Available online at www.sciencedirect.

com

Acta Materialia 56 (2008) 4390–4401

www.elsevier.com/locate/actamat

In situ indentation testing of elastomers

Julia K. Deuschle a,*, Gerhard Buerki b, H. Matthias Deuschle c, Susan Enders a,1,
Johann Michler b, Eduard Arzt a,2
a
Max-Planck-Institute for Metals Research, Stuttgart, Germany
b
EMPA – Materials Science and Technology, Thun, Switzerland
c
Institute of Statics and Dynamics of Aerospace Structures, University of Stuttgart, Germany

Received 31 January 2008; received in revised form 2 May 2008; accepted 5 May 2008
Available online 7 June 2008

Abstract

The development of the contact is crucial in indentation testing, yet only limited knowledge exists on the true contact size for com-
pliant materials. In this investigation the contact evolution and the deformation behavior of polydimethylsiloxane was studied during
indentation in situ inside a scanning electron microscope and by observation in a light microscope. Since detailed information on the
true contact area and the amount of sink-in can be acquired from finite element analysis, simulations on the indentation process have
been performed in order to complement the in situ testing. Comparison of results revealed that the contact areas calculated according to
the standard Oliver–Pharr procedure deviated from the real contact size by approximately 10% for the elastomeric PDMS material.
Ó 2008 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Micro-/nanoindentation; Polymers; Scanning electron microscopy (SEM); Finite element modeling (FEM)

1. Introduction tries, which are often not representative of true experimen-

Mechanical testing by means of nanoindentation has
become a widely accepted tool and even international stan-
dards for instrumented indentation testing have recently
been elaborated [1]. Nanoindentation testing of polymeric
and biological materials has also attracted increasing inter-
est, since this technique allows probing of mechanical prop-
erties of small volume samples, which are usually not
accessible with other methods. However, nanoindentation
is not readily applicable to softer materials like polymers
and biological samples, for which some open questions
remain.
Indentation theory is based on contact mechanics, where
analytical solutions can be derived only for simple geome-
*
Corresponding author. Tel.: +49 711 689 3418; fax: +49 711 689 3412.
E-mail address: deuschle@mf.mpg.de (J.K. Deuschle).
1
Current address: Engineering Mechanics, University of Nebraska,
Lincoln, NE, USA.
2
Current address: INM-Leibniz Institute for New Materials, Saarbrücken,
Germany.
tal conditions. The early work of Hertz [2] and Boussinesq
[3] was extended and modified later for different geometric
conditions. Sneddon [4] considered a punch of arbitrary
profile in contact with an elastic half space and derived a
relation between force and penetration depth, as well as
for the pressure distribution beneath the indenter and the
shape of the deformed surface. These equations of Sneddon
now form the basis of the well-known method of Oliver
and Pharr [5,6], which is commonly used for evaluating
instrumented indentation data. Although this method is
generally accepted to provide accurate and reliable values
of mechanical properties, it has to be kept in mind that it
was derived for purely elastic materials. For materials with
more complex behavior, like strain-hardening or viscoelas-
ticity, additional information is necessary to fully under-
stand the indentation process.
Two approaches are possible to achieve a more compre-
hensive understanding of the indentation process: numeri-
cal simulations [7–14] and in situ techniques [15–25]. The
aim of simulations is to access quantities that are not mea-
surable by experimental techniques, such as deformations

1359-6454/$34.00 Ó 2008 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.actamat.2008.05.003
 J.K. Deuschle et al. / Acta Materialia 56 (2008) 4390–4401
under the indenter. For instance, Larsson et al. [7] pre-
sented a detailed finite element (FE) analysis of Berkovich
indentation for elastic and elastic–plastic materials. Buca-
ille and co-workers [8,9] used a broader spectrum of mate-
rial behavior for FE simulations of indentation and scratch
testing. These studies addressed the determination of
stress/strain fields under the indenter and how the contact
evolution is influenced by the material’s behavior. Determi-
nation of the contact area from simulations [10,11,14] is
important, since the contact area is crucial for the calcula-
tion of mechanical properties from indentation data. Fur-
ther, simulation offers the possibility of varying the
indentation conditions in a systematic and controlled
way, such that the influences of, for example, surface
roughness [12], indenter geometry [13] or sink-in and pile-
up effects [14] on the contact area can be studied.
Experimentally, the contact area can be determined by
optical observation of the contact. This technique is widely
used in tribological studies [15,16], but also finds applica-
tion in indentation testing. In most cases a transparent
sample or substrate is used [16,17]. In contrast, Miyajima
presented some work that used a transparent indenter
[18]. This allows testing of non-transparent media as well,
which makes it a useful tool for investigating indentation
contact formation and retraction.
Another technique is in situ testing in a scanning elec-
tron microscope [20–22]. This technique provides princi-
pally qualitative information on the contact area, but
enables mechanistic insight. Moser et al. [20,21] demon-
strated that pop-ins in the force–indentation depth data
are associated with the activation of shear bands in metallic
glasses. Further interesting results on the deformation
behavior and cracking modes during scratching were
obtained through in situ scanning electron microscopy
(SEM) scratch testing [22]. Also, several in situ transmis-
sion electron microscopy studies have been performed in
order to get a deeper insight into the deformation mecha-
nisms at smaller scales, e.g. dislocation behavior in mar-
tensitic steels [23] and silicon [24], and cracking in
polyethylene [25].
A drawback of electro-optical methods is the sensitivity
of polymeric and biological materials to influences of the
electron beam [26–37]. In polymeric materials, irradiation
changes their chemical structure. Chain scission occurs
and free radicals form. This can lead either to polymer
decomposition or additional crosslinking, depending on
the intensity of the beam [28,32]. Several studies have inves-
tigated the influences of radiation on polydimethylsiloxane
(PDMS) [32–36], which is of particular interest for the pres-
ent paper. In the work of Russell and co-workers [32], indi-
cations for chain scission and even fragment removal
during irradiation with electron beams (15–30 kV accelera-
tion voltage) were found. Under the influence of higher
energy electrons (on the MeV scale) or c-radiation, PDMS
tends to be further crosslinked [33–36].
In the present paper, the indentation process of very soft
elastomeric materials is investigated with the help of in situ
4391
indentation testing (scanning electron and optical micro-
scopes) and FE simulations. The aim was to provide insight
into contact evolution and deformation behavior during
indentation of soft elastomeric materials. The results will
be discussed with respect to the applicability of common
contact descriptions to polymers.
2. Experimental
2.1. Sample material
PDMS of the Sylgard 184 type (Dow Corning Corp.,
Midland, MI) was used for this study. For specimen prep-
aration the liquid mixtures of 1:10 and 1:30 weight ratios of
crosslinking agent to pre-polymer were poured onto silicon
wafers and cured in a vacuum furnace at 150 °C. The cur-
ing temperature was maintained until the crosslinking reac-
tion was completed [38] to ensure constant mechanical
properties.
For SEM investigations the specimens were sputter-
coated with a thin (10 nm thick) AuPd layer in order to
avoid charging effects. For the observation of the true poly-
mer surface small areas were shielded during sputter-coat-
ing. Subsequent indentations were performed on these
uncoated areas.
2.2. Instrumentation
For standard indentation, a Nano Indenter SA2 with a
dynamic contact module (DCM) measurement head from
MTS Nano Instruments (Oak Ridge, TN, USA) was used.
Testing was carried out in the continuous stiffness measure-
ment (CSM) mode with a frequency of 75 Hz using dia-
mond Berkovich and cube corner tips. Contact area and
contact depth were determined according to the Oliver–
Pharr calibration procedure [5,6]. For a more detailed
investigation of the sample surfaces at the indent locations,
atomic force microscopy (AFM) imaging was conducted
on a Dual Scope DS 40-45 system from Danish Micro
Engineering (DME, Hervev, Denmark). The calibration
of the atomic force microscope was performed on silicon
calibration gratings (TGX01 and TGZ03, MicroMasch,
Wilsonville, OR).
In situ indentation tests were performed inside a Zeiss
DSM 962 scanning electron microscope operated at accel-
eration voltages between 5 and 15 kV. The chamber was
equipped with a home-built indenter which allows observa-
tion of the indentation process at various angles, depending
on the indenter tip in use [22]. Again diamond cube corner
and Berkovich tips were chosen. Unfortunately, the force
resolution of the system was not sufficient for these soft
materials, therefore no quantitative force–depth curves
could be recorded during indentation.
A Nano Bionix system (MTS Nano Instruments, Oak
Ridge, TN) was used for the optical in situ observations.
For this purpose a specially designed sample holder made
of glass was used, which allows the sample to be mounted
4392 J.K. Deuschle et al. / Acta Materialia 56 (2008) 4390–4401
between the indenter tip and the video microscope camera.
The video imaging could then be performed from behind
through the transparent glass slide/specimen couple per-
pendicular to the test surface. Thus, it was possible to mon-
itor the true projected contact area continuously during an
indentation. The images from the video microscope were
analyzed with the software Sigma Scan Pro 5.5.
2.3. Simulations
Comparative three-dimensional (3-D) finite element
analysis (FEA) was performed with the commercial code
ABAQUS Version 6.61 [39]. Making use of the threefold
symmetry of the pyramidal indenter tips (cube corner and
Berkovich), the FE model was reduced to one-third of
the actual problem for convenience. Thus the modeled
geometry incorporates the contact between one face of
the tip and the specimen. Expecting more than 5% strain,
a hyperelastic constitutive model was chosen to represent
the mechanical behavior of PDMS. ABAQUS provides a
tool to select the material model which best fits the exper-
imental test data [40]. Uniaxial tensile and compression
tests were conducted on the same PDMS specimens that
have also been used for the indentation experiments. Based
on the obtained stress–strain relations, the material evalu-
ation process determined the parameters of numerous
hyperelastic material models. The neo-Hooke strain energy
potential [41] was identified as the most adequate descrip-
tion. Its two coefficients C10 and D1 are defined as:
C 10 ¼ l0 =2 ð1Þ
D1 ¼ 2=K 0 ð2Þ
where l0 is the initial shear modulus and K0 is the initial
compression modulus. For the two differently crosslinked
PDMS, 1:10 and 1:30, the coefficients were determined to
be D1 = 0.662 and 0.0922 and C10 = 0.255 and 1.123,
respectively. Assuming isotropy for the 3-D randomly
crosslinked PDMS network, these values correspond to
Young’s moduli of approximately 4 MPa for PDMS 1:10
and 0.6 MPa for PDMS 1:30, if Poisson’s ratios are taken
to be 0.42 and 0.45 according to Ref. [38].
To optimize discretization, extensive studies concerning
the mesh density, the element type and the contact formu-
lation were carried out. While the geometric order of the
elements showed no significant influence on the results
(thus eight-node linear elements were chosen for efficiency),
mesh density did matter. The result of the mesh conver-
gence study is a strongly biased mesh (denser at the inden-
ter tip) with 23275 elements overall, while the indenter face
is represented by an analytically rigid plane. Tangential
frictionless contact behavior was assumed and finite sliding
was enabled to allow any possible relative movement of the
surfaces. Having the same results when calculating contact
with a rigid body, the computationally less expensive node-
to-surface contact formulation was preferred to the sur-
face-to-surface alternative and the Lagrange method was
chosen to avoid penetration.
3. Results
3.1. In situ indentation in the scanning electron microscope
During the in situ indentation experiments inside the
scanning electron microscope, the deformation behavior
was monitored via recording of movies. In the following
section the results will be presented by showing individual
frames from these movies.
In Fig. 1a–d a typical loading and unloading cycle on a
fully coated PDMS 1:10 specimen using a cube corner tip is
shown. The maximum indentation depth was 15 lm, corre-
sponding to a maximum test force of approximately 850
lN. The square cracks occurred during sample mounting
and pre-existed before the test. During loading, the tip pen-
etrated into the material with hardly any deformation vis-
ible in the vicinity and a contact contour close to a straight
line. On unloading, two different types of cracks became
visible in the metal layer. The first type of crack formed
along the edges of the tip, where a stress concentration
occurred. The others were concentric circular cracks. Fur-
ther, it is interesting to see Fig. 1c, where the PDMS touch-
ing the tip through the crack along the edges adhered to the
edge of the tip. No residual impression was formed during
the test. Fig. 1e–h shows the same experiment for an
uncoated surface of PDMS 1:10. During loading a clear
sink-in effect was visible as indicated by the bowed contact
contour. The shape of the impression was more conical
than pyramidal, thus the center of the faces did not come
into contact. This is proven by the dirt particle on the left
side of the tip, which remained visible even when it was
below the original surface position (see circles in Fig. 1e
and f). During unloading (Fig. 1g), adhesion between the
tip and the PDMS occurred at the rear face and caused
the steep groove-like impression with hillocks one each
side.
A Berkovich tip was used for the 5 lm deep indentations
on PDMS 1:10 in Fig. 2. The micrographs on the left side
(a)–(c) show the test on the uncoated surface, whereas (d)–
(f) show the indent on a coated specimen. Due to the large
faces of the non-conductive Berkovich tip, the charging
during these tests was more pronounced than for the cube
corner tip. Direct comparison of (a) and (d) suggests that
no sink-in effect occurred in the coated sample, whereas
the dark area along the contact edge in (a) denotes a gap
between tip and surface. No significant residual impres-
sions were created with the Berkovich.
Fig. 3 summarizes a series of tests on uncoated PDMS
1:10 with the cube corner tip, where two parameters were
varied: the indentation depth, ranging from 5 up to 20
lm, and the loading rate, between 9 and 200 nm s1. In
all tests permanent deformation took place, and in some
cases even cracks formed in the PDMS. For a more
detailed characterization, AFM imaging was performed.
The four images in Fig. 4 show impressions after tests to
5 and 10 lm at loading times of 100 and 600 s. The
corresponding profiles reveal that the size of the residual
 J.K. Deuschle et al. / Acta Materialia 56 (2008) 4390–4401 4393

Fig. 1. Comparison of 15 lm indentations into PDMS with a cube corner tip. (a–d) Coated surface; (e–h) uncoated surface. The micrographs in (a) and (e)
show the initial state of the loading segment, (b) and (f) show the fully loaded state, (c) and (g) unloading revealing adhesion between the PDMS and the
tip, and (d) and (h) fully unloaded.

impressions increased slightly with indentation depth and
decreased with loading rate. For higher penetrations and
lower loading rates the shape of the impressions
approached the pyramidal tip shape.
3.2. Simulation
The force–indentation depth data are used to check the
quality of the material model in order to describe the
behavior of PDMS accurately. In Fig. 5 the experimentally
determined force–depth curves are plotted together with
the data obtained from the simulations. The data sets show
good agreement. The experimental curves are the average
of 10 individual tests.
The experimental values of the contact area result from
the Oliver–Pharr calibration procedure according to Refs.
[5,6]. To illustrate the results, Fig. 6a shows an optical
micrograph of the projection of a contact between PDMS
4394 J.K. Deuschle et al. / Acta Materialia 56 (2008) 4390–4401
Fig. 2. Indentations with a Berkovich tip. (a–c) Uncoated; (d–f) coated. Direct comparison of (b) and (e) reveals that the uncoated sample shows a
stronger sink-in effect than the coated one.
and a cube corner tip. The white triangle corresponds to
the upper limit of the indenter cross-section in the absence
of sink-in. In Fig. 6b and c the same area is also denoted by
white lines. Thus, results from the three different methods
can be compared through this maximum cross-sectional
area. The comparison of results is given in Table 1. In
Fig. 6b the simulated contact area between one side of
the pyramidal tip and the surface is given. The curved con-
tact contour illustrates the difference between the contact
depth along the tip edges (white line) and the face centers
(black line). Fig. 6c shows a frame of an in situ test inside
the scanning electron microscope in which the sinking-in
behavior is visible qualitatively.
Fig. 7 compares the contact areas from FEA and in situ
testing for different tip geometries and materials. The grey
lines indicate the limits of the cross-sectional area as shown
by the black and white line in Fig. 6b. The star symbols
represent the area determined from the in situ testing
(according Fig. 6c). Thus, the upper grey lines and the star
symbols are to be compared and give reasonable agreement
between in situ tests and FEA. Further, the experimental
results according to Oliver–Pharr [5,6] are marked by grey
dots and the true contact areas obtained from the simula-
tions by black triangles. The error bars denote one stan-
dard deviation of 10 tests. For the experimental results,
the size of the error bars was approximately the symbol
size, thus they are not visible. Noticeable for the Berkovich
tip (Fig. 7b) is that, although the results from in situ testing
and FEA agreed well, the true FEA derived contact areas
were smaller than calculated from the Oliver–Pharr
method.
4. Discussion
In the first paragraph some issues regarding the observa-
tion of PDMS inside a scanning electron microscope will be
addressed. In the subsequent section the results from differ-
ent methods of contact area determination will be com-
pared and discussed.
4.1. In situ indentation in the scanning electron microscope
When investigating PDMS inside a scanning electron
microscope, it has to be taken into account that PDMS is
 J.K. Deuschle et al. / Acta Materialia 56 (2008) 4390–4401 4395

Fig. 3. Residual impressions on PDMS as a function of loading rate and indentation depth. The left column (a–d) was performed at 200 nm s1, the right
column (e–i) at 9 nm s1. The rows represent indentation depths of (a) and (e) 5 lm; (b) and (f) 10 lm; (c) and (g) 15 lm; and (d) and (h) 20 lm.

a non-conductive polymeric material. The imaging of poly-
meric materials is difficult due to several reasons. Charging
effects can occur; they exhibit a low contrast and are prone
to damage from the electron beam [32–36]. In the course of
our investigations two major factors turned out to deter-
mine the deformation behavior of the specimens: the coat-
ing of the specimen, and the interactions with the electron
beam if the samples were uncoated.
4.1.1. Influence of the Au–Pd coating
How the indentation testing was affected by the thick-
ness of the conductive Au–Pd layer can be seen in Fig. 8.
The three micrographs show significant differences between
samples with a 10 nm thick coating and a 1 nm thick coat-
ing, and an uncoated sample. In the case of the thicker
metal layer, cracking and delaminating of the layer
occurred because of the comparatively high stiffness and
therefore the small degree of deformability of the Au–Pd
layer. In contrast, the thin 1 nm layer seemed to adhere
and deform well. Here an extreme sink-in happened with
a presumably very small contact area between the sample
and the tip.
The deformation behavior of hard films on soft sub-
strates was investigated with FE simulations by Laursen
and Simo [14]. They found that hard film–soft substrate
systems generally tend to show a sink-in behavior. Assum-
4396 J.K. Deuschle et al. / Acta Materialia 56 (2008) 4390–4401

Fig. 4. AFM images and surface profiles of impressions after in situ testing: (a) 5 lm deep with 100 s loading time; (b) 5 lm, 600 s; (c) 10 lm, 100 s; (d) 10
lm, 600 s. As can be seen from the profiles, the residual indent size is more dependent on the loading time than on the penetration depth.

ing that the amount of sink-in is determined by the ratio of
film stiffness to substrate stiffness [14] and that the stiffness
of the gold film increased with increasing thickness, our
observations can be understood. Further, they showed that
for a constant film thickness increasing the penetration
depth leads to reduced sink-in, which explains the very
close contact between tip and surface in the case of the
10 nm thick coated sample. For the uncoated PDMS in
Fig. 8c the rubber-like deformation of PDMS alone led
to a steep, conically shaped indentation. Thus, it can be
argued that, due to the large differences in stiffness between
PDMS and Au–Pd, the deformation of the coated speci-
 J.K. Deuschle et al. / Acta Materialia 56 (2008) 4390–4401 4397

450 experiment
FEA
400

350

300
test force [µN]

250

200

150

100

50 cube corner tip

0 PDMS 1:10
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
indentation depth [nm]

250
experiment
225
FEA
200

175

150
test force [µN]

125

100

75

50

25
Berkovich tip
0 PDMS 1:10
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
indentation depth [nm]

Fig. 5. Experimental force–indentation depth curves measured on PDMS

1:10 in comparison with simulated curves (a) for the cube corner tip and
(b) for the Berkovich tip.

mens did not display the true PDMS behavior. Therefore, Fig. 6. (a) Optical micrograph of the contact perpendicular to the surface.
the specimens used for the determination of the contact The contour is bowed due to sink-in effects, thus the indenter cross-
sectional area indicated by the white triangle is larger than the true contact
were only partially coated, so that the indentations could area. Micrographs of this type were used for the optical determination of
be performed on uncoated spots. This ensured that the true the contact area (see Table 1). (b) The contact of one side of the indenter
PDMS deformation behavior was observed. tip as obtained from FEA. The white line represents the maximum
indenter cross-section, the black line the minimal cross-section. (c) SEM
4.1.2. Electron-beam irradiation effects micrograph from in situ testing with a cube corner tip. The black line
denotes the determinable cross-sectional area, which should theoretically
From a mechanical point of view, the free PDMS sur- be the same as in (a) and (b).
face was necessary for the indentation testing. However,
the effects of subjecting PDMS directly to the electron
beam have to be considered. The consequences of the elec- Table 1
tron irradiation during imaging were obvious. The scanned Comparison of maximum cross-sectional areas (lm2) from different
area was clearly distinguishable from the non-irradiated methods
area through its considerably lower position compared to Method FEA In situ SEM In situ optical
the original, i.e. non-irradiated, surface. AFM profiles of 2500 nm 108 116 ± 15 105 ± 8
the boundary between radiated and non-irradiated surface 5000 nm 360 368 ± 12 366 ± 10
revealed that the irradiated area was 150 nm below the ori-
ginal surface. This led to the conclusion that the radiation
caused a permanent volume reduction of the PDMS. This low-molecular-weight PDMS fragments [32] or with a den-
can be associated either with the evaporation of volatile sity increase due to crosslinking [35]. Evidence has been
4398 J.K. Deuschle et al. / Acta Materialia 56 (2008) 4390–4401

180
experiment (O&P)
FEA
160 in-situ testing (SEM)
limits
140
contact area [µm ]
2

120

100

80

60

40

20
cube corner tip
0 PDMS 1:10
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
indentation depth [nm]

350 experiment (O&P)

FEA
300 in-situ testing (SEM)
limits
250
contact area [µm ]
2

200

150

100

50
Berkovich tip
0 PDMS 1:10
0 1000 2000 3000 4000 5000
indentation depth [nm]

200 experiment (O&P)

FEA
180
in-situ testing (SEM)
160 limits
contact area [µm ]
2

140
Fig. 8. Deformation state at a penetration of 8 lm for differently coated
120
specimens. (a) Coating thickness 10 nm AuPd; (b) coating thickness 1 nm;
100 (c) uncoated.
80

60
a limited amount of volatile fragments. Also, the chamber
40
vacuum was stable, thus contradicting a significant degra-
20 cube corner tip dation and degassing of the sample. Considering these
PDMS 1:30
0 arguments, further crosslinking appears to be the mecha-
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
nism more likely to explain the volume reduction during
indentation depth [nm] electron beam irradiation.
Fig. 7. Comparison of contact area values from ex situ and in situ testing Another interesting finding concerning the beam–speci-
with simulations for different tips and two PDMS samples. O&P = Oliver men interaction is that permanent deformation in the
and Pharr method. PDMS could only be achieved when PDMS was exposed
to the electron beam. Considering Fig. 9, no sign of plastic-
ity or permanent deformation could be found after a 20 lm
found in favor of both phenomena in different studies [32– deep ex situ indentation with a cube corner tip, whereas
36]. In our case, the voltages between 5 and 15 kV used for Fig. 4 clearly shows residual impressions, even after the
the tests are close to those reported in Ref. [32], where frag- indentations to 5 and 10 lm. This once again highlights
ment removal occurred, and are relatively low compared to the strong influence of the electron beam on PDMS. Under
the voltage values reported in refs. [33–36], which caused atmospheric conditions PDMS behaved truly rubber-like
PDMS to further crosslink. However, fully cured PDMS and the recovery of indentations was 100%, whereas per-
samples were used for the study, which should contain only manent impressions were created when PDMS was exposed
 J.K. Deuschle et al. / Acta Materialia 56 (2008) 4390–4401 4399

Fig. 9. (a) AFM image of indent location before testing. (b) AFM image of the same location after ex situ testing. No permanent impression was created.

to electron radiation. The reason for the permanent defor-

mation under the electron beam seems to be associated
with the sample heating, since it is known that PDMS
undergoes a hardening process and becomes increasingly
brittle [42] when it is kept at temperatures above 200 °C.
This embrittlement with prolonged temperature exposure
also explains the results displayed in Fig. 4. Here a residual
impression was found for the 5 lm (Fig. 4a and b) as well
as the 10 lm deep indentations (Fig. 4c and d). A compar-
ison of the depth of the residual impressions after these
experiments reveals that they were almost independent of
maximum penetration depth, but approximately the same
for the same time of testing. In Fig. 4a and c the loading
segment took 100 s (the total cycle duration was 160 s);
the indent depth is roughly 400 nm in both cases, corre- Fig. 10. Comparison of surface profiles in the loaded state at 1.1 lm and
sponding to 92% and 95% recovery respectively. Accord- of a permanent impression of 1.1 lm depth. The surface displacements are
ingly, in Fig. 4b and d the load segment duration was given along the tip edges and in the center of the tip face in order to see the
600 s. This resulted in an indent depth of approximately effects of the tip edges.
1100 nm, which was 80% recovery for the 5 lm test
and 90% for the 10 lm test. Relatively speaking, an
increase in the maximum penetration of 100% led to no in the inclined angle with respect to the opening angle of
change in the residual impression depth, whereas a sixfold the cone. Further, a slight reduction in the indent diameter
increase in loading time increased the indent depth by a is found. These features of elastic recovery can also be
factor of 2.5. This clearly indicates that the beam exposure found in Fig. 10. Noticeable also is that the deformation
time is the decisive factor for the size of the permanent and the recovery are different for the material deformed
impression, whereas the indentation depth itself plays a by the face than for the material deformed by the edge of
minor role. the tip. In the loaded state the material adapted the straight
shape of the tip in the regions of contact; therefore the
4.1.3. Surface profiles slope along the face is higher than along the edge. In the
Considering the creation of the residual indent and the recovered state, however, the difference in the slopes is
recovery mechanism, a comparison between the surface much smaller for the edge areas, indicating less recovery
profiles of a residual impression (measured after testing and hence more plastic deformation in the area of the edge
by AFM) and profiles during loading (extracted from FE compared to the face areas. This is due to the stress concen-
simulations) is given in Fig. 10 for a cube corner. Further, tration at the edges, which led to a greater amount of plas-
this graph contains a comparison between the profiles tic deformation.
along the edge and the face center of the tip. The first con-
clusion from Fig. 10 is that the recovery for a pyramidal tip 4.2. Contact area determination
happened in a similar way to that assumed for conical tips
[43]. For conical impressions, a significant recovery of the In view of the strong interactions between PDMS and
depth is expected, which is accompanied by an increase the electron beam, the validity of the in situ SEM results
4400 J.K. Deuschle et al. / Acta Materialia 56 (2008) 4390–4401
may be in doubt. Therefore, we had to check carefully
whether the contact area determination using the SEM
micrographs delivered reliable results. In order to clarify
whether the deformation behavior of PDMS is actually
influenced by the electron irradiation, we performed a sec-
ond type of in situ indentation test, based on light-micro-
scope contact observation. With this setup, the PDMS
specimens were tested in their original, unaffected state.
Qualitatively, there was no obvious difference in the way
PDMS deformed when indented under the electron beam
or under atmospheric conditions. This is supported by
Fig. 6, where all three techniques applied in this study
are contrasted with each other. The similarities are obvi-
ous, most strikingly in the considerable amount of sink-
in. Quantitative evidence is given in Table 1. Here the
results for the upper limit of the cross-sectional area from
the three different techniques are summarized for the Ber-
kovich tip, since this area is available from FEA (Fig.
6b), in situ SEM (Fig. 6c) and in situ testing using light
microscopy (Fig. 6a). The range of experimental error for
in situ measurements is indicated by one standard devia-
tion of 10 individual measurements. The consistency of
simulations and experimental results points out the ade-
quacy of the chosen material model. The deformation of
PDMS could be represented satisfactorily by the neo-
Hookeian strain energy potential. Of further note is that
the results of the two in situ methods were well matched,
leading to the conclusions that the SEM micrographs allow
the quantitative calculation of the upper contact area limit
and that the deformation is not altered significantly by the
effects of the electron beam. Regarding the two in situ test-
ing techniques, it is worth mentioning that both techniques
are highly complementary because they offer completely
different perspectives for the contact observation, while still
delivering the same results for the cross-sectional area.
We now reconsider Fig. 7, where experimental results
after Oliver–Pharr are compared with the simulations and
the in situ SEM results. In all cases, the results are in rea-
sonable agreement. In Fig. 7a and b the differences between
the cube corner and the Berkovich indentation can be seen
for PDMS 1:10. An important finding is that FEA gave a
contact area about 8% smaller than the Oliver–Pharr
method did in the case of the Berkovich tip and an approx-
imately 5% smaller value for the cube corner tip. This
implies that the Oliver–Pharr method underestimates the
sink-in effect compared to the FE modeling and hence
slightly overestimates the contact area. This result contra-
dicts the common assumption (e.g. [44,45]) that the Oli-
ver–Pharr analysis underestimates the contact area for
soft polymeric (or biological) materials. Our results indi-
cate that the Oliver–Pharr method provides a slight overes-
timate of the contact area for penetrations of a few
micrometers.
Regarding the influence of crosslinking density and thus
the stiffness of the specimen material on the contact evolu-
tion, the results for PDMS 1:10 and PDMS 1:30 shown in
Fig. 7a and c should be compared. For the more compliant
PDMS 1:30 both the experimentally determined and the
simulated contact area and depth values were 5% higher
than for the stiffer PDMS 1:10. From the experimental data
of the two PDMS types, the relative force increase was by a
factor of 5.8, whereas the relative stiffness increase was six
times. According to the Oliver–Pharr calculations the con-
tact depth hc is the difference between the total indentation
depth and the depth of the original surface hs. The surface
depth is determined by the ratio of force to contact stiff-
ness. Thus, the relatively stronger increase in contact stiff-
ness compared to the force leads to a larger contact
depth and area, respectively.
In order to get a more quantitative evaluation of the
amount of sink-in, several attempts have been presented
in the literature, e.g. by Choi et al. [46] and by McElhaney
et al. [47]. Choi et al. [46] used the ratio of contact depth to
total indentation depth as a measure for sink-in. The case
of no sink-in corresponds to a ratio of unity; the smaller
the ratio, the stronger the sink-in. The ratios hc/htot
obtained in this study were the following: for PDMS 1:10
and a cube corner, hc/htot was 0.79 ± 0.01, for the Berko-
vich it was 0.72 ± 0.03; for PDMS 1:30 and the cube corner
tip, 0.78 ± 0.005 was obtained. In Ref. [47] another ratio is
proposed as a measure for the sinking-in or piling-up of
materials, which is the ratio of the contact area to the tri-
angle area in Fig. 6a. The values calculated for PDMS
1:10 and the cube corner were 0.75 ± 0.04, for the Berko-
vich 0.74 ± 0.02, and for the cube corner and PDMS
1:30, 0.74 ± 0.03. These measures of sink-in suggest that
the sink-in for the Berkovich was slightly more pronounced
than for the cube corner. Further, there was very little dif-
ference between the two PDMS specimens, indicating no
significant influence of the crosslinking density on the
sink-in behavior.
5. Conclusions
We have applied in situ indentation testing inside a scan-
ning electron microscope in combination with FE simula-
tions to investigate the indentation process in soft
elastomeric materials. The aim was to study the deforma-
tion and sink-in behavior and the evolution of the contact
during indentation. From this study the following conclu-
sions can be drawn:
For the coated samples the layer thickness of the Au–Pd
coating determined the deformation behavior. The true
behavior of the PDMS specimens could only be monitored
if the surface was uncoated.
Under the direct influence of the electron beam, perma-
nent impressions were created, whereas no residual indents
were found for indentations under atmospheric conditions.
However, an influence of the electron beam on the contact
area evolution could be excluded with the help of in situ
indentation based on light microscopy.
The FE results for the cross-sectional area could be con-
firmed by the values obtained from in situ testing inside the
scanning electron microscope as well as from in situ testing
 J.K. Deuschle et al. / Acta Materialia 56 (2008) 4390–4401
using light microscopy; therefore the FE values for the pro-
jected area can be considered as a reliable measure for the
real contact size.
Since the contact area determination following Oliver
and Pharr deviated from the simulations by several per-
cent, the results of this study indicate that the standard Oli-
ver–Pharr method gives an overestimated but acceptable
value of the contact area during indentation of soft elasto-
meric materials.
Acknowledgements
The authors would like to thank Dr. E. de Souza for
providing the PDMS specimen material. We acknowledge
the help of R. O’Hagan, Dr. H. Pfaff and M. Fajfrowski
for support regarding the nanoindentation equipment.
We are grateful to Dr. B. Gore-Clark for a critical reading
of the manuscript.
References
[1] ISO standard 14577: metallic materials – instrumented indentation
test for hardness and materials parameter. Part 1, part 2 and part 3,
2003; part 4, 2007.
[2] Hertz H. Journal für die reine und angewandte Mathematik
1881;92:156.
[3] Boussinesq J. Application des potentials à l’Étude de l’Équilibre et du
mouvement des solides Élastiques. Paris: Gauthier-Villard; 1885.
[4] Sneddon IA. Int J Eng Sci 1965;3:47.
[5] Oliver WC, Pharr GM. J Mater Res 1992;7:1564.
[6] Oliver WC, Pharr GM. J Mater Res 2004;19:3.
[7] Larsson PL, Giannakopoulos AE, Söderlund E, Rowcliffe DJ,
Vestergaard R. Int J Solid Struct 1996;33:221.
[8] Bucaille JL, Felder E. Phil Mag 2002;82:2003.
[9] Bucaille JL, Felder E, Hochstetter G. Wear 2001;249:422.
[10] Li K, Wu TW, Li JCM. J Mater Res 1997;12:2064.
[11] Bouzakis KD, Michailidis N. Thin solid films 2006;494:155.
[12] Gupta S, Carillo F, Balooch M, Pruitt L, Puttlitz C. J Mater Res
2005;20:1979.
[13] Li M, Chen WM, Liang NG, Wang LD. J Mater Res 2004;19:73.
[14] Laursen TA, Simo JC. J Mater Res 1992;7:618.
[15] Nitta I. Wear 1995;181–183:844–9.
[16] Ovcharenko A, Halperin G, Verberne G, Etsion I. Trib Lett
2007;25:153.
4401
[17] Morris DJ, Cook RF. J Am Ceramic Soc 2004;87:1494.
[18] Miyajima T, Sakai M. Phil Mag 2006;86:5729.
[19] Ruffell S, Bradby JE, Williams JS, Warren OL. J Mater Res
2007;22:578.
[20] Moser B, Kuebler J, Meinhard H, Muster W, Michler J. Adv Eng
Mater 2005;7:388.
[21] Moser B, Löffler JF, Michler J. Phil Mag 2006;86:5715.
[22] Michler J, Rabe R, Bucaille JL, Moser B, Schwaller P, Breguet JM.
Wear 2005;259:18.
[23] Ohmura T, Minor A, Tsuzaki K, Morris Jr JW. Mater Sci Forum
2006;503–504:239.
[24] Minor AM, Lilleodden ET, Jin M, Stach EA, Chrzan DC, Morris
JW. Phil Mag 2005;85:232.
[25] Zhou J, Komvopulos K, Minor AM. Appl Phys Lett 2006;88.
181908-1.
[26] Oshima A, Washio M. Nucl Instrum Meth Phys Res B 2003;208:380.
[27] Song ZG, Ong CK, Gong H. Appl Surf Sci 1997;119:169.
[28] Loo SCJ, Ooi CP, Boey YCF. Biomaterials 2005;26:3809.
[29] Drummy LF, Yang J, Martin DC. Ultramicroscopy 2004;99:247.
[30] Safrany A, Wojnarovits L. Radiat Phys Chem 2004;69:289.
[31] Lovinger AJ, Padden FJ, Davis DD. Polymer 1991;32:3086.
[32] Russell MT, Pingree LSC, Hersam MC, Marks TJ. Langmuir
2006;22:6712.
[33] Maxwell RS, Chinn SC, Solyom D, Cohenour R. Macromolecules
2005;38:7026.
[34] Bradley DA, Dalhan KZ, Roy SC. Appl Radiat Isotopes 2000;53:921.
[35] Pankratova LN, Dubovik II, Rabikina AY, Bugaenko LT. High
Energy Chem 2003;37:423.
[36] Zhu YC, Zhang XH, Qiao JL, Wei GS. Chinese J Polym Sci
2004;22:147.
[37] Stevens MR, Chen Q, Weierstall U, Spence JHC. Microscopy
Microanal 2000;6:368.
[38] Deuschle J, de Souza EJ, Enders S, Arzt E. Crosslinking and curing
kinetics of PDMS studied by dynamic nanoindentation, unpublished
results.
[39] ABAQUS analysis user’s manual 6.6, ABAQUS Inc., Providence, RI,
USA, 2006.
[40] ABAQUS theory manual, ABAQUS Inc., Providence, RI, USA,
2006.
[41] Treloar LRG. The physics of rubber elasticity. 3rd ed. Oxford: Clar-
endon Press; 1975.
[42] Dow Corning Corp. Sylgard 184 Silocone Elastomer Data Sheet
2001.
[43] Stilwell NA, Tabor D. Proc Phys Soc 1961;78:169.
[44] Ebenstein DM, Wahl KJ. J Colloid Interface Sci 2006;298:652.
[45] Gupta S, Carillo F, Li C, Pruitt L, Puttlitz C. Mater Lett 2007;61:448.
[46] Choi Y, Lee HS, Kwon D. J Mater Res 2004;19:3307.
[47] McElhaney KW, Vlassak JJ, Nix WD. J Mater Res 1998;13:1300.