By Christopher H. Childers
Download full paper
Over the past several years, as Boeing has greatly increased its usage of structural thermoplastic composite materials, the potential design space for thermoplastic composites has grown. So new materials have been required to meet this expanding need.
Historically, the processing of thermoplastic composite materials was determined by a trial-and-error approach, used to determine the consolidation parameters for the multiple consolidation methods, where each method requires a unique set of parameters. While this process window could be used to create high-quality laminates or parts in a stable manner, it might not have been the most optimal cycle possible.
For the legacy structural thermo-plastic composite material with a poly aryl ether ketone (PAEK) polymer matrix, this iterative approach was used to generate the processing window for three unique processes used for thermoplastic composites: continuous compression molding; press consolidation; and thermal stamp forming.
However, this iterative approach is both expensive and time consuming. As new materials are developed, a more rapid, cost-effective approach to determining the process window is required.
This paper interrogates the relation-ship between composite laminate consolidation (or thermal forming) cooling process parameters and crystallization kinetics of thermoplastic polymer composite matrix materials.
In general, all three of the processing methods mentioned rely on the use of high temperatures and pressures (greater than 350°C and 15 bar) to convert the thermoplastic pre-preg or semi-preg to a consolidated laminate.
After some time at elevated temperature, the now-consolidated laminates must be cooled. For semi-crystalline polymer materials the cooling step is critical. Cooling controls the percent crystallinity that determines the mechanical performance of the composite. The crystallinity also directly influences any residual stresses in the composite part. To generate crystallinity upon cooling, the three processes use both dynamic (non-isothermal) and isothermal cooling. Thus understanding how the two cooling methods impact crystal growth is critical to developing a rapid, low-cost methodology to define process conditions for new materials.
One approach to understanding thermoplastic composite crystallization kinetics and processing is to model crystallization behavior at various cooling rates and isothermal hold temperatures. Multiple kinetic models exist, including the Flynn-Wall-Ozawa, Nakamura and Avrami models. The Avrami crystallization kinetic model is the most often selected for PAEK materials, such as poly ether ether ketone (PEEK) and poly ether ketone ketone (PEKK).
One of the main factors responsible for the common use of the Avrami Crystallization Kinetics Model is the inclusion of both dynamic and isothermal kinetics models. Both models are based on crystallization rates determined by heat flow measurements made using differential scanning calorimetry (DSC). During DSC cooling, the percent crystallinity, α, is measured by monitoring the change in heat flow over time/ temperature. When correlated to the time component of the experiment, the following relationship has been established and is the basis for the Avrami crystallization kinetics model:
ln( – ln(1 – α) ) = nln(t) + ln(Z)
Where: α = percent crystallinity
t = time
Z = Avrami temperature constant
n = Avrami correction factor
In the case of isothermally cooled processes, Z is directly related to the isothermal hold temperature. For dynamically cooled processes, a correlation factor to correct for cooling rate must be applied:
= ln(Z) |β|
Where: β = cooling rate (°C/time)
The n and Z terms are determined by a plot of ln(-ln(1-α)) versus ln(t) where the slope is equal to n and the intercept equal to ln(Z)13. These terms are unique for each isothermal hold temperature and cooling ramp rate but the values tend to become more similar at very slow cooling ramp rates of high isothermal temperatures as very slow cooling rates force the slowest crystallization kinetics. To complete the kinetic model, the exponential terms are applied to the crystallization half-time equation shown below in Equation 3. The relationship between half-time and processing for the thermoplastic composites is discussed in this paper.
1 t1 = (ln(2))n 2 Z
For this work, DSC was used to generate dynamic and isothermal Avrami crystallization kinetics models for thermoplastic composite consolidation and thermal forming processes. These models were then applied to determine the crystallization half-time and establish robust process windows for all three for a new structural thermo- plastic composite material. This model and approach will ultimately reduce the cost and time frame for development and implementation of consolidation/ forming processes for new thermo-plastic composite materials.
For the experiments, unidirectional carbon-fiber-reinforced thermoplastic composite pre-preg was used as received. The pre-preg material was manufactured to and is in compliance with industry standard requirements. Specifically, the pre-preg utilizes a PAEK polymer matrix with an intermediate modulus carbon fiber.
Dynamic Cooling Differential Scanning Calorimetry
Dynamically cooled differential scanning calorimetry experiments were conducted on thermoplastic unidirectional pre-preg materials to determine crystallization kinetics at various cooling rates. Experiments were completed on a PerkinElmer DSC 8500 with helium purge gas. The instrument was calibrated to measure heat flow up to a cooling rate of 750°C/min using Indium and Sapphire standards. (Temperature calibration was completed by the manufacturer.)
For all dynamic experiments, the thermoplastic composite pre-preg samples of approximately 8-15 mg in size were placed in a crimped aluminum sample pan. The sample was heated to 410°C at 10°C/min in the DSC to achieve complete melt.
Samples were then cooled at a variable rate to 140°C using the cooled helium purge gas. The cooling rates targeted were 5, 10, 20, 30, 40, 50, 60, 80, 100, 150, 250, 350 and 500°C/min. After cooling to 140°C, the samples were heated back to 400°C at 10°C/min to ensure that no cold crystallization was present. (Cold crystallization is defined as any crystallization event that occurs during the heating of the material below the melt temperature.) Transitions were identified and integrated using the standard PerkinElmer analysis software and Origin 9.0.
Isothermal Cooling Differential Scanning Calorimetry
Isothermally cooled DSC experiments were conducted on the same instrumentation as described for the dynamic cooling experiments. Also as previously described, 8-15 mg samples of thermoplastic composite pre-preg material were placed in a crimped aluminum sample pan. The samples were heated to 410°C at 10°C/min in the DSC to achieve complete melt.
Samples were then cooled at 650°C/min to the isothermal hold temperature and held for one hour. After the hold, the samples were cooled to room temperature and then heated to 400°C at 10°C/min to ensure that no cold crystallization was present. Due to the small mass of the samples and instrumental setup, minimal thermal lag was observed between the set point temperature and the sample temperature as measured by the instrumentation, with a maximum differential of 6.95°C between the sample and set point (at a set point of 140°C). Isothermal hold temperatures targeted were 150, 175, 200, 225, 250, 270, 280, 290, 300 and 325°C. Transitions were then identified and integrated using the standard PerkinElmer analysis software and Origin 9.0.
Experiment results are explained and discussed in the full report available online.
Thermal-based process window limitations were recommended for both dynamically and isothermally cooled processes. These process windows are based solely on the melt and crystallization kinetics of the polymer and must be validated mechanically.
However, this methodology eliminates entirely the guess and check methods that were previously used. Under the guess and check system, parts would have to be fabricated at multiple tool temperatures and evaluated thermally and mechanically until the process window was determined.
By the implementation of Avrami kinetic modeling, we can determine the extremes of the process and then optimize, reducing cost and implementation time for all semi-crystalline thermoplastic composite materials. Furthermore, this work demonstrates that for full implementation of this specific PAEK, changes to the tooling concepts of the various consolidation processes are required to more efficiently manage the heat transfer to and from the composite material. Lastly, the methodology developed in this work has been used to provide processing parameters via continuous compression molding, press consolidation and thermal stamp forming.