Вы находитесь на странице: 1из 12

Mechanics of Materials 43 (2011) 127138

Contents lists available at ScienceDirect

Mechanics of Materials
journal homepage: www.elsevier.com/locate/mechmat

Inuence of thermoviscoelastic properties and loading conditions on


the recovery performance of shape memory polymers
Xiang Chen, Thao. D. Nguyen
Department of Mechanical Engineering, Johns Hopkins University, Baltimore, MD 21218, USA

a r t i c l e

i n f o

Article history:
Received 5 January 2010
Received in revised form 18 September
2010
Available online 25 January 2011
Keywords:
Shape memory polymers
Unconstrained recovery
Constrained recovery
Constitutive model
Thermoviscoelasticity
Viscoelasticity

a b s t r a c t
This work investigated the inuence of material properties and loading conditions on the
recovery performance of amorphous shape memory polymers using a recently developed
thermoviscoelastic model. The model incorporated the time-dependent mechanisms of
stress and structural relaxation and viscoplastic ow to describe the glass transition of
the material from a soft viscoelastic rubber to a hard viscoplastic glass. The model captured
many important features of the unconstrained strain recovery response and of the stress
hysteresis observed in the constrained recovery response. A parameter study was developed that varied the model and loading parameters one-by-one to compare their effects
on the start and end temperatures and recovery rate of the unconstrained recovery
response and on the stress hysteresis of the constrained recovery response. The loading
parameters included the cooling rate, the annealing time, and the high and low temperatures of the programming stage and the heating rate of the recovery stage. The results conrmed experimental observations that viscoelasticity is the underlying mechanism of the
unconstrained recovery response. In contrast, the constrained recovery response was inuenced by the interaction of many different mechanisms, including thermal expansion and
structural and stress relaxation. For the loading parameters, the cooling rate of the programming stage and the heating rate of the recovery stage had the largest inuence on
both the constrained and unconstrained recovery response.
2011 Elsevier Ltd. All rights reserved.

1. Introduction
Thermally activated shape memory polymers (SMPs) are
a diverse group of materials that can be manufactured to
memorize a permanent shape and programmed thermomechanically to hold a temporary shape. The material recovers
its permanent shape when exposed to a specic temperature stimulus. The simplest type of SMPs are covalently
crosslinked amorphous networks, class I in the taxonomy
of Liu et al. (2007). The permanent shape is preserved in
the equilibrium conguration of the crosslinked network,
while shape storage and recovery stem from mechanisms
of the glass transition. An example of a class I SMP is a family
of photopolymerized networks produced by the copolymer Corresponding author. Tel.: +1 410 516 4538
E-mail address: vicky.nguyen@jhu.edu (Thao. D. Nguyen).
0167-6636/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mechmat.2011.01.001

ization of tert-butyl acrylate (tBA) monomers with di


(ethelyne glycol) dimethacrylate (DEGDMA), and/or poly
(ethelyne glycol) dimethacrylate (PEGDMA) crosslinking
agents developed by Gall et al. (2005) and Yakacki et al.
(2007) for cardiovascular applications. The tBA, DEGDMA,
and PEGDMA homopolymers have widely different thermomechanical properties, including the glass transition
temperature and rubbery modulus. Consequently, the thermomechanical properties of the copolymer can be tailored
to meet specic stress and strain requirements during
recovery by varying the weight fraction and molecular
weight of the crosslinking agents (Yakacki et al., 2008;
Ortega et al., 2008).
While the relationship between the polymer structure
and thermomechanical properties have been demonstrated for a number of class I SMPs (see also Liu et al.
(2007) and Safranski and Gall (2008)), few studies have

128

X. Chen, Thao. D. Nguyen / Mechanics of Materials 43 (2011) 127138

examined the relationship between the thermomechanical


properties and the shape memory performance. Yakacki
et al. (2008) undertook this investigation for the free and
xed strain recovery response of methyl methacrylate
(MMA) copolymerized with PEGDMA. The MMA-co-PEGDMA materials exhibited glass transition temperatures
ranging from Tg = 5692 C and rubbery moduli varying
from 9.323 MPa. The study found that the strain recovery
response was insensitive to the rubbery modulus. In contrast, a similar study by Yakacki et al. (2007) for tBA-coPEGDMA materials with a lower range of rubbery modulus,
1.511.5 MPa, achieved faster strain recovery for higher
rubbery moduli. The strain recovery initiated near the onset temperature of the glass transition, while the constrained recovery response ended near the mid-point
temperature of the glass transition region. The dependence
of the recovery behavior on the glass transition temperature suggests that viscoelasticity strongly inuences the
recovery response (Yakacki et al., 2008). Buckley et al.
(2007) studied the effect of crosslink density on the unconstrained recovery response of triol-crosslinked polyurethanes. Their results showed that an increase in the
crosslink density increased the retardation times and the
temperature of the peak recovery rate for free strain recovery. Crosslinking agents with higher molecular weight also
decreased the temperature span of free strain recovery.
The present work investigates the effects of thermomechanical properties and loading conditions on the unconstrained and constrained recovery behavior of class I
SMPs by applying the recently developed thermoviscoelastic model of Nguyen et al. (2008). Efforts to model the performance of SMPs have accelerated in recent years,
motivated by the need for an efcient design tool for
increasingly sophisticated materials and devices (Mather
et al., 2009). The majority of SMP models conform to either
a phase transition or thermoviscoelastic approach. In the
phase transition approach, the SMP is conceptualized as a
mixture of a rubbery phase, which fully occupies the material at temperatures T  Tg, and a glassy phase, which
dominates at T  Tg (e.g., Liu et al. (2006), Qi et al.
(2008), Chen and Lagoudas (2008)). The glass transition
is modeled as the temperature driven change in the volume fraction of each phase. Shape storage occurs when
deformation incurred by the compliant rubbery phase at
high temperatures becomes locked in the stiff glass phase
during cooling. The permanent shape is recovered when
the stored deformation is released back into the rubbery
phase during heating. In contrast, the thermoviscoelastic
approach uses rheological concepts to model shape memory behavior. A temperature dependence is prescribed for
the viscosity and moduli parameters to describe the transition between a compliant, mobile rubbery behavior at high
temperature and a stiff, immobile glassy behavior at low
temperatures. The earliest thermoviscoelastic models were
one-dimensional small-strain models. Tobushi et al.
(1997), Tobushi et al. (2001), Morshedian et al. (2005),
and Khonakdar et al. (2007) used the Arrhenius equation
to describe the decrease the mobility during cooling
through the glass transition, while Buckley et al. (2007)
used the an empirically determined equation. Generalized
thermoviscoelastic models have been developed for nite

deformations by Diani et al. (2006) and Nguyen et al.


(2008) using an internal state variable framework. Nguyen
et al. (2008) incorporated the time-dependent effects of
viscoelasticity, structural relaxation, and viscoplastic ow
below the glass transition temperature. The nonlinear
AdamsGibbs model was used to model the transition in
the temperature dependence of the viscosity from WLF
above Tg to Arrhenius below Tg. This allowed the model
to predict many important features of the recovery response, including the start temperature of strain recovery,
and the peak stress and its associated temperature of the
constrained recovery stress response.
For the present work, the thermoviscoelastic model of
Nguyen et al. (2008) was applied to study the effects of
material properties describing structural relaxation, high
temperature viscoelasticity, low temperature viscoplasticity, and thermal expansion on the recovery performance.
The study also examined the effects of the thermomechanical loading conditions during the programming and recovery stage, such as strain rate, cooling and heating rate, and
the annealing time. The results conrmed experimental
observations that stress and structural relaxation were
the underlying mechanisms of free strain recovery. In contrast, constrained recovery was inuenced by a complex
combination of stress relaxation, structural relaxation,
thermal expansion. Neither recovery response was inuenced by the viscoplastic ow mechanism of the glassy
material. For the loading parameters, the applied strain
and cooling rate during the programming stage and the
heating rate during the recovery stage had the largest impact on the recovery performance.
2. Methods
2.1. Model formulation
A detailed development of the generalized, nite deformation, thermoviscoelastic model was presented by Nguyen et al. (2008). Here, we summarize the important
features of the model for uniaxial stress simulations of the
recovery response. The model formulation is an extrapolation of the rheological model in Fig. 1 to nite deformation.
The rheological model consists of a thermal element in
series with a mechanical element. The thermal element
experiences structural relaxation upon a temperature
change DT, and the time-dependent thermal strain
response is described by a spring in series with a Voigt
element. In nite deformation, this arrangement is

Fig. 1. A rheological representation of the constitutive model for class I


SMPs. The model has thermal and mechanical elements. The thermal
element experiences structural relaxation upon a temperature change
while the mechanical element undergoes stress relaxation at high
temperature and viscoplastic ow at low temperatures.

129

X. Chen, Thao. D. Nguyen / Mechanics of Materials 43 (2011) 127138

1
d_  d  ar  ag T  T 0 ;

sR

dt 0 0;

where ar and ag are the coefcients of thermal expansion


(CTE) of the rubbery and glassy materials and T0 is the initial
temperature. The nonlinear AdamGibbs model for the
structural relaxation time sR was used to describe the transition in the temperature dependence of the chain mobility
from the rubbery WLF behavior to the glassy Arrhenius
behavior (Adam and Gibbs, 1965; Scherer, 1984; Hodge,
1987). The model for sR can be expressed in terms of the
WLF constants, C1 and C2, and the reference structural relaxref
ation time, sref
R , at the glass transition temperature T g as,


13
0
C 2 T  T f T T f  T ref
g
C
1
4
@
A5;


sR T; T f sref
R exp 
log e
T C 2 T  T ref
2

1
Tf
d T0;
ar  ag

where Tf is the ctive temperature rst introduced by Tool


(1946) to describe the effects of heat treatments on glass.
Here, the ctive temperature scales linearly with d and acts
as an equivalent representation of the nonequilibrium volumetric deformation. The parameter T ref
g in Eq. (2) is the glass
transition temperature measured at a reference cooling rate.
In general, the glass transition temperature will change with
the cooling/heating rate because of structural relaxation.
The mechanical element, which consists of a spring in
parallel with a Maxwell element, describes the time and
temperature dependent viscoelastic behavior of the rubbery material and the viscoplastic behavior of the glassy
material. In nite deformation, the mechanical stretch for
the deformation of the Maxwell element is decomposed
into elastic and viscous parts, kMi keMi kvMi . We further
decompose the mechanical stretch into volumetric parts,

HM kM1 kM2 kM3 and HeM keM1 kMe2 keM3 , and distortional
e1=3

parts, 
kMi H1=3
kMi and 
keMi HM
M

keMi . Correspondingly,

the stress response is decomposed additively into an equilibrium distortional part, seq
i , a nonequilibrium distortional
, and an elastic volumetric part p. The Arruda and
part, sneq
i
Boyce (1993) network model with Langevin chain statistics
is applied for the equilibrium part to describe the compliant entropic elastic behavior of the rubbery material, while
a NeoHookean model is applied for the nonequilibrium
part to describe the stiff response of the glassy material.
Introducing the effective stretch of a representative volume element of the polymer network and a corresponding
elastic effective stretch as,

 
1=2
nM 1 k2 k2 k2
;
M2
M3
3 M1
 
1=2
ne 1 ke2 ke2 ke2
;
M1
M2
M3
M
3

we can express the stress response as,

1
1
1
2
2
ri leq k2Mi  n2M lneq keMi  neM jHM  1;
J
|{z}
eq

si

J
|{z}
neq

si

 
nM
kL
;
lN  L1
kL
nM

leq

J
|{z}
p

where J = HMHJ = k1k2k3 is the total volumetric deformation. The parameters lN is the shear modulus of the
Table 1
Parameters of the thermoviscoelastic model
Model
parameter

Baseline
value

Physical signicance


T ref
g C

25

Glass transition temperature for


qcool = 1 C/min

sref
R (s)
sref
S (s)

1100

Structural relaxation time at T T ref


g

C1
C2(C)
ar(104/C)

17.44
90
7.67

leq (MPa)
j (MPa)
kL

0.88
611
4.0

ag(104/C)

3.85

neq

34.9

Deviatoric stress relaxation time at


T T ref
g

(MPa)

406

sy0 MPa
Q S =sy0 K=MPa

40
101

syss =sy0

0.43

h (MPa)

250

First WLF constant


Second WLF constant
Rubbery coefcient of volumetric
thermal expansion
Equilibrium shear modulus
Bulk modulus
Limiting chain stretch of equilibrium
network
Glassy coefcient of volumetric
thermal expansion
Non-equilibrium shear modulus of
glassy material
Steady-state yield strength
Activation parameter for viscous
ow
Ratio of initial to steady-state yield
strength
Flow softening modulus

Tonset

10

Storage Modulus (MPa)

described by the multiplicative decomposition of the stretch


into thermal and mechanical parts, ki kTi kMi , where i = 1, 2,
3 for the loading and two lateral directions. The thermal
deformation is assumed to be isotropic and kTi H1=3
T ,
where HT is the volumetric thermal deformation. The thermal element experiences structural relaxation upon a temperature change DT. Thus, the thermal dilatation is
decomposed additively as, HT = 1 + agDT + d, where d(t, T)
is an internal variable related to the nonequilibrium volumetric deformation. The internal thermal deformation
evolves to equilibrium according to the following rst order
rate equation,

10

10

Tterm

10

20

40

60

80

100

Temperature (oC)
Fig. 2. Dening the onset and termination temperatures, Tonset and Tterm,
and slope S of the glass transition region for the temperature dependence
of the storage modulus.

130

X. Chen, Thao. D. Nguyen / Mechanics of Materials 43 (2011) 127138

rubbery network, kL is the limiting stretch corresponding


the contour length of the network chains, j is the bulk
modulus, and leq + lneq is the shear modulus of the glassy
material. For the uniaxial stress problem where i = 1 is the
loading direction, the traction free boundary conditions,
r2 = 0 and r3 = 0, are used to solve for the lateral strains
k2 and k3.
A modied Eyring ow rule is used to model time evolution of the viscous strains to capture both the viscoelastic
relaxation at high temperatures and viscoplastic ow at
low temperatures,

and the compliant entropic elastic behavior of the rubbery


material. As the material cools, the molecular mobility
decreases and the stress response begins to exhibit noticeable viscoelasticity. The decrease in the molecular mobility
is modeled by the increase in the relaxation times sR and sS
with decreasing temperature. From a rheological point of
view, this temperature dependence causes the deformation imposed at high temperatures to become locked in
the dashpot of the Maxwell element at low temperature.
Structural relaxation changes the temperature dependence
of the relaxation times from a WLF to Arrhenius depen-


11
0
0
2
3

 neq
C 2 T  T f T T f  T ref
g
k_ vi
1
C
s
T
Q
s
1
y
S
@
AA
5 si ;


4 ref exp @
sinh
ref
Q
s
log e
T
s
kvi
2gS
y
S
T C2 T f  T g
|{z}
s

 
1=2
2
2
1 neq2
s1 sneq
sneq
:
2
3
2

c_ v

The variable s is the ow stress, sy is the yield strength, and


QS scales with the activation energy. The resulting stress
relaxation time, sS = gS/lneq, where gS is the shear viscosity, has the same temperature and structure dependence
as the structural relaxation time sR. To model post-yield
softening, the yield strength is assumed to evolve with
the viscous strain rate as follows,



sy
c_ v ;
s_ y h 1 
syss

kv 0 1

sy t 0 syss ;

Applied Compressive Strain

where sy0 is the initial yield strength, Syss is the steady state
yield strength, and h is the softening modulus.
In summary, the material model exhibits thermoelastic
behavior at high temperatures, such that for T  Tg, the
thermal strain and stress response is given by the CTE

Table 2
Thermomechanical loading parameters
Loading
parameter

Baseline
values

Signicance

qcool (C/min)
qheat (C/min)
e_ s1

1
1
102

emax

30%

Thigh (C)
Tlow (C)
teq (min)

80
10
1

tan (min)

30

Cooling rate during programming


Heating rate during recovery
Compressive true strain-rate during
programming
Applied compressive strain during
programming
Initial and nal temperature
Annealing temperature
Equilibration time at Thigh before
cooling
Annealing time at Tlow before
recovery

Constrained Recovery

qcool

qheat

Fig. 3. Applied strain and temperature history. The compressive strain was constrained to be greater than or equal to emax during the heating stage in a
constrained recovery simulation, while no constraints were placed in the a unconstrained recovery simulation.

131

X. Chen, Thao. D. Nguyen / Mechanics of Materials 43 (2011) 127138

dence during the glass transition. The nonlinear dependence of sS on the ow stress allows the material to yield
then ow plastically when deformed in the glassy state.
2.2. Parameter study
The model parameters were determined from a set of
thermomechanical experiments for a tBA-co-PEGDMA

SMP using a procedure described in detail in Nguyen


et al. (2008). The experiments included dynamic mechanical analysis (DMA) at 1 Hz for the temperature dependence
of the storage modulus, thermal expansion experiments at
a cooling rate of qcool = 1 C/min, and isothermal compression experiments at strain rates of e_ 0:1=s and
e_ 0:01=s to characterize the rate-dependent and temperature-dependent viscoplastic behavior of the glass. The

end

Tpeak

1.2

0.83

Cooling
Heating

1
0.66

(MPa)

Recovered strain ratio (1-/max)

0.5

0.6
0.4

0.33
T

0.2

start

0.17
0.0

peak

T0.5peak

0.8

0
0

20

40

60

0.2

80

20

40

60

80

Temperature ( C)

Temperature ( C)

C1

C2

100

C1

C2

45
90
180

60

8.72
17.44
34.88

Fig. 4. Recovery behavior calculated for the baseline parameters: (a) strain-temperature curve for unconstrained recovery, and (b) stresstemperature
curve for constrained recovery.

Rref

80

30
20

60

8.72
17.44
34.88

45
90
180

20

550
1100
2200

C2

1
0.8

C1

0.4

Sref

Rref

550
1100
2200

0.6

17.45
34.9
69.8

S (log(MPa)/oC)

Rref

40

10
17.45
34.9
69.8

Sref

550
1100
2200

Sref

17.45
34.9
69.8

40

Tterm (oC)

Tonset (oC)

50

45
90
180

8.72
17.44
34.88

0.2

Fig. 5. Effects of stress and structural relaxation parameters on (a) Tonset, (b) Tterm, and (c) slope S of the glass transition of the storage modulus.

X. Chen, Thao. D. Nguyen / Mechanics of Materials 43 (2011) 127138

glass transition temperature T ref


g was determined from the
thermal expansion experiments for a cooling rate of
qcool = 1 C/min. The WLF parameters C1 and C2 should be
determined from timetemperature superposition tests
for the master curve of the frequency dependence of the
storage modulus. However, the model assumed that the
viscoelastic behavior was described by a single characteristic relaxation time. In reality, the SMP exhibited a broad
spectrum of relaxation times. Because of this simplication, the WLF parameters were treated as phenomenological parameters determined along with the characteristic
relaxation time to t the temperature range of the glass
transition of the storage modulus. We are currently working to incorporate a broad relaxation spectrum in the model to more accurately describe the viscoelastic behavior of
SMPs (Nguyen et al., 2010).
Table 1 summarizes the parameters of the thermoviscoelastic models, their physical signicance, and values
determined for the tBA-co-PEGDMA material. These were
used as the baseline values for the parameter study, which
varied the parameters one-by-one from 0.5 to 2 times the
baseline value. The effects of the CTE, moduli parameters,
and limiting stretch on the thermal dilatation and stress
response of the SMP are clear. The viscoplasticity parameters (sy0 ; syss ; Q s , h0) determine the stress response for large
deformations and temperatures below the glass transition

ref
temperature. The parameters C1, C2, sref
S , and sR describes
the time-dependence of the stress response for temperatures near and greater than the glass transition temperature. To understand the effects of the WLF parameters
and characteristic stress and structural relaxation times
on the viscoelastic behavior of the SMP, the temperature
dependence of the storage modulus was calculated also
for parameter variations of a factor of 0.5 and 2 from their
baseline values. Fig. 2 plots the storage modulus calculated
for uniaxial compression at 1 Hz using the baseline parameters. For the purpose of comparison, we dened three features of the glass transition region: the onset temperature
Tonset, termination temperature Tterm, and the slope S as
shown in Fig. 2.

2.3. Numerical simulations of recovery


The thermoviscoelastic model was implemented
numerically into Matlab and used to simulate the unconstrained strain recovery and the constrained recovery
stress response under uniaxial compression. Fig. 3 illustrates the strain and temperature loading history for both
recovery simulations. Both began with the material equilibrated at the temperature Thigh = 80 C. At time t = 0, the
strain was applied at a constant rate to a compressive true
value of emax = 0.3. The compressed material was held at

C2

neq

eq

50

sy0

C2

ref
R

neq

eq

sy0

40

Tend(oC)

30

20

30
20

10

8.72
17.44
34.88

neq

eq

45
90
180

17.45
34.9
69.8
550
1100
2200

20
40
80

0.44
0.88
1.75

1.92
3.85
7.70
203
406
812

3.84
7.67
15.34

45
90
180

8.72
17.44
34.88

C2

1.2
1.0

k (oC-1)

0.8
0.6
0.4

C1
Sref

Rref

sy0

0.44
0.88
1.75
20
40
80

203
406
812

1.92
3.85
7.70

3.84
7.67
15.34

45
90
180

0.0

8.72
17.44
34.88

0.2

17.45
34.9
69.8
550
1100
2200

10

17.45
34.9
69.8
550
1100
2200

Tstart(oC)

C1
ref
S

20
40
80

ref
R

0.44
0.88
1.75

ref
S

203
406
812

40

1.92
3.85
7.70

60

C1

3.84
7.67
15.34

132

Fig. 6. Effects of model parameters on (a) Tstart, (b) Tend, and (c) the maximum slope k of the unconstrained recovery response.

133

X. Chen, Thao. D. Nguyen / Mechanics of Materials 43 (2011) 127138

Thigh for teq = 1 min, cooled at a rate qcool = 1 C/min to


Tlow = 10 C, then annealed at Tlow for tan = 30 min. Thermal contraction during cooling caused the compressive
strain to exceed emax. In an experiment, this would correspond to the SMP specimen contracting away from the
compression platens. For the constrained recovery simulations, the material was reheated at a rate of qheat = 1 C/min
to Thigh while constraining the compressive strain to be
e P emax. This corresponded to allowing the specimen to
expand thermally back into contact with the compression
platen. In the strain recovery simulations, the sample
was reheated without any constraints placed on the strain.
Table 2 summarizes the loading parameters and their baseline values. These were varied one-by-one by a factor of 2
and 0.5 in the parameter study, except for the high and low
temperatures. The ranges of those were set to Thigh =
(52.5, 80, 135) C and Tlow = (45, 10, 7.5) C.
Fig. 4(a) shows the recovered true-strain ratio during
the heating stage of the unconstrained recovery simulation, and Fig. 4(b) plots the true-stress response from constrained recovery for the entire temperature cycle. The
baseline model and loading parameters were used for the
simulations. Both results agreed well with experiments
as discussed in Nguyen et al. (2008). In particular, the model was able to predict the onset temperature of strain

recovery, and the peak stress and its temperature of the


constrained recovery case. To compare the impact of the
different parameters, three characteristic features were dened for each recovery response. For the unconstrained response, the maximum tangent k described the strain
recovery rate. The start and end temperatures were dened from the intersection of a line denoting the thermal
expansion of glassy material for Tstart and of the thermal
expansion of the rubbery with material for Tend with a line
of slope k for the recovered strain. For the constrained
recovery simulations, Drpeak referred to the maximum
stress overshoot encountered during the heating stage at
the temperature Tpeak. The DT0.5peak described the temperature span of the stresstemperature hysteresis curve calculated where Dr = 0.5Drpeak. The parameter study
compared the sensitivity of these features of the recovery
response to the parameter variations.
3. Results and discussion
3.1. Impact of thermoviscoelastic model parameters
Fig. 5 compares the effects of the WLF parameters, C1
and C2, and the characteristic relaxation times, sref
and
R
sref
on the storage modulus. The parameter sref
S ,
S

0.8

ref
R
ref
R

Recovered strain ratio (1 /max)

Recovered strain ratio (1 /max)

ref

R = 550 s
= 1100 s
= 2200 s

0.6

0.4

0.2

0
20

25

30

35

40

45

ref = 17.45 s
S

0.8

S
ref

S = 69.8 s
0.6

0.4

0.2

0
20

50

ref = 34.9 s

25

30

Recovered strain ratio (1 /max)

max

Recovered strain ratio (1 /

40

45

50

0.8

0.6

0.4

C =8.72
1

C1 = 17.44

0.2

C =34.88
1

0
20

35

Temperature (oC)

Temperature ( C)

30

40

Temperature ( C)

50

60

0.8

0.6

C2=45oC

0.4

C =90 C
2

C =180oC

0.2

0
20

30

40

50

60

Temperature ( C)

Fig. 7. Comparing the effects of different (a) structural relaxation times, (b) stress relaxation times, and WLF parameters (c) C1 and (d) C2 on the
unconstrained recovery response.

134

X. Chen, Thao. D. Nguyen / Mechanics of Materials 43 (2011) 127138

in response to an applied stress that was much smaller


than the yield strength. This viscous deformation was
stored in the dashpot of the Maxwell element during cooling then released during heating (Fig. 1). The stress state
remained signicantly smaller than the yield stress
throughout the programming and recovery process. As a
result, both were insensitive to the parameters for yielding.
Both the rubbery and nonequilibrium shear modulus, leq
and lneq, had a noticeable effect on the unconstrained
recovery response. A higher rubbery modulus increased
the recovery rate, which agreed with the experimental
observations of Yakacki et al. (2007), and caused strain
recovery to start at a lower temperature. The bulk modulus
had no effect on the response (not shown) because the rubbery material was modeled as quasi-incompressible at
Thigh.
Fig. 7 compares the unconstrained recovery response
for different values of the WLF parameters and relaxation
times, sref
and sref
S
S . Increasing the structural relaxation
time sref
caused
the stress relaxation time to respond more
R
sluggishly to a temperature rise. This shifted strain recovery to higher temperatures and resulted in a higher recovery rate k. Similar features were observed in Fig. 5 for the
temperature dependence of the storage modulus. The
stress relaxation time, sref
S , and WLF parameters, C1 and

corresponds to the molecular resistance to viscous ow. A


larger sref
produced a higher glass transition temperature
S
and broader glass transition temperature range (smaller
S). The structural relaxation time sref
describes how
R
quickly the nonequilibrium structure, represented by d,
and thus the structure and temperature dependent stress
relaxation time responds to a temperature change. A larger
sref
produced a more sluggish response, which increased
R
the onset temperature Tonset and the slope S of the glass
transition region. The storage modulus was most sensitive
to the WLF parameters, though C1 and C2 had opposite effects on the glass transition. A lower C2 and higher C1
shifted the glass transition to lower temperatures and narrowed its temperature range. Decreasing C2 by half increased S by more than a factor of ve.
Fig. 6 shows the effects of some key thermoviscoelastic
parameters on the unconstrained recovery response. As expected, the rubbery and glassy CTE did not contribute to
the unconstrained recovery response. The viscoplastic
parameters also had negligible impact as shown in Fig. 6
for the initial yield strength, sy0. The same result was obtained for the steady state yield strength, syss, the activation parameter, QS, and the softening modulus, h (not
shown). The programmed deformation was applied at
Thigh  Tg where the material could ow viscoelastically

55

40
38

50

Tend (oC)

34

45

40

32

35

30
28
30

35

40

45

50

55

30
20

40

60

Tonset (oC)

80

100

Tterm (oC)

(a)

(b)
1.2
1.0

k (1/oC)

Tstart (oC)

36

0.8
0.6
0.4
0.2

0.2

0.4

0.6

0.8

S (log(MPa)/oC)

(c)
Fig. 8. Comparing the temperature dependence of the storage modulus and strain recovery response.

120

135

X. Chen, Thao. D. Nguyen / Mechanics of Materials 43 (2011) 127138

neq

eq

sy0

203
406
812

0.44
0.88
1.75

20
40
80

15

sy0

10

0.4

45
90
180

17.45
34.9
69.8
550
1100
2200

0.44
0.88
1.75
20
40
80

203
406
812

1.92
3.85
7.70

3.84
7.67
15.34

45
90
180

8.72
17.44
34.88

8.72
17.44
34.88

0.2

C1

C2

Rref

neq

eq

sy0

20
40
80

Sref

0.44
0.88
1.75

15

203
406
812

20

T0.5peak (oC)

10

1.92
3.85
7.70

3.84
7.67
15.34

45
90
180

8.72
17.44
34.88

17.45
34.9
69.8
550
1100
2200

C2

1.92
3.85
7.70

eq

C1

Tpeak (oC)

C2

Rref

17.45
34.9
69.8
550
1100
2200

peak (MPa)

Sref

Rref

20

neq

C1

0.8

0.6

Sref

3.84
7.67
15.34

inuenced by thermal expansion as well as stress and


structural relaxation mechanisms. Fig. 10 compares the
constrained stress response during the heating stage for
different values of the glassy CTE and nonequilibrium
modulus, two parameters which produced a large effect.
The peak stress overshoot, Drpeak, was strongly correlated
with the glassy CTE and was dependent on the nonequilibrium shear modulus. These results showed that the overshoot in the stress early in the heating stage was caused
by the constrained thermal expansion of the stiff glassy
material. At higher temperatures near the glass transition,
the material became more compliant and the stress relaxed from the peak stress to the programmed stress at
Thigh. The peak stress increased with the rubbery modulus
since the programmed stress scaled with leq. However, the
increase was not proportional, and a larger leq produced a
smaller ratio of the peak stress rpeak to the programmed
stress. The peak stress overshoot also increased with
ref
decreasing C1 and C2 and increasing sref
and
S . Larger sS
smaller C1 either shifted the glass transition temperature
to higher temperatures or produced a broader glass transition temperature range. Both resulted in the constrained
thermal expansion of a stiffer material early during heating. The temperature Tpeak of the peak stress also strongly
depended on the structural and stress relaxation times

C2, also had similar impact on the unconstrained recovery


response as observed for the storage modulus. The recovery rate k was most sensitive to the WLF parameters C1
and C2, which controlled the temperature range of the glass
transition region S. These results suggested that the features of the unconstrained recovery response correlated
with those of the temperature dependence of the storage
modulus.
Fig. 8 compares the features of the temperature dependence of the storage modulus and strain recovery response
ref
for cases involving the WLF parameters, sref
S , and sR . The
Tend of the unconstrained recovery response increased
nearly linearly with the termination temperature Tterm of
the glass transition of the storage modulus. Similarly, the
recovery slope k correlated well with the slope S of the
glass transition region of the storage modulus, and Tstart
of the recovery response generally trended with Tonset.
The correlation between Tonset and Tstart were not as good
as for the other features because of the signicant inuences of structural relaxation, which was most active at
the start of recovery. These results strongly suggests that
structural and stress relaxation are the underlying mechanisms of unconstrained shape recovery.
The results of the parameter study for constrained
recovery in Fig. 9 showed that the stress response was

Fig. 9. Effects of model parameters on (a) Drpeak, (b) Tpeak, and (c) DT0.5peak of the constrained recovery response.

136

X. Chen, Thao. D. Nguyen / Mechanics of Materials 43 (2011) 127138

1.8
o

neq=203MPa

neq=406MPa

g=1.92e4/ C
=3.85e4/ C
g

(MPa)

(MPa)

neq=812MPa

1.4

g=7.7e4/ C

1.5

0.6
0.5
0.2
0
0

10

20

30

40

10

Temperature (oC)

20

30

40

Temperature (oC)

Fig. 10. Comparing the effects of different (a) glassy CTEs and (b) nonequilibrium moduli on the constrained recovery response.

and the WLF parameters. A larger sref


increased the glass
S
transition temperature, which caused the material to remain glassy longer during heating. This led to a higher
peak stress at a higher temperature, Tpeak. The parameter
C1 had the largest effect on the temperature span DT0.5peak.
As with the unconstrained recovery response, the

37

viscoplasticity parameters, sy0, syss, QS, and h, did not affect


the constrained recovery response, because the stress state
of the deformed specimen remained signicantly below
the yield strength during the programming and heating
stages for the selected model parameters and loading
condition.

43

qheat

36

42

qcool

max

Thigh

Tlow

qcool

41

tan

34

Tend(oC)

Tstart(oC)

35

qheat

33

max

Thigh

Tlow

tan

40
39

32

15
30
60

-45
-10
7.5

52.5
80
135

0.15
0.30
0.60

0.5
1.0
2.0

15
30
60

-45
-10
7.5

52.5
80
135

36

0.15
0.30
0.60

29

0.5
1.0
2.0

37

0.5
1.0
2.0

30

0.5
1.0
2.0

38

31

k (oC-1)

2.0

max

Thigh

52.5
80
135

qheat
qcool

0.15
0.30
0.60

2.5

Tlow

tan

1.5

1.0

15
30
60

-45
-10
7.5

0.5
1.0
2.0

0.0

0.5
1.0
2.0

0.5

Fig. 11. Effects of the thermomechanical loading parameters on (a) Tstart, (b) Tend, and (c) the maximum slope k of the unconstrained recovery response.

137

qcool

qheat

20

max

18

Thigh

0.6

Tlow

tan

qcool

qheat

max

Thigh

Tlow

tan

15
30
60

0.7

-45
-10
7.5

0.8

52.5
80
135

X. Chen, Thao. D. Nguyen / Mechanics of Materials 43 (2011) 127138

16

Tpeak (oC)

peak (MPa)

14
0.5
0.4

12
10
8

0.3

0.2

4
0.1

0.5
1.0
2.0

0.15
0.30
0.60

max

14

Thigh

Tlow

tan

-45
-10
7.5

15
30
60

qheat

0.5
1.0
2.0

qcool

52.5
80
135

16

-45
-10
7.5

52.5
80
135

0.15
0.30
0.60

0.5
1.0
2.0

2
0.5
1.0
2.0

T0.5peak (oC)

12
10
8
6
4

15
30
60

0.15
0.30
0.60

0.5
1.0
2.0

0.5
1.0
2.0

Fig. 12. Effects of thermomechanical loading parameters on (a) Drpeak, (b) Tpeak, and (c) D T0.5peak of the constrain recovery response.

3.2. Impact of thermomechanical loading parameters


Both the unconstrained and constrained recovery responses were unaffected by the strain rate e_ and equilibration time teq of the programming stage. The relaxation time
was negligibly small at Thigh = 80 C such that stress response attained equilibrium nearly instantaneously.
Fig. 11 shows the effects of the remainder of the thermomechanical loading parameters on the unconstrained
recovery response. The start and end temperatures and
the recovery rate were all signicantly affected by the
heating and cooling rate, which demonstrated the importance of structural relaxation to the strain recovery response. A faster heating rate shifted the glass transition
and thus the onset of strain recovery to higher temperatures. It also increased the recovery rate because relaxation
occurs faster at higher temperatures. A lower Tstart and
slower recovery rate were also observed for faster cooling
rate. A slower cooling rate allowed the material structure
to evolve more towards equilibrium during programming.
This caused the molecular mobility to be lower (i.e., the
relaxation times to be higher) at the start of the heating
stage. The parameters Thigh and Tlow also signicantly affected the start and end temperatures of the unconstrained
recovery response, while the applied strain emax had little
effect.

The cooling and heating rates also had a large effect on


the constrained recovery response as shown in Fig. 12. A
faster heating rate increased the glass transition temperature, while a slower cooling rate allowed the material
structure to evolve closer to equilibrium during cooling.
Both effects delayed the softening of the material during
heating, which led to a greater stress overshoot. These results were consistent with the experimental measurements of Castro et al. (2010), which showed a slight
increase in the stress overshoot with decreasing cooling
rate and a larger increase in the stress overshoot with
increasing heating rate. As expected, the peak stress overshoot increased with the programmed strain emax, though
not proportionally. In general, emax and leq produced similar effects on the constrained recovery response as shown
in Figs. 9 and 12.

4. Conclusions
A parameter study was developed to investigate the
relationship between the thermomechanical material
properties and loading conditions during programming
and recovery on the unconstrained strain recovery response and the constrained recovery stress response of
amorphous shape memory polymers. The study used a

138

X. Chen, Thao. D. Nguyen / Mechanics of Materials 43 (2011) 127138

thermoviscoelastic model recently developed by Nguyen


et al. (2008) that incorporated the time-dependent effects
of stress and structural relaxation and viscoplastic ow of
the glassy material. The model parameters were tted in
a prior study to the thermomechanical properties of a
tBA-co-PEGDMA material. The loading parameters included the cooling and heating rate, strain rate, anneal
time, and Thigh and Tlow of the thermomechanical cycle.
The main conclusions of the parameter study were:
 Stress and structural relaxation were the dominant
mechanisms of unconstrained recovery. The start and
end temperatures of strain recovery corresponded with
those of the glass transition of the storage modulus, and
the sharpness of the strain recovery corresponded to
the sharpness of the glass transition region.
 A larger rubbery modulus and smaller glassy modulus
resulted in a faster strain recovery rate and lower start
temperature.
 Constrained recovery was inuenced by thermal expansion in addition to relaxation mechanisms. The features
of the constrained stress response generally did not correlate with features the glass transition of the storage
modulus.
 The stress overshoot observed in the constrained recovery experiments was caused by the constrained thermal
expansion of the stiff glassy material. The peak stress
and its temperature, Tpeak, can be controlled by varying
the glassy CTE and glassy modulus as well as the characteristic relaxation times and the temperature range of
the glass transition region.
 The viscoplasticity parameters for the yield strengths,
activation volume, and softening modulus had no effect
on the free and constrained recovery performance. For
both recovery processes, the programmed deformation
was applied at high temperatures to the soft rubbery
material. This produced internal stresses that were signicantly smaller than the yield strength. The viscoplastic ow mechanism would be more important to
the recovery behavior if the material was programmed
by cold drawing.
 The cooling rate of the programming stage and the
heating rate of the recovery stage had the largest impact
on the unconstrained and constrained recovery
response, which demonstrated the importance of structural relaxation. In particular the rate of strain recovery
and the peak stress overshoot of the constrained recovery response can be tailored by varying the cooling and
heating rates, which has important implications for
designing the work capacity of SMPs.
Many of these conclusions agree with experiments,
which demonstrates the predictive potential of the thermoviscoelastic model. Currently, we are working to incorporate the effects of a broad viscoelastic and structural
relaxation spectrum to improve the model prediction of
the unconstrained recovery rate.

References
Adam, G., Gibbs, J.H., 1965. On the temperature dependence of
cooperative relaxation properties in the glass-forming liquids. J.
Chem. Phys. 43, 139146.
Arruda, E.M., Boyce, M.C., 1993. A three-dimensional constitutive model
for the large stretch behavior of rubber elastic materials. J. Mech.
Phys. Solid. 41, 389412.
Buckley, C.P., Prisacariu, C., Caraculacu, A., 2007. Novel triol-crosslinked
polyurethanes and their thermorheological characterization as shapememory materials. Polymers 48, 13381396.
Castro, F., Westbrook, K.K., Long, K.N., Shanda, R., Qi., H.J., 2010. Effects of
thermal rates on the thermomechanical behaviors of amorphous shape
memory polymers. Mech. Time-Dependent Mater. 14, 219241.
Chen, Y., Lagoudas, D.C., 2008. A constitutive theory for shape memory
polymers. Part I large deformations. J. Mech. Phys. Solid. 56, 17521765.
Diani, J., Liu, Y., Gall, K., 2006. Finite strain 3d thermoviscoelastic
constitutive model for shape memory polymers. Polymer Eng. Sci.
46, 486492.
Gall, K., Yakacki, C.M., Liu, Y., Willet, N., Anseth, K.S., 2005.
Thermomechanics of the shape memory effect in polymers for
biomedical applications. J. Biomedical Mater. Res. Part A 73, 339348.
Hodge, I., 1987. Effects of annealing and prior history on enthalpy
relaxation in glassy polymers: Adamgibbs formulation of
nonlinearity. Macromolecules 20, 28972908.
Khonakdar, H.A., Jafari, S.H., Rasouli, S., Morshedian, J., Abedini, H., 2007.
Investigation and modeling of temperature dependence recovery
behavior of shape-memory crosslinked polyethylene. Macromol.
Theory Simulat. 16, 4352.
Liu, C., Qin, H., Mather, P.T., 2007. Review of progress in shape-memory
polymers. J. Mater. Chem. 17, 15431558.
Liu, Y., Gall, K., Dunn, M.L., Greenberg, A.R., Diani, J., 2006.
Thermomechanics
of
shape
memory
polymers:
uniaxial
experiments and constitutive modeling. Int. J. Plasticity 22, 279313.
Mather, P.T., Luo, X., Rousseau, I., 2009. Shape memory polymer research.
Ann. Rev. Mater. Res. 39, 445471.
Morshedian, J., Khonakdar, H.A., Rasouli, S., 2005. Modeling of shape
memory induction and recovery in heat-shrinkable polymers.
Macromol. Theory Simulat. 14, 428434.
Nguyen, T.D., Qi, H.J., Castro, F., Long, K.N., 2008. A thermoviscoelastic
model for amorphous shape memory polymers: incorporating
structural and stress relaxation. J. Mech. Phys. Solids 56, 27922814.
Nguyen, T.D., Yakacki, C.M., Brahmbhatt, P.D., Chambers, M.L., 2010.
Modeling the relaxation mechanisms of amorphous shape memory
polymers. Adv. Mater. 22, 34113423.
Ortega, A.M., Kasprzak, S.E., Yakacki, C.M., Diani, J., Greenberg, A.R., Gall,
K., 2008. Structure property relationships in photopolymerizable
polymer networks: effect of composition on the crosslinked structure
and resulting thermomechanical properties of a (meth)acrylate-based
system. J. Appl. Polym. Sci. 110, 15591572.
Qi, H.J., Nguyen, T.D., Castro, F., Yakacki, C., Shandas, R., 2008. Finite
deformation thermo-mechanical behavior of thermally-induced
shape memory polymers. J. Mech. Phys. Solids 56, 17301751.
Safranski, D.L., Gall, K., 2008. Effect of chemical structure and crosslinking
density on the thermo-mechanical properties and toughness of
(meth)acrylate shape memory polymer networks. Polymer 49,
44464455.
Scherer, G.W., 1984. Use of the AdamGibbs equation in the analysis of
structural relaxation. J. Am. Cerm. Soc. 67, 504511.
Tobushi, H., Hashimoto, H., Hayashi, S., Yamada, E., 1997.
Thermomechanical constitutive model of shape memory polymer of
polyurethane series. J. Intell. Mater. Syst. Struct. 8, 711718.
Tobushi, H., Okamura, K., Hayashi, S., Ito, N., 2001. Thermomechanical
constitutive model of shape memory polymer. Mech. Mater. 33, 545
554.
Tool, A.Q., 1946. Viscosity and extraordinary heat effects in glass. J. Am.
Ceram. Soc. 29, 240253.
Yakacki, C.M., Shandas, R., Lanning, C., Rech, B., Eckstein, A., Gall, K., 2007.
Unconstrained recovery characterization of shape-memory polymer
networks for cardiovascular applications. Biomaterials 28, 22552263.
Yakacki, C.M., Shandas, R., Safranski, D., Ortega, A.M., Sassaman, K., Gall,
K., 2008. Strong, tailored, biocompatible shape-memory polymer
networks. Adv. Functional Mater. 18, 24282435.

Вам также может понравиться