3.2.1. Tensile Properties of 3D Printed Samples
Overall, all the filaments had good printability with no signs of nozzle blockage (clogging) during printing. Before printing the samples used for tensile testing, DMA, and XRD, the multiplier factor of the 3D printer software was adjusted to achieve acceptable surface and dimensional quality (as shown in
Figure S4). Although the smooth surface of lyocell fibres and their short length after processing negatively affect the mechanical properties of the filaments, these characteristics are beneficial for printing, which enabled the use of small nozzles for printing. The filament with 10% of fibres could be printed with nozzles down to 0.25 mm with no apparent problems. However, in this work, we only report the mechanical test results of samples printed with a 0.5 mm nozzle, since with the smaller nozzle, clogging started to occur with the L20% and L30% formulations.
The resulting tensile properties of 3D printed samples are given in
Table 5. Representative stress/strain curves of all the formulations can be visualized in
Figure S5. The first thing to notice is that the ultimate tensile strength and Young’s modulus of the composites with 10% and 20% of fibres improved with the printing process. Since the samples were printed with a raster angle of 0°, which is parallel to the loading direction, this effect is attributed to increased fibre alignment caused by the shear stress during the printing process (better discussed in the next section) and possibly to PLA molecular alignment. The fibre alignment in fibre-reinforced composites plays an important role in fibre/matrix stress transfer and is a well-known method to improve mechanical properties, especially tensile strength [
52].
The resulting fracture surfaces of 3D printed PLA and the L10% composite are presented in
Figure 6. The fracture surfaces revealed a well-consolidated cross-section, with the presence of inter-bead porosity (voids between layers), that are usually observed in 3D printed samples by FDM [
16,
53]. In the L10% sample, it is possible to observe a combination of fibre pull-out and fibre breakage, indicated by the white and yellow arrows, respectively, in
Figure 6. In fibres that are longer than the critical fibre length, the interfacial forces between fibre and matrix are high enough to prevent fibre pull-out, leading to fibre fracture. This load transfer between fibre and matrix up to fibre breakage improves Young’s modulus and tensile strength of the composite. On the other hand, part of the fibres will be shorter than the critical fibre length and will be prone to pull-out, especially if the smooth surface of the lyocell fibres is taken into consideration, which is translated to limited load transfer between fibre and matrix.
By comparing the tensile test results of the PLA filament and the 3D printed sample, it is possible to observe that the printing process resulted in high-quality samples, with minimal influence of printing defects on the tensile properties. The formulation containing 10 wt% fibres had the highest overall mechanical performance, achieving an ultimate tensile strength, Young’s modulus, and strain at break of 64.2 MPa, 4.56 GPa, and 4.93%, respectively. Although the tensile strength is slightly higher than neat PLA, the difference is not statistically relevant. The Young’s modulus, however, had a significant increase of 40% in comparison with neat PLA. Interestingly, the strain at break also increased with the addition of 10% of fibres and resulted in values with lower standard deviation. Tekinalp et al. (2019) reported similar tensile properties for their 3D printed samples using PLA reinforced with 10% of nanofibrillated cellulose (UTS ≈ 61–64 MPa; YM ≈ 4.2–4.8), but with strain at breaks below 2.5%, which is lower than that of the L10% composite [
16].
The properties of the 3D printed lyocell/PLA composites are also superior to those of a commercial wood/PLA filament with 10% of fibres (values given in
Table 5) printed with the same conditions. Nevertheless, the samples with 30% of fibres presented lower values of tensile strength, Young’s modulus and strain a break in comparison with the filament counterpart and 3D printed PLA. These results are attributed to the high porosity of the filaments and high-stress concentrations in regions outside the free-span which resulted in premature brittle failures. Therefore, only the L10% and L20% formulations were used for the DMA and XRD analyses presented in the next sections.
3.2.2. Fibre Alignment in 3D Printed Samples
As discussed in the previous section, some of the 3D printed samples had better tensile properties than the filaments. This difference is attributed to the increased alignment of the fibres with the printing direction caused by the shear stresses involved during the printing process. This effect is clearly observed in
Figure 7, where optical microscopy images of the filament and 3D printed sample of L10% formulation are presented. In the filament, the fibres are partially aligned with the extrusion direction, but due to the high-intensity mixing during the twin-screw extrusion, part of the fibres have a random orientation. However, when 3D printing is used, the shear stresses in the nozzle promote the alignment of the fibres in the printing direction. The shear rate in the nozzle during the printing process can be calculated according to Equations (S1)–(S4), shown in the
Supplementary Information file.
In order to evaluate and quantify the alignment of the fibres during the printing process, XRD analysis was conducted in transmission mode. By using this method, it is possible to have a comprehensive volumetric analysis of the samples, complementing traditional 2D image analysis. The Herman’s parameter and degree of ordering were calculated for samples 3D printed with different printing speeds, and hence extrusion shear, to evaluate if printing speed may improve the alignment of the fibres. A summary of the obtained results is given in
Table 6. A complete description of the method and Equations used to calculate Herman’s parameter and degree of ordering is given in the
Supplementary Information file (Equations (S5) and (S6)).
The degree of ordering and Herman’s parameter are used to describe how well-aligned cellulose crystallites are in a defined direction. High values of degree of ordering (π) (up to 100%) and Herman’s parameters (
f) close to 1 indicate a maximum orientation, i.e., all the crystallites of the analysed phase are aligned. Low values of π and
f = 0 indicate random orientation. From the values presented in
Table 6, it is possible to observe that all the samples analysed had high values of π (average of 79%) and
f above 0.6. These results corroborate the micrographs shown in
Figure 6, where the fibres are aligned with the printing direction. However, the printing speed used in this study and the corresponding shear rate induced in the nozzle, does not seem to have a relevant influence on the alignment of the fibres, even with almost a three-fold difference. It is hypothesised that the dimensions of the fibres are sufficiently large for the shear rate induced by 0.5 mm nozzle to align the fibres during printing, even at relatively low printing speeds. The induced alignment of the fibres in the composites using FDM can be used to manufacture parts with controlled directional stiffening, which is an advantage in comparison with other manufacturing methods.
3.2.3. Thermo-Mechanical Stability and DSC Analysis
Reinforcing PLA with fibres generally improves its thermo-mechanical properties, which can broaden its use in applications where thermo-mechanical stability is necessary [
16,
54,
55,
56]. The storage modulus (
E’) and
tan δ curves of neat PLA and composites with 10 and 20 wt% of fibres are presented in
Figure 8a. The storage modulus curves have three distinct regions [
54]. In the first one, between room temperature and 50 °C the materials are in a glassy state (below the
Tg) and are characterized by a linear plateau in this temperature range. In the transition region, between 50–70 °C, there is a sharp drop in the storage modulus, and the materials are transitioning from a glassy state to a rubbery state. Above approximately 70 °C the PLA and the composites are characterised by a rubbery behaviour with minimum storage modulus.
Tan δ is the ratio between loss modulus (
E’’) and storage modulus (
E’) and is used to identify the glassy to rubbery transition. The temperature with the highest tan δ value is often stated as the glass transition (
Tg) temperature. When the lyocell fibres were added, the intensity of tan δ peak was considerably reduced, which indicates less viscous behaviour and better thermomechanical stability. In addition, above approximately 90 °C, it is possible to observe a slight increase in the storage modulus in all the samples (see insert of
Figure 8a) related to the beginning of crystalisation of PLA. This effect starts at lower temperaure in the composites, which demonstrates that the addition of the fibres facilitates PLA crystallisation.
The presence of crystalline regions decreases the mobility of PLA molecules. Therefore, an increase in PLA crystallinity is translated into higher storage modulus at higher temperatures improving thermal-mechanical stability. Additional samples were heat-treated at 105 °C for 2 h to evaluate the effect of the increase in PLA crystallinity in the thermo-mechanical behaviour of the 3D printed samples. The samples with 10 and 20 wt% of fibres did not have dimensions affected (below 1%) after heat treatment. The 3D printed PLA sample, on the other hand, had a decrease of 13 % in length and an increase of 5% and 13% in width and thickness, respectively (see
Figure S7). This effect is expected in some grades of PLA and is more evident and anisotropic in 3D printed samples.
The storage modulus and tan δ curves of the heat-treated samples are presented in
Figure 8b. The heat-treated samples had a considerable improvement in the thermo-mechanical stability, characterised by the lower values of
tan δ and higher storage modulus above the transition temperature. The heat-treated neat PLA sample was tested only as a reference, but it is worth mentioning that this type of treatment would not be possible for 3D printed objects using this unreinforced PLA grade since the anisotropic shrinkage/swelling caused by the increase in crystallinity would cause considerable changes in the object’s dimensions. This is considered a major advantage of using fibres in PLA, enabling post-printing heat treatment to increase PLA crystallinity while preserving 3D object shape and dimensions.
The overall properties obtained from the DMA tests of all the specimens (as printed and heat-treated) are summarised in
Table 7. First, there is a continuous decrease in the maximum
tan δ values with the addition of the fibres. In the heat-treated samples, a similar trend is observed, but with lower absolute values. The maximum
tan δ temperature and loss modulus (
E’’) peak temperature were not affected by the addition of the fibres, around 67 and 61 °C, respectively. Nevertheless, when heat-treated all the samples had an increase in the
tan δ and
E’’ peak temperatures, 72 and 65 °C, respectively. The storage modulus (
E’) at different temperatures is also given in
Table 7. At 30 °C, the composite samples have a gradual increase in E’ with the addition of fibres and are characterised by a more pronounced difference between the as-printed and heat-treated samples. This effect is more evident in the values of E’ at higher temperatures. At 80 °C, the
E’ of heat-treated neat PLA is 22.5× higher than the sample without treatment, while for the L10% and L20% this difference is 64.7× and 40×, respectively. Additionally, the
E’ at 80 °C of heat-treated L10% and L20% samples are 60.6× and 72.5× higher than the
E’ of neat PLA, respectively. It was demonstrated that the thermo-mechanical stability of the composites with lyocell fibres could be drastically improved by post-printing heat-treatment without any hurdles associated with dimensional changes.
The 3D printed samples used for the DMA tests were also analysed by DSC to verify the crystallinity of PLA and thermophysical properties of the composites and complement the information given in
Table 7. Representative DSC curves during heating of the as-printed and heat-treated samples are presented in
Figure 9. In
Figure 9a it is possible to observe that all the samples have a typical DSC curve for the 2003D PLA grade, with an endothermic
Tg peak at 62–64 °C, exothermic
Tcc (cold crystallisation temperature) peak between 114–123 °C and endothermic Tm peak between 150–153 °C. In contrast, with heat treatment, only the Tm peaks can be observed. Since most of the available PLA chains are already crystalline, there are no signs of
Tcc in these samples.
A summary of all the results is shown in
Table 8. First of all, as observed in the DMA results, the
Tcc temperature is lower for the composite materials which is an indication of the influence of the fibres on the crystallisation behaviour of PLA. However, after 2 h at 105 °C all the samples had similar PLA crystallinity, between 28–29%. It is worth mentioning that this PLA grade (2003D) has higher D-isomer content than other PLA grades and as an injection moulding grade, it is more difficult to crystallise than other grades [
57]. In fact, during cooling at 10 °C/min (same rate used for heating in the DSC analysis), this grade does not have any signs of crystallisation. The endothermic
Tg peaks had little influence by the addition of fibres and although the
Tg peak of the composites is slightly lower than neat PLA, the starting temperature of the transition is more or less the same for all the samples. The Tg temperatures obtained by DSC are in agreement with the DMA results and are between the
tan δ and
E’’ values (shown in
Table 7). In addition, the heat-treated L10% and L20% composites present a bimodal melting endotherm (
Figure 9b) that is attributed to the melting of the metastable α form of PLA (α’) potentially formed during 3D printing [
53,
57,
58,
59,
60]. The first peak of fusion is related to the simultaneous melting of the primary α crystals formed during printing and recrystallisation of α’ to α crystal form. The second peak of fusion is attributed to the melting of α crystal form generated in the recrystallisation process [
61].