The Influence of Processing Temperature on the Tensile Properties of Melt-Spun PLA Fibres and their Self-Reinforced Composites

The temperature during the consolidation of self-reinforced PLA composites has a significant effect on their mechanical properties. In the present work, the effect of the consolidation or processing temperature on the properties of melt-spun, highly oriented PLA fibres is experimentally studied through single fibre tests. It is shown that the Young’s modulus and strain to failure of PLA fibres increases with exposure to consolidation / processing temperature, and the strength decreases more drastically. Using these data and findings from earlier studies, it is demonstrated that the dependence of the tensile properties of self-reinforced PLA composites on the processing temperature can be directly predicted from the single PLA fiber properties as a function of the processing temperature. This prediction holds true provided that the tensile properties of both the PLA fibers and self-reinforced PLA composites are measured using the same cross-head speed or strain rate.


Introduction
As awareness for sustainable and environmentally friendly materials grows, bioplastics are gaining increasing interest.Polylactic acid (PLA), derived from renewable resources such as corn or sugar beets, is perhaps the most widely used bioplastic / biopolymer.It has relatively attractive mechanical properties compared to the other bioplastics, it is biodegradable and the industrial production capacity is high [1].PLA, however, is primarily used as disposable or packaging material [2] due to its lower mechanical properties when compared to petro-based polymers.
One way to improve the mechanical properties of neat PLA is to develop PLA composites by adding reinforcement such as cellulose in various forms [3][4][5][6].The addition even of bio-based reinforcement does not, however, necessarily means that the environmental performance of these composites is equal to that of neat PLA i.e. the eco-performance of the composites may be lower [7][8][9].
An alternative approach is based on the concept of self-reinforced composites, where the matrix is reinforced with oriented fibres made from the same polymer but with a higher melting temperature [10][11][12].The first self-reinforced composites were based on polyethylene [13] and later the concept was extended to various others material systems such as polypropylene [14][15][16][17], polyamide [18], aramid [19] and cellulose [20][21][22].However, it is only recently that limited number of papers have applied this concept to develop self-reinforced PLA composites [9,[23][24][25] for fully biodegradable composites based on renewable resources.
The mechanical properties of self-reinforced composites can be enhanced by increasing the orientation of the polymer chains in the reinforcement [26] i.e. develop fibres of high stiffness and strength.Oriented PLA fibres produced by solution spinning can have very high properties, such as a tensile strength of 1 GPa [27] or 2.1 GPa with a Young's modulus of 16 GPa [28].Commercial PLA fibres, however, are, typically produced by melt spinning to achieve higher production rates and to avoid the use of solvents [2].Melt-spun PLA fibres have lower properties; the reported strength varies from 280 to 870 MPa and the Young's modulus from 5.2 to 9.2 GPa [2,[29][30][31][32].The wide range of strength and modulus values reported originates from the strong dependence of the polymer fibre properties on: a) the PLA grade used, b) the extrusion process parameters, and c) the drawing parameters (Fig. 1a).Optimisation of the manufacturing parameters has been an active topic of research including PLA e.g.[2,25,27].
After the manufacturing of the fibers, the production of self-reinforced composites typically involves hot press-consolidation [9,25].This process involves heating of the materials to melt the PLA that will form the matrix but not the PLA fibres.However, during this process, the PLA fibres are also heated and their properties will be significantly changed.Therefore, the fibre properties that determine the properties of the self-reinforced PLA composites are in general quite different from the properties usually measured (after drawing but before composite processing).The aim of the present work is to study the influence of self-reinforced composite processing conditions on the PLA fibre properties and subsequently on the self-reinforced PLA composites.The paper is structured as follows.First a statistical analysis of the fibre strength distribution is performed using various gauge lengths and testing speeds at room temperature.Then, the fibre properties are measured after subjecting them to different process temperatures at different testing speeds for a fixed gauge length.It should be emphasised that the draw ratio of the PLA fibres is kept constant.The results are then discussed in terms of their implications on composite properties.Lastly, the major findings are summarized.

Materials
A high melting temperature and medium flow homopolymer PLA (L130 PLA grade) was purchased from Total Corbion PLA.The PLA material was first compounded and then melt spun into multi-fibre yarns (Fig. 1a) using a pilot line.The draw ratio was equal to 2.76.More details about the manufacturing process of the PLA yarns can be found elsewhere [33].The focus of the current work is on the effect of the composite process parameters (to fabricate the self-reinforced PLA composites) on the PLA fibre properties and subsequently on the composite properties i.e. the starting point is after the PLA fibres have been melt-spun and drawn from the PLA compound.
After the PLA yarns were produced (Fig. 1a), the yarns were wound from the bobbins in a metal frame (see Fig. 1b) using a custom made winding machine.Subsequently, the frame was placed overnight in an oven under vacuum and temperature equal to 35 • C to dry the PLA fibres.
The PLA yarns were then press-consolidated in a custom-made press facility using a two-step process.In the first section of the press facility, the frame with the wound PLA yarns was heated for 10 min under vacuum to the process temperature, T p .The heating was applied by the contact of two metal plates as shown in Fig. 1b).Subsequently, in the second step, the frame was quickly moved to the second section of the press facility and cooled down to 30 • C within 1 min under a pressure of 2 MPa.These process parameters are the same with the parameters used to manufacture self-reinforced PLA composites [25,33,34] and thus approximate the conditions that the PLA fibres experience during composite manufacturing.The process temperature, T p , was varied from 155 • C to 180 • C in increments of 5 • C. In the process window from 155 • C to 175 • C, the PLA fibres do not melt, whereas at T p = 180 • C the PLA fibres melt.It should be noted that in the present study there were fewer fibres in the frame compared to the case of composite manufacturing [25,33,34] and the pressure that the fibres experience may be less than 2 MPa.However, it can be argued that the potential difference in the applied pressure does not have a significant effect on the reported results.

Single Fibre Mechanical Testing
After press-consolidation, individual PLA fibres were carefully extracted from the PLA yarns to prevent any damage to the PLA fibres.For the different process temperatures, except the case of T p = 180 • C, it was straightforward to extract single fibres.For the pro- cess temperature equal to 180 • C, only a limited number of fibres were extracted since at this temperature the fibres melt and stick to each other.
The single PLA fibre were tested in tension on a Favimat & Robot single fibre tester (Textechno H. Stein GmbH & Co. KG).Individual fibres were first loaded into a magazine (up to 25 samples) with a pretension weight of 100 mg attached to the bottom end of each fibre.Then, the robot moved one by one the fibres to the testing chamber.Each fiber was first clamped, and then a pretension ranging between 0.5 cN (for small gauge lengths) and 1.5 cN (for large gauge lengths) was applied.Subsequently, the linear density (mass per length) of the fiber was measured using a vibration method following the ASTM standard D1577-07 [35].In this method, the resonance frequency was initially measured at a constant gauge length and pretension, allowing for the evaluation of the linear mass density.Following the linear density measurement, the fiber was subjected to tension testing.For all mechanical tests, a pretension of 0.1 cN was applied, and the load-displacement curves were recorded.By utilizing the linear density and the recorded loads, the fiber stress could be calculated.
The section of the paper that examines the fibre strength distribution, the fibres had not been press-consolidated and the gauge length was varied from 10 mm to 75 mm.Two sets of experiments were performed.In the first set, the cross-head speed was 20 mm/min for all gauge lengths.In the second set, the cross-head speed was varied, depending on the gauge length, to have a nominal strain rate, ε , equal to 0.25 s −1 .For each testing condition approximately 100 fibres were tested.These fibres were randomly selected from different bobbins to more accurately capture the variability in strength among the fibers.
For the investigation of the effect of the press-consolidation temperature on the mechanical properties of the fibres, the gauge length was constant and equal to 50 mm.The tests were performed at constant cross-head speed of 0.2, 20 and 200 mm/min for the different press-consolidation temperatures.In most cases, approximately 50 fibres were tested.All the fibres were taken from the same bobbin.

Fibre Strength Distribution
The fibre strength distribution, in particular of brittle fibres, is described by a Weibull distribution [36] and it can be typically a two or three-parameter distribution [37,38].By setting the threshold stress equal to zero, then the two parameter distribution, under uniform stress, is: where the P f is the probability of failure of a fibre of length L at stress .The characteristic length is denoted as L o .The Weibull modulus is m and the characteristic strength is o .
To experimentally determine m and o , Eq. 1, with L o set equal to L, is re-written as: Then, the experimental data are fit with a straight line in a ln -ln ln 1∕(1 From the coefficients of the straight line, the Weibull parameters can be determined [39].Then, from m and o , the mean fibre strength, m , can calculated from: (1) where Γ is the gamma function.The probability of failure of the ith fibre, P f i , in Eq. 1 is calculated by: where N is the total number of successfully tested fibers for the specific testing conditions.Prior to applying Eq. 4, the fibres are sorted from lowest to highest strength.After sorting the fibres, the median strength, M , can be also calculated.A number of other probability estimators than Eq. 4 have been proposed [40,41].Some estimators are less biased than others [40,42,43], however, for the purposes of the present study Eq. 4 is sufficient for determining the Weibull parameters of the PLA fibers.
In addition to utilizing the Weibull distribution, the fibre strength is also characterized by a Gaussian distribution.A Weibull distribution mainly applies to brittle materials, whereas for ductile materials such as the PLA fibres, used in the present study, a Gaussian or normal distribution can be used.

Thermal Analysis
Temperature-modulated Differential Scanning Calorimetry (DSC) (DSC 214 Polyma, NETZSCH-Gerätebau GmbH) was used to measure the thermal transitions of the PLA fibres.The samples, of approximately 10 mg, were heated from 25 • C to 200 • C with a heating rate of 5 • C/min, an amplitude of 0.8 • C/min and a period of 60s, using N 2 as purg- ing gas.
The crystallinity X c (%) of the PLA fibres was calculated by: where ΔH M represents the melting enthalpy and ΔH c is the cold crystallization enthalpy which was assumed to be zero as no cold crystallization was observed in the measurements.ΔH 100 is the melting enthalpy of 100% crystalline PLA and is set equal to 93.0J/g based on previous studies [44,45].

Weibull Analysis
Figure 2 shows the Weibull plots for two different gauge lengths, L, equal to 15 and 25 mm, respectively.For each gauge length two sets of data are presented, one obtained under constant a cross-head speed and the other obtained under a constant nominal strain rate.From such plots for all gauge lengths, the data for Tables 1 and 2 are obtained.Table 1 shows the Weibull parameters for the PLA fibres tested at a constant cross-head speed, u, equal to 20 mm/min and independent of the gauge length, L, i.e. for higher gauge lengths, the strain rate is lower.The median strength, M , is the strength of the median fibre after the fibres are sorted as described in Sect.2.2.1.It can be seen that the difference between the median strength and the average strength, av , ranges from approximately 3.3% for L = 75 mm and 0.4% for L = 15 mm.The characteristic strength, o , is higher (4) than M or av for all gauge lengths.The biggest difference of about 6% is obtained for L = 10 mm.The lowest Weibull modulus, m, is 7.2 and obtained for L = 10 mm.For all gauge lengths, the Weibull modulus is large and secondly there is one to one correspondence with the standard deviation of the Gaussian distribution.The smallest standard deviation is for L = 60 mm, for this gauge length, the largest Weibull modulus was measured.Table 2 is similar to Table 1 with the difference that for each gauge length, the crosshead displacement is adjusted to achieve a nominal strain rate, ε equal to 0.25 s −1 i.e. u = ε L .It can be seen that the variation of the Weibull modulus and of the standard deviation are smaller compared to Table 1.The median strength is approximately equal to the average strength; the difference is less than 1% except for the case of L = 50 mm with a 3% difference.As in Table 1, the characteristic strength is higher than the median and average strengths; the difference is about 5% for all tested gauge lengths.A more accurate estimation of the Weibull modulus can be obtained by the linear fit of the median stress (from Tables 1 and 2) as a function of the gauge length (see Fig. 3).The Weibull modulus for the fibres tested at a constant cross-head speed is 9.3, whereas for the the fibres tested at a constant nominal strain rate, the Weibul modulus is found to be 15.5.
Figure 4 summarises the Young's modulus and failure strain of the melt-spun PLA fibres after drawing as a function of the gauge length.As expected the Young's modulus is independent of the gauge length and approximately equal to 7.37 GPa (for u = 20 mm/min or ε = 0.25 s −1 ).The failure strain increases with the gauge length.However, the depend- ence of the failure strain to the gauge length is weak.As can be seen the in Fig. 4b, the average failure strain is approximately 33.8% for L = 10 mm and 29.1% for L = 75 mm.Fig. 3 Logarithmic plot of the median stress of melt-spun PLA fibres after drawing, tested at constant cross-head speed and nominal strain, as a function of the gauge length (see Tables 1 and 2).The slope of the straight line is −1∕m

Effect of Processing Temperature on Thermal Properties and Crystallinity
Figure 5 displays typical examples of the DSC curves for melt-spun, drawn PLA fibres before and after press-consolidation, under a process temperature, T p .The DSC results of the effect of the process temperature are summarized in Table 3.It can be seen that the onset temperature, melting temperature and the crystallinity level are all gradually increase as the process temperature, T p , rises.The rate of increase is significantly higher as T p approaches a temperature equal to 175 • C. Beyond this characteristic temperature, the onset temperature, melting temperature and the crystallinity level drop substantially due the melting of the PLA fibres during processing as it is clear from Table 3.

Effect of Processing Temperature on Mechanical Properties
The effect of the processing temperature, T p , is graphically illustrated in Fig. 6 for the fibre's Young's modulus, in Fig. 7 for the fibre strength, and in Fig. 8 for fibre failure strain.In these graphs, each symbol represents the average value obtained from several tests (see Table 4).
The results for the Young's modulus, strength and failure strain are also tabulated in Tables 5,  6, and 7, respectively, allowing for a clearer observation of the standard deviation as well.
It can be seen that the Young's modulus of the PLA fibres increases with the processing temperature and the maximum value is obtained for T p between 160-165 • C, an increase of approximately 10% with respect to the fibres without processing.Higher processing temperatures result in lower Young's modulus and a T p equal to 180 • C (higher than the fibre melting temperature) gives an E f equal to 3.62 GPa (for u = 2 mm/min) which is approximately equal to the Young's modulus of the PLA material (see Sect. 2.1) without any drawing.The effect of the processing temperature is stronger on the fibre strength.A process temperature of 165 • C results in a fibre strength reduction by 1/3, whereas when T p is 175 • C the fibre strength is reduced b a factor of 2. The failure strain of the fibres (Fig. 8 and Table 7) increases with increasing the process temperature up to T p equal to 175 • C. It is interesting to observe that the PLA fibre

(a) (b)
Fig. 4 Young's modulus and failure strain of melt-spun PLA fibres after drawing tested under constant cross-head speed and constant nominal strain rate.The error bars are based on standard deviation failure strain is nearly 2 times higher than the fibres which have not been subjected to a post process after drawing.
From the results presented above, a strong effect of the cross-head speed can be observed for all the characteristic mechanical properties (Young's modulus, strength and failure strain) even if the cross head speed increases from 2 to 20 mm/min.The importance of this will be discussed in the next Section.

Prediction of UD Composite Properties
It is possible to use the results presented in the previous Section to predict the mechanical properties of unidirectional self-reinforced PLA composites.In Sect.3.1 it was  shown, as it should be expected, that the failure of the highly oriented PLA fibres does not depend on defects.Then, the average stress-strain curve (based on engineering stress and strain) of the PLA fibres can be obtained as the average of the individual single fibre tests as shown in Fig. 9. Figure 9 refers to fibres tested with u equal to 20 mm/min and the fibres have not been processed after drawing.As can be seen from Table 4, 45 fibres are used to calculate the average PLA fibre stress-strain curve.
As the applied displacement increases, the average stress-strain curve is based on a 6 Effect of processing temperature, T p , and cross-head speed, u, on the PLA fibre Young's modulus, E f .The gauge length, L, is 50 mm Fig. 7 Effect of processing temperature, T p , and cross-head speed, u, on the PLA fibre strength, u .The gauge length, L, is 50 mm continuous decreased number of single fibre tests, since some fibres break earlier than the other fibres.
The procedure described in Fig. 9 is repeated for the different processing temperatures, T p , and the results are plotted in Fig. 10.It can be clearly seen that the average PLA fibre stress-strain curves for fibres that have been thermally processed are significantly different from the average PLA fibre stress-strain curve after drawing.More importantly the average PLA fibre stress-strain curves are remarkably qualitatively similar with the stress-strain curves of unidirectional PLA self-reinforced composites [34].
Using the linear part of the average PLA fibre stress-strain curves or the results from Table 5, the Young's modulus of the self-reinforced composites is predicted using the rules of mixtures and compared with experimental measurements [34].It should be noted, that in the experiments reported in Ref. [34], the fibre volume fraction was 50%.Therefore, in the rule of mixtures, the fibre volume fraction is assumed to be equal to 50% and the modulus of the PLA matrix is equal to 3.62 (Table 5 -T p = 180 • C).The predictions are Fig. 8 Effect of processing temperature, T p , and cross-head speed, u, on the PLA fibre strength, u .The gauge length, L, is 50 mm given in Table 8.It can be seen that the predictions are in agreement with the measurements.Therefore, it is clear that one should not use the fibre properties after drawing (e.g.ignoring the changes that occurs during manufacturing or processing of the composites) to predict or understand the behaviour of the PLA self-reinforced composites.In a similar fashion, the strength of unidirectional composites is predicted by utilizing the results of Fig. 10 as a function of T p and compared with experiments in Table 9.As mentioned before, it is assumed that the fibre volume fraction is 50%.Additionally, Fig. 9 Average PLA fibre stress-strain curve from individual single fibre tests.The gauge length, L, is 50 mm and the cross-head speed, u, is 20 mm/min Fig. 10 Average PLA fibre stress-strain curves for different processing temperatures, T p .The gauge length, L, is 50 mm and the cross-head speed, u, is 2 mm/min based on earlier findings [33], the PLA matrix fails early in a brittle manner at low applied strains and therefore the strength of the composites primarily depends on the PLA fibers.Consequently, the strength of the unidirectional fibre composites can be estimated by multiplying the maximum stress from Fig. 10 with 0.5 (fibre volume fraction).As can be seen from Table 9, the predictions agree with the experiments taking into account uncertainties in the fibre volume fraction, process temperature ( T p ), and variations in the manufacturing of the PLA fibres (melt spinning and drawing).
Similar strength predictions could be obtained if the fibre strength were taken from Table 6, however, this approach is less rigorous.This becomes evident when considering the failure strain.The failure strain from Table 6 is in the order of 60% for T p in the range of 155-175 • C. Beauson et al. [34] found that the failure strain, for similar T p , ranges between 90 and 150%.The failure strain from Fig. 10 is between 90 and 100% and closer to the experimental measurement.One should expect that increasing the number of fibres tested (Table 4) would lead to better agreement with the experiments since the failure strain depends in the single PLA fibres with the higher strain-to-failure.
Quite often single fibre testing or in particular yarn testing is performed at relatively high cross-head speeds, between 100 to 500 mm/min.On the other hand, composites are tested at lower speed in the order of 1 to 5 mm/min.Figure 11 shows the average fibre stress-strain curve after drawing and after processing at 165 • C for three different cross head speeds (2, 20, and 200 mm/min).As can be clearly seen the effect of the cross-head speed is significant.If the case of the fibre processed at 165 • C is considered, following the method presented above, the predicted composite strength is 149.8MPa when u is 200 mm/min.This value is significantly higher than the experimental  9) or the predicted value using u equal to 2 mm/min (97.7 MPa, Table 9).Thus, it is advantageous to test the polymer fibres at low speeds, corresponding to speeds used in composite testing, in order to directly predict the composite behaviour without performing experiments at the composite level.

Fibre Properties as Function of the Processing Temperature
The processing temperature, T p , is higher than the glass transition temperature, T p , of PLA, which is in the order of 60 • C. The heat treatment or annealing can promote various simultaneous microstructural effects, including polymer chain relaxation, orientation, and crystallization (Table 3), which in turn significantly impact the mechanical properties of the oriented PLA fibres themselves as shown in Sect.3.3 and their self-reinforced composite properties (Sect.4.1).An in-depth understanding of the interplay between these microstructural changes would require modelling at the molecular level which is beyond the scope of the current work.Here, it should be emphasised that increasing the processing temperature results in an increase of the Young's modulus and strain to failure (increased deformation) but at the same time the fibre strength decreases more drastically.

Concluding Remarks
The effect of the processing temperature on the mechanical properties of melt-spun oriented PLA fibres was examined through single fibre tests.It was shown that increasing the processing temperature increases the Young's modulus and strain to failure of the PLA fibres but at the same time it decreases strength at higher rate their.After a certain processing temperature, the PLA fibres melt completely and their properties reduce to those of amorphous PLA.It was also shown, that once the properties of the oriented PLA fibres are known, it is possible to predict the tensile properties of the unidirectional PLA self-reinforced composites.This prediction requires that the fiber properties account for the influence of the processing temperature, and that both the fibers and composites are tested at a similar strain rate.

Fig. 1
Fig. 1 Schematic of a solid-state drawing pilot line and b press-consolidation of the PLA yarns

Fig. 2
Fig. 2 Weibull plots for u = 20 mm/min and ε = 0.25 s −1 for a gauge length equal to 15 mm and b gauge length equal to 25 mm

Fig. 5
Fig. 5 Typical DSC curves of a melt-spun and drawn PLA and of PLA fibres post processed at T p equal to 165 • C and 175 • C, respectively

Fig.
Fig. Effect of testing speed on the average PLA fibre stress-strain curves after drawing and after processing at T p equal to 165 • C. The gauge length, L, is 50 mm

Funding
Open access funding provided by NTNU Norwegian University of Science and Technology (incl St. Olavs Hospital -Trondheim University Hospital).The work has received funding from the European Unions Horizon 2020 Research and Innovation under Grant Agreement No 685614 (BIO4SELF).

Table 1
Tensile strength of melt-spun PLA fibres after drawing as function of the gauge length.Tension under constant cross-head speed, u = 20 mm/min

Table 2
Tensile strength of melt-spun PLA fibres after drawing as function of the gauge length.Tension under constant nominal strain rate, ̇ = 0.25 s −1

Table 3
Thermal properties and crystallinity of the PLA fibres as a function of the processing temperature