You are on page 1of 8

Composites: Part A 50 (2013) 65–72

Contents lists available at SciVerse ScienceDirect

Composites: Part A
journal homepage: www.elsevier.com/locate/compositesa

Creep behavior of polyurethane nanocomposites with carbon nanotubes


Zhen Yao a, Defeng Wu a,⇑, Chong Chen b, Ming Zhang b
a
School of Chemistry & Chemical Engineering, Yangzhou University, Jiangsu 225002, PR China
b
Testing Center, Yangzhou University, Jiangsu 225002, PR China

a r t i c l e i n f o a b s t r a c t

Article history: The polyurethane (PU) nanocomposites containing carbon nanotubes (CNTs) were prepared through
Received 12 January 2013 in situ polymerization for the creep study. The results show that the presence of CNTs leads to a signif-
Received in revised form 21 March 2013 icant improvement of creep resistance of PU. However, this creep resistance does not increase monoto-
Accepted 23 March 2013
nously with increase of CNT contents because it is highly dependent on the dispersion of CNTs. Several
Available online 2 April 2013
theoretical models were then used to establish the relations between CNT dispersion and final creep
and creep–recovery behaviors of nanocomposites. The as-obtained viscoelastic and viscoplastic parame-
Keywords:
ters of PU matrix and structural parameters of CNTs further confirmed the retardation effect by CNTs dur-
A. Polymer–matrix composites (PMCs)
B. Creep
ing creep of the nanocomposite systems. Besides, the time–temperature superposition (TTS) principle
C. Analytical modeling was also employed in this work to make a further evaluation on the creep of PU/CNT nanocomposites
D. Mechanical testing with long-term time scale.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction materials, creep and creep–recovery of the polymer/CNT compos-


ite systems have attracted much attention recently [30–37]. It
As an important engineering material, polyurethane (PU) exhib- has been reported that the presence of CNTs restrains the creep
its many attractive properties, such as adjustable mechanical deformation of matrix polymer significantly, but the improvement
strength, unique shape-memory ability and superior chemical of creep resistance diminishes with the growing stress and in-
resistance, because of its special soft–hard segment structure creased temperature [34–36]. The recovery property is also highly
[1,2]. In order to further extend its applications, several techniques, improved, especially at high temperature [31,35]. Such improve-
including interpenetrating [3,4], blending [5] and copolymeriza- ment is attributed to two possible reasons: (1) the presence of
tion [6], have been developed to control hierarchical structures CNTs leads to the formation of a significant interphase zone with
and final properties of the PU material. Besides, hybridation with altered polymer mobility, namely chain immobilization, which re-
nanofiller [7,8] is also an effective strategy to improve the perfor- sults in the initial reduction of the creep compliance [33–35]; (2)
mance of PU. good CNTs–matrix interfacial bonding further reduces creep
In recent years, the carbon nanotubes (CNTs) have become the through frustrating chain disentanglement, stretching and frag-
next-generation reinforcements for nano-structured polymeric mentation of the macromolecule [33].
composites, owing to its extraordinarily high elastic modulus, The creep study is also very important to the PU/CNT nanocom-
strength, and resilience [9–11]. Hitherto, many CNTs based poly- posites. There are merely few reports on the creep of PU/CNT nano-
mer nanocomposites [12–14] have been successfully prepared composites found in the literatures, however. Jia et al. [30]
via various methods, also including the PU/CNT nanocomposites performed a preliminary study on the creep properties of the ther-
[15–30]. The structure and properties of PU/CNT systems have moplastic polyurethane (TPU)/CNT nanocomposites and found that
hence been extensively studied and the focus of attention are: the presence of CNTs (pure or functionalized) could improve the
(1) the dispersion of CNTs and the microstructure of nanocompos- creep resistance through their good interfacial adhesion to the
ites [15–19]; (2) the thermal and rheological behaviors in the pres- TPU matrix. The interesting results reveal a common creep behav-
ence of CNTs [20,21]; (3) the improvement of active shape memory ior of the PU/CNT systems, similar with that reported on the other
with addition of CNTs [22–24]; (4) the mechanical properties, CNTs based polymeric composites. To deeply understand how the
especially the reinforcing effect of CNTs [20,25–30]. presence of CNTs affects the time dependent properties of PU,
As an interesting time dependent behavior and as an important however, it is necessary to further explore the creep and creep–
evaluation standard on the dimensional stability of engineering recovery behaviors of PU/CNT nanocomposites. Therefore, in this
work, the PU/CNT nanocomposites with different contents of CNTs
⇑ Corresponding author. Tel.: +86 514 87975590x9115; fax: +86 514 87975244. were prepared through in situ polymerization for the creep and
E-mail address: dfwu@yzu.edu.cn (D. Wu). creep–recovery studies at different temperatures and load levels.

1359-835X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.compositesa.2013.03.015
66 Z. Yao et al. / Composites: Part A 50 (2013) 65–72

Some constitutive models were then used to establish the relations 2.5. Dynamic mechanical analysis
between CNT dispersion and creep of nanocomposites. Besides, the
creep with long-term time scale was also explored by time–tem- Dynamic mechanical analysis (DMA) were performed on a DMA
perature superposition (TTS), aiming at further evaluating the time 242 C analyzer (NETZSCH Instruments, Germany) with tensile
dependent properties of PU/CNT nanocomposites. mode over a temperature range from 100 °C to 200 °C at various
frequencies (1 Hz, 5 Hz and 10 Hz). The heating rate is 3 °C/min.
2. Experimental Three rectangular specimens (30 mm  5 mm  1 mm) were
tested for each composition.
2.1. Materials and treatment
2.6. Creep and creep–recovery measurements
The purified multi-walled carbon nanotubes (CNTs, purity
>95 wt%) with the outside diameter of 10–30 nm and the special Creep tests were conducted on a DMA Q800 analyzer (TA
surface area of 250 m2/g were supplied by Chengdu Organic Chem- Instruments, USA) with tensile mode at various temperature and
istry Institute, Chinese Academy of Science. 2,4-Toluene diisocya- stress levels. The applied stress levels (1.5, 2.5, 3.0, 5.0 and
nate (TDI) (C. P.) was purchased by Shanghai Lingfeng Chemical 7.0 MPa) were determined by the tensile tests, which are in the lin-
Reagent Co., Ltd. Castor oil (CO) with hydroxyl number of 163 mg ear viscoelastic region of all samples. The duration of measure-
KOH/g was purchased from Sinopharm Chemical Reagent. The ments was determined as 10 min and 30 min for the creep and
other reagents, including 1,4-butanediol (1,4-BDO) (chain exten- creep–recovery tests, respectively. The creep/recoverable strain
der), xylene (solvent) and dibutyltin dilaurate (DBTDL) (catalyst) and compliance were recorded as a function of the time. Three
were analytically pure and purchased from Sinopharm Chemical rectangular specimens (30 mm  5 mm  1 mm) were tested for
Reagent. CO and 1,4-BDO were dried under a vacuum before use each composition. To obtain master curves of creep, time–temper-
and the other reagents were directly used without any treatments. ature superposition was employed to deal with the results from
The equivalent weight per AOH group of CO is 344.83 g and per short-term creep tests performed at various temperatures. The
ANCO group of TDI is 87.08 g. These two values were used to cal- temperature ranged from 30 °C to 90 °C, and then 2.5 MPa creep
culate the NCO/OH ratio for the synthesis of the PU. stress was applied for 10 min at each temperature level. Before
each measurement, the specimen was equilibrated for 5 min at
each temperature in order to evenly adjust for the correct temper-
2.2. Preparation of PU and PU/CNT composites
ature of the sample.

A reported two-step bulk polymerization method [38,39] was


2.7. The Soxhlet extraction experiments
employed in this study to synthesize PU with a 2.25:1:1 M ratio
of ANCO (TDI):AOH (CO):AOH (1,4-BDO). Firstly, a weighed
The Soxhlet extraction is used here to determine the degrees of
amount of castor oil was placed in around-bottomed flask, heated
crosslinking of PU and its nanocomposites. The sample in the
to 60 °C, and thoroughly mixed with a predetermined amount of
weight of W1 was extracted with acetone for 48 h using Soxhlet
TDI, in which the ANCO/AOH ratio is 2.25. The reaction system
extractor and then, dried in a vacuum oven to constant weight
was then stirred vigorously with mechanical stirring for 60 min
(W2). The degree of crosslinking, Cd, can be determined by:
to prepare a urethane pre-polymer. Secondly, a stoichiometric
amount of DBTDL and 1,4-BDO were added to the stirred system W1
Cd ¼  100% ð1Þ
before addition of certain amount of xylene to control the viscosity W2
of the system. Thirdly, the mixture was degassed under a vacuum
for about 30 min (until bubbling ceased) and then poured into pre-
heated molds (150 mm  120 mm  1 mm glass cavity). Finally, 3. Results and discussion
the system was cured at 120 °C for 6 h.
The PU/CNT nanocomposites were prepared by in situ polymer- 3.1. Creep and creep–recovery behaviors
ization. Firstly, the calculated amounts of CNTs were added to CO
(60 °C). Then the mixtures were stirred vigorously with magnetic Fig. 1 shows the creep compliance–time curves of the neat PU at
stirring for 24 h, followed by 2 h of ultrasonic oscillation to obtain various temperatures. As expected, the creep compliance increases
a stable suspension system. The subsequent steps were the same
1.0x105
as those for the preparation of neat PU. The weight contents of 40 oC 50 oC 60 oC
o o
CNTs were 0.5%, 1%, 2% and 3%. 70 C 80 C 90 oC
creep compliance (μm/N2)

8.0x10 4 secondary creep


2.3. Morphological characterization
primary creep
6.0x104
The dispersion of CNTs in PU matrix was investigated using a S-
4800 field-emission scanning electron microscope (FE-SEM, Hit- instantaneous deformation
achi, Japan) with a 15 kV accelerating voltage. The sheet samples 4.0x104
were frozen in liquid nitrogen and fractured. The fractured surfaces
were then coated with gold using an SPI sputter coater for en-
hanced conductivity before observation. 2.0x104

2.4. Mechanical property characterizations 0.0


0 2 4 6 8 10

The tensile properties of PU and its nanocomposites were deter- time (min)
mined by an Instron Mechanical Tester (ASTM D638) at a cross- Fig. 1. Creep compliance–time curves for the neat PU at various temperatures
head speed of 2 mm/min at 30 °C using the dog-bone shaped under 2.5 MPa. (For interpretation of the references to color in this figure legend,
specimens. the reader is referred to the web version of this article.)
Z. Yao et al. / Composites: Part A 50 (2013) 65–72 67

3.5 25 0.0 wt%


0.8
0.8 0.5 wt%
3.0 7.0 MPa 20
0.6

stress (MPa)
1.0 wt%

strain (%)
15 2.0 wt%
2.5 0.6 0.4

10 0.0 wt% 3.0 wt%

strain (%)
strain (%)

2.0 0.2
5.0 MPa 0.5 wt%
5 linear region 1.0 wt% 0.4
0.0
1.5 2.0 wt%
3.0 MPa 0 10 11 12 13
0.00 0.05 0.10 0.15 0.20 0.25 0.30
0.2 time (min)
1.0 stain (mm/mm)

0.5 2.5 MPa


0.0
1.5 MPa
0.0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
time (min) time (min)

Fig. 2. Creep and creep-recovery strain–time curves for the neat PU under various Fig. 3. Creep and creep-recovery strain–time curves for the PU/CNT nanocompos-
stress levels at 30 °C. The stress–strain curves of the neat PU and its nanocompos- ites with different CNT contents at 30 °C and 2.5 MPa. (For interpretation of the
ites with different CNT contents are given in the inset graph. (For interpretation of references to color in this figure legend, the reader is referred to the web version of
the references to color in this figure legend, the reader is referred to the web version this article.)
of this article.)

strong van der Waals interactions among the nano-scaled CNTs


with temperature, and three typical creep stages (instantaneous [43,44]. In this case, the CNTs cannot be distributed throughout
deformation, primary and secondary creeps) are clear. At the the PU matrix and as a result, their constraining effect on the creep
experimental stress level, the tertiary creep, i.e. creep rupture, of matrix reduces. This is further confirmed by the DMA results. As
which would require longer time, does not appear. This is in favor can be seen in Fig. 5, all nanocomposites show modulus and a tran-
of the following modeling study. Fig. 2 gives the creep and creep– sition temperature values higher than the neat PU. This is attrib-
recovery strain–time curves of the neat PU at various stress levels. uted to the presence of high-modulus CNTs and their impeding
The inset graph shows the stress–strain curves of the neat PU and effect on the chain mobility [45,46]. But with increasing CNT con-
its nanocomposites obtained from tensile tests. Clearly, all the ap- tents, the modulus and transition temperature decrease somewhat,
plied stress levels are in the linear viscoelastic region. Similar with indicating that the sample with higher CNT contents even has low-
other polymers [31,34], the creep and creep–recovery of the neat er effective volume filling ratio than the samples with lower CNT
PU sample present typical viscoelastic–viscoplastic characteristics: contents. This agrees with the creep and SEM results.
the creep strain increases with the time and applied stress level Besides, the presence of CNTs may influence the bulk structure
monotonously, and the residual strains become noticeable under of PU such as crosslinking structure, which is also highly relevant
high loads even after a long time period exceeding the time of ac- to the final creep properties. To a crosslinkable system, the addi-
tive loads. tion of nanofiller may affect crosslinking process in different ways.
The PU/CNT nanocomposites show similar viscoelastic–visco- For instance, the presence of filler can enhance the viscosity of
plastic characteristics on their creep and creep–recovery behaviors, reaction system, changing molecule diffusion ability. Moreover,
as can be seen in Fig. 3. It is clear that the creep strain level of nano- the surface group of filler may change or even participate in the
composites is lower than that of the neat PU, indicating that the curing reactions. Both will affect the final crosslinking level. Srik-
creep of PU matrix is restrained by the presence of CNTs. However, anth et al. [47] found that in the epoxy/amino functionalized
the creep strain level is not reduced monotonously with the in- CNT composites, the presence of CNTs improved the crosslinking
crease of CNT contents. As the CNT contents achieve up to density of epoxy evidently because the surface amino groups of
1.0 wt% and above, the nanocomposites show higher creep level CNTs participated in the crosslinking reactions of epoxy. Similar re-
and higher unrecoverable strain than those containing lower con- sult has also been reported by Dubey et al. [48] on the CNT based
tents of CNTs (indicated parts in Fig. 3). Similar results have also PCR/EPDM blend composites. They found that the radiation could
been reported on the other CNTs based polymer nanocomposites induce interface crosslinking between matrix polymers and CNT,
[32]. resulting in an increased crosslinking density. For the PU/CNT sys-
This CNT content dependence is closely related to the disper- tem in this work, it is hard to directly describe the effect of CNTs on
sion of CNTs. Fig. 4 gives the SEM images of PU/CNT nanocompos- the crosslinking structure of PU through chemical approaches or
ites. At the lower content level (0.5 wt%), CNTs are dispersed spectrum analysis because the used CNTs are merely purified but
randomly in the form of single nanotube or small bundles, present- not surface-functionalized. Hence the Soxhlet extraction method
ing good dispersion and distribution throughout the PU matrix, as was employed here to detect the degrees of crosslinking (Cd) of
can be seen in Fig. 4(a and a0 ). The presence of CNTs can impede the the PU and its nanocomposites roughly. The results are listed in
movements of polymer chain segments and restrain the relax- Table 1.
ations of chain coins as a result [40–42]. Therefore, the creep of It is seen that the values of Cd increase to some extent with CNT
PU matrix, especially viscoelastic and viscoplastic parts, are con- contents, especially at the lower content levels. Further addition of
strained evidently in the presence of well dispersed CNTs. CNTs leads to a decrease of Cd values. However, considering the er-
At the higher content level (3 wt%), however, CNTs are mainly ror range, the difference of crosslinking degrees between PU and its
dispersed as big bundles or even small aggregates, showing poor nanocomposites is not evident. This is reasonable because the used
distribution in the PU matrix (Fig. 4(b and b0 )). This indicates that CNTs are merely purified but not surface-functionalized. Therefore,
the primary CNT aggregates cannot be fully detached at the higher the presence of CNTs may have little effect on the crosslinking of
contents during pre-mixing and in situ polymerization because of PU matrix during reaction and, the final dispersion and distribution
68 Z. Yao et al. / Composites: Part A 50 (2013) 65–72

(a) (a')

(b) (b')
small aggregates

small bundles

Fig. 4. SEM micrographs of PU/CNT nanocomposites with different CNT contents: (a0 and a) 0.5 wt% and (b0 and b) 3.0 wt%. The scale bar for (a and b) and (a0 and b0 ) are 10 lm
and 5 lm, respectively.

4 Table 1
10 The calculated values of crosslinking degree (Cd) of the neat PU
(a) 0.0wt% and its nanocomposites.
0.5wt%
CNT contents (wt%) Cd (%)
1.0wt%
0.0 97.45 ± 0.15
3 3.0wt%
10 0.5 97.65 ± 0.25
log [E' (MPa)]

1.0 98.41 ± 0.30


2.0 97.89 ± 0.26
0.0wt%
0.5wt% 3.0 97.78 ± 0.28
1.0wt%
log [E' (MPa)]

2 3.0wt%
10
of CNTs is dominant on the final creep performance of PU/CNT
nanocomposites.
o
1 temperature ( C)
10 3.2. Theoretical modeling
-100 -50 0 50 100 150 200
o
temperature ( C) In order to further understand how the presence of CNTs affect
the creep process of nanocomposites, several classic creep models
0.30
were used here to describe the relations between CNT contents and
0.0wt% (b) final creep behavior qualitatively. The power law model is one of
0.5wt%
0.25 most commonly used constitutive models, which was developed
1.0wt%
by Findley et al. [49] based on a power law equation:
3.0wt%
0.20 eðtÞ ¼ e0 þ Atn ð2Þ
tan δ

where e(t) is the creep strain at time t, e0 the instantaneous initial


0.15 strain, A the amplitude of transient creep strain and n the time
exponent. The values of e0 can be obtained from creep curves, and
0.10 A and n can be used as the adjusting parameters to apply Eq. (2)
to fit e(t) of the neat PU and its nanocomposites, by which the val-
ues of those parameters can be obtained. The results are listed in
0.05
Table 2. It is seen that the Findley power law model can be well
used to describe the creep of PU/CNT nanocomposites (solid
-100 -50 0 50 100 150 200
o
curves), as shown in Fig. 6.
temperature ( C) Compared with those of the neat PU, the values of A and n of the
Fig. 5. (a) The dynamic storage modulus (E0 ) and (b) loss factor for the neat PU and
nanocomposites decrease more or less. Similar results have also
its nanocomposites under 10 Hz. (For interpretation of the references to color in this been observed on the PA-66/TiO2 [50] and the PCL/SiO2 composite
figure legend, the reader is referred to the web version of this article.) systems [37]. The decrease of A is indicative of reduced creep levels
Z. Yao et al. / Composites: Part A 50 (2013) 65–72 69

Table 2
Average values of parameters obtained from Findley power law model prediction on
the neat PU and its nanocomposites.

CNT contents (wt%) A n


0.0 0.22 ± 0.03 0.24 ± 0.03
0.5 0.14 ± 0.06 0.17 ± 0.07
1.0 0.13 ± 0.04 0.21 ± 0.07
2.0 0.17 ± 0.05 0.19 ± 0.05
3.0 0.17 ± 0.06 0.17 ± 0.06

0.90
__ Findley power law model prediction

0.75
Fig. 7. Typical strain–time curve in a creep and creep-recovery test for polymer
material.
0.60
strain (%)

Table 3
0.45 Average values of parameters obtained from Burgers model prediction on the neat PU
0.0wt%
and its nanocomposites.
0.5wt%
0.30 1.0wt% CNT contents (wt%) EE (MPa) EVl (GPa) gV1 (GPa s) gVp (GPa s) s (s)
2.0wt% 0.0 514.45 1.59 12.9 7.66 8.13
0.5 668.32 2.19 17.2 14.8 7.85
3.0wt%
0.15 1.0 727.76 2.58 24.6 17.6 9.53
0 2 4 6 8 10 2.0 679.68 2.45 19.6 14.3 8.00
3.0 640.91 2.26 15.9 12.4 7.03
time (min)

Fig. 6. Prediction on the creep (30 °C, 2.5 MPa) of the neat PU and its nanocom-
posites by Findley power law model. (For interpretation of the references to color in
this figure legend, the reader is referred to the web version of this article.)
1.0
__Burgers model prediction
of PU in the presence of CNTs and, the decreased values of n sug-
0.8
gest that the nanocomposites would spend more time than the
neat PU to achieve to identical creep level. Both confirm the imped-
ing effect on creep by the CNTs. Theoretically, the values of A and n 0.6
strain (%)

should depend on the CNT volume fraction monotonously. Consid-


ering error ranges, however, these two parameters are nearly inde-
pendent of CNT contents. One possible reason is that the CNTs 0.4
0.0wt%
show poor distribution and dispersion at higher content levels,
resulting in irregular variations of their effective filling volume, 0.5wt%
as discussed above. 0.2 1.0wt%
Another model commonly used is the Burgers model, which is a 2.0wt%
combination of Maxwell and Kelvin–Voigt elements [51]. Accord- 3.0wt%
ing to the Burgers model, the total strain is the sum of three essen- 0.0
0 2 4 6 8 10
tially separate parts, as schematically shown in Fig. 7:
time (min)
eðtÞ ¼ eE þ eV1 þ eVp ð3Þ
Fig. 8. Prediction on creep curves (30 °C, 2.5 MPa) of the neat PU and its
nanocomposites by Burger’s model. (For interpretation of the references to color
where subscripts E, Vl and Vp correspond to the elastic, viscoelastic
in this figure legend, the reader is referred to the web version of this article.)
and viscoplastic contributions. The time dependence of e(t) is:
  
r r tEV1 r
eðtÞ ¼ þ 1  exp  þ t ð4Þ than those of the pure PU. The parameter EE associated to the Max-
EE EV1 gV1 gVp
well spring establishes instantaneous creep strain which would be
where EE and EVl are elastic moduli, gV1 and gVp viscosities, r the recovered after stress elimination. The result shows that the CNTs
applied stress and t the creep time. When the r is applied, the initial are able to improve the elastic modulus of PU matrix, which agrees
deformation takes place in the spring with the modulus EE. Later well with the results of DMA shown in Fig. 5. The maximum of EE
deformation comes from the spring EVl and dashpot gV1, in parallel, for the nanocomposites with 1 wt% CNTs implies highest volume
and from the dashpot with the viscosity gVp. The four unknown filling ratio of CNTs in this content level. Further addition of CNTs
physic variables EE, EVl, gV1and gVp can be used as the adjusting leads to decrease of EE rather than increase, again confirming a de-
parameters to apply Eq. (4) to fit e(t) of the neat PU and its nano- graded dispersion of CNTs indirectly.
composites. The as-obtained values are listed in Table 3. The retardant modulus EVl is related to the stiffness of amor-
It is clear that Burgers model prediction results are in good phous polymer chains. It shows the same tendency with EE, which
accordance with the experimental data for the neat PU and its reflects the reinforcement of CNTs on the Kelvin–Voigt unit. gV1 is
nanocomposites, as can be seen in Fig. 8. All the viscoelastic and related to the viscosity of the Kelvin–Voigt unit and the ratio of
viscoplastic parameters of the nanocomposites show higher values gV1/EV1 is defined as relaxation time s. Minor increase of s also
70 Z. Yao et al. / Composites: Part A 50 (2013) 65–72

1.0 0
10
__Weibull model presiction 0.0wt% 0.0wt%
0.5wt% 0.5wt%
0.8 1.0wt% 1.0wt%
2.0wt% -1
3.0wt%
10

log [strain (%)]


3.0wt%
0.6
strain (%)

0.4
-2
10

0.2

-3
0.0 10
-2 0 2 4 6 8 10 12
10 15 20 25 30 35 40 10 10 10 10 10 10 10 10
time (min) log [t/α T (s)]

Fig. 9. Prediction on creep–recovery curves of the neat PU and its nanocomposites Fig. 10. The master curves for the neat PU and its nanocomposites to a reference
(30 °C, 2.5 MPa) by Weibull distribution function. (For interpretation of the temperature of 40 °C. (For interpretation of the references to color in this figure
references to color in this figure legend, the reader is referred to the web version legend, the reader is referred to the web version of this article.)
of this article.)

recovery, suggesting that the experimental time is not nearly en-


means that the presence of CNTs does not have strong impeding ef- ough for full recovery of those samples after creep.
fect on the PU chain segments movements. But the evidently in-
creased gV1 especially at the content level of 1 wt%, is indicative
3.3. The long-term creep behavior
of strong inhibition on the relaxation of chain coins or local cross-
linking network because their sizes are in same scale with CNT
The time–temperature superposition (TTS) is a convenient way
bundles. These results are hence in accord with the discussion in
to obtain the results with long time scale through shift of the
Section 3.1.
experimental data at various temperatures. The horizontal shift
gVp is related to the unrecoverable creep strain, which is caused factor aT can be calculated according to the Arrhenius equation:
by the permanent deformation. The increased values of gVp indi-  
cate an increased level of permanent deformation for the nano- Q 1 1
log aT ¼  ð6Þ
composites relative to the neat PU. Similar phenomenon has also 2:303R T T r
been observed on the PP/CNT systems [31]. In order to further
where Q is the activation energy and R the universal gas constant. Tr
illustrate the role of CNTs acted during unrecoverable creep defor-
is reference temperature. The as-obtained master curves are shown
mation, the Weibull distribution equation [52] was applied here.
in Fig. 10. The experimental temperatures (30–90 °C) are lower than
On removing the loads, the instantaneous recovery strain er(t) is:
" a transition temperature and the applied stress are in the linear vis-
 b !#
t  t0 r coelastic region of the PU and its nanocomposites. It is clear that the
er ðtÞ ¼ eV1 exp  þ eVp ð5Þ neat PU sample reveals an evident strain platform next to the strain
gr
failure stage after about 109 s of creep at present stress loads, while
where eV1 is the viscoelastic strain recovery determined by the all nanocomposites are still in the secondary creep stage with lower
parameters gr and br over recovery time t. gr characteristic life creep strain level. Beside, the slope of primary stage in master
parameter and br shape parameter. t0 is the time of stress removed. curves, which is indicative of creep rate, reduces evidently in the
eVp is the permanent strain caused by viscous flow effects. The fit-
ting results using Eq. (5) are shown in Fig. 9 and the obtained
parameter values are list in Table 4. 0
0.0wt% slope -0.147
Clearly, the values of viscoelastic strain recovery eV1 decreases
in the presence of CNTs, which is predictable because of inhibiting 0.5wt% slope -0.134
-2 1.0wt% slope -0.143
effect of CNTs. The decrease of characteristic life (gr) values is
indicative of higher recovery rate of nanocomposites than that of 3.0wt% slope -0.122
the neat PU. In other words, the nanocomposites tend towards sta-
shift factor

-4
bility more rapidly during creep–recovery in comparison with the
neat PU. Besides, the values of permanent strain (eVp) through
model prediction are lower than those obtained after 30 min of -6

Table 4 -8
Average values of parameters obtained from Weibull distribution function Burgers
model prediction on the neat PU and its nanocomposites.

CNT contents (wt%) eV1 (%) gr (s) br eVp (%) -10


40 50 60 70 80 90 100
0 0.749 1.237 0.137 0.073 o
0.5 0.553 0.287 0.096 0.009
temperature ( C)
1 0.461 0.067 0.140 0.068
2 0.512 0.162 0.140 0.070 Fig. 11. The shift factor for the neat PU and its nanocomposites as a function of
3 0.508 0.070 0.123 0.081 temperature. (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
Z. Yao et al. / Composites: Part A 50 (2013) 65–72 71

presence of CNTs, especially at the content of 1 wt%. This confirms [8] Aranguren MI, González JF, Mosiewicki MA. Biodegradation of a vegetable
oil based polyurethane and wood flour composites. Polym Test 2012;31:
impeding effect of CNTs on the creep of PU in long-term time scale.
7–15.
For the nanocomposite with 1 wt% CNTs, it may need additional [9] Subramoney S. Novel nanocarbons – structure, properties, and potential
105–106 s to reach the strain failure stage in comparison with the application. Adv Mater 1998;10:1157–71.
neat PU. It is notable that the long-term creep level of the nanocom- [10] Salvetat JP, Kulik AJ, Bonard JM, Briggs GAD, Stockli T, Metenier K, et al. Elastic
modulus of ordered and disordered multiwalled carbon nanotubes. Adv Mater
posites with 3 wt% CNTs becomes higher refer to the nanocompos- 1999;11:161–5.
ite with 1 wt% CNTs. As discussed above, the degraded dispersion of [11] Salvetat JP, Briggs GAD, Bonard JM, Bacsa RR, Kulik AJ, Stockli T, et al. Elastic
CNTs is one of the most important reasons. But there may be some and shear moduli of single-walled carbon nanotube ropes. Phys Rev Lett
1999;82:944–7.
other possible reasons, such as alteration of bulk structure or other [12] Spitalskya Z, Tasis D, Papagelis K, Galiotis C. Carbon nanotube–polymer
unknown factors. This is worthy of further study. composites: chemistry, processing, mechanical and electrical properties. Prog
Moreover, the shift factors at various temperatures reveal good Polym Sci 2010;35:357–401.
[13] Moniruzzaman M, Winey KI. Polymer nanocomposites containing carbon
linearity for all samples, as can be seen in Fig. 11. This suggests that nanotubes. Macromolecules 2006;39:5194–205.
Arrhenius equation can be well used to deal with the time–temper- [14] Valentini L, Kenny JM. Novel approaches to developing carbon nanotube based
ature superposition of creep of those samples. Considering linear polymer composites: fundamental studies and nanotech applications. Polymer
2005;46:6715–8.
deviation, the neat PU and its nanocomposites show almost identi-
[15] Jiang FD, Wu SZ, Wei YJ, Zhang LQ, Hu GH. The study on the microstructures
cal slopes on their shift factor–temperature curves, which indicates and high performance of melt blending polyurethane/multi-walled carbon
that all samples have the relaxation active energy (Q) values close nanotubes composites. Polym Polym Compos 2008;16:471–80.
[16] Ryszkowska J. Quantitative image analysis of polyurethane/carbon nanotube
with one another. Similar results have also been observed on the
composite microstructures. Mater Charact 2009;60:1127–32.
linear dynamic rheological responses of clay based polymer com- [17] Xiong J, Zhou D, Zheng Z, Yang X, Wang X. Fabrication and distribution
posite system [53]. This illustrates that the interactions between characteristics of polyurethane/single-walled carbon nanotube composite
polymer chain segments and CNTs is not the root of creep imped- with anisotropic structure. Polymer 2006;47:1763–6.
[18] Ryszkowska J, Jurczyk-Kowalska M, Szymborski T, Kurzydłowski KJ. Dispersion
ing effect, and the formation of mesoscopic structure of CNTs with- of carbon nanotubes in polyurethane matrix. Physica E 2007;39:124–7.
in the crosslinking network of PU, maybe account for such [19] Ryszkowska J. Applications of quantitative image analysis to the description of
retardation behavior during creep. the morphology of polyurethane/carbon nanotubes composites. Inzy _ Mater
2008;164:204–7.
[20] Kuan HC, Ma CM, Chang WP, Yuen SM, Wu HH, Lee TM. Synthesis, thermal,
mechanical and rheological properties of multiwall carbon nanotubes/
4. Conclusions waterborne polyurethane nanocomposites. Compos Sci Technol
2005;65:1703–10.
The PU/CNT nanocomposites prepared by in situ polymerization [21] Xiong J, Zheng Z, Qin X, Li M, Li H, Wang X. The thermal and mechanical
properties of a polyurethane/multi-walled carbon nanotube composite.
show enhanced creep resistance behavior with decreased creep Carbon 2006;44:2701–7.
strain level in comparison with that of the neat PU. However, the [22] Raja M, Shanmugharaj AM, Ryub SH, Subhac J. Influence of metal nanoparticle
creep strain level is not reduced monotonously with the increase decorated CNTs on polyurethane based electro active shape memory
nanocomposite actuators. Mater Chem Phys 2011;129:925–31.
of CNT contents, which also highly depends on the dispersion of [23] Cho JW, Kim JW, Jung YC, Goo NS. Electroactive shape-memory polyurethane
CNTs. At identical stress level, the nanocomposite containing composites incorporating carbon nanotubes. Macromol Rapid Commun
1.0 wt% CNTs shows the lowest creep level. The creep behaviors 2005;26:412–6.
[24] Ni QQ, Zhang CS, Fu Y, Dai G, Kimura T. Shape memory effect and mechanical
of the neat PU and its nanocomposites can be well described by properties of carbon nanotube/shape memory polymer nanocomposites.
the Findley model and the Burgers model, and the recovery behav- Compos Struc 2007;81:176–84.
iors can be well predicted by the Weibull model. The as-obtained [25] Chen W, Tao XY. Carbon nanotube-reinforced polyurethane composite fibers.
Compos Sci Technol 2006;66:3029–34.
parameters further confirm the impeding effect on the creep by
[26] Meincke O, Kaempfer D, Weickmanna H, Friedrich C, Vathauer M, Warth H.
the presence of CNTs. Mechanical properties and electrical conductivity of carbon-nanotube filled
polyamide-6 and its blends with acrylonitrile/butadiene/styrene. Polymer
2004;45:739–48.
Acknowledgements [27] Frogley MD, Ravich D, Wagner HD. Mechanical properties of carbon
nanoparticles-reinforced elastomer. Compos Sci Technol 2003;63:1647–54.
[28] Xu M, Zhang T, Gu B, Wu JL, Chen Q. Synthesis and properties of novel
This work was supported by the Research Grants from the Na- polyurethane–urea/multi-walled carbon nanotube composites.
tional Natural Science Foundation of China (No. 51173156). Macromolecules 2006;39:3540–5.
[29] Kwon J, Kim H. Comparison of the properties of waterborne polyurethane/
multi-walled carbon nanotube and acid-treated multi-walled carbon
References nanotube composites prepared by in situ polymerization. J Polym Sci Part A
Polym Chem 2005;43:3973–85.
[30] Jia Y, Jiang ZM, Gong XL, Zhang Z. Creep of thermoplastic polyurethane
[1] Lemos VA, Santos MS, Santos ES. Application of polyurethane foam as a sorbent
reinforced with ozone functionalized carbon nanotubes. Express Polym Lett
for trace metal pre-concentration – a review. Spectrochim Acta Part B At
2012;9:750–8.
Spectrosc 2007;62:4–12.
[31] Jia Y, Peng K, Gong XL, Zhang Z. Creep and recovery of polypropylene/carbon
[2] Li FX, Zuo J, Song DH, Li YT, Ding LH, An YL, et al. The design, synthesis and
nanotube composites. Int J Plast 2011;27:1239–51.
characterization of polyurethane with super macromolecular size. Eur Polym J
[32] Starkova O, Buschhorn ST, Mannov E, Schulte K, Aniskevich A. Creep and
2001;137:193–9.
recovery of epoxy/MWCNT nanocomposites. Composites Part A
[3] Kumar H, Kumar A, Siddaramaiah A. Physico-mechanical, thermal and
2012;43:1212–8.
morphological behaviour of polyurethane/poly(methyl methacrylate) semi-
[33] Eitan A, Fisher FT, Andrews R, Brinson LC, Schadler LS. Reinforcement
interpenetrating polymer networks. Polym Degrad Stab 2006;91:1097–104.
mechanisms in MWCNT-filled polycarbonate. Compos Sci Technol
[4] Kostrzewa M, Hausnerova B, Bakar M, Dalka M. Property evaluation and
2006;66:1162–73.
structure analysis of polyurethane/epoxy graft interpenetrating polymer
[34] Zhang W, Joshi A, Wang Z, Kane RS, Koratkar N. Creep mitigation in composites
networks. J Appl Polym Sci 2011;122:1722–30.
using carbon nanotube additives. Nanotechnology 2007;18:185703.
[5] Madhavan K, Reddy BSR. Synthesis and characterization of poly(dimethyl
[35] Alig I, Lellinger D, Dudkin SM, Potschke P. Conductivity spectroscopy on melt
siloxane-urethane) elastomers: effect of hard segments of polyurethane on
processed polypropylene–multiwalled carbon nanotube composites: recovery
morphological and mechanical properties. J Polym Sci Part A Polym Chem
after shear and crystallization. Polymer 2007;48:1020–9.
2006;44:2980–9.
[36] Gan M, Satapathy BK, Thunga M, Weidisch R, Potschke P, Janke A. Temperature
[6] Roeder J, Oliveira RB, Becker D, Goncalves MW, Soldi V, Pires ATN.
dependence of creep behavior of PP–MWNT nanocomposites. Macromol Rapid
Compatibility effect on the thermal degradation behaviour of polypropylene
Commun 2007;28:1624–33.
blends with polyamide 6, ethylene propylene diene copolymer and
[37] Perez CJ, Alvarez VA, Vazquez A. Creep behaviour of layered silicate/starch–
polyurethane. Polym Degrad Stab 2005;90:481–7.
polycaprolactone blends nanocomposites. Mater Sci Eng A 2008;480:259–65.
[7] Kaushik AK, Podsiadlo P, Qin M, Shaw CM, Waas AM, Kotov NA, et al. The role
[38] Liu TM, Bui VT. Instrumented impact testing of castor-oil-based
of nanoparticle layer separation in the finite deformation response of layered
polyurethanes. J Appl Polym Sci 1995;56:345–54.
polyurethane–clay nanocomposites. Macromolecules 2009;42:6588–95.
72 Z. Yao et al. / Composites: Part A 50 (2013) 65–72

[39] Raymond MP, Bui VT. Epoxy/castor oil graft interpenetrating polymer [47] Srikanth I, Kumar S, Kumar A, Ghosal P, Subrahmanyam C. Effect of amino
networks. J Appl Polym Sci 1998;70:1649–59. functionalized MWCNT on the crosslink density, fracture toughness of epoxy
[40] Wu DF, Wang JH, Zhang M, Zhou WD. Rheological properties of carbon and mechanical properties of carbon–epoxy composites. Composites Part A
nanotubes filled poly(vinylidene fluoride) composites. Ind Eng Chem Res 2012;43:2083–6.
2012;51:6705–13. [48] Dubey KA, Bhardwaj YK, Rajkumar K, Panicker L, Chaudhari CV,
[41] Wu DF, Sun YR, Wu L, Zhang M. Kinetic study on the melt compounding of Chakraborty SK, et al. Polychloroprene rubber/ethylene–propylene diene
polypropylene/multi-walled carbon nanotube composites. J Polym Sci Part B monomer/multiple walled carbon nanotube nanocomposites: synergistic
Polym Phys 2009;47:608–18. effects of radiation crosslinking and MWNT addition. J Polym Res
[42] Wu DF, Wu L, Zhang M, Zhao YL. Viscoelasticity and thermal stability of 2012;19:9876–85.
polylactide composites with various functionalized carbon nanotubes. Polym [49] Findley WN, La JS, Onaran K. Creep and relaxation of nonlinear viscoelastic
Degrad Stab 2008;93:1577–84. materials with an introduction to linear viscoelasticity. New York: Dover
[43] Wu DF, Wu L, Sun YR, Zhang M. Rheological properties and crystallization Publications Inc.; 1989.
behavior of multi-walled carbon nanotube/poly(e-caprolactone) composites. J [50] Yang JL, Zhang Z, Schlarb AK, Friedrich K. On the characterization of tensile
Polym Sci Part B Polym Phys 2007;45:3137–47. creep resistance of polyamide 66 nanocomposites. Part II: modeling and
[44] Yuan LJ, Wu DF, Zhang M, Zhou WD. Rheological percolation behavior and prediction of long-term performance. Polymer 2006;47:6745–58.
isothermal crystallization of poly(butylenes succinate)/carbon nanotube [51] Ward IM. Mechanical properties of solid polymers. Weinheim: John Wiley and
composites. Ind Eng Chem Res 2011;50:14186–92. Sons Ltd.; 1983.
[45] Wu DF, Wu L, Zhang M. Rheology of multi-walled carbon nanotube/ [52] Fancey KS. A mechanical model for creep, recovery and stress relaxation in
poly(butylenes terephthalate) composites. J Polym Sci Part B Polym Phys polymeric materials. J Mater Sci 2005;40:4827–31.
2007;45:2239–51. [53] Wu DF, Zhou CX, Zheng H, Mao DL, Zhang B. Study on rheological behavior of
[46] Wu DF, Wu LF, Zhou WD, Yang T, Zhang M. Study on physical properties of poly(butylene terephthalate)/montmorillonite nanocomposites. Eur Polym J
multi-walled carbon nanotube/poly(phenylene sulfide) composites. Polym 2005;41:2199–207.
Eng Sci 2009;49:1727–35.

You might also like