PLEASE SCROLL DOWN FOR ARTICLE

This article was downloaded by: [Technische Universiteit - Eindhoven]

On: 4 April 2009
Access details: Access Details: [subscription number 907217927]
Publisher Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK
Philosophical Magazine
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713695589
Numerical simulation of flat-tip micro-indentation of glassy polymers: Influence
of loading speed and thermodynamic state
L. C. A. van Breemen
a
; T. A. P. Engels
a
; C. G. N. Pelletier
a
; L. E. Govaert; J.M.J. den Toonder
a
a
Department of Mechanical Engineering, Eindhoven University of Technology, 5600MB Eindhoven, The
Netherlands
Online Publication Date: 01 March 2009
To cite this Article van Breemen, L. C. A., Engels, T. A. P., Pelletier, C. G. N., Govaert, L. E. and den Toonder, J.M.J.(2009)'Numerical
simulation of flat-tip micro-indentation of glassy polymers: Influence of loading speed and thermodynamic state',Philosophical
Magazine,89:8,677 696
To link to this Article: DOI: 10.1080/14786430802441188
URL: http://dx.doi.org/10.1080/14786430802441188
Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf
This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
Philosophical Magazine
Vol. 89, No. 8, 11 March 2009, 677696
Numerical simulation of flat-tip micro-indentation of glassy polymers:
Influence of loading speed and thermodynamic state
L.C.A. van Breemen, T.A.P. Engels, C.G.N. Pelletier,
L.E. Govaert
*
and J.M.J. den Toonder
Department of Mechanical Engineering, Eindhoven University of Technology,
P.O. Box 513, 5600MB Eindhoven, The Netherlands
(Received 8 May 2008; final version received 27 August 2008)
Flat-tip micro-indentation tests were performed on quenched and annealed
polymer glasses at various loading speeds. The results were analyzed using an
elasto-viscoplastic constitutive model that captures the intrinsic deformation
characteristics of a polymer glass: a strain-rate dependent yield stress, strain
softening and strain hardening. The advantage of this model is that changes in
yield stress due to physical aging are captured in a single parameter. The two
materials studied (polycarbonate (PC) and poly(methyl methacrylate) (PMMA))
were both selected for the specific rate-dependence of the yield stress that they
display at room temperature. Within the range of strain rates experimentally
covered, the yield stress of PC increases linearly with the logarithm of strain rate,
whereas, for PMMA, a characteristic change in slope can be observed at higher
strain rates. We demonstrate that, given the proper definition of the viscosity
function, the flat-tip indentation response at different indentation speeds can be
described accurately for both materials. Moreover, it is shown that the model
captures the mechanical response on the microscopic scale (indentation) as well as
on the macroscopic scale with the same parameter set. This offers promising
possibilities of extracting mechanical properties of polymer glasses directly from
indentation experiments.
Keywords: instrumented indentation; polymer glasses; yield stress; mechanical
properties
1. Introduction
Instrumented indentation is a versatile technique to probe local mechanical properties of
films and/or bulk materials [1,2]. In principle, a well-defined body is pressed into the
surface of a material while measuring both the applied load and penetration depth. The
obtained data can subsequently be analyzed to determine the mechanical properties of the
indented material. Regarding elastic modulus, in particular, quantitative analytical
analysis methods are available [3,4]. With the aid of the elastic-viscoelastic correspondence
principle, these methods are also applicable to quantitatively assess viscoelastic properties
[59]. With respect to large strain mechanical properties, the analysis of indentation data is
less straightforward. Even for the determination of yield strength, a direct analytical
method of analysis is not available and an estimate can only be obtained using empirical
*Corresponding author. Email: l.e.govaert@tue.nl
ISSN 14786435 print/ISSN 14786443 online
2009 Taylor & Francis
DOI: 10.1080/14786430802441188
http://www.informaworld.com
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
scaling laws. Although these have been proven to be quite useful, the scaling factor
between hardness and yield strength is not universal for all materials [1012].
The development of FEM-based analysis methods has opened up new possibilities.
Supported by the development of appropriate finite-strain constitutive relations, detailed
analysis of local deformation and stress fields is feasible. An excellent example is the work
of Larssons group on Vickers [13] and Berkovich [14] indentation of elasto-plastic
materials. In the case of polymeric materials, the analysis of such a contact problem is
complicated due to the complex large strain behavior, characterized by a pronounced
strain rate and pressure dependence of the yield stress and a post-yield response, which is
governed by a combination of strain softening and strain hardening. In the case of
amorphous polymer glasses, considerable effort has been directed towards the develop-
ment of 3D constitutive models capable of capturing the experimentally observed intrinsic
behavior, especially by, for example, the group of Mary Boyce at MIT [1517], the group
of Paul Buckley in Oxford [1820], and in our Eindhoven group [2123]. These
developments have enabled a quantitative analysis of localization and failure in polymer
glasses [22,2429] and revealed the crucial role of the intrinsic post-yield characteristics on
macroscopic strain localization.
Van Melick et al. [28] were the first to apply such a constitutive model to spherical-tip
indentation of polystyrene (PS) for the analysis of radial craze formation. They
demonstrated that the loadpenetration depth curves could be well reproduced for
different indentation speeds by numerical simulations, using the Eindhoven Glassy
Polymer Model (EGPM) [21]. In a subsequent study, Swaddiwudhipong et al. [30] showed
that the same model was unable to describe the Berkovich indentation response of another
glassy polymer; polycarbonate (PC). To reproduce the response at different indentation
speeds correctly, they required an additional strain gradient effect. It should be noted,
however, that they adopted the parameters for polycarbonate from Govaert et al. [21]
without verifying that this set was appropriate for the thermodynamic state of their own
polycarbonate samples. In a more recent study, Anand and Ames [31] presented an
extension of the BPA model [16], which proved successful in describing the conicaltip
indentation of PMMA, albeit at a single indentation speed.
In the present study, we demonstrate that the constitutive model (EGPM), developed
by our group [22], is also capable of quantitatively describing the indentation response of
PC and PMMA over a range of indentation speeds. A flat-ended cone was chosen as the
indenter body, since this specific tip geometry results in a loadpenetration depth curve, in
which elastic and plastic ranges are clearly distinguishable. At low indentation depths, the
response is governed by elastic deformation, whereas, at large depths, plastic deformation
sets in, leading to a marked change in slope and resulting in a characteristic knee-shaped
loaddisplacement curve [11,32,33]. Moreover, we will show that this is accomplished by
using a parameter set, which also quantitatively describes the materials mechanical
response in macroscopic testing.
2. Finite strain deformation of glassy polymers
2.1. Phenomenology
To study the intrinsic stressstrain response of polymers, an experimental set-up is
required in which the sample can deform homogeneously up to large plastic deformations.
678 L.C.A. van Breemen et al.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
Examples of such techniques are uniaxial compression tests [15,34] or video-controlled
tensile tests [35]. An illustrative example of the intrinsic stressstrain response of a polymer
glass is presented in Figure 1a. Typical features are strain softening, the decrease in true
stress that is observed after passing the yield point, and strain hardening at large
deformations. Strain hardening is generally interpreted as the result of a stress
contribution of the orienting molecular network [15,34,36,37]. Strain softening is closely
related to the fact that polymer glasses are not in a state of thermodynamic equilibrium.
Over time, the glass will strive towards equilibrium, a process usually referred to as
physical aging [38] and, as a result, its mechanical properties change. This is demonstrated
in Figure 1a, which compares the intrinsic response of two samples with different
thermodynamic state. It is clear that physical aging results in an increase of both modulus
and yield stress but, upon plastic deformation, the differences between the curves
disappear and eventually they fully coincide at a strain of approximately 0.3. Apparently,
all influence of thermal history is erased at that strain and both samples are transformed to
a similar, mechanically rejuvenated state. From Figure 1a, it is clear that an increase of
yield stress, due to a thermal treatment, will directly imply an increase in strain softening.
The influence of molecular weight on the intrinsic response is usually small and negligible
[20,22], which makes thermal history the key factor in influencing the intrinsic properties
of a specific polymer glass. The thermal history is also reflected in the long-term failure
behavior of polymer glasses. This was demonstrated for PC, where an annealing
treatment, leading to an increase in yield stress, improved the life-time under constant
stress by orders of magnitude [39].
The intrinsic stressstrain response of glassy polymers also displays a pronounced
dependence on the time-scale of the experiment. This is illustrated in Figure 1b, where the
strain-rate dependence of the compressive stressstrain response of poly(methylmethacry-
late) (PMMA) is shown [40]. It is clear that, with increasing strain rate, the overall stress
level in the yield and post-yield range increases. Also, the amount of strain softening and
strain hardening appears subject to change. At strain rates over 3 10
2
s
1
, the material
heats up due to viscous dissipation and, as a result, strain hardening disappears [41].
The strain-rate dependence of the yield stress for PMMA and PC is shown in
Figure 2a. For the latter, the yield stress increases linearly with the logarithm of strain rate,
which indicates that in this range of strain rates the deformation of PC is governed by a
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
0
20
40
60
80
100
120
comp. true strain []

c
o
m
p
.

t
r
u
e

s
t
r
e
s
s

[
M
P
a
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
0
20
40
60
80
100
120
140
comp. true strain []
viscous heating
c
o
m
p
.

t
r
u
e

s
t
r
e
s
s

[
M
P
a
]
(a)
(b)
Figure 1. Stressstrain response of PMMA in uniaxial compression: (a) influence of thermal history;
annealed sample (- -) and quenched sample (-); (b) influence of strain rate.
Philosophical Magazine 679
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
single molecular relaxation process [42,43], i.e. the amorphous o-transition (main-chain
segmental motion). Although this paper only focuses on the isothermal response, it is
relevant to note that the o-stress contribution displays an Arrhenius-type temperature
dependence, which leads to a horizontal shift of the yield stress characteristic along the
logarithmic strain rate axis. This type of behavior is generally referred to as thermo-
rheologically simple behavior.
In the case of PMMA, the strainrate dependence of the yield stress displays a clear
change in slope, which was shown to be related to onset of a stress contribution of a
second molecular process the [-transition [44,45] a secondary glass transition related to
side-chain mobility. A successful description of such a yield response is obtained using a
ReeEyring approximation, where, as schematically represented in Figure 2b, it is assumed
that each process can be described with an Eyring flow rule, whereas the stress
contributions of both molecular mechanisms are additive [46]. In the case of PMMA, it
should be noted that each process possesses its own characteristic activation energy,
implying that curves measured at different temperatures will no longer coincide with
horizontal shifting. A correct translation to another temperatures can only be achieved by
application of rate-temperature superposition on each contribution separately: this is
generally referred to as thermo-rheologically complex behavior.
2.2. Numerical model
In the present study, we employ a 3D elasto-viscoplastic constitutive model that accurately
captures the intrinsic deformation characteristics of polymer glasses [21,23,39,47]. The
basis of this constitutive model is the decomposition of the total stress into two separate
contributions, as first proposed by Haward and Thackray [36]:
p p
s
p
r
. 1
Here, p
r
denotes the strain hardening contribution that is attributed to molecular
orientation of the entangled network, described here with a simple Neo-Hookean elastic
expression [21,37]:
p
r
G
r
~
B
d
, 2
10
5
10
4
10
3
10
2
10
1
60
90
120
150
PMMA
PC
y
i
e
l
d

s
t
r
e
s
s

[
M
P
a
]
strain rate [s
1
]
log (strain rate)
y
i
e
l
d

s
t
r
e
s
s

+
(a)
(b)
Figure 2. (a) Yield stress of PMMA and PC in uniaxial compression as a function of strain rate;
(b) decomposition of the strain rate dependence of the yield stress into two separate molecular
contributions.
680 L.C.A. van Breemen et al.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
where G
r
is the strain hardening modulus and
~
B
d
the deviatoric part of the isochoric
left-CauchyGreen strain tensor.
The driving stress p
s
is attributed to intermolecular interactions on a segmental scale
[23,39] and is split into a hydrostatic (p
h
s
) and a deviatoric part (p
d
s
) [47,48]:
p
s
p
h
s
p
d
s
K J 1 I G
~
B
d
e
, 3
where K is the bulk modulus, J is the relative volume change, G is the shear modulus, and
~
B
d
e
is the deviatoric part of the elastic isochoric left CauchyGreen strain tensor. The
evolution of J and
~
B
e
are implicitly given by
_
J J tr D 4
~
B
o
e
D
d
D
d
p
_ _

~
B
e

~
B
e
D
d
D
d
p
_ _
, 5
where
~
B
o
e
is the Jaumann derivative of
~
B
e
, D
d
is the deviatoric part of the rate of
deformation tensor. The plastic part of the rate of deformation tensor D
p
is crucial for
adequate evolution of the driving stress, which is related to the deviatoric driving stress by
a non-Newtonian flow rule:
D
p

p
d
s
2j " t, p, S
a

. 6
A correct expression for the solid state viscosity j is essential in obtaining an accurate
description of the 3D stressstrain response. For glassy polymers, this choice mainly
depends on the number of molecular relaxation mechanisms that contribute to the stress.
In the simplest case, only a single molecular mechanism is active, the o- process, and an
expression for j can be obtained by taking the pressure-modified Eyring flow equation
[22,49,50] as a starting point. In the isothermal case, this leads to
_
" ,
p
" t, p, S _ ,
0, or
sinh " t,t
0, o

.,,.
I
expjp,t
0, o

.,,.
II
exp S
.,,.
III
, 7
where
_
" ,
p
represents the equivalent plastic shear rate (
_
" ,
p

2trD
p
D
p

_
). The part
marked (I), where " t is the equivalent shear stress ( " t

1,2trp
d
p
d

_
), represents the stress
dependence of the viscosity governed by the parameter t
0
. Part (II), where p is the
hydrostatic pressure ( p 1,3trp), yields the pressure dependency governed by
parameter j. Part (III) represents the dependence of viscosity on the thermodynamic
state of the material expressed by the state parameter S. Finally, _ ,
0, or
is a pre-exponential
factor representative for the rejuvenated, unaged state, where the index o refers to the
identity of the contributing molecular process.
Equation (7) leads to the following expression for the stress-dependence of the solid
state viscosity:
j
o
" t, p, S
" t
_
" ,
p
" t, p, S
j
0, or

" t,t
0, o
sinh " t,t
0, o
_ _
_ _
exp jp,t
0, o
_ _
exp S , 8
Philosophical Magazine 681
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
where j
0,or
(t
0, o
, _ ,
0, or
) is the zero-viscosity for the rejuvenated state. The part between
brackets can be regarded as a stress-dependent shift factor, which equals one at equivalent
shear stresses lower than t
0,o
and decreases exponentially with increasing equivalent shear
stress.
The dependence of the viscosity on physical aging and rejuvenation (strain softening) is
included by defining [22]:
S S
a
t R
,
" ,
p
_ _
. 9
Here, parameter S
a
can be regarded as a state parameter that uniquely determines the
current thermodynamic state of the material. Evolution of S
a
allows us to capture
the time-dependent change in mechanical properties as a result of physical aging [22]. In
the present investigation, however, we will only consider materials with different initial S
a
values (obtained by application of different thermal histories). The function R
,
describes
the strain softening process, i.e. the erasure of thermal history with plastic deformation.
It is expressed as
R
,
" ,
p
_ _

1 r
0
exp " ,
p
_ _ _ _
r
1
_ _
r
2
1,r
1
1 r
r
1
0
_ _
r
2
1,r
1
, 10
where r
0
, r
1
and r
2
are fit parameters, and " ,
p
denotes the equivalent plastic strain. The initial
value of R
,
equals unity and, with increasing equivalent plastic strain, R
,
decreases to zero.
The essence of the influences of physical aging and strain softening, modelled within
the state parameter S (Equation (9)), is illustrated in Figure 3a, which shows the strainrate
dependence of the yield stress resulting from Equations (8) and (9). In the reference state,
i.e. the fully rejuvenated state, parameter S
a
will be equal to zero. With physical aging, also
taking place during processing, the value of S
a
will increase, which leads to a shift in the
yield stress versus strain rate characteristic along the logarithmic strain rate axis. At a
constant strain rate, the result will be an increase in yield stress compared to that of the
rejuvenated state. Upon deformation, the increasing equivalent plastic strain " ,
p
triggers
strain softening (Equation (10)) and the yield stress shifts back to that of the rejuvenated
state. As a result, the yield stress drops with increasing strain and the intrinsic stressstrain
curve evolves back to that of the rejuvenated state (see Figure 3b).
log (strain rate)
y
i
e
l
d

s
t
r
e
s
s
aged
rejuvenated
S
a

p
true strain
t
r
u
e

s
t
r
e
s
s
aged
rejuvenated
S
a
(a) (b)
Figure 3. (a) Schematic representation of the influence of thermal history and strain softening on the
strain-rate dependence of the yield stress; (b) model prediction of the intrinsic stressstrain curve
indicating the influence of physical aging.
682 L.C.A. van Breemen et al.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
It is important to note that, in the case of PC, the parameters in the model (see Table 1
[39]) proved to be independent of the molecular weight distribution, and the key
parameter, required to adjust for differences in thermal history (illustrated in Figure 1), is
the initial value of the state parameter S: S
a
. This parameter can, in principle, be
determined directly from the yield stress value of a single tensile test [22,39] or, if the
thermal history during processing is known, be calculated directly [51,52].
Equations (7) and (8) both describe a linear increase in yield stress with the logarithm
of the strain rate: a situation that can be observed for all glassy polymers, albeit over a
limited range of temperature and strain rate. When studied over a sufficiently large range
of temperature and strain rate, most glassy polymers reveal a change in slope that is related
to the stress contribution of a secondary relaxation mechanism (generally referred to as the
[-process) [43,45,53]. A successful description of such a yield response can be obtained
using the ReeEyring model [46], where it is assumed that both molecular relaxation
mechanisms act in parallel. This implies that the total stress can be additively decomposed
into their individual contributions:
t
tot
t
o
t
[
. 11
This approach was successfully employed to describe the strainrate dependence of the
yield stress of various amorphous and semi-crystalline polymers [44,45,5356]. An
expression for the individual stress contributions can be obtained by rewriting the
pressure-modified Eyring flow expression (Equation (7)) in terms of equivalent shear
stress:
" t
tot
t
0, o
sinh
1
_
" ,
p
_ ,
0, or
exp
j
o
p
t
0, o
_ _
exp S
a

_ _
t
0, [
sinh
1
_
" ,
p
_ ,
0, [e
exp
j
o
p
t
0, o
_ _
_ _
. 12
While the expression for the [-contribution is equivalent to that of the o-process, there
are two amendments. To a first approximation, we assume that the pressure dependence of
the [-contribution is identical to that of the o-process. Moreover, since for the materials
under investigation (PC and PMMA), the [-contribution is already in its equilibrium
state [57] and does not change position during aging, there appears to be no use for a state
parameter S
a,[
. The state is fixed by the equilibrium value of
_
" ,
0, [
;
_
" ,
0, [e
.
Here, we follow an alternative route, where we capture the slope change in a single
viscosity expression. To accomplish this, we first approach the response of the material in
the o[-range, where both the o- and the [-process contribute to the stress, as a single
flow process:
_
" ,
pl
" t, p _ ,
0, o[
sinh
" t
t
0, o[
_ _
exp
j
o
p
t
0, o
_ _
, 13
Table 1. Material parameters used for the numerical simulation of tensile and compression tests
on PC.
K [MPa] G [MPa] G
r
[MPa] j
0,or
[MPa/s] t
0,o
[MPa] j S
a
[]
3750 321 26 2.1 10
11
0.7 0.08
Philosophical Magazine 683
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
where again it is assumed that the pressure dependence is identical to that of the o-process.
Changes in thermodynamic state are captured in the pre-exponential factor _ ,
0, o[
.
Equation (13) leads to an expression for the stress- and temperature-dependence of the
solid-state viscosity in the o [-range:
j
o[
" t, p
" t
_
" ,
pl
" t, p
j
0, o[

" t,t
0, o[
sinh " t,t
0, o[
_ _
_ _
exp j
o
p,t
0, o
_ _
, 14
where
j
0, o[
S
a

t
0, o[
_ ,
0, o[
S
a

; t
0, o[
t
0, o
t
0, [
, 15
and
_ ,
0, o[
S
a
exp
t
0, o
ln _ ,
0, or
_ _
S
a
_ _
t
0, [
ln _ ,
0, [e
_ _
t
0, o[
_ _
. 16
To obtain a single viscosity function that covers both the o-range as well as the
o[-range, we define the total viscosity as
j
tot
" t, p j
o
j
o[
j
0, or

" t,t
0, o
sinh " t,t
0, o
_ _
j
0, o[
S
a

j
0, or
exp S
a

" t,t
0, o[
sinh " t,t
0, o[
_ _
_ _
exp j
o
p,t
0, o
_ _
exp S
a
R
,
" ,
p
_ _ _ _
. 17
A schematic representation of this decomposition is given in Figure 4a. The expression
between brackets is again a stress-dependent shift factor. Its value equals one at shear
stresses well below t
0
, and decreases towards zero with increasing shear stress. An essential
consequence of Equation (17) is that the o- and the [-contribution will both display identical
softening. As illustrated in Figure 4b, upon plastic deformation, the yield stress charac-
teristic will shift horizontally along the logarithmic strain rate without any shape change.

+
equivalent stress
l
o
g

(
v
i
s
c
o
s
i
t
y
)

+
original rejuvenated

p
log (strain rate)
y
i
e
l
d

s
t
r
e
s
s
(a) (b)
Figure 4. (a) Schematic representation of the decomposition of the stress-dependence of the viscosity
of a thermo-rheologically complex material into two separate parts, the o and [ viscosity functions;
(b) influence of strain softening on the strain-rate dependence of the yield stress.
684 L.C.A. van Breemen et al.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
3. Experimental
3.1. Materials and sample preparation
3.1.1. Materials
The materials used in this study were polycarbonate (PC; Makrolon

, Bayer) obtained in
the form of extruded sheet of 3 mm thickness and poly(methylmetacrylate) (PMMA;
Perspex

, ICI) obtained in the form of extruded rods of 6 mm diameter.

3.1.2. Sample preparation: PC
For planar extension tests, rectangular samples with a dog-bone-shaped cross-section were
milled from the sheet [50]. The testing region has a thickness of 1.7 mm over a length of
10 mm and a width of 50 mm. Owing to the large width-to-length ratio, contraction of the
material is constrained, creating a plane strain condition [58]. For simple shear tests,
samples similar to those used in the planar tests were used, now with a width of 100 mm
instead of 50 mm. With a gauge length of 10 mm, this results in an aspect ratio of 10.
To avoid any influence of orientation effects due to the extrusion process, all samples were
taken from the same direction. For uniaxial tensile tests, samples according to ASTM
D638 were milled from the extruded sheet. To avoid the influence of a processing-induced
yield stress distribution over the thickness of the samples [52], the tensile bars were milled
to a thickness of 1.7 mm, i.e. identical to that of the test section of the samples for planar
extension and shear.
To enable a direct comparison between the indentation tests and conventional
macroscopic tests, indentation experiments were performed on a cross-section of a planar
extension sample. A small specimen was cut from the gauge-section of the sample and
subsequently the cross-sectional surface was cryogenically cut, using a microtome, to
obtain a smooth surface. Flat-tip indentation tests were performed in the middle of the
sample area. For other indentation tests, samples of 10 10 mm were cut from the
extruded PC sheet.
To change the thermodynamic state of the material, some of the samples were annealed
at 120

C for 48 h in an air-circulated oven and subsequently slowly air-cooled to room

temperature.
3.1.3. Sample preparation: PMMA
Cylindrical samples of 6 6 mm were cut from the extruded rod. The end-faces of the
cylinders were machined to optical quality employing a precision-turning process with a
diamond cutting tool. Indentation and uniaxial compression tests were performed on the
same samples. To vary their thermodynamic state, some samples were annealed at 95

C
for five days in an air-circulated oven and subsequently slowly air-cooled to room
temperature.
3.2. Techniques
Indentation experiments were performed using a nanoindenter XP (MTS Nano-
Instruments, Oak Ridge, TN, USA) under displacement control. The geometry of the
tip was a flat-ended cone, chosen for the fact that the elastic and the plastic regions in the
Philosophical Magazine 685
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
loaddisplacement curve can be clearly distinguished. Unfortunately, this flat-tip geometry
has the drawback in that the forcedisplacement response is very sensitive to tip-sample
misalignment. This problem was solved by sample re-alignment using a specially designed
alignment tool. Details on the alignment procedure can be found elsewhere [59]. The
geometry of the tip was characterized using SEM and AFM and proved to have a tip-
diameter of 10 mm (Figure 5a), a top angle of 72

(see Figure 5b), and an edge radius of

1 mm (Figure 5b).
Uniaxial compression tests were performed on a servo-hydraulic MTS Elastomer
Testing System 810. The specimens were cylindrical-shaped and compressed under true
strain control, at constant true strain rates of 10
4
to 10
2
s
1
between two parallel, flat steel
plates. Friction between samples and plates was reduced by an empirically optimized
method. Onto the sample ends, a thin film of PTFE tape (3M 5480, PTFE skived film
tape) was applied and the contact area between steel and tape was lubricated using a 1:1
mixture of liquid soap and water. During the test, no bulging of the sample was observed,
indicating that the friction was sufficiently reduced.
Uniaxial and planar tensile tests were performed on a Zwick Z010 tensile tester, at
constant linear strain rates of 10
5
10
1
s
1
. Shear tests were performed on a Zwick 1475 at
shear rates from 10
5
to 10
2
s
1
. Stressstrain curves were recorded and, where
appropriate, true stresses were calculated assuming incompressible deformation.
3.3. Numerical simulations
Axi-symmetric simulations were performed using MSC Marc/Mentat, a finite element
package. The constitutive model, as outlined in Section 2.2, was implemented in this
package by means of the user-subroutine HYPELA2. The axi-symmetric mesh consists of
3303 linear quad4 elements, using full integration. The size of the mesh, which is
0.05 0.05 mm, is chosen such that the edges do not influence the stress distribution. The
indenter, a flat-ended cone with geometrical specifications as determined by SEM and
AFM (see Figure 5), is modelled as an impenetrable body where no friction between
indenter and sample is taken into account.
0 2 4 6
2
4
R=1m
h
e
i
g
h
t

[

m
]
distance [m]
(b) (a)
Figure 5. Characterization of the tip: (a) top view SEM picture; (b) side view SEM picture with tip
profile obtained by AFM (lower picture).
686 L.C.A. van Breemen et al.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
The finite element mesh used for the simulation is shown in Figure 6. To exclude any
mesh-dependence, a stepwise element refinement was performed until the solution
converged to a steady, mesh-independent result. To prevent excessive computation times,
the mesh refinement was restricted to areas of interest (Figure 6).
4. Results and discussion
4.1. Thermorheologically simple behavior: PC
4.1.1. Material characterization
In the case of PC, only a single molecular process contributes to the yield stress, which
implies that the viscosity function defined in Equation (8) can be applied. Besides the
parameters in this expression (j
0,or
, t
0,o
, j, r
0
, r
1
, r
2
), the model requires the determination
of the strain-hardening modulus G
r
, the elastic shear modulus G and the bulk modulus K.
Most of these parameters can be determined from fitting the results of uniaxial
compression tests at different strain rates. A proven strategy is to start by fitting the
response of a rejuvenated material (S
a
0) on the strain-hardening regime of the
experimental curves, which yields the values for t
0,o
, j
0,or
and G
r
. Next, the softening can
be added and r
0
, r
1
, r
2
and S
a
can be determined.
To enable these model simulations, we first need the values of the elastic bulk modulus
K, the shear modulus G, and the pressure dependence j
o
. The value of K was calculated
from the values of the elastic modulus E and the Poisson ratio v. The latter were
determined in a uniaxial tensile test, yielding values of 2250 MPa for the elastic modulus E
and a value of 0.4 for he Poisson ratio v [47]. Using the interrelation
K
E
3 1 2v
, 18
Figure 6. Mesh used to simulate indentation tests.
Philosophical Magazine 687
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
a value of 3750 MPa was found for the bulk modulus K. In the present, single mode
approach, the elastic shear modulus G is first chosen slightly too low such that the
predicted yield strain approximately equals that experimentally observed; this facilitates
the characterization of the post-yield response. For polycarbonate, a value of G320 MPa
proved optimal.
An excellent method of obtaining pressure dependence j is to perform experiments
directly under superimposed hydrostatic pressure [6062]. Therefore, j
o
was determined
by numerically predicting the yield data obtained from compression tests at different true
strain rates and, finally, from the tensile tests under superimposed hydrostatic pressure, as
reported by Christiansen et al. [60]. Figures 7a and b show that an excellent description
was obtained using the material parameters given in Table 1 with an initial S
a
-value of 27.0
for the compression and 34.0 for the yield experiments, representing the difference in
thermal history between the two material sets used.
4.1.2. Macro-scale simulations
Since both rejuvenation and aging kinetics proved to be independent of the molecular
weight of the polymer, the only unknown parameter in the model is the initial value of the
state parameter, S
a
, which can be directly determined from the yield stress, as measured in
a simple tensile test at a single strain rate. This is demonstrated in Figure 8a, which shows
the results of uniaxial tensile tests at a strain rate of 10
3
s
1
, for the as-received polymer
sheet as well as for a sample annealed for 48 h at 120

C. As a result of this thermal

treatment, the yield stress of the material increased substantially [22,63,64]. In both cases,
the samples showed necking shortly after reaching the yield stress.
For both thermodynamic states, the S
a
-value is determined by matching the
experimental yield stress with the yield stress of a FEM-simulation using an axisymetric
model of a tensile bar with a small imperfection in the middle (for details, see [65]). The
simulations, shown in Figure 8a, yielded S
a
31.7 for the as-received material and S
a
39
for the annealed material. With these values we have now obtained a complete parameter
set for both materials.
0 0.1 0.2 0.3 0.4 0.5 0.6
0
20
40
60
true strain []
t
r
u
e

s
t
r
e
s
s

[
M
P
a
]
0 250 500 750 1000
0
50
100
150
200
250
hydrostatic pressure [MPa]
t
r
u
e

y
i
e
l
d

s
t
r
e
s
s

[
M
P
a
]
(a) (b)
Figure 7. (a) Compression tests, experiments (open symbols) compared with numerical simulations
(-) using material parameters as presented in Table 1 for three different true strain rates: 10
2
s
1
(^), 10
3
s
1
(), 10 nms
1
(); (b) yield stress versus superimposed hydrostatic pressure, model
prediction (-) compared to experimental results () by Christiansen et al. [63] at a strain rate of
1.710
4
s
1
.
688 L.C.A. van Breemen et al.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
To demonstrate the capability of the model to describe the mechanical response in
different loading conditions, we present, in Figure 8b, the strain rate dependence of the
yield stress of the as-received material in planar extension, shear and uniaxial extension.
The solid lines are predictions of the model, using the parameter set presented in Table 1
with the S
a
value of 31.7 that was determined in Figure 8a. The predictions were obtained
by performing FEM-simulations on the actual sample geometries (see also [39]). It is clear
that an accurate, quantitative description is obtained. In the next section, we will
investigate the predictive capabilities of the model in micro-indentation.
4.1.3. Micro-scale simulations
Before we apply the model to numerical simulations of micro-indentation tests, we first
have to address an imperfection within the model. In the fitting procedure described
above, we employed a shear modulus of 320 MPa, which implies an elastic modulus of
900 MPa, considerably smaller than the 2250 MPa observed in uniaxial extension. In the
case of indentation, where the elastic deformation will also significantly contribute at
larger depths, the low value of the modulus will lead to a drastic underestimation of the
materials resistance to deformation. The most elegant method to improve the description
of the pre-yield behavior is by means of a multimode extension of the present approach. In
a previous study, we demonstrated that the pre-yield response can be accurately captured
by a parallel arrangement of 18 modes [23]. Unfortunately, this solution would
tremendously increase the required computational time. Instead, we chose to simply
increase the shear modulus G until the initial slope of the compressive stressstrain curves
was properly described. This was achieved with G784 MPa. To ensure that the yield and
post-yield response remain identical, this change of G subsequently requires an adaption of
the rejuvenated zero viscosity j
0,or
. The resulting new data set is provided in Table 2.
The consequences of these changes are demonstrated in Figure 9. Figure 9a shows the
influence of an increase in shear modulus G on the compressive stressstrain curve. The
initial modulus increases and the strain-at-yield decreases, leading to a shift in the yield
0 0.02 0.04 0.06 0.08 0.1
0
10
20
30
40
50
60
70
80
strain []
s
t
r
e
s
s

[
M
P
a
]
10
6
10
5
10
4
10
3
10
2
10
1
10
0
10
20
30
40
50
60
70
80
90
strain rate [s
1
]
y
i
e
l
d

s
t
r
e
s
s

[
M
P
a
]
planar extension
uniaxial extension
simple shear
(a)
(b)
Figure 8. Experiments (open symbols) compared with the numerical simulation (-) on PC: (a) tensile
tests at a strain rate of 10
3
s
1
for two different thermal histories with S
a
31.7 for the as-received
(), and S
a
39 () for the annealed material; (b) predicted yield stress at different strain rates,
S
a
31.7, for planar extension (^), uniaxial extension () and shear ().
Philosophical Magazine 689
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
and post-yield response to lower strain values. As mentioned before, the value of the yield
stress as well as the shape of the post-yield behavior remains unaffected.
The influence on the numerical simulation of a flat-tip indentation test is shown in
Figure 9b where experimental data of the as-received material is compared to numerical
simulations using the S
a
-value of 31.7 determined previously (Figure 8a). From Figure 9b,
it becomes clear that a correct modulus leads to an accurate, quantitative prediction of the
loading path of the indentation test, whereas the low modulus value leads to a large
underestimation of the indentation resistance. Figure 9b also shows that the current
approach is reasonably successful in capturing the unloading path. We expect that this
prediction could even be improved if a multimode approximation would be employed.
In Figure 10d, a characteristic loading curve of an indentation measurement is shown.
In this figure, three points are marked: A, B and C. In Figures 10ac, graphical
representations of the development of the plastic deformation under the tip are given for
these three points. From the numerical evaluations, it is derived that the plastic
deformation starts at the edge of the indenter, as can be seen in Figure 10a, and grows in
the form of a hemisphere towards the symmetry axis. This is a result of the fact that stress
localizes at the edge of the indenter. Around point B (Figure 10b), the plastic deformation
zone concludes the formation of the hemisphere. From point C onwards (Figure 10c), this
hemisphere then starts to expand in thickness. These results correspond well with
experimental observations made by others [11].
The strength of our approach is further demonstrated in Figure 11, which shows the
influence of thermodynamic state (Figure 11a) and strain rate (Figure 11b) on the
0 0.1 0.2 0.3 0.4 0.5
0
10
20
30
40
50
60
70
80
comp. true strain []
c
o
m
p
.

t
r
u
e

s
t
r
e
s
s

[
M
P
a
]
0 1 2 3 4 5
0
10
20
30
40
50
displacement [m]
l
o
a
d

[
m
N
]
(a)
(b)
Figure 9. Effect of the elastic modulus; (a) simulated compression tests for two different values of
the elastic modulus: G = 321 MPa (- -) and G = 784 MPa (-) compared with the experiments ();
(b) the two different moduli, but now for the indentation tests.
Table 2. Material parameters with increased G, used for the numerical simulation of indentation
tests on PC.
K [MPa] G [MPa] G
r
[MPa] j
0,or
[MPa/s] t
0,o
[MPa] j S
a
[]
3750 784 26 2.8 10
12
0.7 0.08
690 L.C.A. van Breemen et al.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
response in a flat-tip indentation test. In Figure 11a, the results of indentation
experiments (at a rate of 50 nm/s) on both the annealed and the as-received PC sheet are
presented. It is clear that the numerical simulations, employing the S
a
-values determined
in Figure 8a, quantitatively predict the loaddisplacement curves. The agreement is so
precise that the flat-tip indentation test can be used to determine the S
a
-value of a PC
sample with an unknown prior thermal history. To achieve this, the loaddisplacement
curve has to be fitted with a numerical simulation. For PC, the S
a
-value can be
determined in this way, in practice, with an accuracy of 1 (corresponding to a yield
stress inaccuracy of 1 MPa).
The feasibility of characterizing the S
a
-value from an indentation test is demonstrated
in Figure 11b, which presents micro-indentation results at different indentation rates (5, 50
and 200 nm/s). The material indented is a PC, which was annealed for a few hours at
120

C. An S
a
-value of 34 was determined by fitting the numerical prediction to the load
displacement curve of 5 nm/s. The loading curves of the other indentation rates were
subsequently simulated with this value. The result is clear: the influence of strain rate is
also quantitatively predicted by our model.
(a) (b)
(d)
(c)
0 1 2 3 4 5
0
10
20
30
40
50
displacement [m]
l
o
a
d

[
m
N
]
A
B
C
Figure 10. Simulation of the development of plastic deformation at different indentation depths for
PC: (a) at 460 nm; (b) 965 nm; (c) 3 mm; (d) here, the points a, b and c indicate the load-displacement
response for the different stages of plastic deformation.
Philosophical Magazine 691
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
4.2. Thermorheologically complex behavior: PMMA
4.2.1. Material characterization
In the case of PMMA, there are two molecular processes that contribute to the yield stress,
which implies that the viscosity function defined in Equation (17) must be applied.
This means that, besides the parameters already discussed in the previous section (j
0,or
,
t
0,o
, j
o
, r
0
, r
1
, r
2
, G
r
, G, and K), we also have to determine the values of j
0,o[
and t
0,o[
.
For the characterization, we make use of the fact that the [-contribution is only present at
high strain rates and, therefore, the stressstrain response in the low strain rate range will
be determined by the o-process only. As a consequence, we can use the same
characterization strategy at low strain rates as employed in the previous section for PC.
For the elastic properties of PMMA, we used a bulk modulus K of 3 GPa [66], the
appropriate value of the shear modulus G was determined to be 760 MPa from the initial
slope of the compressive stressstrain curves. To facilitate the fitting procedure, we initially
adopted a lower value for G (630 MPa). After characterization of the o-parameters, the
values for j
0,o[
and t
0,o[
are subsequently determined by fitting numerical predictions
to the compressive stressstrain curves obtained at higher strain rates (o[ region).
To obtain the true value of the pressure-dependence parameter j
o
, we used a method
inspired by the work of Bardia and Narasimhan [67], who employed a spherical
indentation test to characterize the pressure sensitivity index of the DruckerPrager
constitutive model. Here, we follow a similar route. Since the compression tests and the
indentation tests were performed on the same sample, the S
a
value is identical in both
cases. In the o-range, the only unknown parameter is, therefore, the pressure dependence
j
o
. Similar to the approach for polycarbonate, we again generated different parameter sets
by fitting the compression data for different values of j
o
. Here, it should be noted that
each set described the compressive stressstrain curves equally well (Figure 12a). With
these data sets, we subsequently predicted the loaddeformation curve for an indentation
rate of 5 nm/s (see Figure 12b) and found that a value of j
o
0.13 is in good agreement
with the experiment. The complete data sets, used for the predictions in Figure 12, are
0 0.5 1 1.5 2 2.5
0
5
10
15
20
25
30
l
o
a
d

[
m
N
]
displacement [m]
0 0.5 1 1.5 2 2.5
0
5
10
15
20
25
30
l
o
a
d

[
m
N
]
displacement [m]
(a)
(b)
Figure 11. Flat-tip indentation experiments (open symbols) compared with the numerical prediction
(-): (a) as-received (S
a
31.7) () and for annealed (S
a
39) () PC at an indentation speed of
50 nms
1
and (b) for speeds of 5 nms
1
(), 50 nms
1
() and 200 nms
1
() on the annealed PC
(S
a
34).
692 L.C.A. van Breemen et al.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
tabulated in Table 3 (compression) and Table 4 (indentation). The corresponding S
a
value
was determined to be 7.4.
Finally, indentation tests were performed at indentation rates of 5, 10, 20 and 40 nm/s.
The results are compared to numerical predictions (employing the parameters listed in
Table 4) in Figure 13a. It is clear that our numerical predictions are in excellent agreement
with the pronounced rate-dependence observed in the experimental forcedisplacement
curves. To demonstrate the presence of a [-contribution in the indentation response, we
performed simulations of indentation tests at rates of 0.1 and 40 nm/s, with, as well as
without, a [-contribution (this implies j
0,o[
0). The results are presented in Figure 13b
and show that, at low indentation rate (0.1 nm/s), the [-contribution is negligible, whereas
at higher rates a significant contribution is visible.
0 0.1 0.2 0.3 0.4 0.5 0.6
0
20
40
60
80
100
120
140
160
true strain []
t
r
u
e

s
t
r
e
s
s

[
M
P
a
]
0 1 2 3 4
0
10
20
30
40
50
l
o
a
d

[
m
N
]
displacement [m]
(a) (b)
Figure 12. Experiments (open symbols) compared to numerical simulation (-) for: (a) compression
tests () performed on PMMA at a strain rate of 10
4
s
1
, 310
4
s
1
, 10
3
s
1
, 10
2
s
1
,
310
2
s
1
and; (b) flat-tip indentation performed at 5 nms
1
().
Table 4. Parameter set with corrected G, used for the numerical simulation of indentation tests on
PMMA.
K
[MPa]
G
[MPa]
G
r
[MPa]
t
0, o
[MPa]
t
0, o[
[MPa]
j
0,or
[MPa/s]
j
0,o[
[MPa/s] j S
a
r
o
R
1
r
2
3000 728 26 2.71 7.05 9.27 10
6
2.22 10
5
0.13 7.8 0.96 30 3.5
Table 3. Parameters used for the numerical simulation of compression tests on PMMA.
K
[MPa]
G
[MPa]
G
r
[MPa]
t
0, o
[MPa]
t
0, o[
[MPa]
j
0,or
[MPa/s]
j
0,o[
[MPa/s] j S
a
r
o
R
1
r
2
3000 628 26 2.71 7.05 8.13 10
6
2.12 10
5
0.13 7.8 0.96 30 3.5
Philosophical Magazine 693
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
5. Conclusion
In the plastic regime, glassy polymers possess a rather complex intrinsic behavior, with a
pronounced pressure- and rate-dependence of the yield stress as well as a post-yield region
displaying both strain softening and strain hardening. We employed a state-of-the-art
constitutive model, previously developed in our group, which describes this intrinsic
behavior, to numerically predict the indentation response. In the model, a single
parameter, the state parameter S
a
, is used to uniquely determine the initial yield stress of
the material and capture all variations in its thermal history. We demonstrated that this
model can capture the rate- and history-dependence of PC and PMMA on both the
macroscopic and microscopic scale. The obtained accuracy of the description also creates
the possibility of extracting the state parameter S
a
directly from micro-indentation
experiments. This offers interesting possibilities with respect to quality control of load-
bearing polymer products. Moreover, it was found that the pressure dependence of the
yield stress can also be obtained by combining indentation and compression tests on the
same samples.
References
[1] M.R. VanLandingham, J.S. Villarrubia, W.F. Guthrie et al., Macromol. Symp. 167 (2001) p.15.
[2] W.C. Oliver and G.M. Pharr, J. Mater. Res. 19 (2004) p.3.
[3] H. Hertz, J. Reine, Angew. Math. 92 (1881) p.156.
[4] W.C. Oliver and G.M. Pharr, J. Mater. Res. 7 (1992) p.1564.
[5] L. Cheng, X. Xia, L.E. Scriven et al., Mech. Mater. 37 (2005) p.213.
[6] L. Cheng, X. Xia, W. Yu et al., J. Polym. Sci. B 38 (2000) p.10.
[7] J.M.J. den Toonder, Y. Ramone, A.R. van Dijken, et al., in Proceedings of the 3rd International
Conference on Benefiting from Thermal and Mechanical Simulation in (Micro)-Electronics
(2002), Paris, p.270.
[8] J.M.J. den Toonder, Y. Ramone, A.R. van Dijken et al., J. Electr. Packag. 127 (2005) p.267.
0 1 2 3 4
0
10
20
30
40
50
displacement [m]
l
o
a
d

[
m
N
]
0 1 2 3 4
0
10
20
30
40
50
0.1 nm/s
40 nm/s
displacement [m]
l
o
a
d

[
m
N
]
(a) (b)
Figure 13. (a) Flat-tip indentation performed on PMMA at a speed of 5 nms
1
(5), 10 nms
1
(),
20 nms
1
(), 40 nms
1
(i) compared with the numerical simulations (-); and (b) numerical
simulation performed using the model, which takes into account the o contribution (- -) only and the
o[ contribution (-) at two different speeds 0.1 nms
1
and 40 nms
1
.
694 L.C.A. van Breemen et al.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
[9] P.L. Larsson and S. Carlsson, Polym. Test. 17 (1998) p.49.
[10] K.L. Johnson, J. Mech. Phys. Solid. 18 (1970) p.115.
[11] Y. Lu and D.M. Shinozaki, J. Mater. Sci. 37 (2002) p.1283.
[12] D. Tabor, Hardness of metals, Clarendon Press, Oxford, 1951.
[13] A.E. Giannakopoulos, P.L. Larsson and R. Vestergaard, Int. J. Solids Struct. 31 (1994) p.2679.
[14] P.L. Larsson, A.E. Giannakopoulos, E. So derlund et al., Int. J. Solids Struct. 33 (1996) p.221.
[15] E.M. Arruda and M.C. Boyce, Int. J. Plast. 9 (1993) p.697.
[16] M.C. Boyce, D.M. Parks and A.S. Argon, Mech. Mater. 7 (1988) p.15.
[17] O.A. Hasan, M.C. Boyce, X.S. Li et al., J. Polym. Sci. B 31 (1993) p.185.
[18] C.P. Buckley, P.J. Dooling, J. Harding et al., J. Mech. Phys. Solid 52 (2004) p.2355.
[19] C.P. Buckley and D.C. Jones, Polymer 36 (1995) p.3301.
[20] J.J. Wu and C.P. Buckley, J. Polym. Sci. B 42 (2004) p.2027.
[21] L.E. Govaert, P.H.M. Timmermans and W.A.M. Brekelmans, J. Eng. Mater. Tech. 122 (2000)
p.177.
[22] E.T.J. Klompen, T.A.P. Engels, L.E. Govaert et al., Macromolecules 38 (2005) p.6997.
[23] T.A. Tervoort, E.T.J. Klompen and L.E. Govaert, J. Rheol. 40 (1996) p.779.
[24] M.C. Boyce and E.M. Arruda, Polym. Eng. Sci. 30 (1990) p.1288.
[25] M.C. Boyce, E.M. Arruda and R. Jayachandran, Polym. Eng. Sci. 34 (1994) pp. 716725.
[26] P.D. Wu and E. van der Giessen, Int. J. Plast. 11 (1995) p.211.
[27] H.E.H. Meijer and L.E. Govaert, Macromol. Chem. Phys. 204 (2003) p.274.
[28] H.G.H. Van Melick, O.F.J.T. Bressers, J.M.J. den Toonder et al., Polymer 44 (2003) p.2481.
[29] H.G.H. van Melick, L.E. Govaert and H.E.H. Meijer, Polymer 44 (2003) pp. 457465.
[30] S. Swaddiwudhipong, L.H. Poh, J. Hua et al., Mater. Sci. Eng. A 404 (2005) p.179.
[31] L. Anand and N.M. Ames, Int. J. Plast. 22 (2006) p.1123.
[32] Y. Lu and D.M. Shinozaki, Mater. Sci. Eng. A 249 (1998) p.134.
[33] S.C. Wright, Y. Huang and N.A. Fleck, Mech. Mater. 13 (1992) p.277.
[34] H.G.H. van Melick, L.E. Govaert and H.E.H. Meijer, Polymer 44 (2003) p.2493.
[35] C. GSell, J.M. Hiver, A. Dahoun et al., J. Mater. Sci. 27 (1992) p.5031.
[36] R.N. Haward and G. Thackray, Proc. R. Soc. Lond. A. 302 (1967) p.453.
[37] T.A. Tervoort and L.E. Govaert, J. Rheol. 44 (2000) p.1263.
[38] J.M. Hutchinson, Prog. Polym. Sci. 20 (1995) p.703.
[39] E.T.J. Klompen, T.A.P. Engels, L.C.A. van Breemen et al., Macromolecules 38 (2005) p.7009.
[40] E.T.J. Klompen, Mechanical Properties of Solid Polymers, Technische Universiteit, Eindhoven,
2005.
[41] E.M. Arruda, M.C. Boyce and R. Jayachandran, Mech. Mater. 19 (1995) p.193.
[42] C. Bauwens-Crowet, J.C. Bauwens and G. Homes, J. Polym. Sci. 7 (1969) p.735.
[43] E.T.J. Klompen and L.E. Govaert, Mech. Time-Depend. Mater. 3 (1999) p.49.
[44] J.A. Roetling, Polymer 6 (1965) p.615.
[45] J.A. Roetling, Polymer 6 (1966) p.311.
[46] T. Ree and H. Eyring, J. Appl. Phys. 26 (1955) p.793.
[47] T.A. Tervoort, R.J.M. Smit, W.A.M. Brekelmans et al., Mech. Time-Depend. Mater. 1 (1998)
p.269.
[48] F.P.T. Baaijens, Rheol. Acta 30 (1991) p.284.
[49] R.A. Duckett, B.C. Goswami, L.S.A. Smith et al., Br. Polym. J. 10 (1978) p.11.
[50] L.E. Govaert, H.J. Schellens, H.J.M. Thomassen et al., Composites A 32 (2001) p.1697.
[51] T.A.P. Engels, L.E. Govaert, G.W.M. Peters et al., J. Polym. Sci. B 44 (2006) p.1212.
[52] L.E. Govaert, T.A.P. Engels, E.T.J. Klompen et al., Int. Polym. Process. XX 2 (2005) p.170.
[53] C. Bauwens-Crowet, J.C. Bauwens and G. Homes, J. Mater. Sci. 7 (1972) p.176.
[54] C. Bauwens-Crowet, J. Mater. Sci. 8 (1973) p.968.
[55] L.E. Govaert, P.J. de Vries, P.J. Fennis et al., Polymer 41 (2000) p.1959.
[56] J.A. Roetling, Polymer 7 (1966) p.303.
Philosophical Magazine 695
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9
[57] C. Bauwens-Crowet and J.C. Bauwens, Polymer 24 (1983) p.921.
[58] W. Whitney and R.D. Andrews, J. Polym. Sci. Polym. Symp. 16 (1967) p.2981.
[59] C.G.N. Pelletier, E.C.A. Dekkers, L.E. Govaert et al., Polym. Test. 26 (2007) p.949.
[60] A.W. Christiansen, E. Baer and S.V. Radcliffe, Phil. Mag. 24 (1971) p.451.
[61] J.A. Sauer, D.R. Mears and K.D. Pae, Eur. Polym. J. 6 (1970) p.1015.
[62] W.A. Spitzig and O. Richmond, Polym. Eng. Sci. 19 (1979) p.1129.
[63] C. Bauwens-Crowet and J.C. Bauwens, Polymer 23 (1982) p.1599.
[64] J.H. Golden, B.L. Hammant and E.A. Hazell, J. Appl. Polym. Sci. 11 (1967) p.1571.
[65] H.G.H. Van Melick, L.E. Govaert and H.E.H. Meijer, Polymer 44 (2003) p.3579.
[66] I.W. Gilmour, A. Trainor and R.N. Haward, J. Appl. Polym. Sci. 23 (1979) p.3129.
[67] P. Bardia and R. Narasimhan, Strain 42 (2006) p.187.
696 L.C.A. van Breemen et al.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
T
e
c
h
n
i
s
c
h
e

U
n
i
v
e
r
s
i
t
e
i
t

-

E
i
n
d
h
o
v
e
n
]

A
t
:

0
3
:
0
8

4

A
p
r
i
l

2
0
0
9