PolymerVol. 38 No. 11, pp.

2819-2828, 1997
(~) 1997 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
ELSEVIER Plh S0032-3861 (96)00826-9 0032-3861/97/$17.00 + 0.00

Structure-property relationships of flexible

polyurethane foams
Dimitrios V. Dounis and Garth L. Wilkes*
Department of Chemical Engineering, Polymer Materials and Interfaces Laboratory, Virginia
Polytechnic Institute and State University, Blacksburg, VA 24061-0211, USA
(Received 11 July 1996)

In this study the viscoelastic and compression set properties were studied as a function of temperature and
humidity for a series of moulded foams based on toluene diisocyanate (TDI) and glycerol initiated ethylene-
oxide-capped propylene-oxide. The results were compared to those obtained on conventional slabstock
foams based on TDI and glycerol initiated propylene-oxide. These comparisons were made to delineate and
clarify distinct differences between these two different but very important systems. It was found that high
temperatures and humidities 'plasticized' the moulded foams to a greater extent than the slabstock foams.
Moulded foams displayed a higher compression set value and higher load decay value in the viscoelastic
measurements than slabstock foams. In an attempt to understand these dramatic differences, the two types
of 'cross-links' (covalent cross-links and urea-based phase separated hard segment domains) were more
directly compared. It was found that the hard segment domains in the slabstock foam had a much higher
level of short range ordering and cohesiveness. This was first confirmed by comparing the wide-angle X-ray
scattering (WAXS) patterns where the amorphous character was much more pronounced in the moulded
systems of equal water/TDI content. The WAXS behaviour of the slabstock system distinctly displayed
short range ordering of the TDI based units. Second, Fourier transform infrared (FTi.r.) data from the
carbonyl region showed that the slabstock foam had a much higher level of bidentate urea (strong hydrogen
bonding) within the hard segments. Conversely, the moulded foam displayed no bidentate urea but only
monodentate urea or weaker hydrogen bonding within the hard segments. This is not to imply that
microphase separation did not occur since small angle X-ray scattering (SAXS) clearly showed evidence of a
phase separated morphology for both types of foam systems. It is thus concluded that the dramatic
differences between the mechanical properties of moulded and slabstock foams are due in part to differences
in the covalent network but mostly due to the lower and weaker ordering of the hard segments in moulded
systems making these physical cross-links more labile at higher temperatures and humidities. These
morphological differences are believed to be due primarily to two differences in the formulation components
and processing between the two studied systems. First, the ethylene-oxide capping used in the polyol of
moulded foams to increase the reactivity is known to also increase the compatibility between the hard and
soft segments. Second, the addition of diethanolamine (DEOA) added in the moulded foam formulation to
decrease demould times by enhancing more rapid cross-linking has been shown to prevent the full local
packing development of the hard-segment domains (physical cross-links). © 1997 Elsevier Science Ltd.

(Keywords: polyurethane foam; slabstock foam; moulded foam; viscoelastic; compression set; morphology; structure-
property behaviour)

INTRODUCTION foams. For example, the network structure of a typical

polyurethane foam is comprised of both chemical and
Polyurethane foams are used in a very wide range of
physical 'cross-links'. The chemical cross-links arise
applications primarily in the areas of insulation, packa-
from the use of a hydroxyl polyol of functionality
ging and load-bearing such as cushioning. Moulded
greater than two while the physical 'cross-links' arise
foams comprise an increasing fraction of polyurethane
from the phase separated hard segment domains (urea
foams because they possess m a n y advantages over
segments). Although both types of 'cross-links' enhance
slabstock f o a m s - - t h e greatest being that the foam can
the foams physical properties, the physical 'cross-links'
be moulded into the desired intricate shape, thus
are liable at high temperatures and high humidity thus
eliminating the need for labour-intensive cutting and
dramatically altering the foams properties. This paper
waste. Also, moulded foams can be produced with
correlates the mechanical behaviour of moulded poly-
multiple zones of hardness, with reinforcements, or with
urethane foams to the morphology and chemistry
a liner such as plastic or fabric skin which reduce labour
through an analysis of the temperature/humidity influ-
costs of final product assembly. In view of this trend, the
ence on the load relaxation behaviour as well as through
need exists for a greater fundamental understanding of
an investigation of other structure-property relation-
the relationships between the structure/morphology and
ships. In addition, qualitative comparisons are made to
physical properties of these two types of polyurethane
previous work out of this laboratory on conventional
slabstock foams specifically performed by Moreland
* To w h o m c o r r e s p o n d e n c e s h o u l d be a d d r e s s e d et al.1.

POLYMER Volume 38 Number 11 1997 2819

Flexible polyurethane foams: D. V. Dounis and G. L. Wilkes

The effects of temperature and humidity on certain They found that the development of urea hydrogen
physical properties of polyurethane foams have been bonding was much more distinct in slabstock foams over
studied and well documented 1-4. For example, More- moulded foams observed with FTi.r. The viscosity
land, et al. of this same laboratory investigated the effects profiles were also very different where initial viscosity
of temperature and humidity on the load relaxation and build was attributed to the development of bidentate
creep behaviour of slabstock polyurethane foams. urea hard segments in the slabstock foam. In the
Specific results from that study will be utilized here for moulded foam, an increase in viscosity was observed
purposes of comparison. In short, they found that an only at the onset of the covalent gel and did not show
increase in temperature both "plasticized' the hydrogen evidence of bidentate urea formation. These differences
bonds between the urea segments and promoted were attributed to the addition of DEOA and ethylene
increases in the exhibited force (rubber elasticity effect) oxide capping of the polyol. Interestingly, SAXS
in view of the covalent network. In the stress relaxation suggested a stronger microphase separation for the
measurements, an increase in stress was observed with moulded foams over the slabstock foam. The results of
increases in temperature (up to ca. 100C) for a low water this later report will be further substantiated in this paper
content foam. In a high water content loam, no change in where additional characterization methods have also
stress was observed with temperature (up to ca. 100'C), been applied to systematically prepared foams of known
which was attributed to the higher HS content providing chemistry.
more hydrogen bonding available for disruption by
temperature and humidity and less soft segment content
EXPERIMENTAL
to promote the 'rubber elasticity' effect. The overall
result of this was a balancing of the rubber elasticity Materials
effect by the 'plasticization' due to increased tempera- The samples of flexible water-blown moulded poly-
ture. Increasing the humidity also 'plasticized' the urethane foams were made with a Hi-Tech RCM 30
behaviour allowing for more chain slippage and thus foam machine at Dow Chemical in Freeport, Texas. This
low stress values and increased stress decays. As with operation consists of two hydraulic pistons to dispense
temperature, the effect was more pronounced with the the liquid components to the mixing head. The formula-
higher HS foams. tion components presented in Table 1 were prepared in
Herrington and Klarfeld also reported a similar trend two storage tanks, A and B. The A side consisted of the
with temperature and humidity on the compression set isocyanate. The B side consisted of the polyols, water,
behaviour of slabstock foams which is a measure of the
surfactants, and catalysts. An aluminium mould having
loss of thickness of a foam following compression at a dimensions of 15" × 15" x 4.5" was used, This mould was
specified environmental condition for a specified length heated to 140°F (60°C) at which point the foam mixture
of time'-. They found that increased temperature and was dispensed into the mould. The mould was placed in
humidity resulted in increased compression set or greater an oven at 2 5 0 F (121°C) for 2.5rain after which the
thickness loss. The effects were more pronounced in the foam pad was removed and mechanically crushed by
higher HS content foams and were attributed to a passing it three times through steel rollers while
'plasticization' by temperature or water which disrupt decreasing the gap.
hydrogen bonds between urea segments. Patten and The samples of the flexible water-blown slabstock
Seefried also observed similar trends with temperature foams were also made at Dow Chemical in a box-
and humidity as well as water content on compression foaming operation and whose components are also
set 3.
shown in Table 1. Here the B-side components were
Skorpenske et al. noted distinct decreases in specific first mixed for 30s using an 8hp electric mixer at
properties such as tensile properties and dynamic 1800rpm. The isocyanate was then added and the
mechanical properties with increases in temperature in components were mixed for 5 s at 1800rpm at which
slabstock flexible polyurethane foams 4. In addition, the
amount of compression set also increased with increasing
test temperature. More importantly, when high humidity
was included, the compression set was even more severe Table 1 Formulation components of the foams studied
and maximum levels were attained at higher tempera- Slabstock
tures. This effect was further accented as the hard Mould Mould
segment content was increased. The observed behaviour ('omponent Fm2 Fm4 Fs2 Fs4 Fs5 Component
was attributed to stress yielding of the hard domains and V4703 50 50 100 100 100 V3100
hydrogen bond disruption. CPP 50 50
Thomas et al. studied the effect of increasing the cross- 1%O 4 2 2 4 5 H~O
DEOA 1.6 1.6
link density on the morphology of flexible moulded
Y-10515 1.08 1.08 1.0 1.0 1.0 BF-2370
polyurethane foams 5. They altered the level of cross- DC-5244 0.54 0.54
linking by changing the functionality of the polyol (2.56 DABCO 33LV 0.43 0.43 0.3 0.3 0.3 DABCO 33LV
2.76) which was accomplished by the addition of NIAX AI07 0.32 0.32
monofunctional polyether molecules. Both SAXS and NIAX A4 0.32 0.32 0.15 0.15 0.15 T-9
TDI 28.32 47.68 30.79 52.06 62.70 TDI
thermal analysis suggested that increasing the polyol
functionality increased the phase mixing. The authors Density (lbft ~) 3.3 1.9 2.85 1.43 1.24 Density (Ibft 3)
added that the morphology was the primary determining All formulation amounts are given as parts per hundred of polyol, pphp
factor as to how these materials responded to varying Description of components: V4703:5000 molecular weight ethylene-
oxide capped triol; V3100:3000 molecular weight triol; CPP: copolymer
environmental conditions.
polyol; DEOA: cross-linking agent; Yl0515, DC-5244, BF-2370: T-9:
A comparison between moulded foams and slabstock surfactants; DABCO 33LV, NIAX AI07, NIAX A4: gel, blow, and
foams has recently been reported by McClusky et al. 6. cure catalysts: TDI: 80/20 blend of 2,4- and 2,6-toluene diisocyanate

2820 POLYMER Volume 38 Number11 1997

 Flexible polyurethane foams: D. V. Dounis and G. L. Wilkes

point they were poured into an open box having Cellular Structure
dimensions of 15" x 15" × 12" and allowed to cure. The
samples here being experimental materials are much Increasing
Mtpification
smaller in size and inherently undergo different thermal
histories than commercial foams. Temperature profiles
of typical slabstock foams of different sizes are well
illustrated in a treatise on antioxidant effects for foams Phase-SemLr+atedStructure
by Skorpenske et al. 7.
As can be seen in Table 1, a greater number of
components were used in the formulation of moulded
foams than for slabstock foams. In addition, the Fo~,oJ( o . s m ) ~ ~ .
components differ significantly in chemistry and
structure. For example, in the slabstock formulation,
(-seA)
the polyol used was a 3000 molecular weight polyether
triol--typical of conventional commercial slabstock
systems. The polyol used in the moulded foams was a Covldemt Netwer~
commonly employed 5000 molecular weight ethylene
oxide capped polyether triol. The ethylene oxide
capping was used in view of its much higher reactivity
(due to the primary hydroxyl groups) over propylene
oxide to allow for low demould times--one of
the principle objectives for the development of
moulded foams. Another major difference between the
formulation components of moulded foams and the
slabstock foams was the inclusion of styrene/acrylonitrile
based copolymer (CPP) particulates of ca. 0.5#m in Figure 1 Simplified diagram of the different levels of structure in
diameter which become dispersed throughout the final mouldedpolyurethanefoamsincreasingin magnification(fromref. 12)
foam matrix. These CPP particulates are added as a
dispersion within the base polyol to aid in cell opening group which links the urea groups to the ether soft
and act as a reinforcer. The final major difference is the segments and provides for a covalently cross-linked
addition of D EO A in the moulded foam formulation which network as illustrated in the lower most portion of
is a cross-linker added to allow the foam to set in a shorter Figure 1 of even higher magnification schematically
amount of time again decreasing demould times. showing the cross-linked polymer chains.
The basic reactions involved in the production of
polyurethanes are often referred to as the blowing and Experimental methods
gelling reactions. The reaction products of the blowing The cellular structure of these foams was evaluated
reaction between the TDI and water are carbon dioxide and compared using scanning electron microscopy
which foams the reacting mixture and a disubstituted (SEM). Thin slices (3-4 mm) of foam were adhered to
amine. The amine produced in this reaction reacts with aluminium stubs using silver paint and allowed to dry. A
additional isocyanate to produce rigid urea groups which thin layer of gold was then applied to the surface of the
when of sufficient size and concentration, phase separate foam using a SPI model 13131 sputter coater. Micro-
into urea rich domains (hard segment domains) primar- graphs were taken using a Cambridge Stereoscan 100
ily due to hydrogen bonding with additional urea groups. SEM operating at 20 kV at a magnification of approxi-
Figure 1 illustrates the three basic levels of structure of mately 30 x.
increasing magnification that are typically found in The procedure used for the compression set experi-
moulded foams with simplified diagrams beginning with ments was a non-ASTM procedure but was designed to
the cellular structure. An enlargement of a cell strut reflect the load relaxation measurements. The samples
would reveal the solid morphology depicted in the were cut into dimensions of 2 x 2 x 1 inch and dried
second diagram in Figure 1. This illustrates a three- under vacuum for approximately 3.5 h. They were then
phase model showing the soft phase, the large CPP placed in the environmental chamber at the designated
particulates and the hard segment (HS) domains. The environmental conditions for ca. 1 h following which
model unrealistically neglects imperfections in the hard they were compressed to 65% for 3 h. The samples were
segment domains as well as single urea segments not then removed and placed in an oven equilibrated at 40°C
found in domains but rather mixed in the soft phase or for 30min after which thickness measurements were
interfacial region. Also not shown are possible 'urea made. Recovery measurements were carried out on the
aggregates' that can sometimes form in higher hard samples which displayed the greatest amount of com-
segment containing foams. The HS domains are com- pression set (100°C-98% RH). After about one month
monly referred to as physical 'cross-links' since they of room temperature storage, these samples were placed
enhance the physical properties at room temperature and in an oven at 100°C for 1 h. The thickness of the foams
low humidity. At elevated temperatures and humidities, was again measured and recorded.
these physical 'cross-links' are significantly labile thereby Compression load-strain measurements were con-
altering the properties of the foams. The hard-segment ducted using the identical experimental setup used in the
domains (structure, order, concentration) play a very load relaxation measurements (described later). The
important role on the final structure, morphology, and 3.5" × 3.5" × 1" samples were first dried under vacuum
properties of the foam. The reaction product between the and at 40°C for 3.5h and placed in an environmental
isocyanate and multifunctional polyol is a urethane chamber at 30°C-35%RH for ca. 60min. The samples

POLYMER Volume 38 Number 11 1997 2821

Flexible polyurethane foams. D. V. Dounis and G. L. Wilkes

were then compressed at 3 5 0 m m m i n ] to 75% strain data were collected using a Kratky camera, with nickel
and released. This was done to compare the level of filtered CuK,~ radiation having a wavelength of 1.542 A
hysteresis as a function of water content used in the passing through a slit collimator (0.03 × 5mm). The
formulation. By numerical integration, the energies upon detector used is a Braun O E D 50 position-sensitive
loading and unloading were calculated leading subse- platinum wire detector. The raw data were corrected for
quently to the fractional hysteresis, 1 - Eu/Ei, where E, parasitic scattering and normalized using a Lupolen
is the energy upon loading and Et is the energy upon standard. The foam samples were cut approximately
loading. 10 mm thick and compressed to approximately 3 ram.
Load relaxation experiments were performed using tt
similar procedure as that used and described by More-
RESULTS AND DISCUSSION
land which was originally designed to mimic the ASTM
procedure used for ILD testing j. Samples, having ('elluhw structure and solid morpholog)'
dimensions of 3.5" × 3.5" x 1", were cut from the foam The cellular structure of the moulded foams was
bun using a band saw equipped with a smooth "wavy compared to the cellular structure of the slabstock foams.
edge' saw blade. Each sample was first dried under Figure 2 shows two micrographs of the 4 p p h water
vacuum and at 40'~C for 3.5h in order to give each content slabstock foam (Fs4), parallel and perpendicular
sample an equal level of moisture. The samples were then to the rise direction. As can be seen, the foam designed to
placed in an environmental chamber preset at the testing be an open-celled structure has very few closed cells.
conditions for ca. 60min. The environmental chamber There also exists a geometric anisotropy, where parallel
was purchased from Russells Technical Products and to the rise direction, the cells appear circular and
was equipped with a Watlow 922 microprocessor which perpendicular to the rise direction the cells appear
controls temperature in the range of 20:'C to 300 C and elliptical with the major axis aligned along the rise
humidity in the range of 0 to 100%. The chamber was fit direction. Since it has been shown that the physical
into the Instron frame equipped with a model MDB-10 properties vary depending on the loading direction, all
compression load cell manufactured by Transducer foams were loaded parallel to the rise direction for
Techniques. Using a 2" indentor, initially at rest, the consistency ~. It was also observed, however, that the
samples were twice compressed to 70% and released at a
rate of 3 5 0 m m m i n ~ After 5min the samples were
compressed to 65% strain at which point the load was
immediately monitored via computer. At 65% compres-
sion, relaxation is believed to occur predominantly
within the solid polymer independent of cellular struc-
ture since the onset of 'densification' is observed here.
Extraction experiments were carried out on selected
samples to compare the level of cross-linking between the
moulded foams and slabstock foam. Samples (<0.15 g)
were cut from the centre of the foam bun, dried and
weighed. The samples were submerged in D M F of
approximately 10 × the foam volume for a period of 48 h
and then taken out, dried under vacuum and at 40°C for
approximately 24h, following which the temperature
was raised to 80~'C for an additional 48 h. The samples
were then weighed again.
In studying the degree of order within the hard
segment domains, WAXS patterns were obtained. The
X-ray source was a Philips X-ray generator model
PW1720 using a Statton camera and a fine focus tube
with nickel filtered CuK~, radiation having a wavelength
of 1.542 A. Foam samples were cut approximately 10 mm
thick and compressed to approximately 3 m m . The
sample-to-film distance was 8cm and all exposure
times were 10 h.
FTi.r. was used to compare the relative degree of order
of the hard segments of the moulded foams and
slabstock foam on a Digilab FTS-40 spectrophotometer
equipped with an attenuated total reflectance (ATR) cell.
For each sample 64 scans were averaged taken in the
range of 400cm I to 4000cm 1 with a resolution of
1
4 cm ~ . All scans were normalized to the absorbance of a
CH stretch at 2945 c m - 1 . The regions studied were: the
amide I region (1620-1800cm J ), the N - H region
(3100-3500cm l), and the isocyanate region (2200
2400 c m - l ).
Phase separation was in part evaluated using SAXS
scans which were obtained with a Phillips model PW 1729 Figure 2 Scanning electron micrographs of slabstock foam, Fs4, (a)
generator operating at 40 kV and 20mA. The smeared parallel and (b) perpendicular to the rise direction (bar = 1 ram)

2822 POLYMER Volume 38 Number11 1997

 Flexible polyurethane foams. D. V. Dounis and G. L. Wilkes

(a Table 2 Percent set (CS) results

Condition % CS Fm4 % CS Fm2 % CS Fs4

30°C-35% R H 2.5 4- 0.1 1.2 4- 0.2

30°C-98%RH 9.5 + 1.5 5.3 4- 0.3
100°C-35%RH 59.0 + 0.3 59.6 4- 0.6
100°C 9 8 % R H 64.3+0.1 63.24-0.1 15.24-1.6
Day after CS % CS Fm4 % CS Fs4

0 63.24-0.1 15.24- 1.6

22 9.5 4- 1.0
After l h (a 100"C 5.94-0.3
89 62.9 4- 0.1
After 1 h Ca 100°C 61.24-0.2

and thus hard segment content and density, appeared to

have very little influence on the compression set.
Comparing the compression set of the slabstock foam
to the moulded foams show that at 100°C 98% R H the
slabstock foam had considerably less set. The compres-
sion set for the slabstock foam was ca. 15% while for the
moulded foams, it was ca. 63%. In addition, the
slabstock foam displayed appreciable recovery. Over a
period of 22 days at room temperature, this foam
recovered to 9%. When subjected to 100°C for 1 h, this
foam recovered even further to ca. 6%. In great contrast,
the moulded foam displayed no recover)'. These dramatic
differences are clearly a result of differences in the solid
morphology and/or covalent network content of the
foams. The slabstock foam either has a better covalent
network or better developed hard segment domains or
both. This hypothesis will be explored later. It should
also be pointed out that the incorporated copolymer
polyol particles also contribute to the compression set of

(a) 4oo0
Figure 3 SEM of moulded foam, Fm4, (a) parallel and (b) 35OO
perpendicular to the rise direction (bar = l mm)

time-dependent properties were independent of loading 2500

direction. Comparing these micrographs to those of the
moulded foam, Fro4, shown in Figure 3a and b, it can be Loading f f /d
1500
seen that the moulded foam, although also open celled,
has somewhat more closed cells which can inherently 1000
alter the foam properties, e.g., air flow. For example, the
500
airflow measurements of these two foams were Unloading
5 ft 3 min-I for Fs4 and 1 ft 3 min 1 for Fm4 confirming 0
that foam Fs4 does indeed have a greater cell openness. 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

The cell struts are also much thicker thereby giving rise to Strain
(b)
a somewhat higher density foam. The geometric aniso-
tropy observed in the slabstock foam is not evident in the 3500

moulded foam due to the reacting mixture being 3000

pressurized from all directions in the moulded foam.
2500
Mechanical and viscoelastic behaviour
92000
The compression set property of polyurethane foams
1500
used as load bearing materials may be the single most
important property. Compression set was utilized to I000
evaluate the foam morphology and network structure
50O
since it is sensitive to both. The compression set
behaviour as a function of temperature and humidity 0
of the moulded foams and slabstock foam is presented in 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Table 2. As the temperature or humidity was increased, Strain

the compression set also increased--especially with Figure 4 Compressionload-strain behaviour illustrating both loading
temperature. The amount of water in the formulation and unloading for moulded forms (a) Fm4 and (b) Fro2

POLYMER Volume 38 Number 11 1997 2823

Flexible polyurethane foams. D. V. Dounis and G. L. Wilkes

the moulded foams a topic that will be addressed in a difference is used to quantify the amount of hysteresis.
separate publication now in preparation 8. The results are given in Table 3 which gives the percent
The compression load-strain behaviour of the hysteresis for each foam. Comparing Figure 4a to Figure
moulded foams is shown in Figures 4a and b where the 4b, shows that Fm4 displayed greater hysteresis than
loading and unloading curves are shown for each foam. Fro2 as do the hysteresis values in Table 3. Since Fm4
As can be seen, the curves can be divided into three does have a higher HS content, it also has a higher
deformation regions as has been well described by hydrogen bonding content which may be mechanically
Gibson and Ashby: the linear bending region, the elastic disrupted. The HS domains are believed to also act as
. . . . . 9
buckling region, and the denslfiCatlon region . The first stress concentrators thereby increasing the amount of
region is representative of elastic bending of the cell localized stress leading to increased relaxation and
struts followed by a buckling of the cell struts where the hysteresis. Figure 5a illustrates the load-strain behaviour
strain increases with small changes in load. As can be of a 2pph water content conventional slabstock foam
seen, the final densification region begins at ca. 65% and Figure 5b illustrates the load strain behaviour of a
strain, the level of compression used for the load 5pph water content foam, Fs5. As can be seen the
relaxation measurements. The area under both the general behaviour is the same as that of the moulded
loading and unloading curves was used to determine foams. The transitions from one deformation region to
the energy of loading and unloading from which the another occur roughly at the same levels of strain. The
slabstock foam, however, displayed higher loads than the
moulded foam of equal water content at any given level
Table 3 First loading unloading cycle hysteresis values of strain. At a strain level of 60%, the load is ca. 2.5 kg
for sample Fs2 and 2.0 kg for the moulded foam Fm2. As
Foam Hysteresis, '!,,
expected the densities of each are slightly different where
Fm4 3(1 the slabstock foam had a density of 2.851bft 3 and the
Fm2 23 moulded foam a density of 3.3 lb ft 3. Since the moulded
Fs5 52
foams density is greater, the lower loads observed for the
Fs2 30
moulded cannot be attributed to differences in density.
As can be seen, the trend with water content is the same;
increased water content resulted in greater hysteresis.
(a) 3.5 More importantly, the amount of hysteresis is higher for
this slabstock foam (Fs2) over that of the moulded
3 counterpart (Fro2). Specifically, the hysteresis associated

2.5

(a)
L o a d l ~ ~
2 ;;-;2

2 1.5

0.5
1
'°i
0
o 0.2 0.4 0.6 O.B

Strain LogTime(n
(b)
-: ~100 T e m p e r a t u r e (~C)
3.5 ~4
(b) -,z~

3 • i i~ s i~lll.
1800

2.5
1400
- ~ (a)
1200
2
1O0O
0

1.5
..... ~i I
~ u I Q't (h)
600
41gl
I
~ J ~ ~ d i n g 2,00
0.001 0.01 0. I I 10 IBO ]lYe0
0,5
Log T i m e (rain)

0 i i i i i t i Figure 6 Load relaxation behaviour for moulded foams Fro4

0 0.2 0.4 0.'5 0.8 illustrated as a function of (a) temperature as well as (b) temperature
Strain and humidity. In (b) the letters designate the following conditions: (a)
3 0 C 35% RH, (b) 3 0 C - 9 8 % RH,(c) 5 0 C - 3 5 % R H , ( d ) 5 0 C 98%
Figure5 Compressionload strainbehaviourillustratingboth loading R H . ( e ) 8 0 C - 3 5 % RH,(13 8 0 C 98% R H , ( g ) t00°C 35% R H , ( h )
and unloading for (a) Fs2 and (b) Fs5 (from ref. I ) 100 C 98% RH

2824 POLYMER Volume 38 Number11 1997

 Flexible polyurethane foams: D. V. Dounis and G. L. Wilkes

with Fm2 was ca. 23% where that associated with Fs2 similar trend was observed at 98% R H in the load
was 30%. The slabstock foam, as will be further relaxation behaviour. Figure 6b shows the effect of tem-
discussed later, has better developed microphase sepa- perature and humidity on the load relaxation behaviour
rated hard segment domains, more hydrogen bonding of Fm4. At a constant temperature an increase in humid-
and therefore even greater localized stress resulting in ity from 35% RH to 98% RH had a similar effect as did
greater hysteresis. an increase in temperature from 30°C to 100°C at 35%
The load relaxation behaviour was evaluated as a RH. At 30°C, the initial load decreased from 2019g at
function of the temperature-humidity history and the 35% R H to 1325 g at 98% RH. Thus, relative humidity
formulation water content. This behaviour at constant as well as temperature 'plasticized' the system thereby
65% compression for Fm4 as a function of temperature promoting a greater time dependent behaviour.
is shown in Figure 6a. As can be seen, an increase in The temperature dependence of the relaxation beha-
temperature systematically shifted the relaxation curves viour at a constant 35% R H for Fm2 is shown in Figure
to lower loads and increased the amount of load decay. 7a. The exhibited behaviour was similar to that of Fm4.
For example, the initial load decreased from 2019g at Foam Fm2, however, displayed higher loads approxi-
30"C 35% RH to 1271 g at 100°C-35% R H while the mately 25% to 33% h i g h e r - - a result of Fm2 having a
percent decay in a 3 h experiment increased from 36% to higher density. Yet, the influence of temperature was
54% for the two conditions respectively. In addition, the similar, i.e. the relaxation curves decreased to lower
curve deviated from linearity at the highest temperature loads with increasing temperature. The initial loads
(100°C) relative to the observed behaviour at 30°C. The decreased from 2389g at 30°C to 1860g at 100°C. The
decrease in load and increase in percent decay with percent decay at 30°C was 32% and at 100°C it was 56%.
temperature is primarily due to hydrogen bond disrup- At temperatures above 80°C, the relaxation curves
tion occurring to a greater extent with temperature deviate from linearity relative to the observed behaviour
therefore increasing the amount of chain slippage that at 30°C. Again this 'plasticization' has traditionally been
occurs. In addition, the increased soft segment mobility attributed to the disruption of hydrogen bonds by
with temperature contributes to the increased percent temperature and/or humidity. Hydrogen bonding is
decays and decreased initial loads. The nonlinearity reported to occur predominantly within the hard
observed at high temperatures, especially 100°C, sug- segment domains but can also occur among urea or
gests that the relaxation mechanism is dramatically urethane groups that interact with ether groups in
enhanced. Here the amount of hydrogen bond disruption the soft segment 6'1°.
is dramatically increased. Retroreaction or reversal of Figure 7b illustrates both the influence of temperature
the urethane reaction is also believed to occur at these and humidity on the load relaxation behaviour of
high temperatures as was shown by Moreland et al. 1. A moulded foam Fro2. The humidity effects are similar as
with Fm4, where, at any temperature, an increase in
(a) humidity resulted in a significant 'plasticization' effect.
Thus, increasing the humidity to 98% from 35%
significantly enhanced the amount of load relaxation.
As depicted in Figure 7b, at temperatures above 80°C, the
relaxation curves deviate from linearity suggesting an
enhancement of the mechanism resulting in accelerated
relaxation. (Similar behaviour was also noted for the
second moulded foam, Fm4.) The degree of 'plasticiza-
i tion' that occurs with temperature and humidity is
evidence suggesting that the physical 'cross-links' play a
significant role in the properties of the foam at ambient
conditions. In addition, the size and perfection of the HS
domains may be different in the moulded foams than
Log Time (
slabstock foams which allows for more dramatic effects
by temperature and humidity. This will be explored
(b) z~o further in a later section.
Compared to slabstock foams of equal water content
studied earlier by Moreland et al., temperature and
relative humidity alter the relaxation behaviour of the
.~18oo ~ ~ (¢) moulded foams more significantly (greater percent decays
(=) in the same 3 h experiment) 1. The variable temperature
load relaxation behaviour of a 2pph water content
(b)
slabstock foam is shown in Figure 8a. Here, increasing
8OO
(r)
(g) the temperature actually resulted in an increase in the
initial loads from 3000g to 3500g up to IO0°C.
(h) In addition, the present decays decreased from 22% to
0.001 0.01 0.1 I 10 100
L o g Time (rain)
20%. At temperatures beyond 100°C, the initial
loads significantly decreased and the percent decays
Figure 7 Load relaxation behaviour for moulded foams Fm2 significantly increased. Recall that moulded foam Fm2
illustrated as a function of (a) temperature as well as (b) temperature not only displayed decreases in load with all increases
and humidity. In (b) the letters designate the followingconditions: (a)
30°C-35% RH, (b) 30°C-98% RH, (c) 50°C 35% RH, (d) 50°C-98% in temperature but the percent decay also significantly
RH, (e) 80°C 35% RH, (f) 80°C 98% RH, (g) 100°C-35% RH, (h) increased. The observed increase in load with increasing
100C 98% RH temperature displayed by the slabstock foams was

POLYMER Volume 38 Number 11 1997 2825

Flexible polyurethane foams. D. V. Dounis and G. L. Wilkes

(a) 3.6 [able 4 Extraction results for three of the foams studied

Foam Sol fraction. %

2.5
Fro4 8.8 d- 0.1
Fro2 8.3 ± 0.2
S~ 3,4
Fs4 6.5 ± 0.5

3.3

the moulded foams and the slabstock loam and in part

-2.75 3.2 to account for the observed differences in the
-].75 ":~'-~.
-075 ":~--~. ,/i/"
compression set behaviour and dramatic viscoelastic
properties. The results presented in Table 4 suggest that
<O~ ~ 1.25 ~':=- the amount of covalent cross-linking is slightly higher
225 " .................... :
150 125
- 100 7{ gll2'
-

in the slabstock foam than the moulded foams as seen

Temperature(C) with the lower sol fraction for the slabstock foam.
(b) 3.5 Between the two moulded foams, Fm2 appears to have
a slightly lower amount of extractables than Fm4.
However, this may be due to Fm4 having a higher
amount of urea groups. Although it can be concluded
that the level of covalent network is somewhat more
developed in the slabstock foam, the dramatic differ-
cnces in the compression set and viscoelastic properties
were believed to arise in part from additional
characteristics, and therefore further features of the
2 z5 3.2- foam morphologies were compared.
1.75 WAXS patterns were obtained to compare the amount
of short range ordering within the HS of each foam that
/.09 0.25 . ... ~ ".
might make one system more labile than another at a
" 225 ~. .
150
. .
125 100 75 50
given environmental condition. It should be recognized
Temperature(C) that while WAXS has been predominantly used to study
crystalline polymers, it has also been used to evaluate the
Figure 8 Load relaxation behaviour illustrated as a function o f packing order in HS domains. Furthermore, the use of
t e m p e r a t u r e for slabstock foams (a) Fs2 and (b) Fs5 (from ref. 1 )
this technique is not intended to imply the existence of
welldeveloped crystalline HS domains, since the high
attributed to an entropically driven elasticity effect of the amount of the 2,4 TD1 isomer certainly prevents this
rubbery covalent network. Thus, in the case where the occurrence The scattering pattern for the 4pph water
HS content is low, the rubber elasticity effect overrides content slabstock foam is shown in Figure 9a. As can be
the softening of the HS domain. In addition, Moreland el seen, this pattern displays, in addition to an amorphous
al. observed that humidity had a relatively small effect on halo typical of liquid-like amorphous systems, a rela-
slabstock foams whereas for the moulded foams, this tively sharp ring with an estimated Bragg spacing of
same variable dramatically softened them I. Again, the approximately 4.5 A and another outer but weaker ring
copolymer polyol particulates were shown to contribute with a spacing of ca. 6.0 A. The exact origin of these rings
to the time dependent response of the foams. The is not yet confirmed absolutely but has been attributed to
amount of decay increased by 30% 50% by their incor- non-crystalline or para-crystalline hard segment order-
poration. As mentioned earlier, a separate publication ing through hydrogen bonding 11. The sharpness or
will address the causes of this observation s . diffuseness of these two rings can be used to qualitatively
The load relaxation of Fs5 is shown in Figure 8b which assess the level of ordering of the hard segments. The
shows the influence of temperature on this 5 pph water rings disappear with only the diffuse amorphous halo
content slabstock foam. While no increase in initial load remaining after extraction in D M F (a solvent known to
was observed with increasing temperature, nor was a interact with urea) as displayed in Figure 9b. Figure lOa
significant decrease noted. For example the initial load show's the WAXS pattern for Fro4 while F~gure 10b
remained at ca. 2700 g from 30:'C to 100'C beyond which shows the WAXS pattern for Fm2 both of which display
the load decreased. Furthermore, the percent decay oHh' the amorphous halo suggesting that very little (if
decreased for 30% at 30"C to 26% at 10OC beyond any) order exists amongst the TDI based hard segments
which the load increased considerably. Again, even for a and hence the HS domains of the moulded foams are
high HS content foam, the observed behaviour was more disordered than in slabstock foams. Thus, based on
significantly different than that displayed by the high this lack of order, the moulded foams and their physical
water content moulded foam. Recall that Fm4 displayed "cross-links" allow for the easier 'plasticization" by heat
a strong sensitivity to temperature and humidity and humidity thereby making them more labile than
exhibiting a strong 'plasticization' trend with increases those in slabstock foams. This lack of structural order
in either observed between the hard segments is believed to be
both kinetic and thermodynamic based. Hard domain
Morphological characterization I%rmation is very sensitive to the solubility of the urea
The foams covalent network was evaluated using segments in the polyol soft phase which is greater in
solvent extraction studies in dimethyl formamide moulded foams due to the stronger more favourable
(DMF) to compare the level of cross-linking between interactions between the hard segments and ethylene

2826 POLYMER Volume 38 Number11 1997

 Flexible polyurethane foams. D. V. Dounis and G. L. Wilkes

(t?

Figure 9 WAXS patterns of slabstock foams Fs4 (a) before extraction, Figure I0 WAXS patterns of moulded foams (a) Fm4 and (b) Fm2
and (b) after extraction

0.06
oxide capped soft segments (used to end cap the 0.058
propylene oxide to decrease demould times in view of L
0.056
its higher reactivity). In addition, for bidentate hydrogen
0.054 / \,
bonded urea to occur, it must principally develop prior ~_ ~ ~:-~,, .Ab) ",
~ 0,052
to the gel point of the covalent network. Recall that
:/
diethanolamine is also added in the formulation of ~ 0,05
/,/ " - " (c) ~ - :

moulded foams to build softness, speed up gelation and 0,048

decrease demould times. The rapid cross-linking that 0.046

occurs early on in the network development makes it 0,044
very difficult for the urea segments to associate
0.042 I I +
(through diffusion) and to pack and form multiple 1800 1780 1760 1740 1720 1700 1680 1660 1640 1620
bidentate hydrogen bonds with other hard segments Wavenumber cm -I

thereby limiting perfection of the hard segment domain

Figure 11 FTi.r. spectra for moulded foams (a) Fm4 and (b) Fm2 as
development. well as slabstock foam (c) Fs4
To further verify the above evidence obtained by
WAXS, FTi.r. analysis of the amide I region (carbonyl
region) was studied which is shown for both moulded This bidentate peak is indicative of hard segment
foams and the slabstock foam in Figure 11. The i.r. ordering strongly supporting the WAXS results. Both
results show significant differences between the slabstock moulded foams also displayed a moderate to strong
foam and the moulded foams. The slabstock foam monodentate absorbance centred at 1700 1650cm -~ as
displayed a strong bidentate absorption peak centred at well as free urea centred at 1715cm 1. Similar compar-
1640 cm-1 which was non-existent in the moulded foams. isons and observations have been reported by McClusky

POLYMER Volume 38 Number11 1997 2827

Flexible polyurethane foams, D. V. Dounis and G. L. Wilkes

slabstock foams in terms of structural and morphologi-

cal features. Extraction studies were initially performed
on both moulded and slabstock foams which showed
that the level of covalent cross-linking was slightly higher
in the slabstock foams. WAXS patterns of the slabstock
"~
foams displayed evidence of short range hard segment
~
= 1
4 structural ordering which was not evident in either of the
moulded foams. The i.r. results confirmed this by
200
illustrating a strong urea bidentate absorbance in the
slabstock profile while the moulded profiles only
0 T t ..... ~ ;
displayed monodentate and free urea absorbances. The
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 structural and morphological as well as ensuing mechan-
S (l/rim) ical property variations are believed to arise from two
major differences in the formulation components with
Figure 12 SAXS profiles of smeared intensity l(s) for the moulded
foams, (a) Fm4, and (b) Fro2, and slabstock foam (c) Fs4 the different processing techniques also having some
influence. First, the use of ethylene oxide end capping in
the moulded foams is also known to favour interactions
et al. 6. For reasons mentioned above (specifically the with soft segment ether groups. Second is the addition of
addition of DEOA and EO capping in moulded foams), DEOA in the formulation with the intent again to
it is concluded that the moulded foams have a much decrease demould times taking advantage of the high
more disordered physical network (hard segment functionality of DEOA thus allowing the foam to reach
domains) and covalent network which allows for greater the gel point in shorter times. The early development of a
'plasticization' with temperature and humidity. Based on network gel prior to the development of hard segment
the WAXS and FTi.r. data, one might be misled into formation subsequently prevents the formation of highly
believing that the moulded foams are phase-mixed. For ordered bidentate structures.
example, a homogeneous morphology would prevent
hydrogen bonding between urea groups and packing in ACKNOWLEDGEMENTS
an orderly fashion. This might then lead to the trends
observed in the WAXS and FTi.r. analysis of the This work has been financially supported by the Dow
moulded foams. To refute this, SAXS analysis was Chemical Company. Their support is gratefully
carried out and is shown in Figure 12 which is a plot of acknowledged. Dr Randolph Grayson is also acknowl-
the normalized smeared intensity vs an angular variable. edged for use of the transmission electron microscopy
As can be seen, all samples display a shoulder (at ca. laboratory as well as Dr Eva Marand and Qingchun Hu
0.1 nm -l) confirming that the samples are microphase for use of the Bio-Rad FTi.r. equipment.
separated on a scale of ca. 100 A. In fact, the shoulders
associated with the moulded foams are more distinct.
The three techniques of WAXS, SAXS and FTi.r. are REFERENCES
crucial for obtaining a complete understanding of the I. Moreland, J. C., Wilkes, G. L. and Turner, R. B., J. Appl,
localized foam morphology as has been shown. It can be Polym. Sci., 1994, 52, 549.
concluded that while all systems are phase separated, the 2. Herrington. R. M. and Klarfeld, D. L., Proc. SPI-6th Interna-
cohesiveness of the domains of the slabstock foams is tional Tech./Marketing Con[~, 1983, p. 177.
3. Patten, W. and Seefried, C. G., J. Cell. Plastics, 1976, 12, 41.
much higher than that of the moulded foams. 4. Skorpenske, R. G., Solis, R., Kuklies, R. A., Schrock, A. K. and
Turner, R. B., 34th Annual Polyurethane Technical/Marketing
('on/~rence, October 1992.
CONCLUSIONS 5. Thomas, O., Priester, Jr., R. D., Hinze, K. J. and Latham, D. D.,
The viscoelastic and compression set behaviour of J. Polym. Sci.. Part B: Polym. Phys., 1994, 32, 2155.
¢~. McClusky, J. V., Priester, Jr., R. D., Willkomm, W. R., Capel,
moulded foams varying in the formulation water/TDl M. A. and Heaney, M. D., Polyurethanes World Congress,
content were studied and compared to those of conven- October 1993.
tional slabstock foams. It was found that systematic 7. Skorpenske, R. G., Schrock, A. K. and Beal, G. E.. J. Cellular
increases in temperature or humidity significantly Plastics, 1991, 27, 560.
8. Dounis, D. V. and Wilkes. G. L., The influence qf C~polvmer
deteriorated the moulded foams physical properties. Polyols on the tqscoelastic Resloonse o[' Molded Polyurethane
These materials displayed great compression set espe- k~mms, to be published.
cially at high temperatures and humidities. In addition, tL Gibson, L. J. and Ashby, M. F., Cellular Solids Structure and
load relaxation results also showed the moulded foams Properties. Pergamon Press, New York, 1988.
to be extremely labile at elevated temperatures and/or 10. Paik Sung, C. S. and Schneider, N. S., Macromoleeules, 1977, 10,
452.
humidities. In no case was an increase in the load with I1. Armistead, J. P., Wilkes, G. L. and Turner, R. B., J. Appl,
temperature observed such as that noted with respect to Po!wn. Sei., 1988, 35, 601.
the slabstock foams. In view of these findings, 12. Herrington, R. and Hock, K., Flexible PoO'urethane Foams,
the moulded foams were compared to conventional Dow Chemical Co., Midland, M1, 1991.

2828 POLYMER Volume 38 Number 11 1997