Study of crystallization kinetics of peek thermoplastics using Nakamura equation

Mochamad Chalid, Muhammad Joshua Y. B., Arbi Irsyad Fikri, Noel Gregory, Dedi Priadi, and Jaka Fajar
Fatriansyah

Citation: AIP Conference Proceedings 1945, 020032 (2018); doi: 10.1063/1.5030254

View online: https://doi.org/10.1063/1.5030254
View Table of Contents: http://aip.scitation.org/toc/apc/1945/1
Published by the American Institute of Physics
 Study of Crystallization Kinetics of Peek Thermoplastics
Using Nakamura Equation
Mochamad Chalid1), Muhammad Joshua Y. B1), Arbi Irsyad Fikri1), Noel
Gregory2), Dedi Priadi 1) & Jaka Fajar Fatriansyah1,*
1
Metallurgical and Materials Engineering, Faculty of Engineering, Universitas Indonesia, Kampus UI
Depok, Jawa Barat, 16424
2
Sciences et Sciences De L’Ingenieur, Universite Bretagne Sud, 2 Rue le Coat St Haouen 56100
Lorient France
*
Corresponding author: fajar@metal.ui.ac.id

Abstract. We have simulated the time evolution of relative crystallization of PEEK at various cooling rates (10, 15, 20
O
C/min) and made comparison with the experiments. The simulation was conducted using Nakamura model which is a
modified Avrami model. The model is a 1 cm radius of circle with the cooling plate which was placed in the upper part of the
circle. The cooling plate temperature was varied in order to obtain particular cooling rates. The measurement point is located
near upper boundary in order to minimize the heat transfer effect. The general trend of time evolution of crystallization was
well captured although some discrepancies occured. These discrepancies may be attributed to the heat transfer effect and
secondary crystallization. Keywords: Crystallization, Avrami model, Nakamura model.

INTRODUCTION
The physical properties of manufactured plastics products are heavily affected by relative degree of
crystallinity. The nature of polymers is that they are semi-crystalline and it is acknowledged that some of the
most used industrial materials are semi-crystalline polymers such as Polyethilene (PE), Polyaniline (PA), PET,
Polypropilene (PP) and Polyether ether ketone (PEEK). Their nature affects the crystallization behavior during
the cooling phase of processing, thus influence the physical properties and processing conditions. For example,
the uneven crystallisation may induce residual stresses and shrinkage and thus reducing product perfromance.
So, it is essential to include the crystallization phenomenon in polymers processing simulation [Galera et. al,
2009]. Therefore, polymers manufacturing simulation softwares should predict the crystallization effects
accurately during the cooling phase particularly. This accurate quantitative (or qualitative) prediction is a must in
order to predict the solidification (or gel) time and this can optimize productivity.
PEEK is one of the industrial polymer which is used in engineering applications. Chemically PEEK has a
stiffer chain and a higher melting point in comparison other polymers, which typically develops only modest
crystallinity (around 30 %) [Bassett et. al, 1988]. It is a colourless organic thermoplastic polymer in the
polyaryletherketone (PAEK) family. It is known for its robustness and thus it is used in various demanding
applications such as: pumps, piston, valves, electrical cable insolator, medical implants, partial replacement
skull and etcs. In addition PEEK can operate in continuous high temperatures (260°C). PEEK melts has relative
high temperature (343 °C / 649.4 °F) in comparison with other thermoplastics and thus it can be processed using
injection moulding or extrusion methods. The investigation of kinetics of crystallisation phenomenon on PEEK
has been conducted by Cebe et. al [Cebe et.al, 1986], Bassett et. al [Bassett et. al, 1988], D’Amore et. al
[D’Amore et. al, 1990], Harris et. al [Harris et. al, 2011] all through the thermal investigation methods. They
determined the kinetics parameters of PEEK using Avrami model using thermal experiments using Differential
scanning Calorimetry (DSC). In the latest development, Levy et. al [Levy, 2017] used Nakamura model to
investigate the crystallisation kinetics and succesfully showed the evolution of relative crystallisation of PEEK
on his COMSOL based simulation [Levy, 2016]. In this paper we incorporate real avrami index/parameter on
Levy simulation based on thermal experiments in order to model crystallisation behavior of PEEK in real
industrial setting.

Proceedings of the 3rd International Conference on Materials and Metallurgical Engineering and Technology (ICOMMET 2017)
AIP Conf. Proc. 1945, 020032-1–020032-5; https://doi.org/10.1063/1.5030254
Published by AIP Publishing. 978-0-7354-1640-6/$30.00

020032-1
 METHODS
Avrami model and Nakamura model of crystallisation
Avrami model is a widely used thermal crystallization model which has assumptions such as: instantaneous
growth, uniform nucleus formation [Avrami, 1939]. The isothermal crystallisation dynamics can be written as
X  t  1  exp   Kavt n  (1)

where X  t  is relative crystallization as function of time, K av is the Avrami crystallization function and n is
the Avrami index. There are many limitations in the application of the Avrami equation to the analysis of
polymer crystallization isotherms. For example, the Avrami index is calculated to be fractional and generally
nonconstant depending the onset of crystallisation temperature definition and cooling rate. In addition, the
isotherms do not always terminate in the manner as predicted by the Avrami equation. This realizes that polymer
crystallization cannot be merely described in simple equation as Avrami model. This is due to the fact that
isotherm model should be divided into two stages: primary and secondary crystallization. Primary crystallization
(which is main part of the process) can be merely described by Avrami model. However, Nakamura [Nakamura
et. al, 1973] realized that in the secondary crystallisation, Avrami model is not sufficient and thus should be
modified considering that secondary crystallization becomes dominant in the end of the stage. Nakamura
expanded equation (1) in order to take into account the non-isothermal crystallization evolution. Nakamura
model can be expressed as follows
ª §t · º
n

X t  1  exp «  ¨ ³ K Nak T  dW ¸ » (2)

«¬ © 0 ¹ »¼
where KNak T  is the Nakamura crystallization which in this work we set the value to be [Levy, 2017]
1

¬exp  2.75 u10 u T  2.818 u T  719.26 ¼

1
K Nak (T ) 3 2
º 3
(3)
This function is connected with the crystallization rate constant of the isothermal crystallization K av , which is
the Avrami crystallization function, through the relation K Nak T  Kav1/ n . For practical purpose, it is easier to
use the Nakamura model in its differential form. The differential form of Nakamura model can be written as
dX  t  n 1
KNak T  n 1  X t   ª¬ ln 1  X t  º¼ n . (4)
dt
In this paper we use the procedure as written by Levy [Levy, 2017] and utilize program/software from similar
source. The avrami index was obtained from DSC data from following sources. The materials used in this paper
is PEEK which has melting temperature 343 oC and latent crystallisation heat 43.84 J/g. PEEK has the chemical
structure as in figure 1.

FIGURE 1. Chemical structure of PEEK

The Avrami indices that we use in this paper are determined from real thermal experiments [Vaconcelos,
2010] and they were chosen spesifically in their respective cooling rates. The geometry of the material in the
simulation is a circle with radius 1 cm. The cooling plate is placed in the upper half of the circle and the
temperature is varied in order to obtain suitable cooling rate. The point of measurement is place near the top of
the circle where the heat dissipates faster in order to eliminate various thermal effects. For heat transfer model
we use the equation of heat transfer in solids.

020032-2
 RESULTS AND DISCUSSION
The evolution of relative crystallinity for each cooling rate (10, 15, 20 OC/min) are shown in figure 2-4
respectively with each bar showing the value of relative crystallinity. The cooling plate, depicted as blue line, is
placed in the upper half of the circle. The measurement point (red) is located in the middle of the top of the circle
where the heat dissipates faster. Here we intentionally put the point in this place in order to remove bias and
various heat effects such as transversal and longitudinal heat wave distribution. In order to achieve spesific
cooling rate, we varied the temperature of cooling plate. The cooling rates were calculated as average of
temperature change in respect of time. Although in the very initial stage of cooling, the temperature gradients
were not stable, after 10 s the cooling rates reached stable change.
Let us discuss the results. First, in ordet to get cooling rate of 10 oC/min, the cooling plate should be set at
293 oC far below melting point and crystallization point temperatures. The initial temperature of the plate was set
around 300 oC and then was cooled in the rate of 10 oC/min. Figure 2 shows the crystallization evolution at
cooling rate 10 oC/min. It can be seen that only after 210 s the upper part of the sample achieves maximum
relative crystallinity although at that time the lower part of the sample achieves half of the maximum relative
crystallinity.
Next, we proceed to the sample that was cooled down at cooling rate 15 oC/min. With similar procedure, in
order to get that cooling rate, the cooling plate was set at temperature of 268 oC with the initial temperature was
around 300 oC. Here, it can be seen from figure 3 that the upper part of the sample achieves maximum relative
crystallinity almost 3 times faster at 80 s in comparison with the sample with cooling rate 10 oC/min, although
the lower part still takes similar time to achieve similar relative crystallinity with the sample with the cooling
rate at 10 oC/min.
The third sample was cooled down at cooling rate 20 oC/min. The cooling plate was set at temperature 243 oC
with initial temperature at 300 oC. The result can be seen in figure 4. The upper part of the sample achieves
maximum relative crystallinity almost twice faster at 50 s in comparison with the sample with cooling rate 15
o
C/min. However the lower part of the sample still take more less similar time to achieve similar relative
crystallinity with the sample with the cooling rate at 15 oC/min. This is maybe due the heat transfer factor where
the heat at the lower part dissipates slower than the upper part of the sample. This is the reason why we put the
point of measurement near the boundary of the upper part of the sample. In order to obtain the effect of cooling
rate to the evolution of relative crystallinity, the heat transfer effect should be minimized.
Next, we plotted the evolution of relative crystallinity of PEEK sample at various cooling rates obtained from
the simulation and from the experimental results derived from Ziaee et. al [Ziaee et. al, 2006]. Here we found
some discrepancies between simulation and experiments. However, general trend of crystallization was captured
well by our simulation. Our simulation demostrated that as the cooling rates are increased then the full
crystallization time (the time needed to achieve relative maximum crystallinity) also increases similar with
experiment. however, at high cooling rate, our simulation shows much faster full crystallization time. This may
be partly attributed to the uneven gradient of cooling rate at the very initial stage of cooling. In addition, the
proper heat transfer condition should be examined. Our simulation shows good agreement at lower cooling rates
at 10 oC/min near the initial stage of crystallization. However at the later stage the discrepancy occurs. This may
be attributed to the secondary crystallization phenomenon.

FIGURE 2. Crystallization evolution at cooling rate 10 OC/min

020032-3
 FIGURE 3. Crystallization evolution at cooling rate 15OC/min

FIGURE 4. Crystallization evolution at cooling rate 20 OC/mins

FIGURE 5. The evolution of relative crystallinity of PEEK samples at various cooling rates (10, 15, 20 OC/min) obtained
from simulation/calculation and from experimental derived from (10, 15, 20 OC/min).

020032-4
 CONCLUSIONS
We have simulated the time evolution of relative crystallization of PEEK at various cooling rates (10, 15, 20
O
C/min) and compared with the experiments. The general trend of time evolution of crystallization was well
captured although some discrepancies occured. These discrepancies may be attributed to the heat transfer effect
and secondary crystallization.

ACKNOWLEDGEMENT

This writing is supported by DRPM (Universitas Indonesia Research Directorate) under PITTA 2017 grant.

REFERENCES
1. Galera, Viviane Verona, Alessandra Lucas Marinelli, and Benjamim de Melo Carvalho., Materials
Research 12.2 (2009): 151-157.
2. Bassett, D. C., R. H. Olley, and I. A. M. Al Raheil., (1988): 1745-1754.
3. Cebe, Peggy, and Su-Don Hong. , (1986): 1183-1192.
4. D'Amore, A. J. M. L., Kenny, J. M., Nicolais, L., & Tucci, V., Polymer Engineering & Science 30.5
(1990): 314-320.
5. Harris, Luke. A Study of the crystallisation kinetics in PEEK and PEEK composites. Diss. University of
Birmingham, 2011.
6. Levy, Arthur., Issue 1 (2017): 7-11
7. Levy, Arthur. "A Novel Physics Node for Nakamura Crystallization Kinetics." Accessed from
https://www.comsol.com/paper/download/.../levy_paper.pdf
8. Avrami, Melvin. , The Journal of Chemical Physics 7.12 (1939): 1103-1112.
9. Nakamura, K., K. Katayama, and T. Amano. , Journal of Applied Polymer Science 17.4 (1973): 1031-1041.
10. Vasconcelos, G. D. C., Mazur, R. L., Botelho, E. C., Rezende, M. C., & Costa, M. L., Journal of Aerospace
Technology and Management, (2010), 2(2), 155-162.
11. Ziaee, Zainab, and Pitt Supaphol., Polymer testing 25.6 (2006): 807-818.

020032-5