**Manufacturing and Properties of Binary Blend from Bacterial Polyester Poly(3-hydroxybutyrate-***co***-3 hydroxyhexanoate) and Poly(caprolactone) with Improved Toughness**

### **Juan Ivorra-Martinez \*, Isabel Verdu, Octavio Fenollar, Lourdes Sanchez-Nacher, Rafael Balart and Luis Quiles-Carrillo**

Technological Institute of Materials (ITM), Universitat Politècnica de València (UPV), Plaza Ferrándiz y Carbonell 1, 03801 Alcoy, Spain; isvergar@epsa.upv.es (I.V.); ocfegi@epsa.upv.es (O.F.); lsanchez@mcm.upv.es (L.S.-N.); rbalart@mcm.upv.es (R.B.); luiquic1@epsa.upv.es (L.Q.-C.) **\*** Correspondence: juaivmar@doctor.upv.es; Tel.: +34-966-528-421

Received: 29 April 2020; Accepted: 12 May 2020; Published: 14 May 2020

**Abstract:** Polyhydroxyalkanoates (PHAs) represent a promising group of bacterial polyesters for new applications. Poly(3-hydroxybutyrate-*co*-3-hydroxyhexanoate) (PHBH) is a very promising bacterial polyester with potential uses in the packaging industry; nevertheless, as with many (almost all) bacterial polyesters, PHBH undergoes secondary crystallization (aging) which leads to an embrittlement. To overcome or minimize this, in the present work a flexible petroleum-derived polyester, namely poly(ε-caprolactone), was used to obtain PHBH/PCL blends with different compositions (from 0 to 40 PCL wt %) using extrusion followed by injection moulding. The thermal analysis of the binary blends was studied by means of differential scanning calorimetry (DSC) and thermogravimetry (TGA). Both TGA and DSC revealed immiscibility between PHBH and PCL. Mechanical dynamic thermal analysis (DMTA) allowed a precise determination of the glass transition temperatures (*T*g) as a function of the blend composition. By means of field emission scanning electron microscopy (FESEM), an internal structure formed by two phases was observed, with a PHBH-rich matrix phase and a finely dispersed PCL-rich phase. These results confirmed the immiscibility between these two biopolymers. However, the mechanical properties obtained through tensile and Charpy tests, indicated that the addition of PCL to PHBH considerably improved toughness. PHBH/PCL blends containing 40 PCL wt % offered an impact resistance double that of neat PHBH. PCL addition also contributed to a decrease in brittleness and an improvement in toughness and some other ductile properties. As expected, an increase in ductile properties resulted in a decrease in some mechanical resistant properties, e.g., the modulus and the strength (in tensile and flexural conditions) decreased with increasing wt % PCL in PHBH/PCL blends.

**Keywords:** bacterial polyesters; poly(3-hydroxybutyrate-*co*-3hydroxyhexanoate)—PHBH; poly(ε-caprolactone)—PCL; binary blends; improved toughness; mechanical and thermal characterization

### **1. Introduction**

Nowadays, awareness of environmental protection, sustainable development, and the use of renewable energies has become a priority for our society. The high volume of wastes generated that are harmful for the environment, oceans, ecosystems, and so on, has become a major problem to be solved. Furthermore, the waste generated in a consumer society, such as the present one, comes mainly from the packaging sector. This need has favoured the development of new environmentally friendly materials [1]. For this reason, the use of the so-called biopolymers is increasing in the packaging

sector. Most of these materials are obtained from renewable resources and they are, in many cases, biodegradable (or compostable in controlled compost soil). Therefore, they positively contribute to minimizing plastic wastes, thus reducing the carbon footprint and also contributing to circular economies by upgrading industrial wastes [2] and/or by-products [3].

In this area, researchers have successfully developed new polymeric materials from renewable and/or biodegradable sources. These important research works have allowed the optimization of interesting biopolymers to be scaled in the industry such as poly(lactic acid)—PLA [4], poly(hydroxyalkanoates)—PHA [5], thermoplastic starch—TPS [6], poly(ε-caprolactone)—PCL [7], and poly(butylene succinate)—PBS [8], among others. Biopolymers can perfectly replace some petroleum-derived polymers [9], since they offer similar performance to most commodities and some engineering plastic. Biopolyesters are an interesting group which includes petroleum-based polymers such as poly(glycolic acid)—PGA, poly(butylene succinate)—PBS, poly(butylene adipate-*co*-terephthalate)—PBAT, and PCL, among others. However, biopolyesters also include bacterial polyesters (PHAs) and some starch-derived polymers such as PLA. The main advantage of polyesters (from both natural or fossil resources) is that they can undergo biodegradation (disintegration in controlled conditions with special compost soil), through the action of microorganisms. This makes composting an important and simple sustainable option for the management of these wastes [10].

Nowadays, biopolymers produced by bacterial fermentation, such as polyhydroxyalkanoates (PHAs), are becoming very promising as there are more than 300 potential PHAs and copolymers. Despite this wide variety, the most commonly used and commercially available PHAs are poly(3-hydroxybutyrate)—P3HB—and poly(3-hydroxybutyrate-*co*-3-hydroxyvalerate)—PHBV [11,12]. PHAs are biologically synthesized polyesters by controlled fermentation with bacteria, such as *Gram-negative* bacteria (*Azobacter*, *Bacillus* and *Pseudomonas*) and *Gram-positive* bacteria (*Rhodococcus*, *Nocardia* and *Streptomyces*). These bacteria, under food stress, can produce energy reserves as intracellular food in the form of granules. These granules are stored in the form of PHAs [13].

Arrieta et al. [10] reported that these bacteria, under feeding conditions of limited macro-elements (such as phosphorus, nitrogen, trace elements or oxygen) and in the presence of an abundant source of carbon (e.g., glucose or sucrose) and/or lipids (e.g., vegetable oils or glycerin), are capable of accumulating up to 60–80 wt % in the form of PHAs. In this way, they can subsist under conditions of food restriction [14,15]. Similar to plants that store energy in the form starch polymer, some bacteria are able to accumulate energy reserves in the form of PHAs [16].

It should be noted that these bacterial polyesters are high-molecular-weight, semi-crystalline, biocompatible thermoplastic polymers. They have very good biodegradability even under environmental conditions. They tend to exhibit rigid behaviour, due to high crystallinity, low thermal stability, and small temperature windows for conventional processing [14,17].

These limitations have been improved by the bacterial synthesis of different copolymers. In this way, a wide range of physical and thermal properties can be tailored, depending on the chemical structure of the used comonomers. More than 150 types of monomers have been successfully synthesized by selecting different raw or modified bacteria and/or the fermentation conditions [13]. The work of Alata et al. [18] reported the effect of medium-length side groups of 3-hydroxyhexanoate (3-HH) units from 5 to 18 mol % on properties of poly(3-hydroxybutyrate-*co*-3-hydroxyhexanoate)—P(3HB-*co*-3HH) or simply PHBH. They observed a remarkable decrease in crystallinity (χc) from 41.6% to 25.1% for copolymers containing 5 mol % and 18 mol % 3-HH, respectively. In addition, due to the reduced crystallinity, secondary crystallization is very low for high 3-HH (above 10 mol %) content in P(3-HB-*co*-3-HH). Other studies have also reported similar results, together with an interesting decrease in the melting temperature of P(3HB-*co*-3HH) [15].

PHBH consists of a random copolymer of 3-HB and 3-HH (see Scheme 1). 3-HH medium-length chains act as short branches of the main 3-HB chains; therefore, stereoregularity is lost, and subsequently, crystallinity is remarkably reduced. Besides this, the presence of aleatory 3-HH chains, broadens the melt peak, but the storage modulus and the overall strength is reduced [17,19]. Nevertheless, PHBH copolymers with low 3-HH content undergo physical aging with time (increase in modulus and strength and reduction of ductile properties such as elongation at break, toughness, and impact strength) [20], which is ascribed to secondary crystallization above the glass transition temperature, *T*<sup>g</sup> [21]. It is worthy to note that typical values of *T*<sup>g</sup> for P3HB are −5 to 5 ◦C, and this interval is remarkably reduced to values as low as −38 ◦C for medium-to-long alkanoate chains, e.g., the *T*<sup>g</sup> of poly(3-hydroxyhexanoate) is close to −28 ◦C [22].

**Scheme 1.** Chemical structure of 3-hydroxyalkanoic acids used to synthesize poly(3-hydroxybutyrate*co*-3-hydroxyhexanoate)—PHBH.

Another key issue in the massive use of PHAs at industrial scale is their "relatively" low cost due to the use of renewable resources such as coconut oil, sugarcane, beet, molasses, other vegetable oils, and, most importantly, carbon-rich industrial wastes such as those obtained from agro-food industry or even sludge coming from sewage treatment plants, by selecting the appropriate bacteria or using bacterial engineering to tailor the desired behaviour of a particular bacteria strain [23,24].

These properties make PHBH an environmentally efficient biopolymer suitable for applications in different packaging applications such as disposable plastic bags, food packaging, catering, agricultural mulch films, and so on [25]. However, as mentioned above, most PHAs undergo secondary crystallization or aging that makes them fragile, thus limiting their possible applications [26]. Xu et al. [27] suggested that one disadvantage of PHBH is that the secondary crystallization process is very slow (for low mol % 3-HH), due to the irregularity of its polymer chain. Large spherulites and secondary crystallization give them poor mechanical properties.

Plasticization of PHAs has been studied with an improvement of ductile properties [28]. Since plasticizers are based on low-molecular-weight compounds, they usually show potential migration problems [29]. An interesting approach to overcome this drawback is blending PHBH with another ductile polymer. However, it is important to bear in mind that the selected polymer for the blend must not compromise biodegradation or disintegration in controlled compost soil. Some researchers have already used poly(ε-caprolactone)—PCL—in blends with PHAs. PCL is a semi-crystalline biodegradable polyester with a very low *T*<sup>g</sup> of about −60 ◦C, which gives an overall ductile behaviour with high elongation at break [26,30]. The addition of PCL decreases the fragility of the PHAs, reducing the elastic modulus and improving the blend processability. However, its low melting temperature (around 50–60 ◦C) means the obtained blends should not be used at temperatures above 50–60 ◦C since dimensional stability could be compromised. Garcia-Garcia et al. [31] reported a noteworthy improvement in the impact behaviour of P3HB by blending it with PCL. In addition, P3HB/PCL blends improved the flexibility and ductility.

The aim of this work is to overcome the intrinsic fragility of a bacterial copolyester, namely poly(3-hydroxybutyrate-*co*-3-hydroxyhexanoate)—PHBH—, by blending with a flexible polyester, namely poly(ε-caprolactone)—PCL. The effect of the incorporation of different amounts of PCL is evaluated by means of mechanical, thermal, thermo-mechanical, and morphological characterization. The evaluation of the results allows the optimum PHBH/PCL blends to be established for applications in the packaging sector that do not compromise the environment at the end-of-life cycle. In this way, it contributes to the reduction of the current serious problem of eliminating the large volume of plastic waste generated by the packaging sector. In addition, these developed formulations could be used in medical applications, as improved toughness is expected with PCL addition and both are resorbable biopolyesters.

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

### *2.1. Materials*

Poly(3-hydroxybutyrate-*co*-3-hydroxyhexanoate)—PHBH—commercial grade ErcrosBio® PH 110 was supplied in pellet form by Ercros S.A. (Barcelona, Spain). This has a density of 1.2 g cm−<sup>3</sup> and a melt flow index of 1.0 g/10 min, measured at 160 ◦C. As indicated by the supplier, it is suitable for injection moulded parts, and it can be melt-blended with other polyesters to obtain tailored properties. Regarding poly(ε-caprolactone)—PCL, commercial grade CapaTM 6800, in pellet form, with a mean molecular weight of 80,000 Da, was supplied by Perstorp (Cheshire, UK). This PCL grade has an MFI (Melt Flow Index) of 3 g/10 min at 160 ◦C.

### *2.2. Manufacturing of PHBH*/*PCL Binary Blends*

PHBH pellets were dried for 8 h at 80 ◦C, while PCL pellets were dried at 45 ◦C for 24 h, in an air-circulating oven CARBOLITE Eurotherm 2416 CG (Hope Valley, UK). As has been reported in other works, the typical weight content (wt %) of flexible polymer blended with brittle polymers to improve toughness is comprised in the 20–40 wt % range. In this work, we selected a maximum PCL content of 40 wt %, since at this composition, PCL is still the minor component in the blend [32–34]. Garcia et al. [31] studied the whole PHB/PCL system and revealed, as expected, that with above 50 wt % PCL, it is PCL which defines the properties of the blend. Ferry et al. [35] also confirmed a maximum loading of 30 wt % PCL to improve the high brittleness of neat PLA. Then, different wt % of PHBH and PCL (see Table 1) were mechanically mixed in a zipper bag to provide initial homogenization. After that, all compositions were extruded using a twin-screw corotating extruder manufactured by DUPRA S.L. (Alicante, Spain) with a temperature profile (four barrels, from the hopper to the extrusion die) of 110, 120, 130, and 140 ◦C respectively and a screw speed of 20 rpm. The extruded material was cooled in air and subsequently pelletized for further processing. After pelletizing, the different blends were subjected to injection moulding in a Meteor 270/75 from Mateu & Solé (Barcelona, Spain). The temperature profile in the injection moulding process was 150 ◦C (hopper), 140, 130, and 120 ◦C (nozzle) in a heated mould at 60 ◦C as recommended by the supplier, since this PHBH has a very low melt strength. The filling time was set to 3 s while the cooling time was 60 s. Standard samples (rectangular and dog-bone shape) were obtained for further characterization. After processing, the specimens were stored at room temperature in a vacuum desiccator for 15 days before characterization, due to the secondary crystallization process or aging that PHBH undergoes with time at 25 ◦C [21,26,27].


**Table 1.** Code and composition (wt %) of binary blends of poly(3-hydroxybutyrate-*co*-3 hydroxyhexanoate)/poly(ε-caprolactone) (PHBH/PCL) blends.

### *2.3. Characterizations Techniques*

2.3.1. Thermal and Thermomechanical Characterization

The thermal transitions of PHBH/PCL binary blends were analyzed using differential scanning calorimetry (DSC) in a Q2000 DSC from TA Instruments (New Castle, DE, USA). The temperature program was scheduled in three different stages: 1st heating, 1st cooling, and 2nd heating. The first heating was scheduled from −50 to 200 ◦C. The second stage consisted of a cooling program from 200 ◦C down to −50 ◦C (this stage is interesting to remove the thermal history and allow crystallization); finally, a 2nd heating cycle identical to the first one (−50 to 200 ◦C) was launched. The heating/cooling rates were all set to 10 ◦C·min−1. All the DSC runs were performed in an inert nitrogen atmosphere with a flow rate of 66 mL·min−1. In addition to parameters such as the melt peak temperature (*T*m), the cold crystallization peak temperatures (*T*cc), enthalpies related to the melting (Δ*H*m), and cold crystallization (Δ*H*cc) processes, the degree of crystallinity, χ<sup>c</sup> (%), was calculated for each polymer in the blend as

$$\chi\_{\rm C}(\%) = \left[ \frac{\Delta H\_{\rm m} - \Delta H\_{\rm cc}}{\Delta H\_{\rm m}^{0} \cdot (1 - w)} \right] \cdot 100 \tag{1}$$

The normalized enthalpy values (Δ*H*<sup>0</sup> m) for a theoretical 100% crystalline PHBH and PCL were taken as 146 and 156 J·g<sup>−</sup>1, respectively, as reported in the literature [19,36,37]. Finally, the term (1 <sup>−</sup> *<sup>w</sup>*) stands for the actual weight of the polymer whose crystallinity is being evaluated.

To study the thermal degradation, thermogravimetry (TGA) was carried out in a Mettler-Toledo Inc. TGA 851-E thermobalance (Schwerzenbach, Switzerland). The thermal program used in this case was a unique dynamic ramp from 30 ◦C up to 700 ◦C at 20 ◦C·min−<sup>1</sup> in an N2 inert atmosphere with a flow rate of 66 mL·min<sup>−</sup>1. In this analysis, the onset degradation temperature was taken as the temperature related to a mass loss of 2 wt % and was denoted as *T*2%.

Dynamic-mechanical thermal analysis, or DMTA, was carried out in a Mettler-Toledo dynamic analyzer DMA1 (Columbus, OH, USA) in single cantilever mode on rectangular samples sized <sup>40</sup> <sup>×</sup> <sup>10</sup> <sup>×</sup> <sup>4</sup> mm3. Samples were subjected to slightly different temperature ramps since PCL melts at 58–60 ◦C. The maximum dynamic deflection was 10 μm and the frequency for the sinusoidal stress wave was set to 1 Hz. Thus, for neat PCL the heating range was from −70 to 50 ◦C to avoid melting. In the case of neat PHBH, the heating ramp was set from −70 to 100 ◦C, and finally, PHBH/PCL blends were subjected to a heating program from <sup>−</sup>70 to 70 ◦C. The heating rate was the same, 2 ◦C·min−1, for all the different scheduled temperature programs

To evaluate the dimensional stability, neat PHBH and PCL, as well as PHBH/PCL blends, were tested in a thermomechanical analyzer (TMA) Q400 from TA Instruments (New Castle, DE, USA) on rectangular samples sized 10 <sup>×</sup> 10 <sup>×</sup> 4 mm3. The temperature sweep was from <sup>−</sup>70 to 70 ◦C, with a constant heating rate of 2 ◦C min−<sup>1</sup> and a constant load of 20 mN. The coefficient of linear thermal expansion (CLTE) of all specimens was determined as the slope for the linear correlation between the expansion and temperature, both below and above *T*g.

### 2.3.2. Mechanical Properties

The mechanical characterization of the PHBH/PCL blends was studied by means of tensile, flexural, impact, and hardness tests on five standardized specimens for each test. The tensile and flexure tests were carried out according to ISO 527 and ISO 178 respectively, in an ELIB 30 universal machine from S.A.E. Ibertest (Madrid, Spain). The load cell was 5 kN for both tests. The crosshead rate was 5 mm·min−<sup>1</sup> for flexural tests and 20 mm·min−<sup>1</sup> for the tensile tests.

The impact resistance (absorbed-energy during impact conditions, per unit area) was determined according to ISO 179, using a Charpy pendulum from Metrotec S.A. (San Sebastian, Spain) with an energy of 1-J. A standardized "V-type" notch was produced on standard rectangular samples.

The hardness properties were measured according to ISO 868. The equipment used was a Shore D hardness tester model 673-D from J. Bot, S.A. (Barcelona, Spain).

### 2.3.3. Morphology Characterization

The surface analysis of the fractured specimens from impact tests was performed with a field emission scanning electron microscope (FESEM) model ZEISS ULTRA55 (Oxford Instruments, Abingdon, UK). The working accelerating voltage was 2 kV. Prior to this analysis, the samples were metallized with a gold-palladium alloy in an EMITECH mod. SC7620 sputter coater from Quorum Technologies Ltd. (East Sussex, UK). In a second analysis, the samples were subjected to a selective PCL extraction in acetone at room temperature for 24 h. In this way, PCL can be extracted, and therefore, it is possible to observe more accurately the phase distribution in the developed binary blends [38].

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

### *3.1. Thermal Properties of PHBH*/*PCL Blends*

Thermal analysis, using DSC of PHBH/PCL binary blends with different amounts of PCL (wt %), and neat PHBH and PCL, was done from the first heating cycle (Figure 1a) to obtain the thermal parameters of the starting material, just after 15 days from its processing, thus allowing secondary crystallization. In addition, the second heating cycle after a slow cooling (Figure 1b) allowed the thermal history of the material to be removed and the main thermal characterization parameters to be obtained.

**Figure 1.** *Cont.*

**Figure 1.** Comparative plot of the differential scanning calorimetry (DSC) thermograms of PHBH, PCL and PHBH/PCL blends with different PCL wt %: (**a**) first heating cycle after processing and aging for 15 days, (**b**) 2nd heating after cooling (**a**) at a controlled rate.

Table 2 summarizes the main thermal parameters corresponding to the DSC thermograms of the first heating cycle plus an additional aging time of 15 days, obtained using DSC runs from Figure 1a. The thermogram of neat PCL showed a single endothermic peak at around 62 ◦C (*T*m\_PCL), which was attributed to the melting of packed crystallites of PCL embedded in an amorphous PCL fraction. Since DSC tests were run from −50 ◦C, the *T*g\_PCL could not be clearly observed. This was because the *T*g\_PCL is very low, with values below −50 ◦C, so that, with this thermal program, it could not be accurately determined. On the other hand, the DSC thermogram of neat PHBH did not allow its *T*g\_PHBH to be identified either. This is because the crystalline fraction of PHBH remarkably increased after aging for 15 days; therefore, the remaining amorphous fraction was noticeably reduced, and then, the step in the baseline (around 0 ◦C), attributed to the *T*g\_PHBH, could not be clearly seen. As one can see in Figure 1a, only the melt process of the crystalline fraction in PHBH could be observed with a peak located at 138 ◦C (*T*m\_PHBH). Regarding binary PHBH/PCL blends, the two above-mentioned endothermal peaks could be seen in all the developed compositions: a first peak at around 60 ◦C, corresponding to melting of PCL, and a second peak at around 135 ◦C, related to the melting of PHBH. One can see that as the PCL wt % increased in the PHBH/PCL blends, the melt peak of PCL became larger, while inversely, the melt peak of PHBH was slightly diluted. These two independent peaks suggest some lack of miscibility between the two biopolyesters, as each polymer melted at its corresponding temperature [26,31].


**Table 2.** Thermal properties of PHBH, PCL, and PHBH/PCL binary blends with different PCL wt % obtained during the 1st heating cycle after processing plus 15 aging days to complete secondary crystallization.

\* Standardized enthalpies based on the actual weight of the polymer present in the samples.

The results in Table 2 indicate that neat PHBH is characterized by a small degree of crystallinity, χc\_PHBH around 13%, even after the 15-day aging process at room temperature. According to the results of Xu et al. [27], this low χ<sup>c</sup> was due to the irregularities in the structure of the polymer chain of the copolymer, which hinders the formation of crystallites, with increasing mol % 3-HH. The addition of PCL wt % slightly increases the crystallinity values, from 15% for the sample with 10 PCL wt % to 18% for 40 PCL wt %. Garcia-Garcia et al. [31] found similar results in the PHB/PCL system with an increase in the degree of crystallization χ<sup>c</sup> from 55.1% to 58.2% with 25 wt % PCL. They attributed this to the fact that PCL can affect the crystallization kinetics of neat PHB. As expected, PHBH showed lower χ<sup>c</sup> due to the hindering effects of 3-HH, as mentioned above. Contrary to this, Antunes et al. [39] reported a decrease in the degree of crystallinity of PHB by increasing PCL wt % up to 20 wt %, while an increase was observed for 30 wt % PCL. The thermograms in Figure 1a, are interesting as they clearly indicate the aging process after 15 days has been able to complete the secondary crystallization. Moreover, the results gathered in Table 2 corroborated the absence of secondary crystallization after 15 aging days.

With respect to the DSC thermograms obtained in a second heating cycle, shown in Figure 1b (after the cooling of the first heating cycle), it is worthy to note that these showed a clear change. In these DSC thermograms a step in the base line at about 0 ◦C could be clearly observed, which was attributable to the *T*g\_PHBH [38]. In this case, as the thermal history is completely different, the results regarding crystallinity were somewhat variable. Przybysz et al. [40] reported a remarkable decrease in PHB/PCL blends due to the addition of different peroxide-based compatibilizers, while Oyama et al. [26] showed completely different results for a PHBH/PCL system with peroxide-based compatibilizers, which showed an increase in χ<sup>c</sup> of PHBH. Nevertheless, Antunes et al. [39] reported a decrease in χ<sup>c</sup> of PHBH without any compatibilizer, which was attributed to changes in crystallization kinetics. In this work, we obtained somewhat varying effects of PCL wt % on the degree of crystallinity of PHBH, as its complex structure (hindering crystallization due 3-HH units) and the additional effects of PCL on crystallization kinetics could overlap with some simultaneous processes and lead to these changes. Obviously, PCL did not show its corresponding step change in the baseline as its *T*g\_PCL is lower than −50 ◦C. The compatibility of a polymer blend can be assessed by changes in *T*<sup>g</sup> as Garcia et al. reported [41]. In this case, the addition of different PCL wt % to PHBH/PCL blends did not affect the *T*g\_PHBH values obtained, as shown in Table 3.


**Table 3.** Thermal properties of PHBH/PCL binary blends obtained during the 2nd heating cycle after a heating-cooling process to remove thermal history.

\* Standardised enthalpies based on the actual weight of the polymer present in the samples. \*\* Melting enthalpies of PCL on PHBH/PCL blends could not be obtained by the overlapping with cold crystallization process in PHBH.

The thermogram corresponding to 100% PHBH showed an exothermic peak at around 49 ◦C which stood for the peak temperature, *T*cc\_PHBH, of the cold crystallization, with a crystallization enthalpy (Δ*H*cc\_PHBH) of 21.14 J·g−<sup>1</sup> [19]. At temperatures close to 113 ◦C, PHBH showed a first and small endothermic peak that corresponded to the melting of PHBH crystalline fraction (*T*m1\_PHBH) and a second melt peak located at 120 ◦C (*T*m2\_PHBH). These two endothermic peaks could be due to two effects. The first is based on the polymorphism presented by some copolymers such as PHBH as observed in other aliphatic polyesters. Due to the heterogeneous composition of the copolymer itself, different crystalline morphologies can be formed, with different thermal stability. The second effect is that crystallization produces primary crystals with low degree of perfection; these may melt and recrystallize to produce crystals of greater perfection or greater thickness, and this could be the explanation for the presence of two overlapped melting peaks [21,23,25,42].

On the other hand, the 100 PCL wt % sample only showed a very marked endothermic peak, at 55 ◦C, which corresponded to its melting temperature, *T*m\_PCL, as mentioned above. Due to the similarity between the cold crystallization process of PHBH and the melting of the crystalline fraction of PCL, the binary PHBH/PCL blends showed two overlapped endothermal (PCL melting)-exothermal (PHBH cold crystallization) peaks in the DSC thermograms (Figure 1b). This overlapping did not allow either the melting enthalpy of the PCL (Δ*H*m\_PCL) or the cold crystallization enthalpy of the PHBH (Δ*H*cc\_PHBH) to be quantified accurately. This effect did not allow the correct calculation of χ<sup>c</sup> in the PHBH/PCL system blends in this 2nd heating cycle as they were overlapped. Nevertheless, these 2nd heating DSC runs were interesting as the samples had undergone a thermal heating-cooling cycle to remove the thermal history, and subsequently, all the thermal transitions could be detected in a clearer way. In particular, the secondary crystallization of PHBH, which disappeared after the aging process (Figure 1a) and could not be detected, was clearly seen in the 2nd heating cycle.

On the other hand, the thermograms of the binary PHBH/PCL blends showed the characteristic melting peak corresponding to PCL and the two melting peaks of PHBH. It should be noted that as the PCL wt % in the PHBH/PCL blends increased, it had virtually no influence on the melting peak temperatures of PHBH (*T*m1\_PHBH and *T*m2\_PHBH), which was the major component in the developed PHBH/PCL blends. The immiscibility between PHBH and PCL suggests that they form two separate phases with almost independent thermal parameters [30].

Regarding the enthalpies of the thermal transitions related to melting, the results obtained are shown in Table 3. The addition of small amounts of PCL (10 and 20 wt %) slightly increased the values of Δ*H*m\_PHBH \*, which indicated a higher amount of energy to melt the crystalline fraction due to there was a slight increase in crystallinity. Higher amounts of PCL (30 and 40 wt %) offer the opposite effect leading to a decrease in crystallinity, probably due to changes in the crystallization kinetics due to the intrinsic structural complexity of PHBH (with 3-HH units which hinder crystallization) and PCL, which could affect crystallization, as has been described previously. When comparing these results with those obtained from samples aged for 15 days, the values of the melting enthalpies were slightly higher. Slow cooling favoured the phenomenon of cold crystallization of PHBH, so that the final melting enthalpy of the crystalline fraction was higher in samples with no previous thermal history.

The thermal stability of binary PHBH/PCL blends was analyzed using thermogravimetric analysis (TGA). Figure 2 presents a comparative plot of the TGA curves for neat PHBH and PCL and PHBH/PCL blends with different PCL wt %.

**Figure 2.** Comparative plot of the thermal degradation of PHBH, PCL, and PHBH/PCL with different PCL wt %. (**a**) TGA degradation curves in terms of mass loss and (**b**) first derivative of TGA thermograms.

The TGA curve of neat PHBH shows a single-step thermal degradation process. The onset degradation temperature (*T*2%) for neat PHB was 264.6 ◦C. Once the thermal degradation/decomposition had started, it proceeded very fast with a maximum degradation rate temperature (*T*deg) of 301.3 ◦C. These results for individual PHBH were in accordance with those reported by Hosoda et al. [43] and Mahmood et al. [19]. Almost all the mass was thermally decomposed very quickly with an endset degradation temperature of 321 ◦C. A small residual char of 2.72 wt % was obtained for neat PHBH. It is worth mentioning that the thermal degradation of PHBH occurred in a very narrow temperature range of 57 ◦C. On the other hand, despite its crystalline fraction melting at relatively low temperature (58–60 ◦C), it is important to note that PCL was much more thermally stable since its onset degradation temperature (*T*2%) was 358 ◦C. In fact, PCL is one of the most thermally stable polyesters. Once the onset degradation temperature was reached, the thermal degradation proceeded in a single-step degradation process, with a maximum degradation rate of 430.9 ◦C (*T*deg), and also occurred in a very narrow range up to 474.8 ◦C, with a residual mass of 0.12 wt %.

The immiscibility between PHBH and PCL was reflected in the TGA curves of the blends. In all of them, two separated degradation steps could be observed: a first step at around 300–310 ◦C, which corresponded to the degradation of the PHBH phase, and a second step, at a higher temperature occurring in the 390–440 ◦C range, which corresponded to the degradation of PCL in the blend, as reported by Garcia-Garcia et al. [31] in P3HB-PCL binary blends. The immiscibility was clearly detected by observing the TGA curves. By observing the TGA curve of the 60PHBH-40PCL sample, the mass loss after the first degradation step was almost 40 wt %, which corresponded exactly to the PCL wt % in the blend. The same effect was observed in all other binary blends, which corroborated the lack of miscibility between PHBH and PCL. On the other hand, the blends also presented low residual char formation, between 2 and 1.13 wt %. Comparatively, there was a very slight shift of TGA thermograms to the right with increasing PCL wt % content.

The first derivative of the thermogravimetric curves (DTG), Figure 2b, shows two clearly differentiated thermal degradation processes without almost any overlapping, which corroborates the above-mentioned immiscibility of the PHBH/PCL binary blends. The first peak observed at temperatures between 301–307 ◦C corresponded to the maximum degradation rate temperature (*T*deg) of PHBH in all blends. On the other hand, the second peak, located between 420–430 ◦C, was attributed to the maximum degradation rate temperature of PCL in all the blends. There were very slight changes in the peak maximum values for each process, thus corroborating this immiscibility.

### *3.2. Thermomechanical Properties of PHBH*/*PCL Blends*

In addition to the thermal stability, it is important to evaluate the effect of temperature on dimensional stability. To this end, samples of the different PHBH/PCL blends were subjected to a thermomechanical characterization, which allowed the coefficients of linear thermal expansion (CLTE) were obtained below and above the *T*g\_PHBH (Table 4).

**Table 4.** Coefficient of linear thermal expansion (CLTE) of PHBH/PCL blends with different PCL wt %, below and above *T*g\_PHBH, obtained using thermomechanical analysis (TMA).


As expected, at temperatures below *T*g\_PHBH, the CLTE values were much lower than those obtained above *T*g\_PHBH. The dimensional expansion of the material was lower below the characteristic *T*g, since the material offers a rigid, brittle, glassy behaviour. On the other hand, above *T*g, the polymer material changed its behaviour to a plastic, rubbery-like behaviour, and subsequently, the dimensional expansion was allowed. Neat PHBH showed a CLTE of 68 <sup>μ</sup>m·m−1· ◦C−<sup>1</sup> below its *T*g. As PCL is a much more flexible polymer, its addition to the PHBH/PCL blends provided increased flexibility, and subsequently, the CLTE increased according to PCL wt % contained in the blends. This increase in CLTE was proportional and increased up to 106.9 <sup>μ</sup>m·m−1· ◦C−<sup>1</sup> for the blend with 40 PCL wt %. This increase in the CLTE, which was directly proportional to the PCL wt %, was representative for a somewhat loss of fragility and an improvement in the ductile behaviour of PHBH at low temperatures. The same effect was observed for the CLTE values obtained at temperatures above *T*g\_PHBH in the blends. In this case, neat PHBH offered a remarkable increase in its CLTE to 172.5 <sup>μ</sup>m·m−1· ◦C−1, which was remarkably higher than the CLTE below its *<sup>T</sup>*<sup>g</sup> (68 <sup>μ</sup>m·m−1· ◦C<sup>−</sup>1). The increasing tendency for the CLTE was similar to that mentioned above, and the PHBH/PCL blend with 40 PCL wt % showed the maximum CLTE of about 198.9 <sup>μ</sup>m·m−1· ◦C<sup>−</sup>1.

The thermal dynamic mechanical analysis of the PHBH/PCL blends allowed the variation of the storage modulus, *E'*, and the damping factor (tan δ) to be obtained. The damping factor was directly related to the phase angle (δ), which is representative for the delay between the applied dynamic stress (σd) and the obtained dynamic elongation (εd). The DMTA technique is much more sensitive to the glass transition temperature, *T*g, detection since this technique measures changes in mechanical properties as a function of temperature, which includes the definition of *T*g with a change from a glassy state to a rubber-like behaviour [30,31,34,37,38]. The dependence of *E'* on temperature is shown graphically in Figure 3a.

**Figure 3.** *Cont.*

**Figure 3.** Comparative plot of the dynamic mechanical thermal analysis (DMTA) properties of PHBH, PCL, and PHBH/PCL blends with different PCL wt %, as a function of temperature, (**a**) storage modulus, *E'*, and (**b**) dynamic damping factor (tan δ).

The values of *E'* obtained for neat PHBH at very low temperatures (e.g., −70 ◦C) were high, about 2 GPa, since this temperature zone corresponds to an elastic-glassy behaviour of the material, which means it was far (below) from its *T*g. For example, at −40 ◦C, the *E'* of neat PHBH was 1903.4 MPa. For this same temperature, the *E'* for neat PCL was much lower, with a value of 596.3 MPa. In this case, the material showed a visco-elastic behaviour, due to it was above *T*<sup>g</sup> (*T*g\_PCL about −60 ◦C). Thus, adding different PCL wt % to the PHBH led to blends with a clear decreasing tendency of *E'* which was proportional to the PCL wt %. As the temperature increased, a remarkable decrease in *E'* values was observed, which represented the change from a glassy state (rigid with high *E'* values) to a viscous, rubber-like behaviour (viscoplastic with low *E'* values). Obviously, this was directly related to the glass transition temperature range from −10 to 20 ◦C. By taking the *T*<sup>g</sup> criterium corresponding to the peak maximum of the dynamic damping factor, *T*g\_PHBH was 9 ◦C. Since the *T*<sup>g</sup> of both PHBH and PCL is relatively low, at room temperature both polymers are in the rubbery-plateau zone, with a relatively flexible viscoelastic behaviour, as shown in the respective *E'* plots. This is the typical behaviour of polymers above their *T*g, as Burgos et al. and Avolio et al. reported [44,45]. At 25 ◦C, the value of *E'* was 653.2 MPa, almost three times lower than neat PHBH below its *T*g. The same ranges of variation were maintained for the studied blends, with values between [539 MPa, 398 MPa] at 25 ◦C vs. [1666 MPa, 1289 MPa] at −40 ◦C. At higher temperatures, *E'* tended to have very low values (close to 0 MPa) for blends with high PCL wt %. This effect was because PCL melted at about 60 ◦C, and once this temperature was overpassed, PCL changed from a rubber-like state to a melt state with extremely low elastic properties. Nevertheless, in blends with 10 and 20 wt % PCL (it is important to bear in mind that PHBH content in these blends was still very high and is not highly affected by PCL addition in terms of dynamic-mechanical behaviour at high temperatures) the effect was less pronounced so that at 60 ◦C, *E'* was 178.5 and 142.0 MPa respectively. These values are typical of a rubber-like material and decrease as the wt % PCL increases.

Figure 3b shows the variation of the dynamic damping factor (tan δ) as a function of temperature for neat PHBH and PCL and for their blends with different PCL wt %. With respect to the neat PCL curve, a marked and broad peak was observed with a peak maximum located at −47 ◦C, corresponding to the glass transition temperature *T*<sup>g</sup> of PCL [30,31] With regard to neat PHBH, a narrow peak (compared to that of PCL) can be seen, with its maximum peak value located at 9 ◦C. As mentioned above, all the applied techniques suggested poor (or even lack) of miscibility between these two polymers. Therefore, the changes in the respective *T*<sup>g</sup> values of neat PCL (−47 ◦C) and neat PHBH (9 ◦C) were not changed in a significant way, therefore corroborating the lack of miscibility of PHBH/PCL blends [38]. In addition, close to 60 ◦C, PHBH/PCL blends show a third peak which was related to the partial melting of one of the components, i.e., PCL. This third peak was much more pronounced in blends with high PCL wt % and shows proportionality to the PCL wt %, but this third peak appears always at the same temperature of about 60 ◦C.

### *3.3. Mechanical Properties and Morphology of PHBH*/*PCL Blends*

Table 5 presents a summary of the mechanical properties obtained from tensile, flexural, and hardness (Shore D) tests, corresponding to neat PHBH and PCL and PHBH/PCL binary blends with different PCL wt %. All the blends of this binary system offered a noticeable decrease in the tensile modulus (*E*t) with an increase in the PCL wt % content compared to neat PHBH. PHBH/PCL blends with 40 PCL wt % showed an *E*<sup>t</sup> value of 722 MPa, which was remarkably lower than that of neat PHBH at about 1022 MPa. This represented a % decrease of almost 30%. The addition of PCL resulted in lowering the overall stiffness of the PHBH/PCL blends. Similar results were reported by Hinüber et al. in PHBH/PCL blends [46]. It should be noted that neat PCL is a biopolymer characterized by a very low tensile strength (σt) of 12.2 MPa (typical of a rubber-like polymer), high elongation at break (εb) (no break means more than 600% as this is the maximum elongation in the used machine), and a low tensile modulus (*E*t) of 386 MPa. All these properties positively contributed to improving the ductility of neat PHBH by blending with PCL. The same tendency could be observed for the tensile strength of the PHBH/PCL blends. The addition of different PCL wt % progressively decreased the values of σt, from 16 MPa for neat PHBH to [13.4, 13.9] MPa with 20 PCL wt % and 40 PCL wt %, respectively. This decrease in the resistance parameters was due to the two-phase structure of the blends, resulting from the lack of (or very poor) miscibility, as mentioned above. The immiscibility between PHBH and PCL forms a dispersed PCL phase that interrupts the continuity of the PHBH-rich matrix phase, making the stress transfer difficult and, subsequently, decreasing the *E*<sup>t</sup> and σ<sup>t</sup> of the PHBH/PCL blends with increasing PCL wt %.

**Table 5.** Summary of the mechanical properties from tensile, flexural, and hardness tests of neat PHBH and PCL and PHBH/PCL blends with different PCL wt %. σ<sup>t</sup> and σ<sup>f</sup> represent the tensile and flexural strength, respectively. *E*<sup>t</sup> and *E*<sup>f</sup> are the respective values for the tensile and flexural modulus.


With respect to the elongation at break, εb(%), the effect was opposite and very positive. The increase in εb(%) in the PHBH/PCL blends also increased ductility with PCL wt %. PHBH is a rather fragile polymer (after aging or secondary crystallization) with only 13.9% elongation at break. Some researchers attributed this fragility to the secondary crystallization or aging on PHBH, which reduced the amorphous fraction [18,19]. With the addition of only 20 PCL wt %, the εb(%) increased up to 67.8% (which represented a percentage increase of almost 387%). Even more, the εb(%) for the blend with the highest PCL wt % considered in this study, i.e., 40 wt %, showed an εb(%) of 461% (this was 33 times higher than neat PHBH).

Table 5 also offers the flexural parameters of neat PHBH and PCL and the PHBH/PCL blends. As in tensile conditions, PHBH/PCL blends became less rigid with PCL addition. This can be confirmed

by a clear decrease in the flexural modulus (*E*f). Neat PHBH had an *E*<sup>f</sup> value of 1029 MPa, and this decreased progressively to 801.7 MPa for the PHBH/PCL blend containing 40 PCL wt %. In this case, the decrease in the flexural strength (σf) was not so pronounced as that observed in tensile conditions.

In addition, the Shore D hardness, as it is a mechanical resistant property, it followed the same tendency as that observed for both modulus and strength. PHBH showed a Shore D of 61, and the addition of the flexible PCL to PHBH decreased the Shore D values down to 55 for the blend with 40 PCL wt %. Despite this, all Shore D values were close to 58 with very slight variations.

Another important property of polymers is toughness. Figure 4 shows the variation of the absorbed energy per unit area (impact resistance) obtained using a Charpy's test. PHBH is rather brittle, and consequently, it offers low toughness. Graphically, an interesting increase in the impact resistance of PHBH/PCL blends could be observed as the PCL wt % increased. It is important to bear in mind that the energy absorption capacity under impact conditions is directly related to the plastic deformation capacity of the material before the breakage occurs, and the supported stress, too [47]. The presence of a biphasic structure (as reported in morphology analysis) could contribute to improving the toughness, as Ferri et al. reported [48]. Thus, the results obtained corroborated those previously analyzed for tensile and flexural characterizations. Neat PHBH showed a low impact resistance of 5.56 kJ·m−2; with the addition of only 10 PCL wt %, it was increased to 8.7 kJ·m−<sup>2</sup> (which represented a % increase of 56%). This same trend was proportionally maintained as the PCL wt % increased. It is worth noting that the impact energy for the PHBH/PCL blend with 40 PCL wt % was around 10.5 kJ·m−2, almost twice as much as neat PHBH. These results were consistent with the above-mentioned decrease in the intrinsic brittleness of neat PHBH by blending with PCL [37].

**Figure 4.** Plot evolution of the impact-absorbed energy of neat PHBH and PHBH/PCL blends with increasing PCL wt %.

Figure 5 shows the FESEM images of the impact fracture surfaces of the PHBH/PCL binary blends with different PCL wt %. All the images show a structure with a continuous and homogeneous matrix phase, which corresponded to the PHBH, and a scattered phase of special morphology, which corresponded to the PCL. This two-phase structure confirmed the lack of miscibility between PHBH and PCL as already concluded with previous thermal analyses. Similar findings were proposed by Quiles-Carrillo et al. [49]. Nevertheless, the typical drop-like structure could not be observed in

this system. In addition, the special morphology of the dispersed phase of the PCL, forming small, thin "sheets or flakes" homogeneously distributed with very regular sizes, must be emphasized. FESEM images also revealed that the higher the PCL wt %, the greater the amount of dispersed phase that could be observed.

**Figure 5.** Field emission scanning electron microscopy (FESEM) images at 1000× of the impact fracture surface morphologies of PHBH/PCL binary blends with different PCL wt %, (**a**) 10, (**b**) 20, (**c**) 30, and (**d**) 40.

To check that this dispersed phase corresponded to PCL present in the binary blends, a selective PCL extraction with acetone was performed for 24 h [38]. Figure 6 shows the FESEM images obtained after this selective extraction. In all of them, the dispersed phase with small flake-like shapes was no longer observed. Instead, small and thin empty voids appeared, which corresponded exactly to the geometric shape of the PCL phase before the selective attack (Figure 5). This observation allowed us to conclude that the dispersed phase indeed corresponded to the PCL present in the binary blend. The lack of miscibility between the biopolymers of the PHBH/PCL system was responsible for the internal biphasic structure formed in the blends. In addition, FESEM images showed how the dispersed PCL-rich phase interrupted the continuity of the PHBH matrix, so that the stress transmission inside the material when subjected to external stresses was not adequate [38]. This reduced the mechanical resistant parameters, which corroborated the results obtained in the mechanical characterization of the PBHB/PCL binary blends.

**Figure 6.** Field emission scanning electron microscopy (FESEM) images at 1000× of the impact fracture surface morphologies of PHBH/PCL binary blends, subjected to PCL selective extraction, with different PCL wt %, (**a**) 10, (**b**) 20, (**c**) 30, and (**d**) 40.

#### **4. Conclusions**

The processing and obtaining of PHBH/PCL binary blends allowed a new environmentally friendly material to be obtained with improved toughness and ductile properties, suitable for industrial use in the packaging sector and for medical applications too.

Considering the intrinsic fragility of PHBH, the addition of different PCL wt %, allowed its toughness, ductility, and, above all, impact resistance to be improved. Blends with a 10 PCL wt % offered a percentage increase in the impact resistance of about 56%. The impact resistance was even improved up to double the initial value of neat PHBH by adding 40 PCL wt %. This effect of increasing the ductility of PHBH by increasing the PCL content in the PHBH/PCL blends had an opposite effect on mechanical resistant properties such as modulus and strength (tensile and flexural). In contrast, the elongation at break was remarkably improved, from 13.9% for neat PHBH up to 461% for the PHBH/PCL blend with 40 PCL wt %.

On the other hand, thermal analysis suggested high immiscibility between PHBH and PCL, since the main thermal parameters of neat PHBH and PCL remained unchanged in blends, which is characteristic of very poor or lack of miscibility. Only a slight increase in the thermal stability of neat PHBH was obtained by adding different PCL wt %, since PCL is much more thermally stable than most biopolyesters. The dynamic mechanical thermal analysis (DMTA) allowed accurate values of *T*g to be obtained, with two clear and unchanged values located at −47 ◦C (*T*<sup>g</sup> of PCL) and at 9 ◦C (*T*<sup>g</sup> of PHBH). These characteristic *T*g values remained unchanged in the PHBH/PCL blends, thus suggesting lack of miscibility. Regarding the morphology of the PHBH/PCL blends, they did not show the typical drop-like PCL phase embedded in a PHBH matrix. PCL appeared in the form of small flakes which could exert a positive effect on ductile properties.

**Author Contributions:** Conceptualization, R.B., L.Q.-C., and J.I.-M.; methodology, L.S.-N., J.I.-M., and O.F.; validation, L.S.-N. and I.V.; formal analysis, L.Q.-C. and R.B.; investigation, I.V. and L.S.-N.; data curation, L.Q.-C., I.V., and O.F.; writing—original draft preparation, L.S.-N.; writing—review and editing, R.B. and J.I.-M.; supervision, R.B. and L.S.-N.; project administration, R.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research work was funded by the Spanish Ministry of Science, Innovation, and Universities (MICIU), project numbers MAT2017-84909-C2-2-R. This work was supported by the POLISABIO program, grant number (2019-A02).

**Acknowledgments:** Juan Ivorra-Martinez is the recipient of an FPI grant from Universitat Politècnica de València (PAID-2019-SP20190011). Luis Quiles-Carrillo wants to thank GVA for his FPI grant (ACIF/2016/182) and MECD for his FPU grant (FPU15/03812). Microscopy services at UPV are acknowledged for their help in collecting and analyzing FESEM images.

**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*

## **Development and Characterization of Sustainable Composites from Bacterial Polyester Poly(3-Hydroxybutyrate-***co***-3-hydroxyhexanoate) and Almond Shell Flour by Reactive Extrusion with Oligomers of Lactic Acid**

### **Juan Ivorra-Martinez \*, Jose Manuel-Mañogil, Teodomiro Boronat, Lourdes Sanchez-Nacher, Rafael Balart and Luis Quiles-Carrillo**

Technological Institute of Materials (ITM), Universitat Politècnica de València (UPV), Plaza Ferrándiz y Carbonell 1, 03801 Alcoy, Spain; jomaoam@epsa.upv.es (J.M.-M.); tboronat@dimm.upv.es (T.B.); lsanchez@mcm.upv.es (L.S.-N.); rbalart@mcm.upv.es (R.B.); luiquic1@epsa.upv.es (L.Q.-C.) **\*** Correspondence: juaivmar@doctor.upv.es; Tel.: +34-966-528-421

Received: 9 April 2020; Accepted: 8 May 2020; Published: 11 May 2020

**Abstract:** Eco-efficient Wood Plastic Composites (WPCs) have been obtained using poly(hydroxybutyrate-co-hexanoate) (PHBH) as the polymer matrix, and almond shell flour (ASF), a by-product from the agro-food industry, as filler/reinforcement. These WPCs were prepared with different amounts of lignocellulosic fillers (wt %), namely 10, 20 and 30. The mechanical characterization of these WPCs showed an important increase in their stiffness with increasing the wt % ASF content. In addition, lower tensile strength and impact strength were obtained. The field emission scanning electron microscopy (FESEM) study revealed the lack of continuity and poor adhesion among the PHBH-ASF interface. Even with the only addition of 10 wt % ASF, these green composites become highly brittle. Nevertheless, for real applications, the WPC with 30 wt % ASF is the most attracting material since it contributes to lowering the overall cost of the WPC and can be manufactured by injection moulding, but its properties are really compromised due to the lack of compatibility between the hydrophobic PHBH matrix and the hydrophilic lignocellulosic filler. To minimize this phenomenon, 10 and 20 *phr* (weight parts of OLA-Oligomeric Lactic Acid per one hundred weight parts of PHBH) were added to PHBH/ASF (30 wt % ASF) composites. Differential scanning calorimetry (DSC) suggested poor plasticization effect of OLA on PHBH-ASF composites. Nevertheless, the most important property OLA can provide to PHBH/ASF composites is somewhat compatibilization since some mechanical ductile properties are improved with OLA addition. The study by thermomechanical analysis (TMA), confirmed the increase of the coefficient of linear thermal expansion (CLTE) with increasing OLA content. The dynamic mechanical characterization (DTMA), revealed higher storage modulus, E', with increasing ASF. Moreover, DTMA results confirmed poor plasticization of OLA on PHBH-ASF (30 wt % ASF) composites, but interesting compatibilization effects.

**Keywords:** PHBH; almond shell flour; mechanical properties; thermal characterization; WPCs

### **1. Introduction**

The current problem related to the negative environmental impact of large volumes of wastes [1] in a consumer society has promoted a significant awareness and sensitiveness about this problem. Some governments are facing this through legislation that protects environment and minimizes the harmful impact on nature. This, in part, has led to the extensive development of new eco-efficient

materials, from the point of view of their renewable origin, low carbon footprint, possibility of composting, biodegradability, and so on [2]. An interesting family of these new eco-efficient materials are the so-called Wood Plastic Composites, WPC. These composites consist on a polymeric matrix in which wood (or whatever lignocellulosic subproduct of the food industry or agroforestry) particles (from 10 to 60 wt %, depending on the manufacturing process) are embedded, leading to an appearance and surface finishing similar to natural wood. As many times, the lignocellulosic fillers are by-products from other sectors, they are cheap and do not increase the cost of the WPC; in addition, they come from natural resources and, subsequently, they represent a sustainable source for use in new and environmentally friendly materials [3–5].

These WPCs are already replacing the use of traditional woods in some applications, which is an important protection of forest resources. WPCs formulations have been optimized in sectors as important as automotive, outdoor furniture, interior design, railings, floors, coatings, decks, fences, pergolas, decking, and so on [3–8].

The fact that they are composed of a polymeric matrix, gives them better behaviour against water or in humid environments. According to Singh et al., WPCs have gained a significant share of the consumer market, becoming the fastest growing segment of the plastics industry [4]. Within the wide range of possibilities that these eco-efficient materials offer as substitutes for wood, those that use thermoplastic polymers as the matrix are of particular interest, precisely because of the ease and versatility of manufacturing processes. Poly(ethylene) (PE), poly(styrene) (PS), poly(vinyl chloride) (PVC) and poly(propylene) (PP) are some of the most widely used polymers in WPCs. Nevertheless, these polymer matrices are petroleum-derived polymers.

Due to the need to protect the environment, the possibility of using biopolymers as matrices in WPCs is currently being studied. The use of a biodegradable thermoplastic polymer (actually, a compostable polymer that disintegrates in controlled compost soil) from natural resources, together with a natural lignocellulosic filler from industrial wastes or by-products, allows the obtaining of totally biodegradable and eco-efficient WPCs [6]. These new green composites represent the new generation of biobased, sustainable, low environmental impact WPCs. Nowadays, there are already a large number of natural biopolymers on a commercial level, among which three main families stand out. The first one, consists on polymers from biomass which include polysaccharides such as starch (and starch-derived polymers such as polylactide), cellulose, chitosan, chitin, and proteins such as casein, keratin, collagen, and so on. The second group includes conventional polymers such as poly(ethylene), poly(urethanes), poly(amides), that are partially or fully obtained from natural resources but they show identical (or very similar) properties to their petroleum-derived counterparts. Finally, a new family of very promising polymers is that of bacterial polyesters which are generally referred to as polyhydroxyalkanoates PHAs. PHAs include more than 300 different polyesters and copolymers such as poly(3-hydroxybutyrate) (P3HB or just PHB), poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV), among others [5,9].

One of the most interesting biopolymers, obtained by bacterial fermentation, is poly(3 hydroxybutyrate-*co*-3-hydroxyhexanoate), PHBH. This copolymer is obtained by incorporating into the polyhydroxybutyrate chain, PHB, 3-hydroxyhexanoate units with medium-length side groups, [P(3HB-co-3HH)]. Mahmood and Corre identify a structure formed by branches of short 3HH units, on the main 3HB chain, thus reducing regularity. Yang and Liao compare the formation of these units by dielectric spectroscopy and melt viscosity [10,11]. Moreover, the addition of the 3HH units extends the temperature range for processing of this copolymer, but the storage modulus and the strength is reduced [12–15]. Watanabe and Oyama, synthesized PHBH from cheap natural resources such as coconut oil, biomass, beet, sugar cane, molasses and vegetable oils [16,17]. These characteristics allow it to be used as a substitute for traditional petroleum-derived polymers in some applications, such as disposable plastic bags, food packaging, catering, agricultural mulch film, and so on [18].

The main objective of this work is to obtain fully biobased WPCs. For this purpose, PHBH was chosen as the thermoplastic matrix; this matrix was reinforced with almond shell flour (ASF). The almond shell flour, ASF, is a waste of the agro-food industry. It is very cheap, fully biobased

and biodegradable. By incorporating ASF into the PHBH matrix it gives a wood-like appearance. In this work, the effect of the amount of ASF on the mechanical, thermal, thermomechanical and water absorption properties of PHBH-ASF composites is investigated. In addition, the optimization of the behaviour of these composites is used by the addition of an oligomer of lactic acid (OLA), to provide some plasticization and to increase toughness. Due to the lack of compatibility between the different elements, reactive extrusion (REX) has been proposed as a strategy to improve the properties of the mixtures. This process will improve the chemical bonding of the biopolymer chains to the surface of the lignocellulosic fillers by the action of reactive molecules with at least two functional sites.

### **2. Experimental Section**

### *2.1. Materials*

The PHBH commercial grade (ErcrosBio® PH 110) used in this study was supplied in pellet by Ercros S.A. (Barcelona, Spain). This polymer has a density of 1.2 g cm−<sup>3</sup> and a melt flow index (MFI) of 1 (g/10 min<sup>−</sup>1) measured at 160 ◦C. Even with this low MFI, this is suitable for injection moulding as it has very low melt strength, so requires an appropriate temperature profile for extrusion and injection moulding. Almond shell powder/flour (ASF) was purchased from Jesol Materias Primas (Valencia, Spain). This powder was sieved in a vibrational sieve RP09 CISA® (Barcelona, Spain) to obtain a maximum particle size of 150 μm. Figure 1 shows the irregular particle size of ASF with average size below 150 μm (the average size is 75 μm). As plasticizer/impact modifier, an oligomer of lactic acid (OLA), commercial grade Glyplast OLA 8 was kindly provided by Condensia Química S.A. (Barcelona, Spain). Glyplast OLA 8 is a liquid polyester (with an ester content above 99%) with a viscosity of 22.5 mPa s measured at 100 ◦C. Its density is 1.11 g cm<sup>−</sup>3; it has a maximum acid index of 1.5 mg KOH g−<sup>1</sup> and a maximum moisture content of 0.1%.

**Figure 1.** Visual aspect of almond shell flour particles obtained by field emission scanning electron microscopy (FESEM) at 100× and a histogram of their size distribution.

### *2.2. Manufacturing of PHBH-ASF Composites*

Before further processing of composites, PHBH pellets and almond shell flour were dried for 6 h at 80 ◦C, in a dehumidifier model MDEO, supplied by Industrial Marsé (Barcelona, Spain). Then, different amounts (see Table 1) of PHBH, ASF (in wt %) and OLA (in *phr*–weight parts of OLA per one hundred weight parts of PHBH) were mechanically pre-mixed in a zipper bag to obtain

pre-homogenization. These six materials were then extruded in a twin-screw corotating extruder from DUPRA S.L. (Alicante, Spain). The four temperature barrels were programmed to the following temperature program: 110 ◦C (hopper), 120 ◦C, 130 ◦C and 140 ◦C (extrusion die) and the screw speed was maintained in the 20–25 rpm range. The extruded material was cooled down to room temperature and then, pelletized for further processing by injection moulding. The injection moulding process was carried out in a Sprinter 11 injection machine from Erinca S.L. (Barcelona, Spain) to obtain standard samples for further characterization. As PHBH has low melt strength, it needs some particular processing conditions. The injection temperature profile was set to 150 ◦C (hopper), 140 ◦C, 130 ◦C and 120 ◦C (nozzle). In addition, it requires a tempered mould at 60 ◦C. The filling and cooling times were set to 1 s and 20 s, respectively. It is well known that bacterial polyesters undergo secondary crystallization or recrystallization with time (sometimes designed as physical aging since this leads to an embrittlement), especially at temperatures above *Tg*. Recrystallization rate is directly related to temperature; therefore, samples have been subjected to a recrystallization process at 25 ◦C for 15 days since it has been reported that almost all recrystallization takes place after two weeks from the processing [17,19,20]. To avoid potential hydrolysis of the polyester surface, samples were stored in a vacuum desiccator with constant moisture.

**Table 1.** Summary of sample compositions according to the weight content (wt %) of PHBH (Poly(3-hydroxybutyrate-*co*-3-hydroxyhexanoate)), and ASF (Almond Shell Flour) and the addition of OLA (Oligomeric Lactic Acid) as parts per hundred resin (*phr*) of PHBH-ASF composite.


### *2.3. Mechanical Properties of PHBH-ASF*/*OLA Composites*

The mechanical characterization of PHBH-ASF/OLA composites was carried out by means of tensile tests according to ISO 527-2:2012 in a universal testing machine, model ELIB-50 from Ibertest (Madrid, Spain). A 5 kN load cell was used and the crosshead rate was set to 10 mm min<sup>−</sup>1. The standardized specimens corresponded to the designation A12 from ISO 20753:2018. Impact resistance was quantified by means of a Charpy test, with a 1-J pendulum from Metrotec S.A. (San Sebastian, Spain), on specimens with a standardized "V" notch, according to ISO 179-1:2010. In addition, the hardness of PHBH-ASF/OLA composites, was obtained using a Shore-D hardness tester model 673-D from J. Bot Instruments, S.A. (Barcelona, Spain) according to ISO 868:2003. All mechanical tests were performed on 5 specimens of each composition.

### *2.4. Morphology of PHBH-ASF*/*OLA Composites*

The morphology study of the impact fractured specimens from impact tests was carried out by field emission scanning electron microscopy (FESEM) in a ZEISS ULTRA 55 microscope from Oxford Instruments (Abingdon, Oxfordshire, UK). The accelerating voltage was 2 kV. Prior to this analysis, the samples were metallized with platinum in a sputtering metallizer EMITECH mod. SC7620 from Quorum Technologies Ltd. (East Sussex, UK).

### *2.5. Thermal Characterization of PHBH-ASF*/*OLA Composites*

The thermal characterization of PHBH-ASF/OLA composites, by means of differential scanning calorimetry (DSC), was performed in a TA Instruments calorimeter mod. Q2000 (New Castle, DE, USA). For the thermal study, a dynamic temperature cycle was scheduled with the following sequence: 1st cycle: <sup>−</sup>50 ◦C to 200 ◦C at a constant heating rate of 10 ◦C min−1, 2nd cycle: 200 ◦C to <sup>−</sup>50 ◦C at a constant cooling rate of 10 ◦C min−1; this step was scheduled to remove the thermal history. Finally, a 3rd cycle from <sup>−</sup><sup>50</sup> ◦C up to 200 ◦C at 10 ◦C min−<sup>1</sup> was programmed. The DSC analysis was performed in an inert nitrogen atmosphere with a flow rate of 50 mL min−1, with samples between (5–10 mg), in standard 40 μL aluminium crucibles. The degree of crystallinity (*Xc*) was calculated by using Equation (1) where Δ*Hm* and Δ*Hcc* (J g−1) are melt enthalpy and cold crystallization enthalpy respectively. Δ*H*<sup>0</sup> *<sup>m</sup>* (J g<sup>−</sup>1) is the theoretical value that corresponds to fully crystalline PHBH; this was taken as 146 (J g<sup>−</sup>1) as reported in Reference [14]. Finally, *w* is the fraction weight of PHBH.

$$X\_{\rm c} = \left[\frac{\Delta H\_{\rm m} - \Delta H\_{\rm c\varepsilon}}{\Delta H\_{\rm m}^{0} \times w}\right] \times 100\tag{1}$$

Thermogravimetric analysis (TGA) was carried out in a Mettler-Toledo TGA/SDTA 851 thermobalance (Schwerzenbach, Switzerland). Samples consisted on small pieces with a total weight of 5–7 mg. These samples were placed in standard alumina pans (70 μL), and then subjected to a heating ramp from 30 to 700 ◦C at a constant heating rate of 20 ◦C min−<sup>1</sup> in nitrogen atmosphere. All the thermal tests were done in triplicate.

#### *2.6. Thermomechanical Characterization of PHBH-ASF*/*OLA Composites*

The dynamic-mechanical-thermal analysis, DMTA, was done in a Mettler-Toledo dynamic analyzer (Columbus, OH, USA), on rectangular samples of 40 <sup>×</sup> 10 <sup>×</sup> 4 mm3. Heating was programmed from <sup>−</sup>70 ◦C to 70 ◦C at a constant rate of 2 ◦C min<sup>−</sup>1; samples were subjected to a single cantilever test in dynamic conditions with a maximum deflection of 10 μm and a frequency of 1 Hz. The coefficient of linear thermal expansion (CLTE) of the PHBH-ASF/OLA composites was determined using a TA Instruments mod. Q400 (New Castle, DE, USA). The heating program was set from −70 ◦C to 70 ◦C, using a constant heating rate of 2 ◦C min<sup>−</sup>1. Rectangular samples with dimensions 10 <sup>×</sup> 10 <sup>×</sup> 4 mm<sup>3</sup> were subjected to a constant force of 0.02 N.

### *2.7. Water Uptake of PHBH-ASF*/*OLA Composites*

The water absorption study was carried out according to the method described in ISO 62:2008, with distilled water at 23 ± 1 ◦C, for 9 weeks. The specimens had rectangular dimensions of 80 × 10 × 4 mm3. Before starting the immersion, samples were dried at 60 ◦C for 24 h in an air circulating oven, model 2001245 DIGIHEAT-TFT from J.P. Selecta, S.A. (Barcelona, Spain).

Samples were extracted periodically from the water every planned period. They were dried to remove any remaining surface moisture and weighed on a precision analytical balance model AG245, from Mettler-Toledo Inc. (Schwerzenbach, Switzerland). After this measurement, they were re-immersed in the distilled water bath.

The amount of absorbed water during the water uptake process can be calculated following this expression:

$$
\Delta m\_t(\%) = \left(\frac{\mathcal{W}\_t - \mathcal{W}\_0}{\mathcal{W}\_0}\right) \times 100\tag{2}
$$

where *wt* stands for the sample weight after an immersion time of t; *w0* corresponds to the initial weight of the dried, before the immersion.

ISO 62:2008 establishes the application of first Fick's Law to determine the diffusion coefficient, *D*, from the collected data regarding the increase of mass by immersion, by means of the expression (2). The calculation of *D* can be done in the linear zone of the water absorption plot. In this initial stage, *wt*/*ws* is a linear function <sup>Δ</sup>*mt* <sup>=</sup> f( <sup>√</sup> *t*), that allows to determine *D* from the slope, θ [21–24].

$$\frac{\mathcal{W}\_t}{\mathcal{W}\_S} = \frac{4}{d} \left(\frac{D}{\pi}\right)^{\frac{1}{2}}\tag{3}$$

where *D* represents the coefficient of diffusion, *d* stands for the initial thickness of the specimen and *ws* stands for the saturation mass in the linear zone. If we plot *wt*/*ws* against <sup>√</sup> *t*, it is possible to obtain the slope (θ) as this condition is met, *wt*/*ws* (≤0.5), then the *D* value can be calculated by following the above-mentioned expression [25].

A correction (Stefan's approximation) is applied to this calculation for the exact calculation of the *D* according to the dimensions of the specimens:

$$D\_c = D\left(1 + \frac{d}{h} + \frac{d}{w}\right)^{-2} \tag{4}$$

where *Dc* is the geometrically corrected diffusion coefficient, *h* is the length, *w* is the width of the sample and *d* is the thickness. This equation is based on the assumption that the diffusion velocities are the same in all directions [23–25].

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

### *3.1. Mechanical Properties of PHBH-ASF*/*OLA Composites*

Table 2 shows the results obtained from mechanical characterization (tensile test, hardness and impact Charpy) of PHBH-ASF/OLA composites. The addition of ASF to the PHBH matrix resulted in composites with greater stiffness with increasing ASF wt %. With only 10 wt % ASF, the elastic modulus in tensile test (*Et*) increases to 1310 MPa from 1065 MPa (neat PHBH without lignocellulosic filler). This means an increase of 23%. This % increase is, obviously higher, for PHBH composites containing 30 wt % ASF. Regarding the maximum tensile strength (σ*max*), the incorporation of natural fillers to the polymeric PHBH matrix, promotes a noticeable decrease. Neat PHBH offers a tensile strength of 20 MPa, which decreases to 16 MPa with only 10 wt % ASF and to 12 MPa with 30 wt % ASF. Singh et al. [4] established that the decrease of tensile strength results from stress concentration at the polymer/filler interfaces. There is a lack of interface interactions between the polymeric matrix (highly hydrophobic) and the lignocellulosic particles (highly hydrophilic), which gets more pronounced with increasing particle content [4]. The mechanical behaviour of these composites highly depends on the potential interactions between the polymer matrix and the surrounding lignocellulosic filler/particle. The lack of (or poor) adhesion leads to formation of microscopic gaps that are responsible for a discontinuous material with the subsequent stress concentration phenomenon [26].


**Table 2.** Summary of the mechanical properties of the PHBH-ASF/OLA composites with different compositions, in terms of the tensile modulus (*Et*), maximum tensile strength (σ*max*), elongation at break (ε*b*), Shore-D hardness and impact strength.

Furthermore, as usual in composites with lignocellulosic fillers/reinforcements [27–34], the plastic deformation capacity of WPCs decreases in a dramatic way. If we focus on the elongation at break (ε*b*), the intrinsic very low ε*<sup>b</sup>* values for neat PHBH (around 8% after an aging time of 15 days), are reduced to half with 20 wt % ASF. Composites with 30 wt % ASF, has a ε*<sup>b</sup>* of only 3.5%. This means a much more fragile and less resistant behaviour of PHBH-ASF composites with increasing wt % ASF without any other component. This mechanical behaviour is like those obtained in other thermoplastic

matrix composite systems with natural fillers [3,26,35–37]. Nevertheless, the addition of OLA leads to an improvement of the elongation at break due to a compatibilization between PHBH/ASF by the interaction of compatibilizer with terminal groups of PHBH and lignocellulosic particles. Additionally, a plasticization effect on the matrix can be expected by OLA acting as lubricant inside de polymer chain. Both effects were reported by Quiles-Carillo et al. with different biobased and petroleum-derived compatibilizers on PLA/ASF [38].

With respect to the hardness values, the increase in the wt % ASF content favours a slight increase in Shore-D hardness as expected since tensile characterization suggested increased stiffness. In fact, the Shore-D hardness increases from 60.2 (neat PHBH) to 66.2 (composite containing 30 wt % ASF) [30]. On the other hand, the impact resistance is one of the properties with the greatest decrease in uncompatibilized PHBH-ASF composites. First, it should be noted that PHBH is a thermoplastic with an intrinsically low impact resistance. This fragile behaviour of PHBH was greatly affected by the addition of ASF (even with low wt % ASF content. The results show how the addition of 10 wt % ASF, decreases the impact strength to values lower to the half (1.8 kJ m<sup>−</sup>2) of neat PHBH (4.3 kJ m−2).

It is worthy to note that WPCs are widely used in applications that include fencing, garden objects, furniture, decking, and so on. The technical requirements will depend on the final application. Considering the mechanical results, these WPCs offer relatively low tensile strength and low elongation at break even without ASF filler. The addition of ASF up to 30 wt % and OLA as compatibilizer, gives interesting materials with a wood-like appearance but they cannot be used for medium technological applications as mechanical properties are low. Moreover, addition of 30 wt % ASF leads to a cost-effective material as PHBH matrix is still an expensive bacterial polyester.

The impact resistance values dropped down to similar values with increasing wt % ASF content. For 30 wt % ASF the absorbed-energy per unit area is around 1.6 kJ m−2, which represents a loss of almost 63% of the capacity to absorb energy during impact conditions, which is representative for the overall toughness. These results showed a clear embrittlement and loss of toughness on PHBH-ASF composites as the wt % ASF content increases. The small lignocellulosic ASF particles form a dispersed phase in the thermoplastic matrix (due to the high hydrophilicity of ASF particles, it is possible to form aggregates which lead to worse properties). This dispersed phase interrupts the continuity of the PHBH matrix; in these conditions, the stress transfer between the particle and the matrix is not allowed. In addition, as observed in Figure 1, ASF particles are not spherical; their shape is very irregular (with angular shapes) and could act as micro-notches that promote formation of microcracks and subsequently affect the crack growth. This phenomenon justifies the decrease in toughness in PHBH-ASF composites [3,4,29,37,39]. Furthermore, since the polymeric matrix is non-polar (hydrophobic) and the ASF particles are highly polar (highly hydrophilic due to its lignocellulosic composition), there is no (or very poor) matrix–particle interaction along the interface. This lack of interface causes a fragilizing effect by concentrating the stresses and decreasing the potential plastic deformation capacity in PHBH-ASF composites [26,35,40–43].

Figure 2 shows in a detailed way the lack of interface between PHBH and ASF particles. Figure 2a,b show a clear micro-gap surrounding the ASF particle. This gap sizes range from 1 μm to 3–4 μm (see white arrows), and the gaps are responsible for lack of interactions in the polymer-particle interface. This gaps are responsible for interrupting the continuity in uncompatibilized PHBH-ASF composites and do not allow stress transfer. This suggests ASF particles do not act as reinforcing material; furthermore, they promote stress concentration leading to a brittle material. Despite this, addition of an oligomer of lactic acid (OLA), could potentially provide improved toughness as observed in Figure 2c,d, which corresponds to the compatibilized composite with 10 *phr* OLA. At higher magnifications ASF particles are completely embedded by the PHBH-rich matrix (Figure 2d, see white arrow with a circle end). This situation is similar to that obtained in composites with 20 *phr* OLA (Figure 2e,f) which shows absence of gap between the ASF particles and the surrounding matrix. This oligomer has carboxylic acid and hydroxyl terminal groups that can readily react (interact) with hydroxyl terminal groups in PHBH and, obviously, with the hydroxyl groups in ASF (mainly,

cellulose, lignin and hemicelluloses). As can be seen, the gap is remarkably reduced (white arrow) and this could contribute to improve toughness and stress transfer [44,45]. Despite high polarity ester groups can establish somewhat interactions with polar groups in cellulose, the main compatibilizing effects are obtained with high reactive groups such as maleic anhydride, carboxyl acids, end-chain hydroxyl groups, glycidyl methacrylate as reported by Pracela et al. [46] by using functionalized copolymers to provide increased interface interactions between a polymer matrix and cellulose particles. Some interactions between ester groups and cellulose particles have been described by Chabros et al. [47] in thermosetting unsaturated polyester resins with cellulose fillers; in particular they describe some interactions between the polar ester groups and hydroxyl groups in cellulose by hydrogen bonding. These small range interactions can also occur in PHBH/ASF composites, but their intensity is lower than that provided by the reaction of carboxylic acid and hydroxyl terminal groups in OLA with both hydroxyl groups in cellulose and PHBH through condensation or esterification reactions. As described by Mokhena et al. [48] the ester groups in PLA are not enough to provide intense interactions between the polyester-type matrix and the cellulose filler. They report the need of different treatments on cellulose such as acetylation, glyoxalization, silylation, treatment with glycidyl methacrylate (GMA), among others to improve polymer-matrix interactions. This could be related not only to the polarity but also with the hydrophilic nature of ASF and the hydrophobic nature of PHBH.

**Figure 2.** Field emission scanning electron microscopy (FESEM) images at 1000× (left side) and 2500× (right side) corresponding to PHBH-ASF composite with 30 wt % ASF with different OLA content, (**a**) & (**b**) 0 *phr* OLA, (**c**)&(**d**) 10 *phr* OLA, (**e**)&(**f**) 20 *phr* OLA.

The PHBH-ASF composite with 30 wt % ASF, showed the worst mechanical properties in terms of ductility and toughness. This was taken as a reference material to improve its properties by the addition of a compatibilizer/plasticizer. An oligomer of lactic acid ester (OLA) was added in different proportions (10 and 20 *phr*) to provide compatibilization and some plasticization [49,50]. In Table 2, the increase in impact resistance for the reference uncompatibilized composite is observed with the addition of 10 *phr* and 20 *phr* of OLA. The addition of small amounts (10 *phr* ≈ 0.1 wt %) of this oligomer significantly improves the impact resistance of the composite [51]. It changes from 1.6 kJ m−<sup>2</sup>

(uncompatibilized PHBH-ASF with 30 wt % ASF) to 2.4 kJ m−<sup>2</sup> for the same composite with 20 *phr* OLA, which represents a % increase of 33%. Considering that this increase in toughness is related to an improvement in ductility, subsequently, Shore-D hardness values decreased, changing from 66.2 Shore-D to 58.6 and 50.0 Shore-D for 10 *phr* and 20 *phr* OLA content, respectively.

This improvement on toughness is corroborated by the capacity of deformation observed in compatibilized composites. By adding only 10 *phr* OLA to the reference uncompatibilized PHBH-ASF composite, its elongation at break is almost doubled. The compatibilization effect reported in FESEM images (gap reduction) provided a more efficient load transfer between PHBH and ASF leading to an improvement of elongation at break as Quiles-Carrillo et al. reported [38]. With 20 *phr* OLA, the ε*b*, increase up to 9.7%, which means an increase of 177% compared to the same composite without OLA. However, the most striking thing is that the ε*<sup>b</sup>* with 20 *phr* OLA is even higher than that of neat PHBH which is a very positive feature, mainly in this highly brittle material. The composite with 30 wt % ASF and 20 *phr* OLA shows a ε*<sup>b</sup>* value of almost 20% higher than neat PHBH without any filler. These results indicate a marked plasticizing effect of this OLA oligomer, which is corroborated by the values of the tensile strength (σ*t*) and the elastic modulus (*Et*). The incorporation of short-chain oligomers OLA increases the free volume of the polymer chains in PHBH, which leads to a reduction in the stiffness and an increase in the ductility of composites. The improvement in ductility these PHBH-ASF composites with the addition of OLA, produces a decrease in σ*<sup>t</sup>* to 8 MPa with 20 *phr* OLA, which is slightly lower than the σ*<sup>t</sup>* compared to the same composite without OLA (12 MPa). On the other hand, the decrease in the *Et* observed by OLA addition, indicated that the compatibilized composites are not as rigid as uncompatibilized materials. The obtained *Et* values for 10 and 20 *phr* OLA are 1158 MPa and 735 MPa respectively. This represents a decrease of 33% and 58%, with regard to the reference uncompatibilized composite with an *Et* value of 1744 MPa [51].

### *3.2. Thermal Properties of PHBH-ASF*/*OLA Composites*

A comparative plot of the DSC thermograms is represented in Figure 3 and the main thermal parameters are summarized in Table 3. All thermograms are characterized by a first change at very low temperature (around 0 ◦C) in the corresponding baseline, which is attributable to the corresponding glass transition temperature (*Tg*). Neat PHBH is a thermoplastic with low *Tg*, close to 0 ◦C; similar values to this have been reported in several studies with PHBH [14]. In a first analysis, it was determined that the addition of lignocellulosic filler, ASF, to PHBH, slightly decreases the *Tg* to values comprised in the −0.5–1.9 ◦C range for all uncompatibilized composites. Considering that *Tg* are not a unique temperature, but a temperature range in which the material undergoes a change from a glassy state to a rubbery state, it can be assessed that these slight variations in *Tg* are not significative. The marked exothermic peak observed in the DSC thermogram of neat PHBH corresponds to the cold crystallization phenomenon, and its peak (corresponding to the temperature in which the crystallization rate is maximum) is located at 54.6 ◦C (*Tcc*). The addition of lignocellulosic ASF particles does not influence the *Tcc* of PHBH in the developed composites. It can only be observed a dilution effect (the cold crystallization peak height is smaller in PHBH/ASF composites, compared to neat PHBH; this is because the PHBH/ASF composite contains 30 wt % ASF which has no thermal transition in this temperature range, and consequently, the intensity of the peak is lower). At higher temperatures, the DSC thermograms show three small and broad endothermic peaks corresponding to the melting process of PHBH [52]. As already indicated in other studies [16,18,19,53–55], due to the polymorphism of the PHBH crystals during the crystallization process, its melting occurs at different temperatures. This situation of PHBH, is identical to other polyhydroxyalkanoates (PHAs) which present three melting peaks at different temperatures *Tm1, Tm2* and *Tm3*, too. Neat PHBH used in this work, show three melting peak temperatures located at 111 ◦C, 130 ◦C and 162 ◦C, respectively. The addition of lignocellulosic ASF particles does not produce significant changes in melting temperatures as observed in other studies [36]. The analysis of the enthalpies corresponding to the cold crystallization process (Δ*Hcc*), indicated that the highest enthalpy corresponded to neat PHBH during the second heating

cycle (Δ*Hcc2* = 26.7 J g<sup>−</sup>1). These values decreased gradually with the addition of ASF particles, down to values of 3.7 J g−<sup>1</sup> for the sample with 30 wt % ASF. The dilution effect (which means considering the actual PHBH content without taking into account the wt % ASF for the cold crystallization enthalpy calculation), would give a theoretical diluted enthalpy of 18.69 J g<sup>−</sup>1, which is remarkably higher than the actual obtained value of 3.7 J g<sup>−</sup>1. These differences are not so pronounced for 10 and 20 wt % ASF.

**Figure 3.** Comparative plot of the second heating curves obtained by dynamic differential scanning calorimetry (DSC) of the different PHBH-ASF/OLA composites with different compositions.

**Table 3.** Main thermal parameters of the PHBH-ASF/OLA composites with different compositions, obtained by differential scanning calorimetry (DSC).


**\*** Δ*Hm1* and *Xc1* correspond to the first heating scan.

In a second stage, an oligomer of lactic acid (OLA) was added to improve toughness of the PHBH-AS composite with worst toughness, i.e., composite with 30 wt % ASF. The DSC thermograms, (Figure 3), show a similar thermal behaviour to the OLA-free composites. The addition of 10 *phr* and 20 *phr* OLA to this composite, shows a decrease of the PHBH *Tg* down to values around −5 ◦C, with respect to −1 ◦C for the same composite without OLA. The low molecular weight OLA chains offer slightly increased mobility in comparison with the polymer chains of neat PHBH; similar results

were reported by Quiles-Carrillo et al. [56]. This phenomenon increases the free volume in the polymeric structure and leads to a poor plasticizing effect. It is true that addition of oligomers of lactic acid to PLA, usually leads to a remarkable decrease in PLA's *Tg* as reported by Burgos et al. [57] from 66 ◦C (neat PLA) down to −10 ◦C. Nevertheless, the decrease in *Tg*, is directly related to the PLA and OLA structure. D. Lascano et al. [51] reported a decrease of PLA *Tg* from 63.3 ◦C (neat PLA; different grade) down to 50.8 ◦C with 20% OLA (different commercial grade of OLA). Armentano et al. [58] reported a dual plasticization effect of PHB and OLA on PLA, but the decrease in *Tg* was not as important as the above-mentioned by Burgos et al. Moreover, Amor et al. [59], reported a slight plasticization effect on PLA/PHBH blends by using OLA in pellet form with a dual plasticization effect of PHBH (which provided a 3 ◦C decrease in *Tg* of PLA with a PHBH loading of 10 wt %) and OLA (which provided a decrease of 1 ◦C with a load of 1 wt %). All these results show the disparity in plasticization of PLA with OLA even they share the same chemical structure. Therefore, the plasticization effect on PHAs is even more complex and has not been studied previously independently of blends with PLA. In this work, this decrease in *Tg* values are very low, but we must bear in mind that PHBH structure contains medium chain hydroxyalkanoates and these, contribute to lowering crystallinity compared to P3HB. Therefore, this slight decrease could be representative of somewhat plasticization effect provided by OLA. Besides this, it corroborates the mechanical results analyzed previously [31,40,51]. *Tcc* values increased from 52 ◦C for the sample without OLA, to 62 ◦C for the sample with 20 *phr* OLA. This is not the typical effect of a plasticizer which increases chain mobility and, therefore, the cold crystallization process is shifted to lower temperatures as reported by Lascano et al. [51] and Ferri et al. [60]. Nevertheless, some additives such as maleinized linseed oil (MLO), maleinized cottonseed oil (MCSO), or even epoxidized vegetable oils, promote an overlapping of several phenomena such as slight plasticization, chain extension, branching and, in some cases, potential crosslinking due to the multifunctionality of these compounds. Some of these vegetable-oil derivatives, produce the same behaviours as observed in this study, i.e., a shift of the cold crystallization process to higher temperatures, due to the disrupted overall structure they provide with branches, chain extension, and so on, as reported by Garcia-Campo et al. [61]. Limiñana et al. [62] reported the potential of these modified vegetable oils as compatibilizers for PBS and lignocellulosic fillers, due to reaction of oxirane, maleic anhydride groups with hydroxyl groups in almond shell flour. In this study, it seems OLA has a similar effect to those modified vegetable oils, since changes in *Tg* are very low (which is representative for very slight plasticization effect, almost inexistent, since the technique itself has some uncertainty in the obtained values depending on the sample size, geometry, surface contact, and so on), and the cold crystallization is shifted to higher temperatures, thus indicating that other phenomena could be occurring, such as compatibilization and/or chain extension.

The Δ*Hcc2* values of PHBH in OLA-compatibilized composites is 11.4 and 15.9.6 J g<sup>−</sup>1, for 10 *phr* and 20 *phr* OLA respectively. It is a striking fact that the same composite without OLA shows a much lower Δ*Hcc2*, 3.7 J g−1. With respect to *Tm1, Tm2* and *Tm3* it was observed that these thermal transitions were slightly decreased with the addition OLA, which could be related to more or less perfect crystals [51].

On the first heating scan the polymer has been recrystallized at 25 ◦C for 15 days. Consequently, during the first heating scan the cold crystallization peak did not appear which means that the polymer structure was not able to form crystallites. Under this conditions PHBH reaches a *Xc1* value of 13.9%. This increases with the amount of ASF filler up to 15.5% with 20 wt % ASF which suggests that ASF (mainly crystalline cellulose fractions) acts as a nucleant agent [62]. Furthermore, with the addition of 30 wt % ASF, *Xc*<sup>1</sup> decreases to 8.3% due to the decrease of free volume necessary for nucleation of polymer as Thomas et al. reported [63]. Mechanical characterization shows no correlation between the degree of crystallinity, while elastic modulus increases up to 30 wt % ASF, the degree of crystallinity is saturated with only 20 wt %. The second scan was performed after a controlled cooling process of 10 ◦C min<sup>−</sup>1, as a result in the second scan the polymer was able to form crystallites due to a cold crystallization process. Under this condition the degree of crystallinity could increase until 30 wt %

ASF. The compatibilizing effect of OLA in both conditions decreased the degree of crystallinity by reducing the gaps between the filler and the matrix as it is reported in FESEM analysis and Gong et al. proposed [64].

Thermogravimetric analysis, TGA, allowed to analyze the thermal stability of PHBH-ASF composites. The TGA curves of the studied materials are gathered in Figure 4, and the main thermal degradation parameters are summarized in Table 4. The thermal degradation process of neat PHBH occurs in a single step. PHBH shows good thermal stability up to 266.8 ◦C (the onset degradation temperature was taken as the temperature for a weight loss of 2%, and it is denoted as *T2%*) of PHBH. Above this temperature, thermal degradation starts, with a very fast weight loss and a temperature of maximum degradation rate, *Tmax*, of 308.9 ◦C, obtained from peak corresponding to the first derivative of its TGA curve or first DTG (Derivative thermogravimetry) (Figure 4b). The results obtained by Singh for PHBV indicated that the degradation process involves the breaking of polymer chains and hydrolysis. Since PHBH presents a similar structure, the mechanism of degradation should be similar [4]. Reaching the endset of the degradation process, located at 371 ◦C, PHBH generates a small residue or ash of 2.4 wt % of its initial weight. These results are in accordance to those obtained in other works [14,65].

**Table 4.** Summary of the main thermal degradation parameters of PHBH-ASF/OLA composites with different compositions, in terms of onset degradation temperature (*T2%*), temperature of maximum degradation (*Tmax*), and residual mass at 700 ◦C.


\* Initial weight loss in ASF due to residual water evaporation.

The thermogram obtained for ASF particles shows different degradation processes corresponding to three different sections [31,35,37]. Since it is an agro-food waste of lignocellulosic nature, it shows a first weight loss around 100 ◦C, which corresponds to the loss of remaining water in ASF, specifically 6.3 wt %. During the dynamic degradation process, when temperature reaches 213 ◦C, a rapid weight loss is observed in two main steps. The first step of weight loss, of about 44.2 wt %, corresponds to the degradation of the cellulose and hemicellulose contained in ASF particles. The first component to start degradation is hemicellulose, followed by cellulose and lignin. Lignin shows a slower (in a wide temperature range) degradation process, so the third section of the TGA curve shows a lower slope, starting at 357.3 ◦C (temperature change of slope in the thermogram) up to 500 ◦C, with a loss of 47.6 wt %. Complete degradation occurs around 500 ◦C, leaving a final carbonaceous residue or ash of 1.5 wt %, mainly from lignin. In Figure 4b, it can be seen how the temperature corresponding to the maximum degradation rate of hemicellulose-cellulose fraction is located at 300.6 ◦C, while the lignin fraction maximum degradation rate is close to 460.7 ◦C. Perinovic determined that degradation of polysaccharides, hemicellulose and cellulose starts between 220–290 ◦C, while lignin degradation range is comprised between 200 ◦C and 500 ◦C [32,33,35,37,41,66].

**Figure 4.** Comparative plot of (**a**) thermogravimetric analysis (TGA) curves and (**b**) first derivative (DTG) of the PHBH-ASF/OLA composites with different compositions.

The TGA curves of PHBH-ASF composites indicated that the thermal degradation process is a linear combination of the two individual degradation phenomena observed in PHBH and ASF, the first part of the curve is identical to neat PHBH, and a small hump at the end of the curve can be detected, which is attributable to residual degradation of ASF which changes with the ASF wt % [4,31]. For any wt % content in almond shells, the TGAs are characterized by presenting degradation start temperatures

(*T2%*) slightly lower than neat PHBH (with a *T2%* of 286.8 ◦C), in the 220–250 ◦C temperature range. The addition of lignocellulosic fillers leads to slightly lower thermal stability, since ASF particles degrade separately from PHBH, and the overall effect is the PHBH-ASF composites has reduced its thermal stability. This factor is due to the initial degradation of low molecular weight components on almond shell flour such as hemicelluloses. Quiles-Carrillo et al. [38] reported similar results with PLA composites with almond shell flour.

TGA curves of composites are very similar to those of the PHBH (as it is the main component), with practically only one step degradation stage, but with a small hump at higher temperatures, corresponding to lignin degradation. As the ASF content increases, this hump becomes more pronounced. This process ends at temperatures around 500 ◦C, generating small amounts of residue close to 2 wt %. The TGA curves of PHBH-ASF composites indicated that from a thermal point of view, the incorporation of lignocellulosic fillers such as ASF, slightly reduces the stability, but even in this case, the processing window is not compromised since all the onset degradation temperatures are above 250 ◦C, and the recommended processing temperature for this polymer is 140–150 ◦C.

On the other hand, composites with 10 *phr* and 20 *phr* OLA, show a very similar thermal degradation behaviour to the reference uncompatibilized composite (PHBH-ASF with 30 wt % ASF) without OLA. The addition of 10 *phr* OLA seems to slightly improve the thermal stability of the developed composites. The characteristic thermal degradation values are delayed by 34 ◦C (onset degradation temperature, *T2%)* and by 10 ◦C (for the maximum degradation rate, *Tdeg*) compared to the unmodified reference composite without OLA. The biggest improvement in thermal stability is observed for the composite with 10 wt % OLA. This improvement is due to the chemical interaction of the compatibilizer with both components of the composite as above-mentioned. The complex structure formed after reaction of OLA with both PHBH and ASF, can act as a physical barrier that obstructs the removal of volatile products produced during decomposition [67].

### *3.3. Thermomechanical Properties of PHBH-ASF*/*OLA Composites*

Figure 5a shows the variation of the flexural single cantilever) storage modulus, *E'*, with respect to temperature, obtained by DMTA analysis. It can be seen graphically how *E'* decreases with increasing temperature in all the developed composites, as expected due to the softening of the polymeric PHBH matrix. At low temperatures, *E'* values are high in all composites, since this temperature range corresponds to the elastic-glassy behaviour of the PHBH matrix. In this first zone, *E'* for neat PHBH is 1869 MPa at −40 ◦C, which is lower than *E'* values of any of the uncompatibilized PHBH-ASF composites (e.g., *E'* is 2019 MPa for the PHBH-ASF composite with 30 wt % ASF at −40 ◦C). These results are in accordance with those obtained by mechanical characterization which suggested a stiffening as the wt % ASF increased [37]. Table 5 shows the numerical comparison of the variation of *E'* as a function of ASF wt % and OLA *phr*, at two different temperatures.

**Table 5.** Main dynamic-mechanical thermal parameters of PHBH-ASF/OLA composites with different compositions: flexural storage modulus (*E'*) measured at −40 ◦C and 25 ◦C and glass transition temperature (*Tg*), obtained by dynamic-mechanical thermal analysis (DMTA).


**Figure 5.** Comparative plot of dynamic-mechanical thermal analysis (DMTA) curves of PHBH-ASF/OLA composites with different compositions: (**a**) flexural storage modulus (*E'*) and (**b**) dynamic damping factor (*tan* δ).

As the temperature increases, *E'* decreases rapidly as it acquires a rubbery state behaviour. This is related to the α-relaxation process or the glass transition temperature (*Tg*). Table 5 shows how at 25 ◦C, the *E'* value for neat PHBH has decreased to 1345 MPa from 1869 MPa at −40 ◦C and, subsequently, the stiff-elastic behaviour changes to a rubber-like behaviour. The same trend can be observed for uncompatibilized PHBH-ASF composites. However, when compared to the neat PHBH matrix, at the same temperature, the higher the wt % of ASF, the stiffer the composite becomes. With only 10 wt % ASF, *E'* at 25 ◦C increases 6.4% with respect to PHBH at the same temperature. Accordingly, the composite with 30 wt % ASF offers higher stiffness (a percentage increase of 19% regarding neat PHBH). The presence of ASF particles finely dispersed in the PHBH matrix restricts the mobility of the polymer chains, thus decreasing their viscous behaviour, which causes an increase in the *E'* value as the ASF loading increases [4,27–29,39]. In addition, at this temperature range, lignocellulosic components are below its *Tg*, which means they show a glassy behaviour that promotes

increased stiffness. Nevertheless, it is important to bear in mind that the main component in PHBH/ASF composites is PHBH and the dynamic behaviour is highly influenced by PHBH behaviour since conditions are not as aggressive as a conventional tensile test up to fracture. These results corroborate those obtained in the mechanical characterization of PHBH-ASF composites.

As in previous analyses, by adding small amounts of OLA to the PHBH-ASF composite with the highest ASF loading, which shows the worst ductile/toughness properties, the DMTA graphs in Figure 5a show the lowest *E'* in the temperature range analyzed. At −40 ◦C, *E'* decreases from 2019 MPa without OLA addition, to 1601 MPa and 853 MPa for the addition of 10 *phr* and 20 *phr* OLA, respectively. The trend is the same at room temperature (25 ◦C). With an addition of only 10 *phr* OLA the *E'* is decreased by 16%, and with 20 *phr* OLA *E'* is reduced even more, thus decreasing the rigidity of PHBH-ASF composites and, subsequently their *E'* [40]. It is worth noting the extremely small changes in *Tg* obtained by DMTA which suggests, as DSC, very poor plasticization effect, suggesting compatibilization is one of the most representative effects of OLA in this PHBH/ASF system.

Figure 5b shows the variation of the dynamic damping factor (*tan* δ) as a function of temperature, for neat PHBH, uncompatibilized PHBH-ASF and PHBH-ASF/OLA composites. Despite there are several criteria to obtain the *Tg* from DMTA graphs, the most used is the peak maximum of *tan* δ. The *Tg* values obtained for all developed materials are gathered in Table 5. It can be observed that *tan* δ peaks are slightly moved towards higher temperatures in uncompatibilized PHBH-ASF composites, compared to neat PHBH (a maximum shift of 3–4 ◦C). This change is not significant as observed in other techniques such as DSC. On the other hand, the addition of OLA leads to slightly lower *Tg* values by DMTA, but once again, these changes are not high enough to give a clear evidence of the chain mobility restriction by ASF particles or increased chain mobility by OLA.

For potential structural/engineering/conventional applications of WPCs, it is very important to know their dimensional stability with temperature. This can be assessed by thermomechanical analysis (TMA) which allows us to obtain the coefficient of linear thermal expansion (*CLTE*). It must be stated that a good dimensional stability involves low *CLTE* values. Table 6 summarizes the CLTE values for the developed composites, at temperatures below and above their corresponding *Tg*. In general, at temperatures below *Tg* the *CLTE* values are much lower than above their *Tg*. The dimensional expansion of the material is lower at low temperatures because the material is more rigid, which is the typical glassy behaviour below *Tg*. Above *Tg*, the behaviour is viscous or rubber-like and so that, the dimensional expansion is favoured, with higher *CLTE* values.


**Table 6.** Summary of the main thermo mechanical properties of neat PHBH and PHBH-ASF/OLA with different compositions, regarding the thermal expansion, obtained by thermomechanical analysis (TMA).

First, the analysis of the *CLTE* values at temperatures below *Tg* shows that increasing wt % ASF gives more dimensional stability, which is in accordance with previous mechanical results that suggested a clear stiffening with ASF addition. Pure PHBH has an initial value of 77.1 μm m−<sup>1</sup> ◦C<sup>−</sup>1, which decreases to 66.8 μm m−<sup>1</sup> ◦C−<sup>1</sup> with 30 wt % ASF particles (which represents a % decrease of 13%), which involves improved dimensional stability. However, OLA addition significantly increases the *CLTE* values, as typical plasticizers do, thus leading to slightly lower dimensional stability. With 10 *phr* and 20 *phr* OLA, *CLTE* is 72.0 and 90.7 μm m−<sup>1</sup> ◦C<sup>−</sup>1, respectively.

Secondly, the results obtained in the study at temperatures above *Tg*, show the same tendency as the results discussed in the previous paragraph. Neat PHBH shows an initial *CLTE* of 160.7 μm m−<sup>1</sup> ◦C−<sup>1</sup> (much higher than below *Tg*), which decreases to 140.3 μm m−<sup>1</sup> ◦C−<sup>1</sup> with 30 wt % ASF particles. Identically as observed previously, *CLTE* becomes greater again with the addition of OLA, reaching values of 194.3 μm m−<sup>1</sup> ◦C−<sup>1</sup> with 20 *phr* OLA. In general, as the ASF content increases, composites show improved dimensional stability [4,29,68]. However, these dimensional expansions are higher than those of neat PHBH in PHBH-ASF composites (30 wt % ASF) with 10 *phr* and 20 *phr* OLA, due to the plasticizing effect [51].

### *3.4. Evolution of the Water Uptake and Water Di*ff*usion Process in PHBH-ASF*/*OLA Composites*

The water absorption capacity of WPCs is an important feature in some applications due to the lignocellulosic component. This creates a three-dimensional path inside the polymer matrix that allows water entering (for example when the composite is subjected to high relative humidity environments) and this causes an expansion. It is possible that after this initial stage, this WPC could be subjected to drying at sun with low humidity; then this 3D-path allows water/moisture removal, promoting a contraction. This situation is quite usual in WPCs such as those used in fences, decking, and so on. This repeated expansion–contraction cycles could lead to formation of microcracks. Figure 6 shows the mass increase (wt %) with respect to immersion time in water for the PHBH-ASF/OLA composites. It can be seen graphically that during the first week of immersion, the developed composites show a rapid increase in mass by water absorption (Δ*mass*). As immersion time increases, the mass increase is slower. Some samples even show an asymptotic behaviour, which indicates that saturation has been reached (Δ*mass*∞). This type of behaviour corresponds to that indicated by the first Fick's Law.

**Figure 6.** Water uptake of PHBH-ASF/OLA composites with different compositions. Evolution of the water uptake for a period of nine weeks.

Obviously, due to the hydrophobicity of neat PHBH, it shows the lowest water absorption for nine weeks. After 35 days of immersion it reaches a constant saturation mass, Δ*mass*∞ of 0.53%, which is maintained practically until 63 days of immersion. As mentioned above, the *Xc* of neat PHBH is 13.9%. The addition of ASF (highly hydrophilic particles due to its composition: cellulose, hemicellulose and lignin as the main components) considerably increases water absorption, but the effects of PHBH crystallinity can also be observed in this behaviour. The composite with 10 wt % ASF reaches a water saturation mass, Δ*mass*∞ of 1.46 wt % after 42 days (*Xc* of PHBH in this composite is 14.9% and this prevents from water entering). In a similar way, uncompatibilized composites containing 20 wt % show a relatively low Δ*mass*∞ of 3.1 wt %. This is almost double the previous value (with 10 wt % ASF). This value is expected since the *Xc* of PHBH in this composite is close to 15.5%. Nevertheless, the composite containing 30 wt % ASF, shows a remarkable increase in Δ*mass*∞ up to values of 7.1 wt % after nine weeks, which represents almost 14 times the value for neat PHBH, representing a typical result in most WPCs [23,36,43]. The result is higher than the extrapolation from the ASF wt % (which should be of about three times the value of the composite with 10 wt % ASF, i.e., 4.5 wt %); despite this, we have to bear in mind that the *Xc* of PHBH in this composite has decreased to 8.3% and this has a negative effect on water absorption as amorphous regions allow water entering [69,70]. Therefore, the increase in water absorption is not only related to the number of lignocellulosic components, but also with the degree of crystallinity of the polymer matrix. Cellulose promotes water absorption due to hydroxyl (–OH) groups that interact with water molecules [21–23,25,41,71].

On the other hand, the addition of OLA to the reference composite (PHBH with 30 wt % ASF), shows an unexpected behaviour. As one can see in Figure 6, the water absorption curve with OLA, moves to lower wt % absorbed water. The values of the mass increase after nine weeks of immersion reach values of 5.8 wt % and 5.5 wt % for OLA contents of 10 *phr* and 20 *phr*, respectively. This means a decrease in water absorption of 18% and 22.6% respectively when OLA is added to the PHBH-ASF system. Typical plasticizers, increase the water absorption as they are responsible for an increase in the free volume, thus allowing water molecules to enter. Nevertheless, OLA not only provides plasticization effects, but also improved polymer-particle interaction among the interface due to the interaction between the hydroxyl groups in OLA and the hydroxyl groups of both PHBH (terminal groups), and cellulose/hemicellulose/lignin in ASF particles. Therefore, in addition to a plasticization phenomenon, we could think on an additional compatibilization effect which was also evidenced in Figure 2 (FESEM characterization), in which the gap size was higher on OLA-free composites than composites with OLA.

Table 7 shows the values of the diffusion coefficient (*D*) or diffusivity of water into the developed composites, by applying the first Fick's Law. The lowest corrected diffusion coefficient, *Dc* value is offered by neat PHBH, as expected due to its hydrophobicity. The only addition of 10 wt % ASF, leads to a *Dc* value, almost triple compared to neat PHBH. Obviously, ASF is responsible for water entering the composite structure; therefore, uncompatibilized composites with 30 wt % ASF, shows an increase of two orders of magnitude. Due to the hydrophilic nature of ASF particles, and possibly accentuated by the capillarity of the micro-gaps between PHBH and the embedded ASF particles, water molecules can easily enter into the composite structure as in most WPCs [51,71]. Finally, the *Dc* values PHBH-ASF/OLA composites remain with similar values to those of the same composite without OLA. Thus, it was deduced that the amount of wt % ASF is the parameter with the greatest influence on the water diffusion process in PHBH-ASF/OLA composites, together with the degree of crystallinity as previously discussed with the relationship of the water absorption and the wt % ASF loading and the degree of crystallinity of the PHBH matrix. Since the *Dc* values for the composites with 10 and 20 *phr* OLA are similar to that of the same composite with 30 wt % ASF, it is possible to conclude the poor plasticization effect of this OLA, suggesting, once again, that compatibilization is the main acting mechanisms of OLA on PHBH/ASF composites.


**Table 7.** Values of the diffusion coefficient (*D*) and the corrected diffusion coefficient (*Dc*) for PHBH and the PHBH-ASF composites processed with OLA.

### **4. Conclusions**

The results obtained in this study indicate that the analyzed system of poly(3-hydroxybutyrateco-3-hydroxyhexanoate) (PHBH) and almond shell flour (ASF), is suitable for the manufacture of fully biobased and environmentally friendly Wood Plastic Composites (WPCs). PHBH-ASF composites present a very interesting set of properties for technical applications as wood substitute materials. The characterization of PHBH-ASF composites showed that the addition of lignocellulosic particles of ASF leads to an embrittlement and reduced toughness. These effects are much more evident with increasing the wt % ASF. To overcome or minimize these negative properties, an oligomer of lactic acid, OLA was added to give PHBH-ASF/OLA composites with improved ductile properties and, subsequently, improved toughness. It is worth noting a remarkable increase in impact strength with 20 *phr* OLA. A higher mobility of the PHBH polymer chains by the addition of OLA is the reason for the improvement on toughness, even on composites with 30 wt % ASF.

The study of the PLA-ASF/OLA composites allowed us to obtain good balanced properties and, therefore, these materials can be used in the WPC industry as they are suitable for technical applications that require certain stiffness and thermal stability, in an interesting range of properties depending on the wt % ASF content. Furthermore, the addition of OLA oligomer decreases the water absorption capacity of PHBH-ASF/OLA, thus broadening potential uses in high humidity environments. Finally, its thermoplastic nature allows it to be easily processed by conventional extrusion–injection moulding, and overall, these composites contribute to a sustainable development and a reduction of the carbon footprint as all the used materials are bio-sourced.

**Author Contributions:** Conceptualization, R.B. and L.Q.-C.; methodology, L.S.-N., J.I.-M. and J.M.-M.; validation, J.M.-M., J.I.-M. and L.Q.-C.; formal analysis, L.Q.-C. and J.I.-M.; investigation, J.M.-M., T.B., L.Q.-C. and J.I.-M.; data curation, L.Q.-C. and J.I.-M.; writing-original draft preparation, L.Q.-C. and T.B.; writing-review and editing, R.B., J.I.-M.; supervision, R.B., L.Q.-C. and T.B.; project administration, R.B. and T.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research work was funded by the Spanish Ministry of Science, Innovation, and Universities (MICIU) project number MAT2017-84909-C2-2-R. This work was supported by the POLISABIO program grant number (2019-A02).

**Acknowledgments:** J. Ivorra-Martinez is the recipient of an FPI grant from Universitat Politècnica de València (PAID-2019). L. Quiles-Carrillo wants to thank GV for his FPI grant (ACIF/2016/182) and MECD for his FPU grant (FPU15/03812). Microscopy services at UPV are acknowledged for their help in collecting and analyzing FESEM images.

**Conflicts of Interest:** The authors declare no conflicts 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/).

### MDPI

St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Polymers* Editorial Office E-mail: polymers@mdpi.com www.mdpi.com/journal/polymers

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com