A test of strength - Thermal analysis of composites shows curing behaviour
Energy efficiency, reduction of exhaust emissions and alternative drivetrain concepts are hot topics in the automotive industry. A promising approach to master these challenges is weight reduction, albeit without compromising comfort and safety.
The most interesting lighter weight materials are carbon fibre-reinforced polymers (CFRPs), which have a comparable stiffness to steel but are about 50% lighter and corrosion resistant. The ideas behind the use of such composites in the aviation industry or motor sports are now to be transferred to the mass production of cars.
Cycle times for parts manufacture are much shorter and require, for example, fully automatic fibre handling, rapid resin injection and fast curing. Key to overcoming these issues is the development of new formulations and the optimisation of processes to meet industry needs. Thermal analysis can help achieve these objectives – in particular, dielectric analysis (DEA) to investigate the curing behaviour of resins, and dynamic-mechanical analysis (DMA) to characterise the dynamic-mechanical properties of cured composites.
Finding a cure
DEA can be used to observe changes in the dielectric properties of a thermosetting resin during curing, as well as to investigate structural changes inside the sample material that are not associated with any exothermic reaction.
The functional principle of DEA is consistent with that of an impedance measurement. In a typical test, the liquid or pasty sample is placed in contact with two electrodes (the dielectric sensor). When a sinusoidal voltage (the excitation) is applied, the ions and dipoles inside the sample are forced to move to the electrodes with opposing polarity, or to align. This movement results in a sinusoidal current (the response) along with a phase shift between voltage and current.
As the curing reaction progresses, the sample material becomes increasingly viscous and as a consequence, the mobility of the charge carriers decreases. This is followed by a corresponding attenuation of the amplitude and an increased phase shift in the resulting current signal. This relationship makes dielectric analysis perfectly suitable for cure monitoring.
From the amplitude and phase shift, the dielectric properties of permittivity (ε') and loss factor (ε'') are calculated. The loss factor is proportional to the ion conductivity (σ), which is the reciprocal value of the ion viscosity (ν). The ion viscosity is the most relevant parameter for curing studies.
In order to be able to monitor the curing behaviour in almost any application – both in laboratories and in manufacturing environments – a DEA instrument should have a modular concept. Laboratory versions of a DEA instrument can be used for R&D in conjunction with a furnace (Tmax = 400°C), a bench-top lab press, UV light and/or a humidity generator. Industrial versions intended for production monitoring and process control are designated with up to 16 DEA modules, while the eight and 16 channel versions can be implemented into a 48cm rack. For in situ measurements, the industrial devices are connected via rugged extension cables and connectors to various sensor types, which are located in moulds, presses, ovens or autoclaves. For composites with carbon fibres, the sensor must be coated to prevent short circuits brought about by bridging of the electrodes by the carbon fibres.
The plot on the graph (below) reflects the curing of the polymer matrix of a CFRP material based on an epoxy resin. The measurement was performed in a mould at 80°C at a frequency of 10Hz using a tool mount sensor coated with a thin glass layer.
Initially, the curve of the ion viscosity decreases as a result of the temperature increase and shows a minimum at 38 seconds. This is the point of the lowest viscosity and, therefore, exhibits the best flow behaviour. Curing begins immediately after, with an increase in ion viscosity by more than four orders of magnitude. The slope during the increase correlates with the reactivity of the used resin – the steeper the slope, the higher the reactivity. At around 170 seconds, the curve reaches its first plateau. After a second small step, a significant change in ion viscosity can no longer be detected (at around 320 seconds). This signals that the curing process is nearly complete and that the cured composite can be de-moulded.
Dynamic-mechanical analysis (DMA) is a technique providing information on the visco-elastic properties of a sample material under a small, sinusoidal dynamic force as a function of temperature, time and/or frequency. DMA is very sensitive for the determination of phase transformations, such as glass transitions.
During a DMA test, a sinusoidal force (stress, σ) is applied to the sample. This results in a sinusoidal deformation (strain, ε). Since polymers are viscoelastic, the corresponding stress and strain curves are mutually shifted. The deviation is the phase shift (δ).
Further calculation results in two modulus values – the storage modulus (E') and the loss modulus (E'') The ratio between the two moduli (E''/E') is termed loss factor tanδ. Generally, the storage modulus refers to the elastic part of the response and represents the material’s stiffness, while the loss modulus is a measure for the oscillation energy transformed into heat. It is tanδ that characterises the mechanical damping or internal friction of a visco-elastic system.
A dynamic-mechanical analyser should be robust and sturdy, come with a variety of sample holders and provide accurate and reliable results in a temperature range between -170°C and 600°C. A single cantilever bending sample holder with free push-rod is ideal for testing CFRPs. As shown below right, the sample is tightly fixed at one end and a free push-rod oscillates at the other end with a superimposed static force. This guarantees quantitatively high storage modulus values (E') at low damping values (E'', tan).
The graph above shows a DMA measurement on a carbon fibre-reinforced polymer carried out in bending mode (single cantilever bending with free push-rod) at a frequency of 5Hz and a heating rate of 3K/min. At around 50°C, the storage modulus (E', representing stiffness) is about 145,000MPa, indicating that this sample is even stiffer than metallic titanium.
During further heating the storage modulus decreases slightly, followed by a drop at 161°C (extrapolated onset temperature of E') due to the glass transition of the epoxy matrix. By passing through a glass transition region, DMA measures the increase in flexibility of the sample material resulting in a decline of the storage modulus curve. This makes DMA much more sensitive in the determination of glass transition temperatures than differential scanning calorimetry (DSC) – especially for materials with low polymer content.
The displayed step in the E' curve is accompanied by a peak in the loss modulus curve (E'') at 169°C and a peak in the tanδ curve at 174°C. In principle, all three points can be used for defining Tg, whereas the standard test method ‘ASTM D7028’ recommends the use of extrapolated onset temperature of the storage modulus curve in logarithmic scaling. Figure 3 (below) represents two heatings on the same CFRP sample, carried out in three-point bending mode at a frequency of 1Hz and a heating rate of 3K/min. The level of the storage modulus at around 40°C within the first heating step (red) is significantly lower than during the second heating (black) due to post-curing of the sample material during the test. This is also reflected by the shift of the extrapolated onset temperature of the E' curve from 139°C to 159°C. A gain in matrix density typically causes an increase in glass transition temperature.
Summarising the above, it can be noted that the DEA measures the curing state as a function of time and/or temperature, and the DMA can be used for mechanical characterisation of the cured samples.
The significant results of these DEA and DMA tests on high-tech composite materials can subsequently be used to optimise formulations and processing of the material.
Authors: G Kaiser, S Schmölzer and T Pflock (NETZSCH Analysing and Testing, Selb, Germany) and P Davies (NETZSCH Analysing and Testing, UK). For more information, contact firstname.lastname@example.org