations as also evidenced by the indentation measurements.

### **2. Experimental**

### *2.1. Materials*

In this paper a polypropylene matrix (Hyosung Topilene PP J640, MFI@230 ◦C, 2.16 kg: 10 g/10 min) provided by Songhan Plastic Technology Co. Ltd. (Shangai, China) and a plain wave basalt fabric (areal weight: 210 g/m2) from Incotelogy GmbH (Pulheim, Germany). The resin was used as received (PP) or pre-modified (PPC) by the inclusion of 2 wt.% of a coupling agent Polybond 3000 (MFI@190 ◦C, 2.16 kg: 400 g/10 min) from Chemtura (Philadelphia, Pennsylvania, USA). This latter is a polypropylene grafted with maleic anhydride (PP-*g*-MA) with 1.2 wt.% of MA.

In more details, the modification of the commercial PP was performed with the aid of a co-rotating twin-screw extruder Collin Teach-Line ZK25T (Ebersberg, Germany), operating with the following temperature profile: 180–190–205–195–85 ◦C, from the hopper to the die, and at a screw speed of 60 rpm. The neat PP and the extruded pellets of modified PP were transformed in flat films with a thickness approximately equal to 35–40 μm using a Collin flat die extruder Teach-Line E20T equipped with a calendar CR72T (Ebersberg, Germany). For this stage, the processing was conducted at a screw speed of 55 rpm, setting the temperature profile along the screw at 180, 190, 200, 190 and 185 ◦C.

### *2.2. Laminates Preparation*

The conventional film stacking technique was used to obtain laminates. Layers of plastic films and basalt fabric, alternatively overlapped, are subjected to a pre-optimized pressure and temperature cycle (see Figure 1a) by press Collin GmbH (Edersberg, Germany) Mod. P400E.

In this way, 380 mm × 380 mm plaques constituted by 16 plies, symmetrically settled with respect to the medium plane, with an average thickness of 2.6 mm and a volumetric content of reinforcement of about 50% (ASTM D 3171-04, Test Method II) have been produced.

As a reference, plates of only PP matrix, with a thickness approximately equal to 2.5 mm, were prepared from PP granules and by superposition of plastic films mentioned above, according to the conditions shown in Figure 1b.

**Figure 1.** Hot-pressing conditions for (**a**) composite and reference (**b**) plates.

### *2.3. Experimental Techniques*

### 2.3.1. Static Mechanical Properties

Flexural tests have been carried out by a universal Instron dynamometer Mod. 5564 equipped with a load cell of 1 kN. Specimens 12.7 mm wide and 100 mm long were cut from each sample laminate and loaded at room temperature in the three-point bending mode, according to the ASTM D790 standard, using a cross-head speed set at 2.5 mm/min and a span equal to 70 mm.

Flexural parameters evaluated by processing typical stress–strain curves were averaged on at least 5 determination for each investigated sample.

### 2.3.2. Impact Properties

Falling weight machine (Ceast Fractovis, Torino, Italy) was used for the impact tests at complete penetration, to obtain and study the full load–displacement curves to get useful information about the response of the laminates and for investigating the effect of the varied parameters. The penetration energy, Up, set to obtain the complete penetration of all composite systems studied is equal to 100 J. Then, different energy levels (U = 3, 8, 15 J) were chosen to carry out the so-called indentation tests, useful to study the damage start and evolution. The rectangular samples, 100 × 150 mm, cut by

a diamond saw from the original panels, were supported by the clamping device suggested by the ASTM D7137 Standard and were loaded in the center by an instrumented cylindrical impactor with a hemispherical nose, 19.8 mm in diameter. Tests were carried out using an impactor (m = 3.640 kg) placed at specific heights to obtain the selected impact energies. After all the impact tests, the samples were observed by a visual 2D stylus profilometer (Dektak XT) to derive quantitative information about step heights, which is indentation depth along the impacted areas.

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

The flexural stress–strain curves of composite specimens based on the neat and pre-modified film of polypropylene (coded as PP and PPC, respectively) are reported in Figure 2. At the same time, the evaluated mechanical parameters (modulus and strength) are summarized in Table 1. Flexural parameters compare quite favorably with those of other studies [5].

**Figure 2.** Flexural stress–strain curves.



With regard to composite systems, previous morphological analyses conducted by electron scanning microscopy (SEM) on the same laminates have shown that for PP/Basalt composites, as expected given the nature of the phases involved, a poor interfacial adhesion is evident with images showing reinforcing fibers predominantly smooth and clean. On the contrary, for the PPC/basalt system, the authors reported that the presence of PP-*g*-MA improves the fiber wetting [25].

In light of the foregoing consideration, the significant improvement of the flexural parameters such as modulus and strength detected for the specimens containing the coupling agent is reasonable. In particular, the flexural strength of PPC/Basalt specimens was about twice the value estimated for composites based on neat PP.

Figure 3 compares reference materials in terms of load–displacement curves at penetration. From the curves, it is possible to note the higher maximum load, Fmax and the corresponding displacement value, d, for the PP film-based specimens (see Table 2).

**Figure 3.** Load–displacement curves at penetration of PP granule and PP film-based samples.

**Table 2.** Impact parameters at penetration for matrix.


In Figure 4, the impact curves at the penetration of composite materials are reported.

Remembering that the slope of the first linear part of the curve is an index of the impact rigidity of the system tested, the compatibilized system showed a slightly lower impact rigidity compared to the PP/Basalt one.

**Figure 4.** Load–displacement curves up to penetration: comparison between BS\_PP and BS\_PPC.

However, as also highlighted by the average data summarized in Table 3, it is capable of supporting higher loads (Fmax) before the reinforcement breaks recording a higher deflection, d, in correspondence of the maximum load.

Furthermore, interestingly, in both cases, it was not possible to penetrate the panels even if a higher value of the maximum load, Umax, is recorded for the PP/Basalt composite sample (Table 3). This behavior is highlighted in Figures 5 and 6. It is evident that both the non-compatibilized and the compatibilized specimens battle to the penetration presenting only a hunching due to the impact event.


**Table 3.** Impact at penetration parameters for both composite systems.

**Figure 5.** Impacted PPC basalt samples at penetration: (**a**) Front; (**b**) back.

**Figure 6.** Impacted PP basalt samples at penetration: (**a**) Front; (**b**) back.

As far as the indentation measurements are concerned, in Figures 7–9 load–displacement curves of PP and PPC samples for three impact energy levels, U, are reported.

**Figure 7.** Load–displacement curves at indentation: comparison between BS\_PP and BS\_PPC; U = 3 J.

**Figure 8.** Load–displacement curves at indentation: comparison between BS\_PP and BS\_PPC; U = 8 J.

**Figure 9.** Load–displacement curves at indentation: comparison between BS\_PP and BS\_PPC; U = 15 J.

Taking into account that the area enclosed in the indentation curves represents the energy absorbed by the laminate to create damage, Ua, this parameter and the maximum impact load, Fmax, increase as the impact energy, U, increases as shown in Figures 10 and 11, respectively. In particular, the Ua values, reflecting the extent of induced internal damage, are more significant for the PP/Basalt samples than for the PPC/Basalt ones for every tested energy indicating the occurrence of minor damage for the latter, under similar impact conditions.

**Figure 10.** Maximum load, Fmax, versus impact energy, U.

**Figure 11.** Absorbed energy, Ua, versus impact energy, U.

The increasing load usually results in a greater maximum deflection (Figure 12). The presence of the compatibilizer does not change this effect even if it gives rise to minor deflections to indicate a small amount of energy spent on bending. For the PP/Basalt system, the trend of the deflection has a stronger rise after U = 8 J, lowering the bending. The latter results in a greater amount of energy absorbed (see Figure 11), indicating global higher damage in the PP/Basalt system.

**Figure 12.** Maximum deflection, d, versus impact energy, U.

The last assertion is confirmed by measurements of the indentation depth representing the footprint impress by the impactor on the impacted side of the sample (residual plastic deformation) shown in Figure 13. The results demonstrated that this parameter increases at the increasing of the impact energy, U, for both types of basalt composite samples.

**Figure 13.** BS-PP and BS-PPC Indentation profiles.

Figure 14 shows the trend of the indentation depth, I, as a function of the impact energy, U. It is possible to note that the indentation, I, measured on the basalt PP samples is higher than that shown by the PPC ones on the entire range of impact energies examined: it was an expected effect since the best interfacial adhesion prevents plastic deformation of the matrix.

**Figure 14.** PP/Basalt and PPC/Basalt Indentation depth, I, versus the impact energy, U.

### **4. Conclusions**

Polypropylene/basalt fabric composite laminates prepared by film piling and compression molding techniques were analyzed. Flexural and low-velocity impact properties were evaluated, taking neat PP plates as references and considering effects related to matrix modification with a pre-optimized amount of a maleic anhydride grafted PP as a coupling agent. Mechanical results demonstrated the preliminary change of the matrix to enhance the interfacial adhesion leads to samples with improved flexural performances especially the strength and ability to withstand higher impact loading with respect to neat PP based composite laminates. Premised that for both investigated systems, no penetration seems to occur, the more significant matrix-reinforcement interaction for compatibilized composite laminates make them less prone to plastic deformations as also evidenced by the indentation measurements. The indentation, I, recorded for the basalt PP samples is higher than the PPC ones confirming the higher absorbed energy, Ua that denotes greater damage due to the impact tests.

**Author Contributions:** Conceptualization, I.P, V.L. and P.R.; methodology, I.P, V.L. and P.R.; validation, I.P, V.L., V.P. and P.R.; investigation, I.P; resources, P.R.; data curation, I.P and V.P. writing—original draft preparation, I.P. and P.R. writing—review and editing, V.L.; supervision, V.L.; funding acquisition, V.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by ONR Solid Mechanics Program, Survivability of marine composites and structures under impact and blast in extreme environments N00014-14-1-0380 ONR.

**Acknowledgments:** The authors gratefully acknowledge the ONR Solid Mechanics Program, in the person of Yapa D.S. Rajapakse, Program Manager, for the financial support provided to this research.

**Conflicts of Interest:** The authors declare no conflicts 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**

1. Fiore, V.; Scalici, T.; di Bella, G.; Valenza, A. A review on basalt fibre and its composites. *Compos. Part B* **2015**, *74*, 74–94. [CrossRef]


© 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* **Falling Weight Impact Damage Characterisation of Flax and Flax Basalt Vinyl Ester Hybrid Composites**

### **Hom Nath Dhakal \*, Elwan Le Méner, Marc Feldner, Chulin Jiang and Zhongyi Zhang**

Advanced Materials and Manufacturing (AMM) Research Group, School of Mechanical and Design Engineering, University of Portsmouth, Anglesea Road, Anglesea Building, Portsmouth, Hampshire PO1 3DJ, UK; elwan.le-mener@port.ac.uk (E.L.M.); feldner.marc@gmail.com (M.F.); chulin.jiang@port.ac.uk (C.J.); zhongyi.zhang@port.ac.uk (Z.Z.)

**\*** Correspondence: hom.dhakal@port.ac.uk; Tel.: +44-(0)23-9284-2582; Fax: +44-(0)23-9284-2351

Received: 28 February 2020; Accepted: 1 April 2020; Published: 3 April 2020

**Abstract:** Understanding the damage mechanisms of composite materials requires detailed mapping of the failure behaviour using reliable techniques. This research focuses on an evaluation of the low-velocity falling weight impact damage behaviour of flax-basalt/vinyl ester (VE) hybrid composites. Incident impact energies under three different energy levels (50, 60, and 70 Joules) were employed to cause complete perforation in order to characterise different impact damage parameters, such as energy absorption characteristics, and damage modes and mechanisms. In addition, the water absorption behaviour of flax and flax basalt hybrid composites and its effects on the impact damage performance were also investigated. All the samples subjected to different incident energies were characterised using non-destructive techniques, such as scanning electron microscopy (SEM) and X-ray computed micro-tomography (πCT), to assess the damage mechanisms of studied flax/VE and flax/basalt/VE hybrid composites. The experimental results showed that the basalt hybrid system had a high impact energy and peak load compared to the flax/VE composite without hybridisation, indicating that a hybrid approach is a promising strategy for enhancing the toughness properties of natural fibre composites. The πCT and SEM images revealed that the failure modes observed for flax and flax basalt hybrid composites were a combination of matrix cracking, delamination, fibre breakage, and fibre pull out.

**Keywords:** polymer-matrix composites (PMCs); composite laminates; low-velocity impact; delamination; X-ray micro CT

### **1. Introduction**

Over the past decade, consumers' increased awareness of and expectations towards environmental sustainability, government legislation, an increased sense of corporate social responsibility (CSR) from industry sectors for achieving sustainable development aspirations through a triple bottom line performance (environment, economic, and social), have inspired research on materials which are renewable and recyclable [1,2].

Natural fibre-reinforced polymeric composites have been used in a wide range of engineering applications in recent years due to their abundant availability, lower density, and much higher specific strength and modulus than conventional glass and carbon fibre-reinforced composites [3–5]. Moreover, these reinforced materials possess a low embodied energy to process and use compared to energy-intensive conventional fibre-reinforced composites. Despite several benefits, there are still some issues which limit the use of natural fibre-reinforced composites in semi-structural and structural applications [6,7]. One of the key issues facing these composites is their hydrophilic nature, which leads to poor fibre matrix interfacial adhesion and lower mechanical properties [8].

Among the natural fibres used in polymeric composites, bast fibres (flax, hemp, jute, and kenaf) stand out as the most promising reinforcements [9,10]. Due to their unique hollow structure, these fibres provide a good damping property, which is very important when it comes to dealing with impact damage and vibration damping behaviours. The good damping properties of bast fibre-reinforced composites make them an attractive alternative to be used in automotive components where impact and damping properties are very important [11,12]. However, their high natural variability, strong affinity to water until saturation, limited processing temperature range, relatively low impact resistance, and low thermal stability negatively influence their long-term durability [13,14]. Moreover, their low impact resistance behaviour under different operation conditions is another concern when these materials are used in automotive and marine sectors.

Mitrevski et al. [15] studied the influence of impactor shape on the impact damage of composite laminates. The results demonstrated that the impactor shape plays a large role in the damage response of composite materials. Composite laminates undergo various impact-induced damage modes under impact loadings. A review carried out by Cantwell and Morton [16] has reported the various impact-induced failure modes of composite laminates. Wisheart and Richardson [17] investigated the impact damage response of complex geometry pultruded glass/polyester composites.

During the past two decades, there have been many reported works, especially those covering the mechanical and thermal properties of natural bast fibre-reinforced composites. This highlights the significant increase in the demand for these fibres. The automotive sector is leading the way towards using natural bast fibres due to the drives to produce lightweight and sustainable materials and reduce health risks during manufacturing and recycling. Many leading original equipment manufacturers (OEMs) in the automotive sector have been using natural fibre-reinforced composites in various parts, such as door linings, seat cushions, door cladding, and mainly non-structural applications. Nevertheless, due to the lack of sufficient mechanical properties for structural applications, natural fibre-reinforced biocomposites are not fully utilised in semi-structural and structural applications [18,19].

In recent years, widespread research has been focused on utilising a hybrid approach which consists of combining two or more reinforcements, in which the synergic effects of both reinforcements are utilised. In order to compensate for the shortcomings of the natural fibre-reinforced composites, glass and carbon fibres, as well as nano particulates, have been used as hybrid constituents [20–23]. There are many reported works where basalt fibres have been introduced as hybrid reinforcements on natural fibre composites owing to their high thermal stability, good mechanical properties, good corrosion resistance, and natural origin (coming from volcano rock). The work carried out by Dhakal et al. [24] investigated the influence of basalt fibre hybridisation on the post-impact mechanical behaviour of hemp fibre-reinforced composites, and their report suggests that basalt fibre hybrid systems significantly improved the post-impact mechanical properties. Similarly, carbon fibre hybridised flax fibre composites were investigated and it was observed that the hybrid system offered excellent mechanical properties compared to flax fibre non-hybrid composites [25]. Paturel and Dhakal [26] investigated the water absorption and low velocity impact damage characteristics of flax/glass fibre hybrid vinyl ester composites. Their findings suggest that glass fibre hybridised composites significantly reduced the water uptake percentage compared to flax fibre vinyl ester composites without hybridisation. It is evident from the various literature that the impact damage characteristics of natural fibre composites have been well-documented. However, not many studies have been focused on investigating the influence of basalt fibre hybrid flax composites subjected to low-velocity impact loading at different incident energy levels. Additionally, there has been limited work on the influence of hybridisation on the moisture absorption and its effects on the low-velocity impact damage mechanisms.

This study aimed to investigate the effect of basalt fibre hybridisation on the water absorption and low-velocity falling weight impact behaviour of flax fibre-reinforced vinyl ester hybrid composites with varying incident impact energies. For this, the flax/VE composites were impacted at low impact energies ranging from 50 to 70 Joules, which was sufficient to create impact damage up to penetration. The impact performance of flax and flax/basalt/VE hybrid composites was evaluated in terms of the load bearing capability, energy absorption, and damage modes. In addition, the water absorption behaviour of flax and flax basalt hybrid composites and its effects on the impact damage were also investigated. The damage mechanisms of impacted composite specimens were characterised using non-destructive evaluation techniques, such as X-ray computed micro-tomography (πCT) and scanning electron microscopy (SEM).

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

### *2.1. Materials and Laminate Fabrication*

The matrix material used was vinyl ester, Scott-Bader Crystic VE676-03, obtained from Scott-Bader. Woven flax and woven basalt fibres were used as the reinforcements (±45) as biaxial stitched non-crimp fabrics of 600 g/m2. Figure 1 shows the flax and basalt fabric used to make flax and flax/basalt hybrid composites. The chemical and mechanical properties of key bast fibres (flax, kenaf, hemp, and jute), along with basalt fibre included for comparison purposes, are presented in Table 1.

**Figure 1.** Reinforcements used, (**a**) flax woven fabric, and (**b**) basalt woven fabric.

**Table 1.** Chemical composition, and physical and mechanical properties of commonly used natural bast fibres [27–30].


\* For comparison purposes.

### *2.2. Composite Laminate Fabrication*

The flax and flax/basalt hybrid laminates were fabricated by the vacuum infusion technique. Two types of samples were fabricated to investigate the influence of hybridisation. In the first set of samples, six layers of flax fibres were used. A second set of samples, including one layer of basalt fibres on the top face and five layers of flax fabric on the rear side, were employed. The main reason for investigating the fibre contents and orientation is to optimise the hybrid materials in the loading direction. The average of the fibre volume fraction of Flax/VE and Flax/Basalt/VE was approximately 31% and 33%, respectively. The void content was approximately 3%. The sample size of a 70 mm by

70 mm square was cut using water jet cutting of the composite panel. Schematics the of flax and flax basalt hybrid composite laminates are shown in Figure 2.

**Figure 2.** Formulation of laminates: (**a**) Flax/VE laminate with a thickness of 6 mm, and (**b**) flax/basalt/VE hybrid laminates with a thickness of 5.5 mm.

### *2.3. Moisture Absorption Measurement*

The moisture uptake behaviour of flax/VE and flax/VE/basalt hybrid laminates was investigated in accordance with BS EN ISO:1999 [31]. Five specimens, consisting of 70 mm by 70 mm squares of flax/VE and flax/VE/basalt hybrid composites, were placed in a desiccator for 48 h and weighted near to 0.1 mg. Then, the specimens were immersed in de-ionised water at room temperature. After 24 h of immersion, the specimens were taken out and the surface was dried with absorbent paper. The process was repeated until the saturation moisture was reached. The percentage of moisture uptake was calculated using Equation (1):

$$M(\%) = \frac{M\_{\rm t} - M\_0}{M\_0} \times 100\tag{1}$$

where *M* (%) is the moisture uptake in percentage, *M*t is the weight of the water-immersed specimen at a given time, and *M*<sup>0</sup> is the initial mass of the specimen in a dry condition.

### *2.4. Low-Velocity Drop Weight Impact Testing*

Low-velocity instrumented falling weight impact testing was conducted by using an instrumented falling weight impact testing (IFWIT) machine. A Zwick/Roell impact test machine (IFW 413) was used for the testing in accordance with the British Standard BSEN ISO 6603-2 recommendations [32]. The hemispherical impact tup used was made of steel and had a 20 mm diameter. The incident energy was tailored by adjusting the release height of the impact mass, i.e., changing the impact velocities (while keeping the other parameters constant). To analyse the damage behaviour of flax and flax/basalt hybrid vinyl ester hybrid composites, the three different incident energies employed were 50, 60, and 70 Joules, respectively (corresponding impact velocities of 2.08, 2.28, and 2.46 m/s). The test specimens were 70 mm by 70 mm squares.

### *2.5. X-Ray Computed Micro-Tomography (*π*CT)*

πCT, XT H 225 was used to assess the barely visible impact damage failure. The samples were subjected to the three different incident energies of 50, 60, and 70 Joules and examined using X-ray (πCT) to effectively evaluate the extent of damage due to impact loadings.

### *2.6. Scanning Electron Microscopy (SEM)*

In order to investigate the damage mechanisms of the impacted samples, the surfaces of dried specimens and room temperature immersed specimens were examined using SEM Zeiss EVO LS10. Before examination, the samples were placed in a desiccator to remove all of the water of the samples, in order to avoid evaporation during characterisation. The specimens were also surface prepared and the damaged area was imaged.

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

### *3.1. Flax*/*Basalt*/*VE Hybrid Composite Lay Up*

One of the main aims of this study was to find out if hybridising one side of the composite plate would provide optimal hybrid effects. Figure 3 shows the hybrid sample with a rear basalt layer. It is clear from the figure that the basalt layer exhibits push-out delamination. As the incident energy increases, the delamination also increases. Delamination is one of the most prevalent failure mechanisms in composite laminates. This phenomenon becomes even greater when two different types of fibres are hybridised. It can be seen that the impactor has perforated and all the flax layers have fractured, but the basalt layer has not fractured; these impacted images show delamination of the basalt layer.

**Figure 3.** Images of damage progression on the front and rear faces of a flax/VE/basalt hybrid composite where basalt was placed on the rear side of the composite panels impacted in the range of (**a**,**b**) 50 J, (**c**,**d**) 60 J, and (**e**,**f**) 70 J.

The influence of the basalt fibre layer on the front and rear side of the laminates is further explain in Figure 4. It is evident from the figure that flax composites with a basalt layer on the top impact face provided optimal properties in comparison to the basalt fibre on the rear side under the three different incident impact energies of 50, 60, and 70 Joules. The main reason for this phenomenon is that when basalt fibre was placed on the rear side of the laminates, a significant amount of delamination was observed, which is shown in Figure 4. Employing this evidence, the remaining investigation was carried out on samples where basalt fibre was placed on the top side of the flax/VE composites. Just placing one layer of basalt fibre on the top of the flax/VE composite laminate provides a good design choice to fabricate high-performance composite laminates using a simple and cost-effective method.

**Figure 4.** Maximum load comparison of different impact configurations of flax/VE, flax/VE/basalt rear, and flax/VE/basalt front for each incident energy employed.

### *3.2. Impact Damage Characteristics*

### Load and Energy Absorption Capabilities

Important impact parameters and corresponding values obtained from the low-velocity testing for flax and flax basalt hybrid samples at three different incident impact energies (50, 60, and 70 Joules) are presented in Table 2.


**Table 2.** Important impact parameters and corresponding values obtained from impact testing.

Load–deformation–energy traces obtained from the impact testing for flax/VE composites are shown in Figure 5.

Two of the most used parameters to assess damage resistance in composites after an impact are the impact energy and absorbed energy. The impact energy represents the maximum energy that the specimen can transform (it is equal to the kinetic energy of the impactor right before dart contact with the sample when the impact takes place), whereas the absorbed energy is the unrecoverable energy dissipated by the system (including energy dissipated by friction and, most importantly, by mechanisms which are peculiar to the material).

**Figure 5.** Load and work vs. deformation for flax/VE composites: (**a**) flax 50 J, (**b**) flax 60 J, and (**c**) flax 70 J. FM, load maximum; FI, incipient damage load; EM, energy maximum; EI, incipient damage energy; ET, energy total.

The absorbed energy can be calculated from load vs. deformation curves. In order to evaluate the laminate's performances, the transient response of each laminate was recorded in terms of the load, energy, and displacement. It can be observed from Figures 5 and 6 that the peak contact force is higher

for the flax/basalt hybrid composite than that of flax/VE without hybridisation, which indicates that the hybrid specimens offer a higher impact resistance during impact events. A similar positive hybrid effect can be observed in load–time traces (Figure 7).

**Figure 6.** Load and work vs. deformation for flax/VE/basalt hybrid composites: (**a**) flax 50 J, (**b**) flax 60 J, and (**c**) flax 70 J. FM, load maximum; FI, incipient damage load; EM, energy maximum; EI, incipient damage energy; ET, energy total.

**Figure 7.** Typical load vs. time traces for 50 J, 60 J, and 70 J impacted (**a**) flax and (**b**) flax/basalt hybrid composites.

The load–deformation–energy traces for flax/basalt/VE hybrid composites are depicted in Figure 6. Flax/VE/basalt hybrid composites absorb more energy than flax/VE, as illustrated in Figure 6. We can observe similar curves for all three energy levels for flax/basalt hybrid composites. It is evident from Figure 6 that the applied incident energies (50, 60, and 70 Joules) were not high enough to penetrate or perforate (showing rebound energy) the hybrid samples, indicating their superior mechanical behaviour and higher energy dissipation potential compared to flax/VE samples without hybridisation. It is clear that flax/basalt hybrid composites exhibited a significantly improved impact performance compared to flax/VE composites with hybridisation.

Figure 6 illustrates the performance of flax basalt hybrid composites in terms of representing different impact parameters. It is clear from the results that flax/basalt hybrid composites exhibited significantly improved impact performances compared to flax/VE composites without hybridisation (Figure 5). It can also be observed that basalt hybridisation contributes to increasing the deformation of composites. This improvement could be attributed to the higher failure strain of basalt fibres compared to commonly used natural fibres such as flax and hemp. This phenomenon provides a balanced property, as one would expect for hybrid systems. These results indicate that basalt fibre hybridisation into natural fibre composites provides a promising strategy for enhancing the overall impact toughness. Such hybrid effects have been reported for improved mechanical properties, such as tensile, flexural, and fracture toughness behaviours [33,34].

Figure 7 shows the load–time traces of impacted flax and flax/basalt hybrid composite specimens. In this case, the peak load is the same as previously shown (Figures 5 and 6). However, the test time required to complete the impact event is important to consider. The time the striker was in contact with the impacted specimens is approximately the same for each composite. However, the time taken to complete the impact event is longer for flax/basalt hybrid composites compared to flax/VE composites. This is an indication that hybrid composites have better impact resistance behaviour as a result of the hybrid effect.

### *3.3. Moisture Absorption Behaviour*

Figure 8 depicts the moisture absorption curves of flax/VE and flax/basalt/VE hybrid composites. It is evident from the curves that the moisture uptake at the beginning is linear and rapidly increases for flax/VE composites compared to flax/basalt hybrid composites. After the initial rapid rise, the moisture uptake slows down and reaches saturation at 768 h (30 days) for flax composites, whereas it takes longer—1008 h (42 days)—for flax/basalt hybrid composites. The longer time taken for hybrid composites to reach saturation moisture absorption can be attributed to the influence of basalt fibres restricting the flow of water molecules, as basalt fibres have better water repellence behaviour compared

to flax/VE composites. Nonetheless, for the side where only flax is exposed, there would still be moisture ingress taking place at a higher rate than where basalt fabric was placed as a hybrid layer. This can be observed by the moisture uptake percentage difference between flax and flax/basalt hybrid composites, which is only 0.5%. If the basalt fabric was placed on both sides of the flax samples, the moisture uptake percentage of hybrid composites would have been far lower. Moreover, the sides of the both types of composites were not sealed, which is another reason for the higher moisture absorption displayed by both composites.

The maximum weight gain reported for vinyl ester matrix is 1.07% at room temperature [30]. The maximum weight gain percentages for flax/VE and flax/basalt/VE hybrid composites were approximately 4% and 3.5%, respectively. The lower moisture uptake percentage for flax/basalt hybrid composites is attributed to the barrier effects of top-layer basalt fibre on flax/VE composites. The moisture absorption behaviour for both composites indicates Fickian behaviour, which is rapid in the beginning and slowly reaches saturation.

By comparing the flax-basalt hybrid specimens with those made entirely of flax, it can be seen that the addition of basalt fibre improves the moisture absorption resistance of the hybrid composite. Since the five layers of flax fibre were sandwiched by one ply of basalt fibre, the total area of flax exposed to the water was decreased for flax/VE composites. The reason for the difference in moisture absorption between the flax and basalt hybrid specimens can be further explained by considering the chemical composition of flax fibres. The cellulose in the flax fibre is what provides the majority of the stiffness and strength; however, the semi-crystalline structure contains a large amount of hydroxyl groups, which give the fibre hydrophilic characteristics. By covering flax fibres with basalt fibres in a hybrid composite, the surface area exposed to water is reduced and therefore absorbs less moisture.

### Influence of Moisture Absorption on the Impact Resistance Behaviour

The influence of moisture absorption on the flax composite in dry and wet conditions was investigated. The effects of moisture absorption on the load bearing capability of flax/VE composites impacted at two different incident energy levels are shown in Figure 9. The wet flax/VE specimens displayed a higher peak load compared to wet specimens, slightly outperforming the dry sample. This could be attributed to water absorption-induced plasticisation of the vinyl ester matrix leading to an increase of deformation and impact energy absorption [5,23].

Generally, when natural fibres absorb moisture, they swell, which promotes the development of adverse effects on the mechanical properties, such as tensile, flexural, and fatigue properties, due to the weak fibre matrix interface. However, as far as the impact performance is concerned, the results from this study suggest that there was no negative influence of moisture absorption on the load. Instead, the wet samples withstood a slightly higher peak load than the dry ones. This could be attributed to

engrossed amounts of water causing swelling of the flax fibres, which could fill the gaps between the fibre and vinyl ester matrix and could have eventually led to an increase of impact load [5].

### *3.4. Damage Characterisation*

### 3.4.1. Damage Behaviour in Dry Conditions

Figure 10 depicts the damage of front and rear faces of impacted flax/VE composite specimens. As can be clearly observed, all the samples impacted at 50, 60, and 70 Joules were fully penetrated. The incident energy of 50 Joules was enough to cause damage to these groups of samples.

**Figure 10.** Images of damage progression on front and rear faces of flax/VE composite samples impacted in the range of (**a**,**b**) 50 J, (**c**,**d**) 60 J, and (**e**,**f**) 70 J.

Figure 11 shows the damage of the hybrid sample with one basalt layer on the top. In the pictures (a) and (b), a 50 J incident energy was not high enough to perforate the sample thanks to the basalt layer on the top. However, in the other images (Figure 11e–f) for 60 and 70 Joules of incident energy, the maximum energy absorbed is exceeded and all the layers of fibres are broken. As a result, the samples are fully penetrated.

**Figure 11.** Images of damage progression on front and rear faces of flax/VE/basalt hybrid composite panels impacted in the range of (**a**,**b**) 50 J, (**c**,**d**) 60 J, and (**e**,**f**) 70 J.

3.4.2. Visual Observations of Damage Behaviour in Wet Conditions

Figure 12 shows impacted front and rear specimens for water-immersed flax/VE samples impacted at 50 and 70 Joules of incident energies. It can be observed that the wet samples do not have as clear holes as those of dry specimens (Figure 10), indicating the increased ductility of water-immersed samples.

**Figure 12.** Images of damage progression on front and rear faces of flax/VE composite panels impacted in the range of (**a**,**b**) 50 J and (**c**,**d**) 70 J after water absorption.

### 3.4.3. Damage Behaviour from SEM Observations

SEM images illustrated in Figures 13 and 14 reveal matrix cracking, fibre fracture, and delamination on the fractured surfaces of impacted flax/VE and flax/basalt/VE hybrid samples. These damages are in close agreement with those observed in previous studies on flax/carbon epoxy and flax/glass vinyl ester matrix hybrid composites [25,26]. The SEM images further reveal that the extent of damage increased with the increase of the incident energy level. These images further suggest that at a higher energy level, the composites undergo severe damage, with evidence of fibre breakage and pull out, especially in the case of flax/VE composites (Figure 13). For flax/basalt/VE hybrid composites, delamination and fibre bending can be observed (Figure 14). It is also worth mentioning that with basalt fibre on the top layer of hybrid composites, the energy dissipation is increased, which allows an enhancement in impact and fracture toughness behaviours [35,36]. Moreover, it can be observed that for higher energy-impacted samples, more severe fibre damage and breakage can be observed. Different failure modes for flax/VE composites are further explained by the Micro-CT scan illustrated in Figure 15, which compliments the observation made via SEM images. From the annotation provided for the SEM images of flax/VE composites, as shown in Figure 13a–d, the following can be observed:

(1) Fibre breakage; (2) delamination; (3) fibre debonding; and (4 and 5) fibre pull out.

**Figure 13.** *Cont*.

(e) (f)

**Figure 13.** SEM images of flax/VE composites: (**a**) 150 (**b**) 1000× magnification with 50 J, (**c**) 150× magnification with 60 J, magnification with 50 J, (**b**) 1000× magnification with 50 J, (**c**) 150× magnification with 60 J, (**d**) 1000× magnification with 60 J, (**e**) 150× magnification with 70 J, and (**f**) 1000× magnification with 70 J.

From the annotation provided for the SEM images of flax/basalt/VE hybrid composites, as shown in Figure 14a–d, the following can also be observed:

(1) Matrix cracking and fibre bending and (2, 3, and 4) basalt fibre breakage and fracture.

(a) (b)

**Figure 14.** *Cont*.

(c) (d)

(e) (f)

**Figure 14.** SEM images of flax/VE/basalt hybrid composites: (**a**) 150× magnification with 50 J, (**b**) 1000× magnification with 50 J, (**c**) 150× magnification with 60 J, (**d**) 1000× magnification with 60 J, (**e**) 150× magnification with 70 J, and (**f**) 1000× magnification with 70 J.

Figure 16 shows a 3D view of the impacted flax/basalt hybrid sample. Unlike the image illustrated in Figure 15 for flax/VE without hybridisation, the hybrid sample shows less damage and different failure modes, such as delamination.

**Figure 16.** Micro CT scan pictures of a flax/basalt hybrid composite after an impact test, with a 50 J impact energy: (**a**) half-view of the specimen with impact failure, in the form of delamination, and (**b**) top view of the damaged area's shape.

Figure 17 illustrates a 3D half-view of a hybrid composite. It can be clearly observed that the basalt hybridised sample exhibited larger delamination within the top and adjacent layer. In comparison to the flax/VE composite, the flax/VE/basalt hybrid composite does not display penetration. Indeed, the basalt layer has absorbed a higher impact energy.

**Figure 17.** Micro CT scan picture of a flax/basalt hybrid composite specimen (half-view) after an impact test, with a 50 J impact energy, with different failures displayed by the specimen.

### **4. Conclusions**

The effects of basalt fibre hybridisation on the low-velocity falling weight impact behaviour of flax/VE bio-based composites have been investigated following water immersion at room temperature at three incident impact energies: 50, 60, and 70 Joules. The results show that the incident energy has a significant influence on the load bearing capability and total energy absorption characteristics. For a short period of water immersion, it was found that water immersion did not result in a reduction in the impact load and energy absorption. Additionally, the basalt fibre hybridisation at the front side of the laminates significantly enhanced the impact load and total energy of flax/VE composites, showing the potential of flax/VE bio-based composites for semi-structural or structural applications. The damage mechanisms following the X-ray micro CT examination and SEM characterisations performed reveal the greater resistance to penetration and perforation by basalt hybrid composites, which is an indication of their higher impact performance, offering balanced properties of environmental benefits and enhanced impact behaviour. The damage modes for flax/VE composites were matrix cracking, fibre breakage, and fibre pull out. Comparatively, for the flax/basalt hybrid composites, the predominant failure mode was matrix cracking and delamination.

**Author Contributions:** H.N.D. contributed in design, conceptualization, overall leadership, project supervision and writing. E.L.M. and M.F. conducted the experimental work, analysis of results and writing of the preliminary report. C.J. contributed in conducting experimental works and data analysis. Z.Z. contributed in experimental design and analysis. All authors have read and agreed to the published version of the manuscript.

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

**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/).
