Torrefaction of Co ﬀ ee Husk Flour for the Development of Injection-Molded Green Composite Pieces of Polylactide with High Sustainability

Featured Application: Green composite pieces of polylactide (PLA) containing 20 wt% of co ﬀ ee waste derived ﬁller can be applied for rigid packaging, such as food trays and containers, beverage cups, and food contact disposables, such as cutlery and plates. Abstract: Co ﬀ ee husk, a major lignocellulosic waste derived from the co ﬀ ee industry, was ﬁrst ground into ﬂour of ﬁne particles of approximately 90 µ m and then torreﬁed at 250 ◦ C to make it more thermally stable and compatible with biopolymers. The resultant torreﬁed co ﬀ ee husk ﬂour (TCHF) was thereafter melt-compounded with polylactide (PLA) in contents from 20 to 50 wt% and the extruded green composite pellets were shaped by injection molding into pieces and characterized. Although the incorporation of TCHF reduced the ductility and toughness of PLA, ﬁller contents of 20 wt% successfully yielded pieces with balanced mechanical properties in both tensile and ﬂexural conditions and improved hardness. Contents of up to 30 wt% of TCHF also induced a nucleating e ﬀ ect that favored the formation of crystals of PLA, whereas the thermal degradation of the biopolyester was delayed by more than 7 ◦ C. Furthermore, the PLA / TCHF pieces showed higher thermomechanical resistance and their softening point increased up to nearly 60 ◦ C. Therefore, highly sustainable pieces were developed through the valorization of large amounts of co ﬀ ee waste subjected to torrefaction. In the Circular Bioeconomy framework, these novel green composites can be used in the design of compostable rigid packaging and food contact disposables.


Introduction
Valorization of food processing by-products and agricultural waste is acquiring increasing importance due to the high interest in the development of novel sustainable materials as part of the Circular Bioeconomy framework [1,2]. Agro-food wastes, such as flax fibers [3,4], rice husk [5], jute fibers [6,7], almond husk [8,9], walnut husk [10], orange peel [11], coconut fibers [12] or sisal fibers [13], represent examples of recently explored fillers to be incorporated into biopolymer matrices. The resultant green composites can successfully show technological advantages over husk flour (TCHF) contents were incorporated into PLA and the mechanical and thermal properties were determined to analyze the influence of the coffee waste derived lignocellulosic filler on the final performance of the green composite pieces and ascertain their potential for sustainable packaging and disposables.

Materials
PLA biopolymer commercial grade Ingeo™ 3001D was supplied in pellet form by NatureWorks LLC (Minnesota, MN, USA). This grade corresponds to a poly(d,l-lactide) (PDLLA) of high optical purity in l-lactic acid isomer, with 1.4 ± 0.2% of d-lactic acid fraction [46]. The number-and weight-average molecular weights were found to be M n = 90.5 kD and M W = 158.3 kD by gel permeation chromatography (GPC), which gives a polydispersity index (PDI = M W /M n ) of~1.75 [47]. It is also characterized by a density of 1.24 g/cm 3 following the American Society for Testing and Materials (ASTM) method D792 and a melt flow rate (MFR) of 22 g/10 min, measured at 210 • C and 2.16 kg, according to the ASTM method D1238, which makes it feasible for injection molding.
Coffee husks were separated from the previously natural dried grains by a threshing process. The husks were obtained from a coffee crop located at 1390 m above sea level in the San José de Motoso district of Girón in Santander, Colombia, and were dried in a laboratory oven at 105 • C for 24 h to remove moisture. Afterwards, the dried husks were milled at a speed of 8000 rpm using a ZM 200 centrifugal mill from Retsch (Düsseldorf, Germany) and finally sieved with a 250-µm mesh filter to the size of 110 microns to obtain the so-called CHF (see Figure 1).
Appl. Sci. 2020, 10, x FOR PEER REVIEW 3 of 18 (TSE) and the resultant compounded pellets were shaped into pieces by injection molding. Different torrefied coffee husk flour (TCHF) contents were incorporated into PLA and the mechanical and thermal properties were determined to analyze the influence of the coffee waste derived lignocellulosic filler on the final performance of the green composite pieces and ascertain their potential for sustainable packaging and disposables.

Materials
PLA biopolymer commercial grade Ingeo™ 3001D was supplied in pellet form by NatureWorks LLC (Minnesota, MN, USA). This grade corresponds to a poly(D,L-lactide) (PDLLA) of high optical purity in L-lactic acid isomer, with 1.4 ± 0.2% of D-lactic acid fraction [46]. The number-and weightaverage molecular weights were found to be Mn = 90.5 kD and MW = 158.3 kD by gel permeation chromatography (GPC), which gives a polydispersity index (PDI = MW/Mn) of ~1.75 [47]. It is also characterized by a density of 1.24 g/cm 3 following the American Society for Testing and Materials (ASTM) method D792 and a melt flow rate (MFR) of 22 g/10 min, measured at 210 °C and 2.16 kg, according to the ASTM method D1238, which makes it feasible for injection molding.
Coffee husks were separated from the previously natural dried grains by a threshing process. The husks were obtained from a coffee crop located at 1390 m above sea level in the San José de Motoso district of Girón in Santander, Colombia, and were dried in a laboratory oven at 105 °C for 24 h to remove moisture. Afterwards, the dried husks were milled at a speed of 8000 rpm using a ZM 200 centrifugal mill from Retsch (Düsseldorf, Germany) and finally sieved with a 250-µm mesh filter to the size of 110 microns to obtain the so-called CHF (see Figure 1).

Torrefaction of CHF
The resultant CHF was torrefied in a laboratory oven Carbolite mod. 2416CG from Keison Products (Barcelona, Spain). The flour was first dried at 105 °C for 30 min, then heated up under nitrogen atmosphere to the torrefaction temperature of 250 °C at a heating rate of 5 °C/min, and finally this temperature was kept constant for 1 h. The resulting torrefied husk was milled and sieved to the size of approximately 90 µm using the same procedure applied in CHF to obtain the so-called TCHF (see Figure 2).

Torrefaction of CHF
The resultant CHF was torrefied in a laboratory oven Carbolite mod. 2416CG from Keison Products (Barcelona, Spain). The flour was first dried at 105 • C for 30 min, then heated up under nitrogen atmosphere to the torrefaction temperature of 250 • C at a heating rate of 5 • C/min, and finally this temperature was kept constant for 1 h. The resulting torrefied husk was milled and sieved to the size of approximately 90 µm using the same procedure applied in CHF to obtain the so-called TCHF (see Figure 2).

Preparation of the Green Composite Pieces
The biopolymer pellets and the fillers were dried separately for 24 h at 60 °C prior to processing. PLA and TCHF were mixed in a zipped bag according to the contents indicated in Table 1. These composition ranges were selected as previous studies with lignocellulosic materials in PLA indicated that excessive loss of properties occurs with higher amounts of lignocellulosic fillers [13,48,49]. The mixtures were melt-compounded in a co-rotating twin-screw extruder (D = 25 mm, L/D = 24) from Dupra SL (Alicante, Spain), an equipment described in previous work [17]. The profile temperature was set, from the hopper to the die, to 165-170-175-180 °C and the rotating speed was 40 rpm. The extruded filaments were pelletized using an air-knife unit after being cooled down in an air stream. A green composite of PLA with 20 wt% of CHF was also prepared under the same conditions for comparison purposes. The compounded pellets were thereafter injection molded to obtain pieces in a Meteor 270/75 from Mateu & Solé (Barcelona, Spain). The profile of temperatures in the barrel was set, from the hopper to the injection nozzle, to 165-170-175-180 °C and the mold temperature at 60 °C. A low injection time of 1 s was employed, whereas the applied clamping force was 20 tons and the pieces were cooled for 20 s in the mold. As a result, 4-mm pieces were produced that, prior to characterization, were stored for at least 48 h.

Mechanical Tests
Tensile, flexural, hardness, and impact standard tests were carried out for mechanical characterization of the green composites. Pieces with a geometry of 80 mm × 10 mm × 4 mm were tested in a universal test machine ELIB 30 from S.A.E Ibertest (Madrid, Spain) equipped with a 5-kN load cell in both tensile and flexural conditions. The guidelines of the International Organization for Standardization (ISO) 527 and ISO 178, were followed for the tensile and flexural tests using cross-

Preparation of the Green Composite Pieces
The biopolymer pellets and the fillers were dried separately for 24 h at 60 • C prior to processing. PLA and TCHF were mixed in a zipped bag according to the contents indicated in Table 1. These composition ranges were selected as previous studies with lignocellulosic materials in PLA indicated that excessive loss of properties occurs with higher amounts of lignocellulosic fillers [13,48,49]. The mixtures were melt-compounded in a co-rotating twin-screw extruder (D = 25 mm, L/D = 24) from Dupra SL (Alicante, Spain), an equipment described in previous work [17]. The profile temperature was set, from the hopper to the die, to 165-170-175-180 • C and the rotating speed was 40 rpm. The extruded filaments were pelletized using an air-knife unit after being cooled down in an air stream. A green composite of PLA with 20 wt% of CHF was also prepared under the same conditions for comparison purposes. The compounded pellets were thereafter injection molded to obtain pieces in a Meteor 270/75 from Mateu & Solé (Barcelona, Spain). The profile of temperatures in the barrel was set, from the hopper to the injection nozzle, to 165-170-175-180 • C and the mold temperature at 60 • C. A low injection time of 1 s was employed, whereas the applied clamping force was 20 tons and the pieces were cooled for 20 s in the mold. As a result, 4-mm pieces were produced that, prior to characterization, were stored for at least 48 h.

Mechanical Tests
Tensile, flexural, hardness, and impact standard tests were carried out for mechanical characterization of the green composites. Pieces with a geometry of 80 mm × 10 mm × 4 mm were tested in a universal test machine ELIB 30 from S.A.E Ibertest (Madrid, Spain) equipped with a 5-kN load cell in both tensile and flexural conditions. The guidelines of the International Organization for Standardization (ISO) 527 and ISO 178, were followed for the tensile and flexural tests using cross-head speeds of 10 and 5 mm/min, respectively. A Shore D durometer 673-D from Instrumentos J. Bot S.A. (Barcelona, Spain) was used to determine Shore D hardness according to ISO 868. The impact absorbed energy was determined following the ISO 179 in a Charpy pendulum of 6 J from Metrotec S.A. (San Sebastián, Spain) using unnotched samples of 80 mm × 10 mm × 4 mm. For each mechanical test, at least six different samples were tested and the values were averaged.

Microscopy
The fracture surfaces of the pieces obtained from the impact tests were coated in a vacuum with an ultrathin aurum-palladium layer in a high vacuum sputter Emitech mod. SC7620 (Quorum Technologies Ltd., East Sussex, UK). Then, field emission scanning electron microscopy (FESEM) was performed to analyze their morphologies in high-resolution Zeiss Ultra55 (Oxford Instruments, Abingdon, UK) operating at 2 kV.

Thermal Tests
Sample pieces with a mass of 6.5 ± 0.5 mg were placed in crucibles and analyzed by differential scanning calorimetry (DSC) in a DSC calorimeter from Mettler-Toledo Inc. (Schwerzenbach, Switzerland) under a nitrogen atmosphere (66 mL/min). The program consisted of a heating ramp from 30 to 250 • C to remove the thermal history, a cooling ramp down to 30 • C, and a second heating ramp to 350 • C. A heating rate of 10 • C/min were used in all cases. The maximum degree of crystallinity (X C_max ) was determined following Equation (1) [50]: where the melting enthalpy of PLA is ∆H m (J/g), while the theoretical value of fully crystalline PLA is ∆H m • (J/g), taken as 93.7 J/g [51], and the weight fraction of PLA in the sample is 1 − w.
Sample pieces with a mass of 15.5 ± 0.5 mg were placed in crucibles and subjected to thermogravimetric analysis (TGA) in a TGA/SDTA 851 thermobalance from Mettler-Toledo Inc. (Schwerzenbach, Switzerland). The samples were heated from 30 to 700 • C with a heating rate of 10 • C/min in nitrogen (66 mL/min). All the thermal tests were performed in triplicate.

Thermomechanical Tests
The Vicat softening point (VST) and heat deflection temperature (HDT) of the green composite pieces were determined using a VICAT/HDT station mod. VHDT 20 (Metrotec S.A., San Sebastián, Spain). VST tests were carried out following the ISO 306 with a load of 5 kg and a heating rate of 50 • C/h, whereas HDT measurements were performed according to the ISO 75 with a load of 320 g and a heating rate of 120 • C/h. Rectangular samples of 40 mm × 10 mm × 4 mm were subjected to dynamic mechanical thermal analysis (DMTA) in torsion mode using an AR-G2 oscillatory rheometer from TA Instruments (New Castle, DE, USA) equipped with a special clamp system for solid samples. The temperature program consisted of a thermal ramp from 30 to 140 • C at a heating rate of 2 • C/min. The maximum deformation percentage (γ) was set to 0.1% at a frequency of 1 Hz. All the thermomechanical tests were performed in triplicate. Table 2 includes the mechanical properties of PLA and its green composite pieces to ascertain the effect of the TCHF content on PLA. One can notice that neat PLA resulted in a brittle piece with an elongation at break (ε b ) of nearly 9%. The tensile strength at yield (σ y ) was approximately 50 MPa and the tensile modulus (E t ) was 3.58 GPa. One can observe that the incorporation of both CHF and TCHF induced a progressive reduction in all the σ y and ε b values and an increase in E t . For instance, with the addition of 20 wt% of CHF, the σ y values diminished to 39.16 MPa and with the addition of 20 wt% of TCHF, the σ y values diminished to 43.29 MPa, which represent a percentage decrease of 21% and 13%, respectively. Further reductions in the mechanical strength were observed at higher TCHF contents. For instance, the addition of 50 wt% of TCHF induced a percentage decrease of~44%. Alternatively, the incorporation of CHF and TCHF improved the mechanical elasticity of the pieces. Thus, E t increased to 3.48 GPa for the PLA/CHF composite and it was in the 3.59-4.45 GPa range for the PLA/TCHF composite pieces. This result suggests than an efficient dispersion and wettability of the torrefied filler in the biopolymer matrix was achieved. With regard to ductility, the addition of the lignocellulosic fillers led to a percentage decrease from 8.91%, for the neat PLA piece, to 6.98% for the PLA/20CH piece, 7.10% for the PLA/20TCH piece, and finally to the lowest value of 5.18% for the PLA/50TCH piece. It can be thus considered that the coffee waste derived fillers acted as a stress concentrator rather than a reinforcing element in the biopolymer matrix. Furthermore, this result highlights the lack of miscibility between both composite components and also the low particle-polymer interfacial interactions. This effect was, however, more noticeable in the piece containing the untreated CHF and, most notably, for the pieces filled with the highest TCHF contents. The latter observation may suggest that TCHF poorly disperses in the PLA matrix at high content [52,53]. Therefore, as also demonstrated by Lessa et al. [54] for chitosan films containing waste coffee grounds (WCG), high amounts of TCHF cannot be efficiently distributed in the biopolymer matrix. Table 2. Mechanical properties of the injection-molded pieces of polylactide (PLA) with coffee husk flour (CHF) and torrefied coffee husk flour (TCHF) in terms of tensile modulus (E t ), tensile stress at yield (σ y ), elongation at break (ε b ), maximum flexural stress (σ f ), flexural modulus (E f ), Shore D hardness, and impact strength. Nevertheless, one should also consider that this mechanical reduction was relatively low in terms of the high amount of lignocellulosic filler when compared with those attained in previously prepared green composites of PLA and coffee waste derived fillers. For instance, in the work reported by Suaduang et al. [55], the tensile strength at break (σ b ) of PLA films decreased from 79.20 MPa (transverse direction, TD) and 72.79 MPa (machine direction, MD) to 20.77 MPa (TD) and 18.31 MPa (MD) after the incorporation of 10 wt% of SCG. In another study, the σ b value of PLA films decreased from 44.3 MPa to approximately 20 MPa after the incorporation of 40 wt% of SCG, which was attributed to the poor filler dispersion in the biopolymer matrix due to its low compatibility [56]. The improvement attained here may be attributed to the removal of impurities and other volatile TCHF components during torrefaction that would otherwise decrease the strength of interfacial interactions and adhesion between the filler and PLA matrix. Furthermore, the resultant increase in hydrophobicity of the lignocellulosic filler after torrefaction can also be a meaningful explanation for the improved mechanical properties attained in the green composite pieces with TCHF in comparison with the pieces based on untreated CHF. In this regard, one can expect that the higher presence of moisture in CHF can promote a larger decrease in the PLA's M W during processing at high temperature due to hydrolysis of the ester groups, which can result in impairment of the final mechanical properties [57].

Mechanical Performance of the Green Composite Pieces
The flexural modulus (E f ) and strength (σ f ) of the unfilled PLA piece were 3.17 GPa and nearly 99 MPa, respectively. Similar to the tensile properties, the incorporation of TCHF yielded a progressive increase in E f and a reduction in σ f . In particular, the green composite pieces filled with 50 wt% of TCHF showed a E f value of 5.53 GPa, which is approximately 1.7 times higher than that of the unfilled PLA, whereas σ f was reduced to 61.42 MPa. Some previous studies have reported similar mechanical properties of green composites based on coffee waste derived fillers. For instance, Dominici et al. [58] showed a worsening of tensile strength and ductility when CSS was incorporated into bio-based high-density polyethylene (bio-HDPE). The described decrease of the mechanical properties was ascribed to the low aspect ratio of the CSS particles as well as their poor adhesion to the biopolymer matrix. In this regard, the use of coupling agents has improved both the tensile strength and ductility of PLA composites prepared with SCG, though the improvements achieved were relatively low and their overall and specific migration limits (OML and SML) should also be carefully determined [59].
One can also observe that, as the TCHF content increased, hardness also increased. The Shore D hardness value increased from 78.65, for the unfilled PLA piece, to 85.05 for the PLA/50TCH piece. This change in the mechanical behavior corresponds to the typical effect of a relatively stiff lignocellulosic filler on a polymer matrix [60]. In contrast to hardness, the impact absorbed energy of PLA progressively decreased after the TCHF incorporation. The impact energy of the neat PLA piece was 25.10 kJ/m 2 and it was reduced down to 7.19 kJ/m 2 for the green composite piece filled with 50 wt% of TCHF. This notable decrease in toughness can be related to the poor matrix-particle interactions, which could promote microcrack formations and stress concentrations during the impact test [60]. Therefore, the addition of the coffee waste derived lignocellulosic filler led to a more elastic and stiffer material, but the matrix-filler continuity was not achieved, and an early fracture was attained. Thus, in terms of mechanical performance, PLA pieces containing 20 wt% of TCHF were slightly more resistant but less ductile, showing characteristics comparable to those of neat PLA. Figure 3 shows the visual aspect of the injection-molded pieces of PLA, PLA filled with 20 wt% of CHF and TCHF, that is, PLA/20CHF and PLA/20TCHF, respectively. The increase in the characteristic dark brown color produced in the pieces of the green composites is due to the Maillard reaction of coffee husk, which took place during the torrefaction and its subsequent melt processing with PLA, leading to formation of melanoidins [35].

Aspect and Morphology of the Green Composite Pieces
Appl. Sci. 2020, 10, x FOR PEER REVIEW 7 of 18 The flexural modulus (Ef) and strength (σf) of the unfilled PLA piece were 3.17 GPa and nearly 99 MPa, respectively. Similar to the tensile properties, the incorporation of TCHF yielded a progressive increase in Ef and a reduction in σf. In particular, the green composite pieces filled with 50 wt% of TCHF showed a Ef value of 5.53 GPa, which is approximately 1.7 times higher than that of the unfilled PLA, whereas σf was reduced to 61.42 MPa. Some previous studies have reported similar mechanical properties of green composites based on coffee waste derived fillers. For instance, Dominici et al. [58] showed a worsening of tensile strength and ductility when CSS was incorporated into bio-based high-density polyethylene (bio-HDPE). The described decrease of the mechanical properties was ascribed to the low aspect ratio of the CSS particles as well as their poor adhesion to the biopolymer matrix. In this regard, the use of coupling agents has improved both the tensile strength and ductility of PLA composites prepared with SCG, though the improvements achieved were relatively low and their overall and specific migration limits (OML and SML) should also be carefully determined [59].
One can also observe that, as the TCHF content increased, hardness also increased. The Shore D hardness value increased from 78.65, for the unfilled PLA piece, to 85.05 for the PLA/50TCH piece. This change in the mechanical behavior corresponds to the typical effect of a relatively stiff lignocellulosic filler on a polymer matrix [60]. In contrast to hardness, the impact absorbed energy of PLA progressively decreased after the TCHF incorporation. The impact energy of the neat PLA piece was 25.10 kJ/m 2 and it was reduced down to 7.19 kJ/m 2 for the green composite piece filled with 50 wt% of TCHF. This notable decrease in toughness can be related to the poor matrix-particle interactions, which could promote microcrack formations and stress concentrations during the impact test [60]. Therefore, the addition of the coffee waste derived lignocellulosic filler led to a more elastic and stiffer material, but the matrix-filler continuity was not achieved, and an early fracture was attained. Thus, in terms of mechanical performance, PLA pieces containing 20 wt% of TCHF were slightly more resistant but less ductile, showing characteristics comparable to those of neat PLA. Figure 3 shows the visual aspect of the injection-molded pieces of PLA, PLA filled with 20 wt% of CHF and TCHF, that is, PLA/20CHF and PLA/20TCHF, respectively. The increase in the characteristic dark brown color produced in the pieces of the green composites is due to the Maillard reaction of coffee husk, which took place during the torrefaction and its subsequent melt processing with PLA, leading to formation of melanoidins [35].   The fracture surfaces after the impact tests of the neat PLA piece and the PLA/CHF and PLA/TCHF pieces at different filler contents are shown in Figure 4. The unfilled PLA piece, shown in Figure 4a, presented a typical brittle fracture with low plastic deformation [61]. In Figure 4b, one can see that the incorporation of untreated CHF led to a similar fracture surface but yet softer, indicating that the toughness of the piece was reduced. One can also observe the formation of a large gap between the coffee waste derived lignocellulosic fillers and the biopolymer PLA matrix, which points to their weak interactions [62,63]. The morphology of the green composite based on 20 wt% of TCHF, shown in Figure 4c, changed to a rougher surface, also showing a reduction in the filler-to-matrix gaps. However, no enhanced yielding was detected, which is in agreement with the ductility decrease reported in Table 2. In the case of the green composites prepared at higher TCHF contents, shown in Figure 4d-f, one can see that the fracture surface roughness increased with the filler addition. Moreover, the FESEM micrographs displayed some large voids, microcracks, and gaps in the interface between both components, that is, PLA and TCHF. This phenomenon led to a stiffer material, but with reduced impact absorbed energy, as also shown during the mechanical analysis. One can also observe in the FESEM micrographs that the PLA/TCHF pieces prepared at the highest filler contents, that is, 40 wt% and 50 wt%, showed the co-presence of single particles and irregular aggregates, being distributed randomly within the PLA matrix. Similar observations were described for pectin/coffee ground composites in which the formation of aggregates was ascribed to the prevalence of particle-particle interactions over the forces between the composite components [64].

Aspect and Morphology of the Green Composite Pieces
Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 18 The fracture surfaces after the impact tests of the neat PLA piece and the PLA/CHF and PLA/TCHF pieces at different filler contents are shown in Figure 4. The unfilled PLA piece, shown in Figure 4a, presented a typical brittle fracture with low plastic deformation [61]. In Figure 4b, one can see that the incorporation of untreated CHF led to a similar fracture surface but yet softer, indicating that the toughness of the piece was reduced. One can also observe the formation of a large gap between the coffee waste derived lignocellulosic fillers and the biopolymer PLA matrix, which points to their weak interactions [62,63]. The morphology of the green composite based on 20 wt% of TCHF, shown in Figure 4c, changed to a rougher surface, also showing a reduction in the filler-to-matrix gaps. However, no enhanced yielding was detected, which is in agreement with the ductility decrease reported in Table 2. In the case of the green composites prepared at higher TCHF contents, shown in Figure 4d-f, one can see that the fracture surface roughness increased with the filler addition. Moreover, the FESEM micrographs displayed some large voids, microcracks, and gaps in the interface between both components, that is, PLA and TCHF. This phenomenon led to a stiffer material, but with reduced impact absorbed energy, as also shown during the mechanical analysis. One can also observe in the FESEM micrographs that the PLA/TCHF pieces prepared at the highest filler contents, that is, 40 wt% and 50 wt%, showed the co-presence of single particles and irregular aggregates, being distributed randomly within the PLA matrix. Similar observations were described for pectin/coffee ground composites in which the formation of aggregates was ascribed to the prevalence of particle-particle interactions over the forces between the composite components [64].  Figure 5 shows the DSC thermograms of PLA and its green composites with CHF and TCHF. Table 3 gathers the main thermal properties obtained from the cooling and the second heating scans, once the thermal history was removed from the samples. The neat PLA sample showed a glass transition temperature (Tg) of 62.3 °C and the incorporation of TCHF led to a slight decrease in the values up to 60.7 °C for the green composite filled at 50 wt%. In all curves, a low-intense endothermic peak was observed, which can be ascribed to the aging process of the biopolyester [42]. It can also be observed in Figure 5a that, during cooling, the PLA samples did not crystallize from the melt. In contrast, in Figure 5b, one can see that the neat PLA samples showed an exothermic peak centered at 122.7 °C for neat PLA, which is ascribed to the biopolyester chain rearrangement during cold  Figure 5 shows the DSC thermograms of PLA and its green composites with CHF and TCHF. Table 3 gathers the main thermal properties obtained from the cooling and the second heating scans, once the thermal history was removed from the samples. The neat PLA sample showed a glass transition temperature (T g ) of 62.3 • C and the incorporation of TCHF led to a slight decrease in the values up to 60.7 • C for the green composite filled at 50 wt%. In all curves, a low-intense endothermic peak was observed, which can be ascribed to the aging process of the biopolyester [42]. It can also be observed in Figure 5a that, during cooling, the PLA samples did not crystallize from the melt.

Thermal Performance of the Green Composite Pieces
In contrast, in Figure 5b, one can see that the neat PLA samples showed an exothermic peak centered at 122.7 • C for neat PLA, which is ascribed to the biopolyester chain rearrangement during cold crystallization. The green composites showed slightly higher values of both cold crystallization temperature (T cc ) and cold crystallization enthalpy (∆H cc ), showing the highest values for TCHF contents of 20 wt% and 30 wt%, respectively. The cold crystallization process is an exothermic process that takes place when the biopolymer chains become flexible, above PLA's T g , which is caused by a primary nucleation, by which an ordered state is achieved. Since more PLA mass could cold crystallize, one can consider that intermediate filler amounts favored crystallization of PLA. A second endothermic peak, corresponding to the melting temperature (T m ), was observed at 151.3 • C for neat PLA, which slightly increased for filler contents of 20 wt% and then decreased for higher contents. An anticipation of the melting phenomena was also observed in HDPE by Arrigo et al. [65] after the addition of biochar particles derived from waste coffee grounds. This behavior was explained by the presence of specific interactions occurring between the biopolymer chains and the particle interfaces. In fact, the lower T m values attained at the highest contents, that is, 40 wt% and 50 wt% of TCHF, in combination with the higher ∆H cc values, suggest that the biopolymer chains were immobilized onto the surface of the lignocellulosic fillers. The maximum degree of crystallinity, that is, X C_max , which does not consider the crystals that are formed during cold crystallization [50], increased progressively up to a TCHF content of 30 wt% and then slightly lower values were observed. This result indicates that the lignocellulosic fillers induced a nucleating effect that favored the formation of crystals in the green composites. In the case of the lower X C_max values for the green composites containing 40 wt% and 50 wt% of TCHF, one can consider that the fillers presence at these contents also supposed an obstacle for the biopolymer chains to form stable crystals, generating a negative effect on crystallinity and leading to the formation of a more amorphous structure [66][67][68]. Furthermore, one can also observe that the torrefaction process did not yield any significant effect on the thermal properties of PLA at the studied filler loading, suggesting that nucleation was based on a physical phenomenon.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 9 of 18 crystallization. The green composites showed slightly higher values of both cold crystallization temperature (Tcc) and cold crystallization enthalpy (∆Hcc), showing the highest values for TCHF contents of 20 wt% and 30 wt%, respectively. The cold crystallization process is an exothermic process that takes place when the biopolymer chains become flexible, above PLA's Tg, which is caused by a primary nucleation, by which an ordered state is achieved. Since more PLA mass could cold crystallize, one can consider that intermediate filler amounts favored crystallization of PLA. A second endothermic peak, corresponding to the melting temperature (Tm), was observed at 151.3 °C for neat PLA, which slightly increased for filler contents of 20 wt% and then decreased for higher contents. An anticipation of the melting phenomena was also observed in HDPE by Arrigo et al. [65] after the addition of biochar particles derived from waste coffee grounds. This behavior was explained by the presence of specific interactions occurring between the biopolymer chains and the particle interfaces.
In fact, the lower Tm values attained at the highest contents, that is, 40 wt% and 50 wt% of TCHF, in combination with the higher ∆Hcc values, suggest that the biopolymer chains were immobilized onto the surface of the lignocellulosic fillers. The maximum degree of crystallinity, that is, XC_max, which does not consider the crystals that are formed during cold crystallization [50], increased progressively up to a TCHF content of 30 wt% and then slightly lower values were observed. This result indicates that the lignocellulosic fillers induced a nucleating effect that favored the formation of crystals in the green composites. In the case of the lower XC_max values for the green composites containing 40 wt% and 50 wt% of TCHF, one can consider that the fillers presence at these contents also supposed an obstacle for the biopolymer chains to form stable crystals, generating a negative effect on crystallinity and leading to the formation of a more amorphous structure [66][67][68]. Furthermore, one can also observe that the torrefaction process did not yield any significant effect on the thermal properties of PLA at the studied filler loading, suggesting that nucleation was based on a physical phenomenon.
(a) (b)   Table 3. Thermal properties of the injection-molded pieces of polylactide (PLA) with coffee husk flour (CHF) and torrefied coffee husk flour (TCHF) in terms of glass transition temperature (T g ), cold crystallization temperature (T cc ), melting temperature (T m ), cold crystallization enthalpy (∆H cc ), melting enthalpy (∆H m ), and maximum degree of crystallinity (X C_max ). The TGA and first derivate thermogravimetric (DTG) curves of the injection-molded pieces of PLA and its green composite pieces with CHF and TCHF are shown in Figure 6, whereas Table 4 shows the relevant values extracted from the curves. In relation to the thermal degradation profiles of CHF and TCHF, in both cases they presented three stages; however, the torrefied fillers showed significantly higher stability. The first mass loss, which occurred below 100 • C, being in both cases less than 5% but slightly higher for CHF, is associated with the evaporation of moisture and volatiles present in the lignocellulosic fillers. The second one occurred between approximately 240 and 380 • C, was the most intense, and was associated with a mass loss of over 60% and 35% for CHF and TCHF, respectively. This degradation stage, which was also more intense for CHF, is mainly caused by the thermal decomposition of the molecular bonds of hemicellulose, cellulose, and small amounts of lignin, which overlapped with the thermal degradation of PLA. The third stage started at 385 • C and continued for the whole thermal range tested. This last mass loss is due to the degradation of the lignin fraction present in CHF and TCHF [11,69]. In this regard, de Brito et al. [70] showed that green coffee cake (GCC), a solid by-product of the coffee processing industry, resulting from the extraction of oil by the cold pressing technique, also presented three regions of mass loss, similar to other lignocellulosic biomasses. The first one was observed between 50 and 100 • C, the second one between 200 and 300 • C, and the last one between 275 and 400 • C. The authors ascribed the second and third mass losses to hemicellulose depolymerization in combination with temperature-induced glycosidic cleavage and to lignin components and cellulose degradation, respectively. Finally, after the degradation steps, a significant residue of approximately 25% and 40% was found for CHF and TCHF, respectively, remarking the higher stability of the carbon formed during the torrefaction step. The TGA results also indicated that TCHF was thermally stable up to more than 250 • C, which makes it useful as a reinforcing agent in most WPCs.
In the case of PLA, it can be observed that the biopolymer degraded in a quick step process at approximately 350 • C. The main products generated during the thermal degradation of PLA have been ascribed to lactic acid, oligomers of lactic acid (OLAs), acetaldehydes, carbon dioxide, carbon monoxide, and ketones [60]. The TGA curves of the PLA/TCHF pieces showed a slight increase in the thermal stability for the lowest filler content. In particular, the DTG curves showed an increase in the pyrolysis peak related to the degradation temperature (T deg ) from 347.4 • C, for neat PLA, up to 354.7 • C, for the PLA/20TCHF composite. The increase in the thermal stability of the green composite pieces can be related to the "barrier" effect of the decomposition of the biopolymer material by the lignocellulosic fillers, which promoted an increase of the degradation temperature [70]. However, a decrease in the temperature at the mass loss of 5% (T 5% ), which corresponds to the onset degradation temperature, was observed for TCHF contents above 30 wt%. The lower T 5 % values attained at higher filler contents can be related to the "chimney" effect of the coffee waste lignocellulosic fillers, which prevails their "barrier" effect, and favored the movement of heat in the PLA matrix [57]. In particular, T 5% was reduced from 320.4 • C, for the unfilled PLA piece, down to approximately 300 • C for TCHF contents of 40 wt% and 50 wt%. Therefore, a slight improvement in the PLA's thermal stability was attained with the addition of 20 wt% of TCHF, whereas a negative effect was observed for the pieces filled with higher TCHF contents. It is noteworthy that a higher thermal stability was observed in comparison to the PLA composites prepared with CSS by Totaro et al. [71], who ascribed the decrease to both the metals present and the moisture released by the fillers, which can catalyze degradation and induce hydrolysis of the ester groups, respectively. Since lower thermal stability values were attained for the PLA/20CHF piece, the improvement attained herein is an indication that the less thermally stable compounds, such as hemicellulose and cellulose, were successfully removed during the torrefaction step [33], leading to a material with higher lignocellulosic content and, hence, improved thermal stability. Furthermore, as expected, one can observe that adding TCHF caused the char residue to increase. Such high char content is also related to the torrefaction step, which can generate a higher percentage of carbon compared with the non torrefied lignocellulosic filler [33].
Appl. Sci. 2020, 10, x FOR PEER REVIEW 11 of 18 comparison to the PLA composites prepared with CSS by Totaro et al. [71], who ascribed the decrease to both the metals present and the moisture released by the fillers, which can catalyze degradation and induce hydrolysis of the ester groups, respectively. Since lower thermal stability values were attained for the PLA/20CHF piece, the improvement attained herein is an indication that the less thermally stable compounds, such as hemicellulose and cellulose, were successfully removed during the torrefaction step [33], leading to a material with higher lignocellulosic content and, hence, improved thermal stability. Furthermore, as expected, one can observe that adding TCHF caused the char residue to increase. Such high char content is also related to the torrefaction step, which can generate a higher percentage of carbon compared with the non torrefied lignocellulosic filler [33].  10.9 ± 0.8 Table 5 shows the VST and HDT values of the injection-molded pieces of neat PLA and PLA/TCHF, which are important thermomechanical parameters that can be used to estimate their upper useful temperature limit for uses in, for instance, packaging or disposables. VST, which reflects the material's softening by temperature rises, increased progressively from 56.6 °C, for neat PLA, to 59.9 °C, for PLA/50TCH, respectively, following an evolution similar to that found in hardness. This result is in agreement with the study of Huang et al. [52], who showed that the VST of HDPE increased with the loading of coffee husk fiber (CHF), which was attributed to a fiber reinforcement that prevented deformation of the composites. Concerning HDT, the values evolved in a similar manner to VST, showing a slight and progressive increase from 51 °C, for neat PLA, to 54.1 °C for  Table 4. Main thermal degradation parameters of the injection-molded pieces of polylactide (PLA) with coffee husk flour (CHF) and torrefied coffee husk flour (TCHF) in terms of temperature at mass loss of 5 % (T 5% ), degradation temperature (T deg ), and residual mass at 700 • C.  Table 5 shows the VST and HDT values of the injection-molded pieces of neat PLA and PLA/TCHF, which are important thermomechanical parameters that can be used to estimate their upper useful temperature limit for uses in, for instance, packaging or disposables. VST, which reflects the material's softening by temperature rises, increased progressively from 56.6 • C, for neat PLA, to 59.9 • C, for PLA/50TCH, respectively, following an evolution similar to that found in hardness. This result is in agreement with the study of Huang et al. [52], who showed that the VST of HDPE increased with the loading of coffee husk fiber (CHF), which was attributed to a fiber reinforcement that prevented deformation of the composites. Concerning HDT, the values evolved in a similar manner to VST, showing a slight and progressive increase from 51 • C, for neat PLA, to 54.1 • C for PLA/50TCHF. This enhanced thermomechanical resistance can be attributed to the higher modulus and amount of crystallinity, since both parameters increased with increasing TCHF content. Similar results were obtained by Sarasini et al. [72], who showed that the softening point of PBAT/poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) blends increased up to 3 • C with the incorporation of CSS. Table 5. Thermomechanical properties of the injection-molded pieces of polylactide (PLA) with torrefied coffee husk flour (TCHF) in terms of Vicat softening point (VST) and heat deflection temperature (HDT).

Piece
VST (  Finally, Figure 7 shows the mechanical performance of the green composite pieces under dynamic load conditions as a function of temperature. Figure 7a displays the evolution of the storage modulus with temperature. In the temperature range between 30 and 50 • C, below the glass transition region, the incorporation of TCHF produced an increase in the storage modulus of PLA. This behavior indicates that the addition of the TCHF filler led to a stiffer material since the lignocellulosic particles restricted the biopolymer chain mobility and also promoted crystallization as described above during the thermal characterization. From approximately 55 • C, a prominent decrease in the storage modulus due the glass transition process can be seen [41]. In addition, from nearly 90 • C, an increment in the storage modulus was detected. This phenomenon is associated with the cold crystallization of PLA. During this process, the biopolymer chains reorganize into a more compact structure, leading to an increase in the storage modulus [73]. Furthermore, as more TCHF powder was added, in agreement with the previous DSC analysis, a slight decrease in the cold crystallization temperature was noticed.

Conclusions
The results achieved in this study demonstrate that coffee husk, a large food processing byproduct, can be used in the form of flour as a reinforcing filler in green composites. Torrefaction, a mild form of pyrolysis at 250 °C in reduced atmosphere, yielded more thermally stable biomass that was also more hydrophobic and, hence, more compatible with PLA. Characterization of the injectionmolded PLA/TCHF pieces showed a positive effect on the overall mechanical improvement and a thermal stability enhancement with the addition of 20 wt% of TCHF. However, the green composites  Figure 7b displays the evolution as a function of temperature of the damping factor (tan δ) for the unfilled PLA piece and the PLA/TCHF composite pieces. The peak for the neat PLA progressively shifted to lower values as more filler was added. It particularly decreased from 68.7 • C, for the neat PLA piece, down to values in the 62-64 • C range. This peak is ascribed to the alpha (α)-transition of PLA, being in relation to the PLA's T g . One can expect that with a greater interaction between filler and biopolymer matrix, which can be achieved by modifying the fiber surface during the torrefaction step, there would be a restriction on the movement of the macromolecular chains located on the surface of the fillers and, hence, an increase in T g [74]. However, the reduction attained in T g , also observed by DSC, suggests that the fillers allowed an increase in the free volume in the green composites, thus indicating low compatibility or poor interaction with PLA [70]. This behavior confirms that the movement of the amorphous PLA was promoted by the coffee waste derived lignocellulosic fillers and it suggests that shorter biopolymer chains were formed due to a chain scission. Furthermore, the nucleating effect of TCHF was further corroborated by the gradual reduction in the tan δ peak intensities of the green composite pieces. The latter indicates that the number of molecules experimenting glass transition was reduced by the fillers' presence [75].

Conclusions
The results achieved in this study demonstrate that coffee husk, a large food processing by-product, can be used in the form of flour as a reinforcing filler in green composites. Torrefaction, a mild form of pyrolysis at 250 • C in reduced atmosphere, yielded more thermally stable biomass that was also more hydrophobic and, hence, more compatible with PLA. Characterization of the injection-molded PLA/TCHF pieces showed a positive effect on the overall mechanical improvement and a thermal stability enhancement with the addition of 20 wt% of TCHF. However, the green composites filled with 30 wt%, 40 wt%, and 50 wt% showed lower performance, although they are still interesting due to the high amount of waste that can be valorized. Both elasticity and hardness increased, whereas ductility and toughness decreased, for all of the TCHF contents due to the lack of interaction as observed by FESEM. In contents of up to 30 wt%, the lignocellulosic fillers induced a nucleating effect that favored the formation of PLA crystals and showed sufficient thermal stability. Furthermore, the addition of TCHF led to slightly higher VST and HDT values and the thermomechanical resistance of PLA was improved, meaning the resultant pieces were able to withstand temperatures of up to nearly 60 • C. In a Circular Bioeconomy context, the resultant PLA/TCHF composite pieces prepared with 20 wt% of the torrefied coffee waste derived filler can be used for rigid packaging, cups and containers, and food contact disposables, such as cutlery and plates, where novel compostable and sustainable materials with high stiffness and low thermal resistance are required.