Thermal expansion coefficient and bulk modulus of polyethylene closed‐cell foams
Abstract
A regular Kelvin foam model was used to predict the linear thermal expansion coefficient and bulk modulus of crosslinked, closedcell, lowdensity polyethylene (LDPE) foams from the polymer and gas properties. The materials used for the experimental measurements were crosslinked, had a uniform cell size, and were nearly isotropic. Young's modulus of biaxially oriented polyethylene was used for modeling the cell faces. The model underestimated the foam linear thermal expansion coefficient because it assumed that the cell faces were flat. However, scanning electron microscopy showed that some cell faces were crumpled as a result of foam processing. The measured bulk modulus, which was considerably smaller than the theoretical value, was used to estimate the linear thermal expansion coefficient of the LDPE foams. © 2004 Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys 42: 3741–3749, 2004
Thermal Expansion Coefficient and Bulk Modulus of
Polyethylene ClosedCell Foams
O. ALMANZA,
1
Y. MASSOMOREU,
2
N. J. MILLS,
2
M. A. RODRÍGUEZPE
´
REZ
3
1
Departamento Fı´sica, Universidad Nacional de Colombia, Santafe´ de Bogota´, Colombia
2
Metallurgy and Materials, University of Birmingham, Birmingham, United Kingdom
3
Departamento Fı´sica de la Materia Condensada, Universidad de Valladolid, 47011 Valladolid, Spain
Received 18 September 2003; revised 14 June 2004; accepted 22 June 2004
DOI: 10.1002/polb.20230
Published online in Wiley InterScience (www.interscience.wiley.com).
ABSTRACT: A regular Kelvin foam model was used to predict the linear thermal
expansion coefﬁcient and bulk modulus of crosslinked, closedcell, lowdensity polyeth
ylene (LDPE) foams from the polymer and gas properties. The materials used for the
experimental measurements were crosslinked, had a uniform cell size, and were nearly
isotropic. Young’s modulus of biaxially oriented polyethylene was used for modeling the
cell faces. The model underestimated the foam linear thermal expansion coefﬁcient
because it assumed that the cell faces were ﬂat. However, scanning electron microscopy
showed that some cell faces were crumpled as a result of foam processing. The mea
sured bulk modulus, which was considerably smaller than the theoretical value, was
used to estimate the linear thermal expansion coefﬁcient of the LDPE foams.
© 2004
Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys 42: 3741–3749, 2004
Keywords: bulk modulus; foams; modeling; polyethylene (PE); thermal expansion
INTRODUCTION
The uses of crosslinked, closedcell polyoleﬁn
foams
1
include packaging, automotive applica
tions, and thermal insulation for pipes. The prop
erties required for these applications include low
thermal expansion, high thermal stability, and
good mechanical properties. In the shoe industry,
polyethylene (PE) foams are bonded to textile
substrates; differences between the thermal ex
pansion of the foam and that of the substrate
could cause tensile stresses at the material inter
face, possibly leading to adhesive failure.
The elastic moduli
2,3
and process expansion
4
of
such foams have been modeled in terms of their
cellular microstructure. A regular lattice of tetra
kaidecahedral cells, known as the Kelvin foam
(Fig. 1), was used to compute the foamcompres
sive Young’s modulus.
2
However, to obtain rea
sonable predictions, faces were assumed to buckle
if compressed in plane. Mahapatro et al.
4
used the
Kelvin foam model, in which the cell faces re
mained ﬂat under biaxial tension, to compute the
equilibrium density of crosslinked PE foams from
the gas content and the modulus of the PE melt.
However, only qualitative models exist for the
thermal expansion coefﬁcient and thermal con
ductivity. These properties vary with the foam
relative density (R), which is deﬁned as the foam
density divided by the polymer density. They tend
to the gaseous value when R tends to 0 and to the
polymer value when R tends to 1. The thermal
conductivity
5–7
varies almost linearly with R for
R ⬎ 0.15. In previous articles,
8 –10
the thermal
Correspondence to: N. J. Mills (Email: n.j.mills@bham.ac.uk)
Journal of Polymer Science: Part B: Polymer Physics, Vol. 42, 3741–3749 (2004)
© 2004 Wiley Periodicals, Inc.
3741
expansion coefﬁcient of PE foams has been
reported
1. To increase slightly between 0 and 40 °C.
2. To be approximately equal to the PE value
for foam densities greater than 80 kg/m
3
but to increase at lower densities.
3. To decrease if Young’s modulus (E)ofPE
increases.
4. To be anisotropic, with lower values in the
direction in which cells are elongated.
These trends have been explained qualita
tively
8,9
in terms of the gas and polymer contri

butions to the total thermal expansion coefﬁcient.
However, the goal has been to develop a model
predicting the thermal expansion coefﬁcient from
the polyhedral cell structure and the material
properties. RodriguezPerez and de Saja
11
showed
that cell faces contain oriented crystalline struc
tures; consequently, the face properties must be
for oriented lowdensity polyethylene (LDPE).
Mills and Gilchrist
12
showed that air diffuses
slowly out of LDPE foams when compressive
stresses are applied. For a cylindrical sample
10 mm in diameter and 10 mm high, it will take
more than 10 h for signiﬁcant gas loss and several
weeks for the foam to reach equilibrium. There
fore, the longterm storage temperature of the
foam affects the equilibrium absolute air pressure
in the cells, but there is no gas loss during the
thermal expansion measurements, which typi
cally last 30 min. Consequently, the model can
ignore gas diffusion.
The Kelvin model can be used to predict the
bulk modulus of the foam (K
F
). Kraynik et al.
13
predicted, using Abaqus ﬁnite element analysis
for a material with E and a Poisson’s ratio (
)of
0.49, that K
F
of R could be obtained as follows:
K
F
⫽ 0.435ER (1)
In general, the same model should be used to
predict the bulk modulus and thermal expansion
coefﬁcient. Both properties involve volume
changes, with isotropic expansion of the foam.
KELVIN FOAM MODEL
General
In the model, the cell faces are assumed to be ﬂat
and to remain ﬂat, although there is scanning
Figure 2. (a) Kelvin closedcell foam unit cell with
mirror symmetry planes at the boundaries (faces meet
ing at G have been omitted for clarity) and (b) left side
surface of the unit cell, on which a force balance is
performed.
Figure 1. Packing of three tetrakaidecahedra with
planar faces in a bodycenteredcubic Kelvin foam
lattice.
3742 ALMANZA ET AL.
electron microscopy (SEM) evidence for some cell
faces being wrinkled or buckled. Although PE
foams have irregular cell shapes, rather than the
uniformsize and shape cells of the Kelvin model,
it is unlikely that the cell shape irregularity
greatly affects the thermal expansion coefﬁcient.
The model is for foams with isotropic cell shapes.
Figure 2(a) shows the chosen unit cell, which is
stacked (repeated by translational symmetry op
erations) to form the complete foam structure. It
contains four halves of hexagonal faces, four
quarters of horizontal square faces (because these
are shared with two structural units, they con
tribute half a square face), and two halves of
vertical square faces. Thus, the 4:3 ratio of hex
agonal faces to square faces is the same in the
unit cell as that in a single isolated tetrakaideca
hedral cell. For
␦
Ⰶ L (where
␦
is the face thick
ness and L is the edge length), if there is negligi
ble polymer in the edges, R is given by
R ⫽
3
冑
2
16
共1 ⫹ 2
冑
3兲
␦
L
(2)
To generalize the analysis either for an external
application of relative pressure P or for thermal
expansion when P is 0, we consider the total com
pressive force acting perpendicularly to a vertical
boundary of the unit cell. This boundary has area
2公2L
2
. A biaxial tensile stress (
f
) acts in a hex

agonal cell face of width 公3L and thickness
␦
and
in two shared, sectioned, halfsquare faces of total
width 0.5L and thickness
␦
/2, both of which are
perpendicular to the boundary plane. The cellair
relative pressure (p
C
) also acts on the area 2公2L
2
(ignoring the correction for the face crosssec
tional area as R Ⰶ 1). The total compressive
force is
P2
冑
2L
2
⫽ p
C
2
冑
2L
2
⫺
f
共
冑
3 ⫹ 0.5兲L
␦
(3)
so
f
⫽
2
冑
2
0.5 ⫹
冑
3
L
␦
共p
C
⫺ P兲 (4)
Combining eqs 2 and 4 gives
f
⫽
3共p
C
⫺ P兲
2R
(5)
Equation 6 in Mahapatro et al.
4
is incorrect be

cause of an error in evaluating the area of the
polymer crossing the left boundary of the unit cell
[Fig. 2(b)]. The term in brackets in eq 3 was
incorrectly taken to be 3, and this led to a face
stress that was 74.4% of that in eq 5.
The polymer is assumed to be linearly elastic,
so the tensile elastic strain in each face (
E
)is
E
⫽ 共1 ⫺
兲
f
E
(6)
The cell face material is anisotropic; E is for
stresses acting in the plane of the face, and the
linear thermal expansion coefﬁcient (
␣
lP
)isfor
the expansion in the face length and width.
is
also appropriate for the anisotropic material.
Bulk Modulus
If eq 5 is divided by the tensile strain in the face
(which is also the tensile strain in the foam), and
eq 6 used to replace
f
/
E
, we obtain
ER
1 ⫺
⫽
3共p
C
⫺ P兲
2
E
(7)
For small volume strains, the foam volume strain
(
V
) is equal to 3
E
, and K
F
is deﬁned by
K
F
⬅ ⫺
P
V
(8)
The bulk modulus of the air (K
A
) in the cells is
K
A
⬅ ⫺
p
C
F
⫽ p
0C
(9)
where p
0C
is the absolute air pressure in the cells.
Because the foam was stored at atmospheric pres
sure (p
a
) for several months before the tests, this
was assumed to be p
a
. Therefore, we obtain
K
F
⫽ K
p
⫹ K
A
⫽
2ER
9共1 ⫺
兲
⫹ p
a
(10)
where K
p
is the polymer bulk modulus.
Figure 3(a) shows a onedimensional (1D) ver
sion of the bulk modulus experiment. The poly
mer and cellair springs act in parallel, so their
bulk moduli are summed, as expressed by eq 10.
The polymer contribution to eq 10 is in agreement
POLYETHYLENE CLOSEDCELL FOAMS 3743
with Kraynik et al.’s
13
eq 1. Abaqus would only
consider the stresses in the solids, so the effect of
the compressed cell air would be ignored. Equa
tion 10 contrasts with the bulk modulus of ER/9
for an opencell Kelvin foam.
14
Opencell foams
are connected by edges running in one dimension,
whereas closedcell foams are connected by faces
running in two dimensions. The doubled connec
tivity causes a doubling of the polymer contribu
tion to the bulk modulus, with an extra effect of
.
As LDPE is viscoelastic rather than linearly
elastic, E in eq 10 can be replaced by the creep
modulus [E(t)] on timescale t of the experiments
to obtain a timedependent bulk modulus.
Thermal Expansion
The linear thermal expansion coefﬁcient (
␣
l
)is
deﬁned as follows:
␣
l
⫽
1
l
0
dl
dT
(11)
where l
0
is the sample length at the reference
temperature of 25 °C and l is the length at tem
perature T. Figure 3(b) shows a 1D version of the
thermal expansion experiment; there is no exter
nal pressure change, and the polymer and cellair
springs act in series. The foam volume expansion
coefﬁcient (
␣
VF
) is the sum of that of the polymer
structure and the excess air volume expansion,
constrained by the foam structure:
␣
VF
⫽
␣
VP
⫹ 共
␣
VA
⫺
␣
VP
兲
K
A
K
F
(12)
The appendix gives the details of the derivation.
When eq 12 is divided by 3, the foam linear ex
pansion coefﬁcient (
␣
lF
) is yielded:
␣
IF
⫽
␣
IP
⫹ 共
␣
IA
⫺
␣
IP
兲
p
0
K
F
(13)
The general form of this equation allows its use
either with the theoretical K
F
value of eq 10 or
with the experimental value. As air behaves as an
ideal gas, its volume thermal expansion coefﬁ
cient (
␣
VA
) is equal to 1/T, where T is the absolute
temperature; consequently, its linear thermal ex
pansion coefﬁcient (
␣
lA
) is 11.4 ⫻ 10
⫺4
K
⫺1
at
293 K.
LDPE FOAM CHARACTERIZATION
The assumptions in the model about the foam
geometry need to be veriﬁed; in particular, it is
necessary to determine whether the cells are
equiaxed, the cell faces are ﬂat, and the fraction of
the polymer in the edges is low. Furthermore, the
appropriate polymer properties must be used in
the model. The modeling
4
of cell expansion during
foam processing suggests that the crystals have a
preferred orientation in the cell faces. Conse
quently, the face properties are anisotropic; the
inplane E value and linear thermal expansion
coefﬁcient differ from those in the direction nor
mal to the cell faces. Hence, polymer properties (E
and
␣
lP
), measured on bulk LDPE with a spheru

litic microstructure, should not be used in the
thermal expansion model. Ideally, the properties
should be measured in the plane of the foam cell
faces, but such micromechanic measurements
Figure 3. 1D representations of (a) the bulk modulus and (b) the thermal expansion
experiments, with the cellair and polymer structure acting as springs. The thermal
expansion is by (
␣
VA
⫺
␣
VP
)⌬T.
3744 ALMANZA ET AL.
have not been made on PE foam faces. Conse
quently, E and
␣
lP
, values measured in the plane
of a biaxially oriented LDPE ﬁlm, which has a
similar microstructure, are used for the modeling.
Foams
Crosslinked, closedcell LDPE foams, manufac
tured by Zotefoams (Croydon, United Kingdom),
were used to check the validity of the theoretical
model because they do not contain foaming agent
residues, they have nearly isotropic cellular
shapes, and the polymer crystallinity is indepen
dent of the foam density. The product codes and
densities of the foams are given in Table 1. The
black foam (LD70B) contains approximately
2 wt % carbon black. The crystallization charac
teristics of the solid LDPE sheet, used in the
Zotefoams process, are included in the table. Dur
ing the process, the LDPE is compounded with a
peroxide crosslinking agent and is extruded as a
thick sheet, which is passed through a hot oven to
effect crosslinking to a gel content of approxi
mately 40%. Slabs, cut from the extruded sheet,
are subjected to several hundred bars of nitrogen
gas pressure in an autoclave for several hours, at
a temperature above the PE crystal melting point,
so that nitrogen dissolves.
15
After the slabs cool,
the pressure is reduced to zero, and the slabs are
placed in a larger autoclave and reheated above
the polymer melting point under a lower pres
sure. When the pressure is released, the foam
expands to its ﬁnal low density. It is taken out of
the autoclave and cooled to room temperature.
Foam Characterization
Density measurements were carried out with the
density determination kit for a Mettler AT261
balance according to Archimedes’ principle.
SEM was used to assess the cellular structure.
Foam samples were microtomed at a low temper
ature and, after being vacuumcoated with gold,
were examined in a JEOL JSM820 microscope.
The foam needed to be kept in vacuo for more
than the time for 90% air loss by diffusion, which
was estimated
16
to be 10 h for a 5mm cubic
sample of LD19 foam, for the artiﬁcial expansion
of cells on the vacuum application to disappear.
RodriguezPerez and de Saja
11
described the
determination of the cell diameter, cell shape an
isotropy, crystallinity, and crystal orientation for
these foams. They showed, by SEM of etched
foam, that twodimensional (2D) spherulites nu
cleated on both cell face surfaces. The average
surface diameter (D
s
) of the 2D spherulites varies
with the foam density, but it is a nearly constant
multiple (6) of
␦
. The mean cell diameter (D
c
)in
Table 1 is the mean of the average cell diameters
in the three directions. The cell shape anisotropy
ratio
17
is deﬁned as the ratio of the largest D
c
value to the smallest D
c
value for the three direc

tions of measurement.
␦
of thirty cell faces, chosen randomly, was
measured directly from the SEM screen, and the
mean value was calculated. The 95% conﬁdence
interval of these measurements was ⫾8% of the
mean. Finally, the mass fraction in the edges (f
s
)
was obtained with the method of Kuhn et al.,
18
which assumes that the cells are regular dodeca
hedra with pentagonal faces and that the faces
have uniform thickness where they meet the
edges. From the average values of D
c
, the edge
diameter (D
e
), and
␦
, the volumes of the polymer
in the edges and faces are obtained as follows:
V
e
⫽
␦
(1.3D
c
2
⫺ 5.4D
c
D
e
⫹ 1.7D
c
2
) (14)
V
f
⫽ 2.8D
c
D
e
2
⫺ 3.9D
e
3
(15)
Table 1. Foam Densities, Cell Size Parameters, and Crystallinity for Zotefoams LDPE Foams
Product
Code
Density
(kg/m
3
)
D
c
(
m)
Cell
Anisotropy
Ratio
␦
(
m)
Edge
Fraction
Melting
Point
(°C)
Crystallinity
(%)
LD15 16.7 313.5 1.00 1.4 0.22 105.9 40.6
LD18 22.5 879.7 1.01 5.8 0.21 108.4 41.8
LD24 24.6 311.9 1.01 1.9 0.16 108.6 43.8
LD29 30.7 528.1 1.02 4.2 0.24 108.5 42.9
LD33 32.0 424.4 1.00 3.6 0.28 105.8 43.9
LD33(1) 32.5 396.9 0.99 2.5 0.36 108.6 41.6
LD60 58.5 773.4 1.02 10.3 0.24 109.0 42.1
LD70B 69.5 528.1 1.04 6.0 0.35 106.6 40.8
LD Solid 910 — — — — 105.9 40.6
POLYETHYLENE CLOSEDCELL FOAMS 3745
Edge crosssectional areas were measured at four
randomly chosen positions on each micrograph.
These and the average cell diameter were used to
calculate f
s
. The 95% conﬁdence error of these
measurements was ⫾8% of the mean.
The linear thermal expansion coefﬁcient was
measured with a PerkinElmer TMA7 thermome
chanical analyzer. The test specimens were cylin
ders 10 mm in diameter and 5–10 mm high. These
were placed between parallel metal plates 10 mm
in diameter. An applied compressive stress of
130 Pa kept the plates in contact with the sample;
it caused a 0.06% elastic strain at room temper
ature for a lowdensity foam. As the thermal ex
pansion strain between 5 and 25 °C was 20 times
higher than the elastic strain, the latter could be
neglected. The sample height direction was the
direction perpendicular to the foam sheet for all
foams.
Two types of measurements were made:
1. The temperature was raised from 20 to
130 °C at 5 °C/min to characterize the over
all thermal expansion response.
2. To measure the linear thermal expansion
coefﬁcient, away from any thermal transi
tions of the polymer,
8
the foam was cooled
from room temperature to 5 °C, at which it
was kept for 15 min. It was then heated
from 5 to 25 °C at 1 °C/min and kept at
25 °C for 15 min. Each material was mea
sured three times with new samples. The
95% conﬁdence limits of the measurements
were ⫾6% of the mean.
RESULTS
Thermal Expansion Coefficient
The linear thermal expansion coefﬁcient of
the LD solid sheet from Zotefoams was 1.3
⫻ 10
⫺4
K
⫺1
. Figure 4 shows typical data for the
variation of the linear thermal expansion coefﬁ
cient of the isotropic foam with the temperature.
The coefﬁcient increases slowly from 20 to 30 °C
and then decreases at higher temperatures as the
melting process of the crystals starts. Figure 5
shows experimental values from 5 to 25 °C as a
function of the density.
Cell Geometry
Figure 6 shows SEM images of an LD19 foam
immediately upon insertion into the SEM instru
ment and after 64 h of storage in vacuo. The cut
faces labeled A, B, and C were initially ﬂat,
whereas they appear buckled after 64 h. The com
plete face D appeared ﬂat initially, whereas it was
wrinkled after 64 h. The sudden exposure to a
vacuum caused the stretching of the faces of com
plete cells at the surface of the foam, but, after
64 h, the majority of the air diffused from the
foam, leaving the faces in the same wrinkled state
in which they had before being placed into the
instrument.
The melting points and crystallinity (ca. 110 °C
and 40%, respectively) of the foams, given in
Table 1, are typical of LDPE. The cell size did not
depend on the density, whereas the edge mass
fraction had a mean value of 0.25.
Figure 4. Linear thermal expansion coefﬁcient, in
the direction perpendicular to the foam sheet, versus
the temperature for Zotefoams LD24.
Figure 5. Experimental linear expansion coefﬁcient
data for Zotefoams foams versus the density, compared
with theoretical predictions for
␣
lP
⫽ 3.3 ⫻ 10
⫺4
K
⫺1
and foam bulk moduli proportional to R.
3746
ALMANZA ET AL.
Linear Thermal Expansion Coefficient of the
Oriented LDPE Films
Mills and Zhu
2
measured the tensile response of a
biaxially stretched LDPE packaging ﬁlm, with a
density of 910 kg m
⫺3
and a thickness of 45
m,
which had a biaxial draw ratio of 6.5. Its E value
at a strain rate of 7 ⫻ 10
⫺4
s
⫺1
, up to 1.5% strain,
was 202 MPa. The E value was consistent with
the range of values (175–225 MPa) measured for
other LDPE blown ﬁlms of different degrees of
orientation.
19
No published values for the linear thermal ex
pansion coefﬁcient of biaxially oriented LDPE
ﬁlm could be found. The anisotropic thermal ex
pansion of 2D PE spherulites was discussed by
Barham and Keller.
20
The crystal b axis appears
to be similar to that of the whole 2D spherulite
and to be very small. For both amorphous and
semicrystalline polymers, the thermal expansion
coefﬁcient along the draw direction decreases
with increasing deformation and increases in the
direction perpendicular to the draw direction.
21
For uniaxially oriented LDPE ﬁlms, the linear
expansion coefﬁcient decreases with drawing, to
values close to 2 ⫻ 10
⫺5
K
⫺1
in the parallel direc

tion and to 2 ⫻ 10
⫺4
K
⫺1
in the perpendicular
direction.
22
COMPARISON WITH THEORY
MassoMoreu and Mills
16
showed that the slightly
wrinkled cell faces of LDPE foams affect their re
sponse to pressure changes. They found that the
bulk modulus of an LD19 foam was 440 kPa for
relative pressures of 0 to ⫺20 kPa, but it increased
to 820 kPa for relative pressures of ⫺40 to ⫺80 kPa.
However, the theoretical value for a Kelvin foam
with ﬂat cell faces (E ⫽ 202 MPa and
⫽ 0.4) is
1550 kPa. Hence, the bulk modulus is lower than
the theoretical value, even for large volumetric
expansions. A temperature increase of 10° causes
a linear strain of 6.7 ⫻ 10
⫺3
in an LD19 foam,
whereas a 20kPa decrease in pressure causes a
linear strain of 15.2 ⫻ 10
⫺3
. The strains are com

parable in magnitude, so the bulk modulus data is
at an appropriate strain level for predicting the
thermal expansion coefﬁcient.
Given the difﬁculty of ﬁnding values for
␣
lP
of
biaxially oriented LDPE cell faces, the asymptotic
value of the foam linear thermal expansion coef
ﬁcient, when the density became large, was used.
This was estimated to be 3.3 ⫻ 10
⫺4
K
⫺1
(dis

cussed later), that is, higher than the 1.3
⫻ 10
⫺4
K
⫺1
value for bulk LDPE of the same
crystallinity percentage. If this is used with E
⫽ 202 MPa and the theoretical K
F
value in eq 12,
the predicted linear expansion coefﬁcient for
LD19 foam is 3.8 ⫻ 10
⫺4
K
⫺1
, much lower than
the experimental value of 6.7 ⫻ 10
⫺4
K
⫺1
. The too
small experimental bulk modulus and the too
large experimental linear thermal expansion co
efﬁcient result from the foam cells having buckled
faces. Cells with heavily buckled faces would re
spond like a bellows and so would have a linear
thermal expansion coefﬁcient almost that of air,
11.4 ⫻ 10
⫺4
K
⫺1
.
It is known that K
F
is 440 kPa for an LDPE
foam with a density of 19 kg m
⫺3
. To estimate the
bulk moduli of LDPE foams with densities
greater than 19 kg m
⫺3
, we assumed that the
bulk modulus was proportional to R, as predicted
by eq 10, being 440 kPa when R was 0.018. With
this information, the data in Figure 5 for the foam
Figure 6. Micrographs of LD19 foam (a) soon after insertion into the SEM instrument
and (b) after 64 h in vacuo. Cut faces A–C and complete face D are (a) ﬂat and (b) buckled.
POLYETHYLENE CLOSEDCELL FOAMS 3747
linear thermal expansion coefﬁcient versus the
density were ﬁtted with a range of
␣
P
values. The
best ﬁt at high densities was for a cell face ther
mal expansion coefﬁcient of
␣
lP
⫽ 3.3 ⫻ 10
⫺4
K
⫺1
.
Figure 5 shows that the ﬁt was good for relative
densities of 0.035 or greater, but it was an under
estimate for the lowest relative density foams.
DISCUSSION
The model, in which the cell faces are ﬂat, overes
timates the bulk modulus and therefore underesti
mates the linear thermal expansion coefﬁcient of
LDPE foams. Its predictions should be better at
foam densities higher than those tested. However,
at high densities (Fig. 5), the predicted foam ther
mal expansion coefﬁcient approaches that of LDPE,
so the test of the model is not demanding.
Mills and Zhu
14
predicted the E values of
closedcell foams with the Kelvin foam model, in
which cell faces could not support inplane com
pressive forces. This underestimated E of LDPE
foams, but an alternative theory,
13
in which the
faces remained ﬂat, overestimated E. The trend
line of E versus the density (Fig. 7) is steeper than
that of either theory, suggesting that face buck
ling becomes easier at lower densities. The data
in this ﬁgure were measured for the Zotefoams in
slow compressive tests.
23
However, although face
buckling inﬂuences the experimental E values,
the pattern of stresses in the cell faces differs
from that in the bulk modulus and thermal ex
pansion experiments.
The foam processing route causes the cell faces
to be slightly wrinkled. It is assumed that the
faces buckle during the cooling of the foam from
the melt, when the cell gas relative pressure be
comes negative. In extruded LDPE foams, for
which pentane is used as a blowing agent, it ap
pears necessary to add permeability modiﬁers to
the LDPE to prevent the collapse of cells after
extrusion because the diffusion rate of pentane
out of the cells is faster than the diffusion rate of
air into the cells.
24
In the future, efforts should be made to directly
measure the linear thermal expansion coefﬁcients
of polymer foam faces. If such data and bulk mod
ulus measurements for a range of foam densities
were available, it would be possible to check the
Kelvin model predictions more rigorously.
CONCLUSIONS
The theoretical model predicts K
F
and linear ther

mal expansion coefﬁcients from the polymer and
air thermal expansion coefﬁcients and E of the
polymer in the cell faces. This model assumes
that the cell faces are all ﬂat, but SEM shows that
many cell faces are buckled. Experimental bulk
moduli and linear thermal expansion coefﬁcients
of LDPE foams are consistent with the presence
of wrinkled and buckled cell faces, which signiﬁ
cantly change the values.
If the experimental K
F
value is used, and the
polymer linear thermal expansion coefﬁcient is
estimated from the highdensity asymptotic be
havior of the foam thermal expansion coefﬁcients,
reasonable predictions of the foam linear thermal
expansion coefﬁcient can be made.
Financial assistance from Junta de Castilla y Leo´n
(VA026/03), from La Secretaria de Estado de Educacio´n
y Universidades (Spain) for a postdoctoral grant
(O. Almanza), and from the Engineering and Physical
Science Research Council for Y. MassoMoreu is grate
fully acknowledged. The authors thank Zotefoams PLC
for supplying the foams.
APPENDIX: DERIVATION OF THE
THERMAL EXPANSION COEFFICIENT
The total (equibiaxial) tensile strain (
T
) in each
cell face, as a result of a temperature increase (⌬T
⫽ T ⫺ T
0
), is the sum of the elastic and thermal
strains (for any closedcell foam model):
Figure 7. Experimental E data for Zotefoams LDPE
foams versus the density, compared with the theoreti
cal models of Mills and Zhu
14
and Kraynik et al.
13
3748 ALMANZA ET AL.
T
⫽
␣
IP
⌬T ⫹ 共1 ⫺
兲
f
E
(A1)
This can be rearranged to yield
␣
IF
⬅
T
⌬T
⫽
␣
IP
⫹
f
共1 ⫺
兲
E⌬T
(A2)
For the Kelvin foam model,
f
is given by eq 5 for
P ⫽ 0. When this is substituted, we obtain
␣
IF
⫽
␣
IP
⫹
3p
C
共1 ⫺
兲
2ER⌬T
(A3)
In terms of the volume thermal expansion coefﬁ
cients, we obtain
␣
VF
⫽
␣
VP
⫹
p
C
K
P
⌬T
(A4)
Figure 3(b) shows a 1D representation of a ther
mal expansion experiment, in which the volume
strain in the cell air must be the same as the
volume strain in the polymer when there is a
temperature rise (⌬T). The air and polymer
springs appear to be in series:
共
␣
VA
⫺
␣
VP
兲⌬T ⫺
p
C
P
0
⫽ 3共1 ⫺
兲
f
E
(A5)
Rearranging this and substituting for
f
with
eq 5, we obtain
␣
VA
⫺
␣
VP
⫽
p
C
⌬T
冉
1
p
0
⫹
9共1 ⫺
兲
2E
冊
⫽
p
C
⌬T
冉
1
K
A
⫹
1
K
P
冊
(A6)
Substituting for p
C
/⌬T from eq A6 into eq A4 gives
␣
VF
⫽
␣
VP
⫹
1
K
p
␣
VA
⫺
␣
VP
冉
1
K
A
⫹
1
K
p
冊
⫽
␣
VP
⫹ 共
␣
VA
⫺
␣
VP
兲
K
A
K
F
(A7)
which is eq 12.
REFERENCES AND NOTES
1. Park, C. P. In Handbook of Polymeric Foams and
Foam Technology; Klempner, D.; Frisch, K. C.,
Eds.; Hanser: Munich, 1991; Chapter 9.
2. Mills, N. J.; Zhu, H. X. J Mech Phys Solids 1999, 47,
669 – 695.
3. Roberts, A. P.; Garboczi, E. J. Acta Mater 2001, 49,
189 –197.
4. Mahapatro, A.; Mills, N. J.; Sims, G. L. S. Cell
Polym 1998, 17, 252–270.
5. Leach, A. G. J Phys D: Appl Phys 1993, 26, 733–
739.
6. Collishaw, P. G.; Evans, J. R. G. J Mater Sci 1994,
29, 486 – 498.
7. Glicksman, L. R. In Low Density Cellular Plastics:
Physical Basis of Behaviour; Hilyard, N. C.; Cun
ningham, A., Eds.; Chapman & Hall: London, 1994;
Chapter 5.
8. Rodrı´guezPe´rez, M. A.; Alonso, O.; Duijsens, A.; de
Saja, J. A. J Polym Sci Part B: Polym Phys 1998,
36, 2587–2596.
9. Rodrı´guezPe´rez, M. A.; Duijsens, A.; de Saja, J. A.
J Appl Polym Sci 1998, 68, 1237–1244.
10. Rodrı´guezPe´rez, M. A.; Alonso, O.; Souto, J.; de
Saja, J. A. Polym Test 1997, 16, 287.
11. RodriguezPerez, M. A.; de Saja, J. A. J Macromol
Sci Phys 2002, 41, 761–775.
12. Mills, N. J.; Gilchrist, A. J Cell Plast 1997, 33,
264 –292.
13. Kraynik, A. M.; Neilsen, M. K.; Reinelt, D. A.;
Warren, W. E. In Foams and Emulsions; Sadoc,
J. F.; Rivier, N., Eds.; NATO ASI Series E 354;
Kluwer: Dordrecht, 1999; pp 259 –286.
14. Mills, N. J.; Zhu, H. X. J Mech Phys Solids 1999,
47, 669 – 695.
15. Eaves, D. E.; Witten, N. Soc Plast Eng ANTEC ’98,
1998, vol. 2, 1842–1849.
16. MassoMoreu, Y.; Mills, N. J. Polym Test 2004, 23,
313–322.
17. Cowin, S. C. J Mater Sci 1991, 26, 5155–5157.
18. Kuhn, J.; Ebert, H. P.; ArduiniSchuster, M. C.;
Bu¨ttner, D.; Fricke, J. Int J Heat Mass Transfer
1992, 35, 1795–1801.
19. Patel, R. M.; Butler, T. T.; Walton, K. L.; Knight,
G. W. Polym Eng Sci 1994, 34, 1506 –1513.
20. Barham, P. J.; Keller, A. J Mater Sci 1977, 12,
2141–2148.
21. Choy, C. L. In Developments in Oriented Polymers;
Ward, I. M., Ed.; Applied Science: London, 1982; pp
121–151.
22. KacarevicPopovic, Z.; Kostoski, D.; Novakovic,
L. J. Radiat Phys Chem 1999, 55, 645– 658.
23. Almanza. O. Ph.D. Thesis, University of Vallado
lid, 1999.
24. Dieckmann, D.; Holtz, B. J Vinyl Addit Technol
2000, 6, 34 –38.
POLYETHYLENE CLOSEDCELL FOAMS 3749
 CitationsCitations13
 ReferencesReferences27
μCTBased Analysis of the Solid Phase in Foams: Cell Wall Corrugation and other Microscopic Features
 "It is difficult to estimate a value for the error of this 3D methodology considering the unclear frontier separating the strut from the wall within a continuous structure. On the other hand, there is no exact reference method for comparison although the values are in congruence with the literature for these materials (Almanza et al., 2004; RodíguezPérez et al., 2008). In fact, alternative methodologies should be less accurate (Kuhn et al., 1992; Glicksman, 1994). "
[Show abstract] [Hide abstract] ABSTRACT: This work presents a series of threedimensional computational methods with the objective of analyzing and quantifying some important structural characteristics in a collection of lowdensity polyolefinbased foams. First, the solid phase tortuosity, local thickness, and surface curvature, have been determined over the solid phase of the foam. These parameters were used to quantify the presence of wrinkles located at the cell walls of the foams under study. In addition, a novel segmentation technique has been applied to the continuous solid phase. This novel method allows performing a separate analysis of the constituting elements of this phase, that is, cell struts and cell walls. The methodology is based on a solid classification algorithm and evaluates the local topological dissimilarities existing between these elements. Thanks to this method it was possible to perform a separate analysis of curvature, local thickness, and corrugation ratio in the solid constituents that reveals additional differences that were not detected in the first analysis of the continuous structure. The methods developed in this work are applicable to other types of porous materials in fields such as geoscience or biomedicine. "The subsequent contraction of the edges probably causedTable 5 Predictions for 2cell high Kelvin model, with cell diameter 1 mm, compressed along [ the faces to wrinkle. These issues were partly addressed by Almanza et al. (2004) who attempted to predict the thermal expansion coefficient of PE foams of different densities. The predicted effect of initiallybowed faces in a LDPE foam model compressed in the [1 1 1] direction, was to reduce the compressive yield stress by 11%, which is not large given the uncertainties in the polymer properties. "
[Show abstract] [Hide abstract] ABSTRACT: Finite element analysis, of regular Kelvin foam models with all the material in uniformthickness faces, was used to predict the compressive impact response of lowdensity closedcell polyethylene and polystyrene foams. Cell air compression was analysed, treating cells as surfacebased fluid cavities. For a typical 1 mm cell size and 50 s1 impact strain rate, the elastic buckling of cell faces, and popin shape inversion of some buckled square faces, caused a nonlinear stress strain response before yield. Pairs of plastic hinges formed across hexagonal faces, then yield occurred when trios of faces concertinaed. The predicted compressive yield stresses were close to experimental data, for a range of foam densities. Air compression was the hardening mechanism for engineering strains < 0.6, with face toface contact also contributing for strains > 0.7. Predictions of lateral expansion and residual strains after impact were reasonable. There were no significant changes in the predicted behavior at a compressive strain rate of 500 s1. [Show abstract] [Hide abstract] ABSTRACT: The purpose of this paper is to review the most significant developments of the last 10 years in the field of crosslinked polyolefin foams. The methods to produce the foams, the relationships between structure and properties, and the main applications of these materials are briefly reviewed. Topics of possible future research are proposed. Polyolefinfoams–Polyethylenefoams–Crosslinking–Foamingagents–Physicalpropertiesoffoams
People who read this publication also read
Data provided are for informational purposes only. Although carefully collected, accuracy cannot be guaranteed. Publisher conditions are provided by RoMEO. Differing provisions from the publisher's actual policy or licence agreement may be applicable.
This publication is from a journal that may support self archiving.
Learn more