Experimental Methods

3.2.1 Materials

The tested fibre is the PAN (Polyacrylonitrile)-based AKSAca A-42 carbon fibre with bulk density of 1.78 g/cm3 and yield of 800 g/km, respectively. The surface morphology and diameter of the carbon fibre were characterised by a high resolution Field Emission Scanning Electron Microscope (FE-SEM JEOL JSM 6340F) at an accelerating voltage of 30 kV and emission current of 12 µA and accelerating voltage of 5 kV. As shown in

Figure 3.1, the fibre diameter was determined, by measuring 20 fibres, to be 7.3±0.4 μm. The A-42-12K fibre tows contain 12,000 fibre filaments.

Figure 3.1 FE-SEM image of A-42 carbon fibres.

Matrix material is Bakelite® EPR-L20 epoxy resin. Bakelite® EPR-L20 epoxy was mixed with EPH-960 hardener at the weight ratio of 100:35, which is suggested by manufacturer’s data sheet, and the mixture was then degassed for approximately 30 minutes before curing. The curing condition consisted of a 24-hour room temperature curing and 15 hours of 60 °C heat treatment.

54

3.2.2 Longitudinal Tensile Test of Carbon Fibre

In the tensile tests, short-gauge-length dry single filaments were tested by a self-made testing system following the ASTM C1557-03 standard. As shown in Figure 3.2, the single filament specimen was prepared by mounting a single carbon fibre to a paper holder with instant cyanoacrylate glue. The sample was placed in the grips of the micro- tester equipped with a 250 gram-force (gf) load cell. Both sides of the paper holder were cut by scissors before testing, leaving the fibre between the grips intact. Single filament specimens with gauge lengths of 5, 10, 15 and 20 mm were tested at the speed of 0.001 mm/s to approach a static test. A minimum of 10 specimens were tested for each gauge length.

Figure 3.2 Single filament specimen with paper holder for tensile test.

Large-gauge-length resin-impregnated fibre bundles were tested by an Instron universal tester under the guidance of ISO 10618:2004 standard. A yarn with 12,000 fibres was impregnated with L20 epoxy resin before aluminium tabs were mounted at its two ends. Samples with gauge lengths of 50, 100, 150 and 200 mm were tested at the speed of 2 mm/min according to the standard.

Figure 3.3 Apparent compliance vs gauge length divided by cross-sectional area of carbon fibre.

55

It should be noted that the strain measurement of the tests was done by measuring the movement of the grip of the tensile machine. Therefore, system compliance correction had to be performed through the procedure described in ASTM C1577-03. Accordingly, the system compliance (𝐶𝑠) is determined by plotting ∆𝐿/𝐹 (𝐶𝑎, apparent compliance)

against 𝑙₀/𝐴 curve in which ∆𝐿 is the grip movement measured from the machine, 𝐹 is the failure load, 𝐴 is the cross-sectional area of carbon fibre and 𝑙0 is the gauge length of

the sample. In Figure 3.3, the intercept corresponds to zero gauge length gives the value of 𝐶𝑠, which is 0.022 mm/N. The compliance of the tensile test system influences the

calculated strain and Young’s modulus, especially when the sample gauge length is small. The results presented in this study have been calibrated for the compliance.

3.2.3 Longitudinal Compressive Test of Carbon Fibre

Unlike longitudinal tensile strength and modulus, various difficulties have been encountered by researchers to measure axial compressive strength of carbon fibre in the past five decades primarily due to the difficulty in introducing a pure axial compressive failure to a carbon fibre without causing buckling [11]. In order to overcome these difficulties, some indirect interpretation methods were reported including tensile recoil method, effect elastic loop method and bending beam method [11-13]. Among the methods proposed, tensile recoil method [12] is preferred by many [13, 14] thanks to its procedural simplicity and reliability to produce repeatable results.

56

In the tensile recoil method, a single fibre was stretched to a predetermined tensile stress level to allow some strain energy to be stored in the fibre. Then the fibre was cut by a sharp surgical scissor in the middle of the gauge length, initiating a recoil effect. When the fibre is cut, the tensile stress in the fibre drops totally, converting the stored strain energy to kinetic energy. An unloading stress wave thus propagates towards the clamped ends. The moment the unloading wave reaches the rigid end, the compressive stress wave propagates from the clamped end reflects toward the free end. The magnitude of the compressive stress wave generated during a specimen recoil is equal in magnitude to but of opposite sign to the initial tensile stress. If the resultant compressive stress exceeds the compressive limit of the fibre, the fibre undergoes recoil compressive damage. As such, by increasing the pre-stress level in a precise manner from a magnitude that is below the compressive strength of the fibre to a magnitude that exceeds the compressive strength, a transition in the damage behaviour of carbon fibre from no recoil compressive damage to some recoil compressive damage will be observed. Hence, the pieces of the filament after testing were carefully examined under a magnifying glass to determine if they have failed (F) or not (NF), as illustrated in Figure 3.4. Hence a threshold stress for observation of recoil compressive damage can be established to approximate the compressive strength of carbon fibre [12]. The samples were tested by a dedicated tensile machine, apt-dc servo controller Thorlabs Z812B, with a load cell of 20 gf. A total of 400 fibre samples, split into 8 batches, were tested.

3.2.4 Torsional Pendulum Tests of Carbon Fibre

The longitudinal shear modulus of the carbon fibre was determined by torsional pendulum test which was invented by Tsai et al. [4]. The theoretical background is that a disk hung by a wire will oscillate about its equilibrium position if it is twisted by a small angle. Although the magnitude of the oscillation will decrease, the frequency of oscillation is a function of the longitudinal shear modulus of the wire.

In this study, a washer, suspended by a single carbon fibre, was set in free torsional oscillation without air turbulence, as shown in Figure 3.5. The longitudinal shear modulus of carbon fibre is calculated by [4]:

57 𝐺_𝑓12=𝜋𝑚𝐿𝑓2[8(𝐷02−𝐷𝑖2)+

32 3ℎ2]

𝑑4 , (3.1)

where 𝑚 is the mass of the hanging washer; 𝐷0 and 𝐷𝑖 are the outer and inner diameter of

the washer, respectively; ℎ is the thickness of the washer; 𝑑 is the diameter of the fibre; 𝐿 is the length of the fibre that has been suspended; 𝑓 is the oscillation frequency; and 𝐺_𝑓12 denotes the longitudinal shear modulus of fibre. The frequency was measured by a stopwatch with resolution of 0.01s. As mentioned, the diameters of the fibres were measured by FE-SEM. The shear modulus tests were performed using 4 washers with different geometries at 3 fibre lengths: 15, 20 and 25 mm. For each weight-length combination, five samples were tested.

Figure 3.5 The experiment setup of torsional pendulum test.

3.2.5 Nano-indentation Test of Carbon Fibre

Although different experimental techniques were reported [6, 8, 15-16], the nano- indentation technique was employed to probe the transverse modulus (𝐸𝑓2) of the carbon

fibre [8]. Such test on the impregnated yarn has two advantages. Firstly, fibres in the sample are restrained so that movements are prohibited, minimising the slippage interference. Secondly, different properties of carbon fibre and the epoxy matrix (and interphase) allow their distinctive load-displacement curves to be distinguished.

58

Comparing the data obtained via nano-indentation with known epoxy properties will crosscheck the validity of carbon fibre properties measured by the same test.

Figure 3.6 Schematic illustration of the nano-indentation experimentation.

The yarn, which consists of 12,000 fibres, was embedded in epoxy resin, and then polished to a flat surface parallel to the longitudinal direction of the fibres. Nano- indentation was made on the flat surface with the fibres revealed on the top surface. The tests were performed on Agilent Nano Indenter G200 with a Berkovich tip. To provide a more precise measurement of initial surface contact, continuous stiffness measurement technique [8], as opposed to the conventional ones which use only the unloading path in the load-displacement curves, was employed in this study. Nano-indentations were performed along six indent lines, as illustrated in Figure 3.6, each consisting of 10 indent points with spacing of 20 μm, along the transverse direction at randomly chosen locations of the impregnated yarn. The tip indentation was controlled by a frequency of 45 Hz at the strain rate of 0.05 s-1. The load was maintained for 30 s to evaluate the errors caused by temperature variations [8]. The Poisson’s ratio of the carbon fibre was assumed to be

𝜈 = 0.2 in the experiments.

3.2.6 Tensile Test of L20 Epoxy

Tensile properties of L20 epoxy were characterised in accordance with ASTM standard D638-10. L20 epoxy was mixed with hardener at the weight ratio of 100:35 and cured in a dumbbell-shaped mould. The dumbbell-shaped tensile samples (see Figure 3.7) used

59

were cut from a cured epoxy panel. The thickness of the samples was approximately 4.5 mm on average.

Figure 3.7 Specimen shape and dimensions in mm [17].

The samples were tested using Instron Universal Tester 5569 at the cross-head speed of 5 mm/min, an Instron static axial clip-on extensometer 2630-105 with 25 mm gauge length was used to measure the strain.

3.2.7 In-plane Shear Test of L20 Epoxy

The in-plane shear strength of L20 epoxy was measured according to ASTM standard D7078. The samples, which had the same curing conditions as tensile samples, were cut to V-notched shape as shown below (see Figure 3.8).

60

The samples were clamped by a self-designed fixture (see Figure 3.9) and were tested using Instron Universal Tester 5567 at the cross-head speed of 2 mm/min. The shear strength S was calculated using equation:

𝑆 =𝑃_𝐴, (3.2)

where 𝑃 is the ultimate load before break and 𝐴 is the cross-sectional are between the two V-notch. The dimensions of the sample were all measured by a digital calliper.

Figure 3.9 Assembled view of fixture and sample (left) and fixture in experimentation (right).