Beruflich Dokumente
Kultur Dokumente
The influence of the thermal history experienced during injection molding on the mechanical properties of polycarbonate is investigated. Distributions of the yield stress as they
result from inhomogeneous cooling during processing, predicted by a previously developed
50
modeling approach, are validated and are in
good agreement with experiment. Predictions
of the mechanical performance for different
40
mold temperatures during processing are also
validated over a range of applied strain and
30
deformation rates and applied stresses and
forces, for simple tensile bars and an actual
20
(open markers)
(closed markers)
product. Good agreement is found. It is shown
that mold temperature has a tremendous influ10
Tm = 130C
ence on the life time of polymer products,
Tm = 30C
model prediction
which can differ by more than two orders of
0 0
1
2
3
4
5
6
7
magnitude.
10
10
10
10
10
10
10
10
timetofailure [s]
Introduction
Catastrophic failure of polymer objects, either upon impact
(e.g., of protective products such as airbags and helmets) or
after prolonged exposure to load (for instance supporting
structures, high-pressure pipes), limits their ultimate useful
lifetime. Hence, understanding of that process and, ideally,
being able to accurately predict when failure occurs is of
critical importance. This reflects on the selection of the
materials employed, the choice of the proper processing
parameters and the selection of the geometrical design of
T. A. P. Engels, L. C. A. van Breemen, L. E. Govaert, H. E. H. Meijer
Materials Technology (MaTe), Eindhoven University of
Technology, P.O. Box 513, NL-5600 MB, Eindhoven,
The Netherlands
E-mail: l.e.govaert@tue.nl
Macromol. Mater. Eng. 2009, 294, 829838
2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/mame.200900227
829
830
Modeling
Multi-Mode EGP-Model
The basis of the EGP-model is the decomposition of the
Cauchy stress into a hardening stress sr and a driving stress
ss, which is split up into a hydrostatic part (superscript h)
and a deviatoric part (superscript d).[23] The deviatoric part
is modeled as a combination of n parallel linked Maxwell
elements,[22,27,28] which, for isothermal conditions, leads to:
s s r s hs
n
X
s ds;i
i1
~ d kJ 1I
Gr B
n
X
~d :
Gi B
e;i
(1)
i1
~ the isochoric
where Gr is the strain-hardening modulus, B
left Cauchy-Green strain tensor, k the bulk modulus, J the
volume change ratio, I the unity tensor, and G is the shear
modulus. The subscript e refers to the elastic part, the
subscript i refers to a specific mode, i 1; 2; 3; . . . ; n.
The evolution of the elastic and volumetric strains is
given by:
J_ JtrD
(2)
~_ e;i ~L Dp;i B
~ e;i B
~ e;i ~Lc Dp;i :
B
(3)
s ds;i
;
2hi t; p; Sa
(4)
DOI: 10.1002/mame.200900227
(5)
(7)
1 r0 exp g p
R gp
r 1=r1
1 r0r1 2
(8)
where
r
1 d
s : sd :
t1
2 s;1 s;1
(9)
a)
a)
Tm
sy
-C
MPa
30
57.1
27.6
90
58.8
29.1
130
62.1
32.2
t0
Sa
r0
r1
r2
MPa
MPa
MPa
26
3 750
0.7
0.08
0.965
50
3
Mode
Gr
(6)
Sa
h0
Pas1
MPa
2.10 1017
3.50 102
3.48 10
16
5.55 101
2.95 1014
4.48 101
2.84 10
13
4.12 101
2.54 1012
3.50 101
2.44 10
11
3.20 101
2.20 1010
2.75 101
2.04 10
2.43 101
1.83 108
2.07 101
10
1.68 10
1.81 101
11
1.51 106
1.55 101
12
1.40 10
1.37 101
13
1.27 104
1.19 101
14
1.10 10
9.80 100
15
1.23 102
1.04 101
16
2.62 10
2.11 100
17
2.14 100
1.64 101
www.mme-journal.de
831
90
10
10
10
Sa
1
10
10 8
10
10
10
10
10
12
10
16
10
20
10
24
10
80
70
60
50
40
30
20
10
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
time [s]
Figure 1. Left: Influence of state parameter on relaxation modulus versus time. Right:
influence of state parameter on true stress versus true strain response. Dashed lines ():
rejuvenated state; solid lines (-): increasing thermal histories, i.e., quenched and
annealed, respectively.
yield stress
(10)
t
0
0
0
a1
T T t dt
(11)
DUa
1
1
T t Tref
R
(12)
teff t; T ta
Sa t; T c0 c1 log
t0
teff t; T ta
t0
(13)
processing
related
service
related
log(time)
Figure 2. Yield stress versus life-time: two regions; processing
related and service related.
832
teff t; T
DOI: 10.1002/mame.200900227
Z
0
tc
t_eff;c 0
0
0
a1
T Tc t dt
(14)
Experimental Part
Materials
The material used in this study was an injection molding grade of
PC: Lexan 141R, supplied as granules by Sabic Innovative Plastics
(Bergen op Zoom, The Netherlands). The number-average moleMacromol. Mater. Eng. 2009, 294, 829838
2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Sample Preparation
All tensile experiments were performed on samples machined from
injection-molded rectangular plates, see Figure 3 (left). The plates
with dimensions 70 70 1 mm3 were molded on an Arburg 320S
all-rounder 500150. The runner of the mold ensured uniform
filling, as proven by several short shot experiments. The only
variable used during the injection molding is the mold temperature. Melt temperature was kept constant at 285 8C, as was the
cooling time at 60 s. From the plates, bars with dimensions
70 10 1 mm3 were cut parallel to the flow direction and fitted
with gauge sections of 33 5 1 mm3, see Figure 3 (right).
To investigate the influence of the temperature history on the
distribution of yield stress over the thickness of a product by means
of micro-indentation, small bars with cross-sections 2 1 mm2
were taken from the centers of the molded square plates by a
precision machining operation. Subsequently, increasingly thinner
layers (1 mm) were removed from the surface of the 2 1 mm2
cross-section by a microtoming operation under cryogenic (liquid
nitrogen) conditions, this to minimize a possible influence of
machining on the thermodynamic state of the sample surface. A
Leica RM 2165 rotation microtome was used; for each sample a
fresh glass knife was taken. To obtain different temperature
histories, three mold temperatures were used 30, 90, and 130 8C.
To verify the performance after injection molding and the
influence of thermal history thereupon, tensile samples were taken
from injection molded plates processed at mold temperatures of 30
and 120 8C.
Finally, as an example of a real product, a thick-walled cupshaped sample, see Figure 4, was used. The bottom ring of the cup
has a diameter of 78 mm. The cup itself starts with an outer
diameter of 65 mm and has a gradually decreasing diameter up till
60 mm at the top. We define Figure 4 (left) as the upright position
(the way in which the samples was loaded in the load-frame). The
thickness was around 3 mm in all cross-sections. The cup-shaped
sample was injection molded on an Arburg 320S all-rounder
500150. The melt temperature was set to 285 8C; the injection
rate to 50 ccms1; and the packing pressure to 500 bars. Mold
temperatures were set to 30 and 130 8C. In both cases a cooling time
of 120 s was used.
All samples used in this study were stored at room temperature
after injection molding. For PC no increase in yield stress was to be
expected at room temperature based on the time-temperaturesuperposition results of Klompen et al.[23] It was also experimen-
www.mme-journal.de
833
Methods
Micro-indentation experiments were performed on a nanoindenter XP (MTS NanoInstruments, Oak Ridge, Tennessee) under
displacement control. The indenter has a flat tip, effectively a flatended cone with a top angle of 708 and a circular contact area with
diameter 10 mm. Correction for tip-sample misalignment was
performed using a specially designed alignment tool. For details on
the experimental technique see;[30,38] and for the alignment tool
see.[38]
Tensile tests were performed on a Zwick Z010 universal tensile
tester at a room temperature of 23 8C. Experiments were performed
_ 0 or engineering
by applying constant linear strain rates "_ x=l
stresses s F=A0 . Unless indicated otherwise, a standard
constant linear strain rate of 103 s1 was used. All tensile yield
stresses listed in the results section are engineering yield stresses,
and taken as the mean value of five experiments.
The cup-shaped samples were tested on a Zwick 1475 tensile
testing machine. The cups were placed in the machine with the
tapered section to the top (see Figure 4 left). The bottom plate used
has a flat circular recess to fit the outer diameter of the bottom ring
of the cup. Samples were loaded with constant displacement rates,
or constant forces. Experiments were performed at a room
temperature of 23 8C and corrected for the finite stiffness of the
setup.
Results
Influence of Thermal History on the Yield-Stress
Distribution
Previously[24,25] distributions of yield stresses throughout a
polymer product were predicted based on the differences in
temperature history experienced during molding. Fast
cooling, e.g., at the mold surface, limits physical aging
leading to a low yield stress, while during slow cooling, e.g.,
in the core of the product, aging is more pronounced as is the
increase in yield stress. Verification was done using a mean
yield stress calculated based on area averaging, and
excellent agreement was found for different mold temperatures and mold thicknesses. Here we will attempt to
also validate the distributions, first by using micro-
834
DOI: 10.1002/mame.200900227
25
center
surface
center
50
0
0
200
400
600
800
55
50
1000
time [s]
10 11 12
sheet #
Figure 7. Left: cooling histories as measured (-) and as calculated (). Right: measured
(*) and predicted (solid drawn line) yield stress distributions.
300
70
temperature [ C]
250
200
surface to center
150
100
Tmold=120C
surface to center
50
Tmold=30C
0
0
10
15
time [s]
20
25
30
temperature [ C]
60
Tmold=120C
50
Tmold= 30C
center
0
0.25
0.5
0.75
surface
1
normalized thickness []
Figure 8. Left: temperature versus time during the cooling of the injection molded
samples. Right: corresponding predicted yield stress distributions.
www.mme-journal.de
835
70
60
50
40
30
20
Tm = 120C
T = 30C
10
model prediction
0
0
0.04
0.06
0.08
0.1
836
y [mm]
y [mm]
strain []
DOI: 10.1002/mame.200900227
45
T = 130C
m
40
Tm = 30C
model prediction
force [kN]
35
30
25
20
15
10
5
0
0
displacement [mm]
Figure 13. Experimental and numerical results of compression
tests of a cup at a loading rate of 0.045 mms1 for samples with
mold temperatures 30 and 130 8C.
50
50
40
40
30
20
10
0 4
10
Tm = 130C
Tm = 30C
model prediction
3
10
10
10
10
10
30
20
Conclusion
(open markers)
(closed markers)
10
0 0
10
Tm = 130C
Tm = 30C
model prediction
1
10
10
10
10
10
10
10
timetofailure [s]
Figure 14. Left: maximum load versus loading rate. Right: applied load versus time-tofailure. Both for samples with mold temperatures 30 and 130 8C.
Macromol. Mater. Eng. 2009, 294, 829838
2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.mme-journal.de
837
838
DOI: 10.1002/mame.200900227