Introduction

Helicoidal composite structures in nature [1,2,3] are fundamentally asymmetrical [4,5,6] and are known to have exceptional tolerance to impact damage [4]. They are essentially, stacked laminates of aligned fibres with individual adjacent laminates rotated by an angle relative to their neighbouring through-thickness laminates, producing a twisted (Bouligand) architecture with multifunctional purposes including e.g. thermal regulation and physical protection [7,8,9]. Carbon fibre reinforced plastic (CFRP) composites are high performance composites exhibiting outstanding specific properties. While CFRP composites exhibit superior specific properties to glass fibre reinforced plastic (GFRP) composites, they are inferior in terms of impact resistance. As such, researchers have started laying greater focus on the development of helicoidal CFRP as a means of more effectively retarding fracture than typical laminated composite structures [10,11,12,13] One area that has received less attention nevertheless, is the dynamic mechanical performance of helicoidal CFRP, even though high performance composites are often used in technologies (such as aerospace and automotive) where their viscoelastic and dynamic thermo-mechanical properties are of importance [14]. Dynamic mechanical thermal analysis (DMTA) is a reliable experimental method by which means independent anisotropic variables can be evaluated [15, 16].While it is known that storage modulus, loss modulus and Tg vary with inter-ply pitch angle [7, 17,18,19], there is no report clearly correlating the characteristics these properties as a function of orientation architecture. Perhaps the most comprehensive report to date is that by Coban et al. [7] who researched the viscoelastic properties of FRP composites with inter-ply stacking angles of 30°, 45°, 75°, 90° and + / − 45°, and reporting that the lowest viscoelastic properties were evident in composites comprised + / − 45° stacking orientations. These stacking angles are nevertheless very different to asymmetric helicoidal structures in nature, where the stacking angles are much smaller. To date, there is no research that relates the dynamic thermo-mechanical behaviour of helicoidal composites with small stacking angles. The research presented herein aims to fill this gap in knowledge comparing CFRP composites of 0° (UD), 0/90° (cross-ply), 5°, 15°, 10°, 20°, 25° and 30° inter-ply stacking orientations.

Experimental procedure

MR70 12P carbon fibre prepreg with Toray E750 toughened epoxy resin (Mitsubishi Chemical Carbon Fibre Composites) was vacuum bag manufactured into composite laminates comprising the following different inter-ply orientation angles: 0° (UD), 0/90° (cross-ply), 5°, 15°, 10°, 20°, 25° and 30°. While UD and cross-ply composites are symmetric composites, helicoidal composites (5°–30°) display an asymmetrical Bouligand structure. All CFRP composite laminates were manufactured using 20 plies and were 1.9 mm thick after curing, which was conducted in a convection oven (Sciquip HT 230), raising the temperature from ambient temperature to 135 °C over 1 h, holding this temperature for another hour, and then cooling the composite over an hour back to ambient (room) temperature. Table 1 provides information on the stacking sequences used, and the relative fractions of plies oriented between 0° and 45° of the loading axis, and the fraction of those oriented at 46°–90° of the loading axis.

Table 1 Laminate stacking sequences for composite sample sets (n = 3) arranged at specific pitch angles
Full size table

Samples for DMTA testing (3-point bending mode) were prepared and thermal sweep testing conducted in accordance with ISO 6721-1 [20] using a Triton 2000 Series DMTA. The specimens (n = 3 for each series) were cut to 42 mm (long) and 8.2 mm (wide), a heating rate of 4 °C/min ramped between 50 and 230 °C at 1 Hz following [17, 19, 21, 22] at an amplitude of 50 µm. The 3-point bending test span was 15 mm.

Results and discussion

Figure 1 shows the storage modulus (E′) and loss modulus (E″) plotted against pitch angle in (a) and (b), respectively, showing median, upper and lower bounds. The range of values is evidently highest for UD composites The Bouligand inter-ply stacking angle influences the ply fraction between 0° and 45° off the loading axis, and samples with a higher fraction of plies within this angle-range have a higher E′ as compared to those with a lower fraction of plies within this angle-range. As per Table 1, E′ is higher when this fraction is higher, with 20°–25° stacking angle being the stiffest. For conventional composites, E′ decays as a function of increasing orientation angle [23, 24] assuming that all laminates are oriented at one set angle relative to the loading axis. Here we note that Bouligand structured composites do not behave in this way. The higher degree to which plies are oriented from the loading axis, the more the behaviour depends on fibre–matrix interfaces where polymer chains are pinned [25] as well as the bulk polymer matrix, to a maximum of 90° from the loading axis, where the fibre contribution is at its lowest. Likewise, E″ rises as a function of inter-ply stacking angle from the loading axis up to 25° (UD25), since the 0°–45° ratio is higher than in any of the other series. E″ defines the ability of the composite to dissipate energy as heat through intermolecular friction during sinusoidal loading. As carbon fibres are stiff, they carry and transfer load into the matrix, but their stiffness can reduce molecular mobility and thus intermolecular friction. When all fibres are oriented in the same manner (e.g. UD composites) the molecules between the fibres are relatively free to move and develop heat through intermolecular friction. As such, UD composites exhibit the highest E″, as well as the highest E′ (which is itself due to stiff fibres storing energy most effectively in the UD orientation). In Bouligand (helicoidally stacked) composites, we postulate that as stiff carbon fibres cross at angles to each other from layer to layer, they restrict molecular mobility more effectively than in the UD composites and as such, we observe lower E″ values in the Bouligand structured composites. In conventional composites, the rate of change of E′ and E″ as a function of increasing orientation angle is reported in many places to decrease dramatically for angles above 45° [23, 24]. Helicoidal stacking arrangements increase the thickness direction tortuosity of the fibre–matrix continuum at the microstructural level as compared to conventional stacking arrangements. This interferes with molecular mobility [11] and we hypothesise that this is because fibres stacked at angles to one another house different sized regions of confined matrix where interfaces play a more significant role during loading than bulk material. Confined polymer regions are essentially regions where bulk is predominantly surrounded by pinned polymer interfaces [25] to fibres. As a result these can impeded the extension of bulk polymer macromolecules and other structural units. It is also possible that crosslinking density is increased through the development of both intermolecular and intramolecular bonding [26] in confined polymer regions created by the presence of a tortuous fibrous helicoid. Variations in the helicoidal orientation therefore affect composite heterogeneity at the microstructural level, which is suggested herein to affect molecular movement and consequently, intermolecular friction between the polymer molecules unlike that seen in conventional composite orientations. Since Bouligand structured composite properties are hard to predict based on conventional composite theories, we propose herein, that the properties of such composites can be speculated based on the fraction of plies oriented between 0° and 45°. This is evidenced by plotting median values of E′ and E″ against 1°–45° fibre orientation in Fig. 1c and d, respectively. There is evidence of positive correlation between E′ and E″ with the fraction of 0°–45° oriented plies since the Pearson’s r values are 0.93 and 0.89, respectively. Scatter about the linear regression best fit for E′ and E″ against the fraction of 0°–45° oriented plies is described by their determination coefficients (R2) which are 0.87 and 0.79, respectively. E′ and E″ are reported to be linearly proportional properties [27, 28]. Figure 1e shows the median E′ values plotted against the median E″ values and positive linear correlation is noted between the two based on the Pearson's r value of 0.84. The determination coefficient indicating the level of scatter about the linear regression line is 0.71.

Fig. 1
figure 1

a Storage modulus and b loss modulus plotted against inter-ply pitch angle showing median, upper and lower bounds, respectively, c storage modulus and d loss modulus median values plotted against 0°–45° fibre orientation fraction, respectively, and e storage modulus plotted against the loss modulus (median values)

Full size image

Figure 2 shows representative tan-δ curves for each of the composite series’ tested. We note that while there is only one peak for UD composites, helicoidally stacked and cross-ply composites all exhibit two tan-δ peaks, one shoulder peak and one primary peak. This indicates increased heterogeneity results from varying the stacking angle of these composites, signifying noticeable changes in the deformation mechanisms of the helicoidal composites as a function of inter-ply stacking angle. Essentially, the dual tan-δ peaks (primary and shoulder peaks) mean that these materials have distinctly dissimilar viscoelastic material phases and two separate glass transition temperatures [29], and median Tg values (Tg1 and Tg2 for lower and higher Tg values, respectively) are provided in Table 2. These Tg values are taken from either the peak of the tan-δ curve, or from the mid point of the slope of the shoulder peak and each indicates the temperature at which a specific phase within the composite changes from a rigid glassy material to a soft rubbery-like material. The variations of Tg between helicoidal stacking angles is notably large (40 °C range between all values of the Tg1 phase and 44 °C range between all values of the Tg2 phase). Varying ply orientation is known to result in large Tg (22–33 °C) ranges in continuous fibre composites [17, 30], however, little is known about how Tg changes from varying the helicoidal stacking angles. Here we show that helicoidal stacking has the ability to restrict macromolecular mobility to a greater or lesser extent, as evidenced by the large Tg ranges observed for each of the phases in the composite. Since all variables except inter-ply stacking angle are the same in each composite, the evidences point to that it can only be the stacking angle that affects macromolecular mobility within the composite. Variations observed between pitch angles are related to some extent, to fibre orientation in plies closer to the outermost faces such that plies oriented closer to 0° from the bending axis exhibit higher levels of bending resistance [17]. Plies oriented away from the bending axis are more dependent on fibre–matrix interfaces [23] and the effects of polymer chain pinning [25], while plies oriented closer to 0° from the bending axis tend to be dominated by the properties of the fibres. Since carbon fibres are significantly stiffer than the polymer matrix they are embedded within, stress transfer is affected not only by the intraply fibre angle relative to the bending axis, but also the interply stacking angles and the interactions between them [6]. Table 2 additionally provides details on the properties of composites showing median values from each series for the width of the tan-δ peak at half height (Pw), which provides detail on the heterogeneity of the composite, and the area under the tan-δ curve (Aδ) which provides information on molecular mobility in the material. Aδ also refers to damping properties, i.e. the effectiveness of energy that the composite absorbs and dissipates under loading (in this case in bending). UD composites here are noted to have the highest damping (Aδ) and the lowest levels of material heterogeneity, as indicated by the lowest Pw value. It is also able to retain its glassy state at higher temperatures than any of the helicoidal composites. ID5, ID10 and ID20 composites also have notably high capacity for damping (Aδ = 8.26, 8.26 and 8.26 °C, respectively) and exhibit relatively high levels of material heterogeneity (Pw = 43.8, 47.2 and 47.1 °C, respectively). This said, it is interesting to note that although both ID5 and ID90 composites have equal fractions of plies oriented within the 0°–45° range (cf. Table 1) and the level of material homogeneity is highly dissimilar (45.8 and 58.3 °C, respectively). The damping of ID5 is moreover higher than for ID90. This most likely reflects the effects of using much higher inter-ply stacking angles in ID90, which are either perfectly oriented for energy storage at 0° or imperfectly oriented for energy storage at 90° [31]. As such, we note that for these properties, there is no correlation between ply orientation fraction and the properties. Rather, stacking angle appears to have a greater effect on the properties of damping and the physical heterogeneity of the composite. Molecular mobility and hence damping, is lowest in ID30 composites (Aδ = 5.27 °C) even though these composites exhibit levels of material heterogeneity similar to that of ID25 composites (Pw = 48.4 °C). There is therefore, no evident correlation between the three parameters of damping, heterogeneity and Tg, that we can currently deduce from this work.

Fig. 2
figure 2

Representative tan-δ plotted against temperature for each composites test series

Full size image
Table 2 Peak and shoulder peak values of glass transition temperature [Tg1 (lower) and Tg2 (higher)], the width of the tan-δ peak at half height, and the area under the tan-δ curve
Full size table

Conclusions

Asymmetrical helicoidally stacked (Bouligand structured) CFRP exhibit a wide range of viscoelastic and physical properties that are hard to predict. We show that the viscoelastic properties E′ and E″ are, linearly correlated to the inter-ply pitch angle when this is expressed in terms of a 0°–45° range relative to the loading axis. While there is evidence that the inter-ply pitch angles have the ability to either restrict to a greater or lesser degree, macromolecular mobility, the values of the intrinsic properties within the composites are identified herein as being much harder to predict (such as damping, material heterogeneity and Tg). Further research is required to ascertain how these intrinsic properties are affected by inter-ply stacking angles. Two further findings from this work so far are (i) that E′ and E″ are not mutually exclusive properties in Bouligand structured composites and (b) that the introduction of above 0° inter-ply stacking angles introduces greater levels of heterogeneity and thus deformation mechanisms in composites, as evidenced by the presence of two glass transitions in each of the non-UD composites.