Available online at www.sciencedirect.

com

Acta Materialia 56 (2008) 31–42

www.elsevier.com/locate/actamat

On the origin of deformation-induced rotation patterns

below nanoindents
N. Zaafarani, D. Raabe *, F. Roters, S. Zaefferer
Department of Microstructure Physics and Metal Forming, Max-Planck-Institut für Eisenforschung, Max-Planck-Straße 1, 40237 Düsseldorf, Germany

Received 22 June 2007; received in revised form 30 August 2007; accepted 1 September 2007
Available online 18 October 2007

Abstract

This study is about the origin of systematic deformation-induced crystallographic orientation patterns around nanoindents (here of
single crystalline copper; conical indenter) using the following approach: first, the rotation pattern is investigated in three-dimensions
(3D) using a high-resolution 3D electron backscattered diffraction (EBSD) technique (EBSD tomography) which works by a serial sec-
tioning and EBSD mapping procedure in a scanning electron microscopy-focused ion beam cross-beam set-up. Second, the problem is
modeled using a crystal plasticity finite element method which is based on a dislocation density-based constitutive model. Third, the
results were discussed in terms of a geometrical model which simplifies the boundary conditions during indentation in terms of a com-
pressive state normal to the local tangent of the indent shape. This simplification helps to identify the dominant slip systems and the
resulting lattice rotations, thereby allowing us to reveal the basic mechanism of the formation of the rotation patterns. The finite element
simulations also predict the pile-up patterning around the indents, which can be related to the dislocation density evolution.
 2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Nanoindentation; 3D EBSD; Crystal plasticity; Gradient; Geometrically necessary dislocations

1. Introduction we must also state that a detailed simulation does not nec-
essarily replace a firm understanding of the main mecha-
The observation of deformation-induced crystallo- nism which is responsible for the fact that such patterns
graphic lattice rotations caused by nanoindentation has exist at all. In other words, although good agreement was
recently attracted attention [1–8]. These orientation obtained in a previous experimental and theoretical study
patterns were observed experimentally [1–4,7] and also in of this phenomenon [4], it remains unclear why these defor-
corresponding simulations [4,6]. Concerning the experimen- mation-induced patterns consist of multiple narrow zones
tal work, it is important that the same rotation patterns of alternating crystalline rotations and counter-rotations
were identified by using three different types of techniques of equal magnitude in close sequence. In this paper, we
performed by different groups, namely, synchrotron wide- therefore try to elucidate the general crystallographic prin-
angle Bragg diffraction [1], electron backscattered diffrac- ciple behind these patterns. For this purpose, we take the
tion (EBSD) in two [7] and three dimensions [4], and following approach: first, the rotation patterns are investi-
transmission electron microscopy (TEM) [2,3]. The visco- gated by a high-resolution three-dimensional (3D) EBSD
plastic crystal plasticity finite element simulations which technique (EBSD tomography) for a nanoindent performed
we published recently [4] could also capture some important by a conical indenter with a spherical tip in a copper single
features of the experimentally observed orientation pat- crystal. Next, we conduct advanced crystal plasticity finite
terns. However, irrespective of some success along this line, element simulations which are based not on a viscoplastic
formulation but on a dislocation density-based constitutive
*
Corresponding author. Tel.: +49 211 6792 333. model [9,10]. Finally, the results are discussed in terms
E-mail address: d.raabe@mpie.de (D. Raabe). of a geometrical model which simplifies the boundary

1359-6454/$30.00  2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.actamat.2007.09.001
32 N. Zaafarani et al. / Acta Materialia 56 (2008) 31–42

conditions imposed during indentation in terms of a com- slip system a are treated as functions of the statistically
pressive state normal to the local tangent of the indent stored dislocation density, qSSD. In Ref. [9] it is shown that
rim. This simplified assumption allows one to identify the the mobile dislocation density can be approximated by
dominant slip systems and the crystallographic principle 2k B T pffiffiffiffiffiffiffipffiffiffiffiffiffiffi
behind the resulting reorientations. The finite element sim- qMa ¼ qFa qPa ð2Þ
c1 c2 c3 Gb3
ulations also predict the pile-up patterning around the
indents which can be related to the dislocation density where c1, c2 and c3 are material constants, kB is the Boltz-
evolution. mann constant, T is the absolute temperature and G is the
shear modulus. The quantities qFa and qPa denote the
2. Modeling and simulation forest dislocation density and the density of parallel dislo-
cations, respectively. Both measures are calculated as
2.1. Motivation projections of the statistically stored dislocation density
qSSD on the slip system a. Based on the cutting process
The deformation and reorientation patterns caused by of the forest dislocations, the average dislocation velocity
nanoindentation motivate the use of a crystal plasticity can be calculated as
 
finite element approach for obtaining a better understand- Q  ðjsa j  spass;a ÞV a
ing of this phenomenon. The crystal plasticity finite va ¼ ka m0 exp  cut ð3Þ
kBT
element method allows one to study the activation of
certain slip systems at different stages of indentation, the where
pffiffiffiffiffiffiffi
change of the crystal orientation, and the formation of spass;a ¼ c1 Gb qPa ð4Þ
pile-up or sink-in patterns around the indent [11]. The
strength of the method lies particularly in properly map- is the shear stress required to move a dislocation past the
ping the strong kinematical influence of crystalline anisot- other parallel dislocations.
pffiffiffiffiffiffiffi
ropy during plastic straining, which as a rule also leads ka ¼ c2 = qFa ð5Þ
to the prediction of the correct tendency in the resulting
is the jump width of the mobile dislocations related to the
orientation changes. This has been shown in a number of
density of forest dislocations qFa.
studies [4,11–16] where predictions obtained by the crystal
The activation volume, Va, resulting from the steps
plasticity finite element method were compared in detail
(kinks) created by the cutting process can be expressed as
with corresponding micromechanical experiments with
respect to strain, texture and forces. Conventional crystal V a ¼ c 3 b2 ka ð6Þ
plasticity models (e.g. [17,18]) use viscoplastic hardening
The quantities m0 and Qcut in Eq. (3) are the attack
laws as a good compromise between computation speed
frequency and the activation energy for the forest cutting
and predictive reliability. The shortcoming of such visco-
process, respectively.
plastic formulations is the absence of an internal variable
The evolution of the statistically stored dislocation den-
concept beyond the incorporation of the crystal orienta-
sity describes the hardening on each slip plane. The evolu-
tion. More advanced variants of the crystal plasticity finite
tion Eq. (7) includes multiplication ðq_ þ Þ and annihilation
element method, therefore, replace the viscoplastic consti-
ðq_  Þ terms for the statistical density of dislocations, qSSD,
tutive description by dislocation-based models [9,10,19–
in the form of a rate formulation,
26]. Since it is likely that some dislocation effects which
cannot be readily captured by viscoplastic hardening laws q_ SSD ¼ q_ þ1 þ q_ þ2  q_ 1  q_ 2 ð7Þ
may play a substantial role in nanoindentation of single þ1 þ2
where q_ and q_ are the production rates of qSSD due to
crystals, we use in this study an advanced crystal plasticity the interaction with forest dislocations and dipole forma-
finite element method which is based on dislocation rate tion, respectively.
formulations for the simulation of nanoindentation.
c4 1
q_ þ1 ¼ c_ a ð8Þ
2.2. A physically based crystal plasticity model b ka
c5
2.2.1. Theory q_ þ2 ¼ d dipole qM c_ a ð9Þ
b
The model which was introduced in Refs. [9,10] adopts
c4 and c5 are fitting parameters,
the Orowan equation as a kinematic equation on each slip
pffiffiffi
system to establish a connection between the shear rate and Gb 3
the mobile dislocation density d dipole ¼ ð10Þ
16pð1  mÞðjsa j  spass;a Þ
c_ a ¼ qMa bva ð1Þ
is the critical distance for dipole formation [24] and m is the
where c_ a is the shear rate on slip system a and b is the mag- Poisson’s ratio. The rates q_ 1 and q_ 2 are the non-
nitude of the Burgers vector. The mobile dislocation thermal and thermal annihilation rates of the dislocations,
density, qMa, and the velocity of the dislocations, va, on a respectively.
 N. Zaafarani et al. / Acta Materialia 56 (2008) 31–42 33

q_ 1 ¼ c6 qSSD c_ a ð11Þ However, as the scale of the two experiments (compres-

2   sion, nanoindentation) differs in size, the fitted values may
Ddf Aac b Qbulk c8
q_ 2 2
¼ c7 bqSSD sa exp  c_ a ð12Þ also be different for the two scales. This applies in the cur-
kBT kBT rent case for the value of the initial dislocation density.
where c6 and c7 are fitting parameters, Ddf is the pre-expo- Although the original single crystal fit procedure provided
nential part of the diffusion coefficient, Aac is the activation a value of 5 · 1010 m2, a higher value of 1012 m2 was
area and Qbulk is the activation energy for self diffusion. As chosen in the actual simulations since the simulated
the deformation takes place at room temperature, the ther- force–depth curves in this case revealed a better match
mal annihilation term, Eq. (12), was neglected. when compared with the ones observed in the nanoinden-
tation experiments. Even though this seems like a large dis-
2.2.2. Parameter sensitivity and fitting procedure crepancy, it can be understood by local deviations from the
The evolution equations involve a number of material average dislocation density and also the rapid increase of
parameters (c1–c6) which have to be fitted using experimen- the dislocation density at the very beginning of plastic
tal data. For the determination of the proper values of the deformation.
material parameters, a compression test was performed on At first sight, the dislocation-based model uses a similar
a copper single crystal. The shape of the compression number of fit parameters as empirical models. However, it
sample was a cylinder with dimensions 3 mm · 4.5 mm. must be emphasized that a major advantage of the disloca-
The compression speed was 0.1 mm s1, resulting in an tion-based model is that all parameters have a physical
average overall strain rate of 4.3 · 104 s1. As the com- meaning and that they can be further coupled to metallur-
pression axis was in the [1 1 1] crystal direction, no rotation gical mechanisms and multiscale models. All parameters
was expected ([1 1 1] is stable under compression). Due to that were used lie within the range expected according to
the high symmetry of this orientation, the deformation is Ref. [9].
achieved by the activation of at least six slip systems. A
shear stress–shear strain curve for a single slip system 2.3. Details of the crystal plasticity finite element model
was extracted from the experiments by using the appropri- implementation
ate Schmid factor and comparing it with the corresponding
curve calculated analytically from the evolution equations. The finite element code MARC was used as a solver for
Fig. 1 illustrates the very good agreement between the the constitutive model discussed above via the user-defined
model and the experiment. Table 1 comprises the fitted subroutine HYPELA2 [28]. The mesh consisted of 4312
parameters and the main constants. All parameters except elements and 5224 nodes (Fig. 2a). The elements were of
the initial dislocation density are dimensionless. The activa- the 3D hexahedral type, with eight integration points. Near
tion energy for the slip of edge dislocations, Qcut, was the center of the indented area a finer mesh was used due to
approximated by 12 Gb3 , which results in a value of the high deformation gradients expected in this zone
6.33 · 1019 J for Cu. The attack frequency of a dislocation (Fig. 2b). The lower face of the cylinder was fixed. Other-
segment is fixed to a constant value of 1010 s1 according to wise, each node possessed three translational and three
Ref. [27]. rotational degrees of freedom. The kinematics were based
on an updated Lagrangian scheme.
The tip with a radius of 1.5 lm was modeled as a rigid
body. The simulation was performed with displacement
control up to an indentation depth of 1 lm in the negative
[1 1 1] crystallographic direction. The initial orientation of
the single crystal sample was chosen by selecting appropri-
ate Euler angles (0, 54.74 and 45, Bunge’s notation) such
that the x, y, z coordinate system represented the ½1 1 0,
½1 1 2 and [1 1 1] crystal directions, respectively (Fig. 4).

3. Experimental procedure

Nanoindents were conducted using different forces on a

Cu single crystal with a Hysitron nanoindenter system
(TriboIndenter). The largest indent was 1 lm deep. The
force direction was in the negative [1 1 1] direction of the
crystal. The tip had a conical shape with a spherical tip
in order to avoid any symmetry in the experiment other
than that of the crystal. A scanning electron microscopy
Fig. 1. Fitted shear stress–shear strain curve: (a) undeformed mesh and (SEM) image of the used indenter tip is illustrated in
(b) mesh after indentation Fig. 3. The EBSD experiments were conducted by using a
34 N. Zaafarani et al. / Acta Materialia 56 (2008) 31–42

Table 1
Summary of crystal constants and parameters used in the model
Material constant Unit Value Parameter Unit Value
Shear modulus, G GPa 75.4 For passing stress, c1 – 0.1
Burgers vector, b m 2.56 · 1010 For jump width, c2 – 1.95
Attacking frequency, m0 s1 1010 For obstacle width, c3 – 0.3
Activation energy of slip, Qcut J 6.33 · 1019 For lock formation, c4 – 0.037
Poisson’s ratio, m – 0.33 For dipole formation, c5 – 0.04
For athermal annihilation, c6 – 50
Initial dislocation density, q0 m2 5 · 1010

Fig. 4. Top view presenting the cutting planes relative to the indent
Fig. 2. The mesh used in finite element simulation (one quarter of the position.
mesh is not shown).

Fig. 5. Schematic drawing of the successive section milling and EBSD

Fig. 3. SEM image of the used diamond indenter tip. measurement.

4. Comparison between experimental results and the

dislocation-based crystal plasticity finite element model
joint high-resolution field emission SEM–EBSD set-up
together with a focused ion beam (FIB) system in the form 4.1. Rotation patterns in crystal misorientation presentation
of a cross-beam 3D crystal orientation microscope (3D
EBSD). Details of this novel experimental approach are The set of successive sections shown in Fig. 6 compares
given in Refs. [4,29]. The FIB was directed normal to the the misorientation values for the crystal in the measured
(1 1 1) plane, producing successive parallel milled areas, as ð1 1 2Þ serial sections (Fig. 6a1–a4) with the simulations
illustrated in Fig. 4. After each milling step an EBSD (Fig. 6b1–b4). All sections shown are near the indent cen-
map was taken (Fig. 5). By the alternating milling and ter. The large accumulated strains in this region allow a
EBSD steps a tomographic 3D EBSD image of the spatial more detailed study of the misorientation changes. The
crystal misorientation distribution underneath the nanoin- color code indicates the magnitude of the orientation
dent was reconstructed. Further details on the indentation changes relative to the initial undeformed crystal
procedure are given in Refs. [4,11]. orientation.
 N. Zaafarani et al. / Acta Materialia 56 (2008) 31–42 35

Fig. 6. Misorientation maps (absolute lattice rotation) underneath the indentation at different cross sections, comparison between experimental and
simulation results. (The seemingly shallower indent near the center is due to the edge effects and milling-induced curvature.)

The measurements reveal a pronounced deformation- ity finite element model shows that it succeeded in captur-
induced crystal rotation pattern underneath the indent, as ing fine details of the crystallography of the rotation
reported previously [4]. The alternating rotation and reori- patterning (Fig. 7b1–b4). Concerning the absolute values
entation zones are arranged directly underneath the indent. of the rotation angles about the ½1 1 2 axis, a maximum
The measurements show that at the sides of the nanoindent rotation angle of 14 was observed in the experiment.
only very small crystallographic rotations took place. The The corresponding maximum in the dislocation-based crys-
pattern underneath the indent near the center section can tal plasticity model is about 23. This deviation could be
be characterized by six rotation zones which are separated due to edge effects at the contact zone and milling-induced
by narrow rotation-free areas (Fig. 6a4). The maximum curvature caused by the ion beam so that no complete
values of the misorientation angles can be observed at the EBSD mapping could be made up to the actual contact
surface directly at the contact layer. interface. In this region, though, maximum rotation values
are likely to occur. Also, the lateral resolution of the mea-
4.2. Rotation patterns in axis-angle presentation surement which was conducted using a 80 nm step size was
probably not fine enough to capture all the effects at the
The orientation maps are also represented in the form of actual contact surface.
rotation angles about the ½1 1 2 crystal direction in the sam-
ple reference system (note that in Ref. [4] the axis rotation 4.3. Pile-up pattern formation around the indent
is given in the crystal coordinate system). The maps are
plotted on the ð1 1 2Þ planes in Fig. 7a1–a4 at different spac- The pile-up pattern surrounding the ½1 1 1 indent is
ing to the center of the indent. The maps reveal pro- characterized by a sixfold symmetry, as already reported
nounced lattice rotations about the ½1 1 2 axis below the by Wang et al. [11]. The simulations reveal the same pat-
indent and lower values at the rims of the indent. The mea- tern (Fig. 8). Comparing the pattern of the mobile disloca-
surements also show frequent and quite abrupt changes in tion density around the indent (Fig. 9) with the pile-up
the sign of the rotation direction along and near the surface reveals some common features. The mobile dislocations
when presented in the form of ½1 1 2 axis rotations. The are arranged around the indent in a pattern similar to the
rotations decrease further away from the contact zone. pile-up distribution. In fact, the peak values for the mobile
The comparison with the dislocation-based crystal plastic- dislocation density are found in almost the same positions
36 N. Zaafarani et al. / Acta Materialia 56 (2008) 31–42

Fig. 7. Comparison between crystal rotation about the direction ½1 1 2 in the sample reference system underneath the indentation at different cross sections
between experimental and simulation results (positive values mean counter clockwise rotation). Note that in [4] the axis rotation is given in the crystal
coordinate system.

Fig. 8. Pile-up pattern around the nanoindent.

 N. Zaafarani et al. / Acta Materialia 56 (2008) 31–42 37

Fig. 9. Distribution of the mobile dislocation density close to the surface.

Fig. 10. Development of mobile dislocation density around the indent on a f1 1 2g midsection at: (a) 10% indentation (0.1 lm depth) (b) 30% indentation
(0.3 lm depth) (c) 50% indentation (0.5 lm depth) and (d) 100% indentation (1 lm depth).

Fig. 10 shows the evolution of the mobile dislocation

density, qM around the indent on a f1 1 2g midsection. In
the first phase of the indentation the dislocation density
shows a local peak directly underneath the indenter tip.
As the indentation proceeds, the peak of the mobile dislo-
cation density migrates towards the sides of the indent. As
a consequence of this dislocation activity, the material
flows along the path of least resistance. This tendency min-
imizes the strain energy and, since the contact area to the
indenter acts as a rigid boundary, the most adequate path
for the excess material is to flow towards the free surface,
which leads to pile-ups.

Fig. 11. SEM image of a serial section: (a) non-relaxed and (b) relaxed. 4.4. Effect of the free surface

as the maximum pile-ups. This observation reflects the One drawback of the FIB technique for serial sectioning
obvious connection between the local shear and the activity is that the method is destructive. This means that any error
of the mobile dislocations. in the milling procedure requires full repetition of the entire
38 N. Zaafarani et al. / Acta Materialia 56 (2008) 31–42

Fig. 12. Effect of free surface milling on the total crystal rotation at the center cross section.

experiment. Another problem of the method lies in the fact in the finite element analysis. On the surface of the sample
that the approach steadily produces new free surfaces for along the ½1 1 2 crystallographic direction the rotations
EBSD characterization (Fig. 11). This might have a small about the ½1 1 0 axis indicate the mentioned phenomenon,
effect on the measured orientations compared with the as shown in Fig. 13a. Three regions are inspected by select-
(hypothetical) state of a bulk crystal without free surfaces. ing corresponding nodes (Figs. 13). Node (a) is chosen very
With the help of the FE simulation, the error caused by close to the center, where the crystal reveals a negative rota-
the free surface can be estimated. For this purpose, the tion. While node (c) represents the neighboring positive
boundary conditions were changed to mimic a free surface rotation area, node (b) stands for the cross-over region
(relaxed model). This approach simulates the effect of cut- between the two counter-rotated areas. This means that at
ting the surface free. The misorientation patterns predicted node (b) almost no crystal rotation is expected to take place.
in the two model variants are compared in Fig. 12. The While Fig. 14 reveals the evolution of the rotation about
comparison shows that, although the misorientation pat- the ½1 1 0 direction (x-axis) during the indentation, Fig. 15
tern remains almost unchanged, some deviations occur in represents the corresponding slip system activity. It can be
the absolute values. The difference varies between 5 and seen from Fig. 15a that close to the center the crystal
8% from the initial model (which is referred to here as undergoes a small positive rotation at the beginning of
the non-relaxed model variant). the indentation, which soon reverses its sense. Moving out-
wards from the center (Fig. 14b and c), the positive rota-
5. Discussion tion increases until it dominates the whole indentation
process. A likely interpretation of this observation is that
The phenomenon addressed in this paper is the gradual the crystal rotation behavior is influenced by two different
formation of systematic orientation patterns in the mechanisms. The first one is caused by the spatial rotations
deformed zone underneath an indent caused by conical due to the geometry of the indenter. This means that cer-
nanoindentation with a spherical tip. The patterns are par- tain rigid body rotations are imposed by the antisymmetric
ticularly characterized by systematic arrangements of rota- portion of the external displacement gradient tensor which
tion and counter-rotation fields, as presented in the figures is generated by the indentation process. The local crystallo-
above. These rotation- and counter-rotation patterns were graphic slip is responsible for the second kind of rotation.
confirmed both by the measurements and by the In a second approach to better understand the reason
simulations. for the rotation patterns we consider as a simple approxi-
The first approach to better understand the rotation pat- mation the compression force exerted by the indenter as
terns is to conduct a detailed analysis of some selected nodes the only loading source. This force acts almost perpendic-
 N. Zaafarani et al. / Acta Materialia 56 (2008) 31–42 39

Fig. 13. Three nodes with different rotations on the ð1 1 0Þ plane are chosen for investigation: (a) ½1 1 0 rotations on a front view at the middle cross section
and (b) Same rotations on the surface of the crystal.

ular to the local contact surface tangent between tool and the indentation process the axis rotates towards the ½1 1 1
crystal (schematical image in Fig. 16). direction for all selected nodes (1 ! 2 in Fig. 16), before
The components of the contact force at the selected it reverses its motion towards the ½0 0 1 pole, (2 ! 3 in
nodes are extracted from the simulation results. Knowing Fig. 16). This confirms the results of Fig. 14.
the orientation of the crystal at each node allows the deter- The question that arises is: why does the compression
mination of the crystallographic direction of the loading axis reverse its sense of rotation? The answer can be given
axis during the whole indentation process. by monitoring the slip systems involved in the process. At
This is shown in the schematical image on the right-hand the beginning of the indentation three pairs of slip systems
side of Fig. 16, where a simplification is described which is are activated, as indicated in Fig. 15. Slip system pair (1)
used to track the changes in the single crystal kinematics consists of ½1 0 1ð1 1 1Þ and ½0 1 1ð1 1 1Þ, pair (2) of
during indentation in terms of a Sachs loading case under ½1 01ð1 1 1Þ and ½0 1 1ð1 1 1Þ and pair (3) of ½1 10ð1 1 1Þ
simple compression tangential to the local contact interface. and ½1 1 0ð1 1 1Þ. It is well known [30] that for a single crys-
In the projection shown in Fig. 16 the corresponding tal under compression the rotation of the compression axis
evolution of the compression axis is plotted for the various towards the slip plane normal of the active slip system is
nodes. For all nodes the compression axis lies on the the primary rotation. Therefore, and also from Figs. 16
½0 0 1  ½1 1 1 boundary. This is also expected due to the and 17, one can see that the slip system pair (1) causes
symmetry of the problem resulting in a rotation about the compression axis to rotate towards the ½1 1 1 pole fol-
the ½1 1 0 axis. The movement of the compression axis lowing the reorientation route (1 ! 2), leading to an initial
along the ½0 0 1  ½1 1 1 boundary results in simultaneous positive crystal rotation. However, this rotation reduces its
slip on pairs of two slip systems, as can be seen in Schmid factor and hence the shear activity is lowered, as
Fig. 15. It can also be observed that at the beginning of can be seen in Fig. 15a.
40 N. Zaafarani et al. / Acta Materialia 56 (2008) 31–42

Fig. 14. The evolution of the total crystal rotation about ½1 1 0 during Fig. 15. The active slip systems during indentation at: (a) node a, (b) node
indentation at: (a) node a, (b) node b and (c) node c. b and (c) node c.

On the other hand, the two other slip system pairs (2) two rotation mechanisms equilibrate and the crystal
and (3) remain fully active with progressing indentation. reverses its sense of rotation. Route (2 ! 3) describes a
Each pair tends to rotate the compression axis according compression axis rotation in the positive direction around
to the route (2 ! 3) (Fig. 16), towards the ½0 0 1 pole, ½1 1 0, i.e. a negative crystal rotation occurs.
which lies on the great circle going through the two slip The same slip mechanism occurs for node (b). The differ-
plane normals of the pairs ð1 1 1Þ and ð1 1 1Þ. Position (2) ence with respect to node (a) is caused by the tip geometry.
for node (a) (see Fig. 16) in Fig. 15a is located where the By moving outward along the tip, the direction of the
 N. Zaafarani et al. / Acta Materialia 56 (2008) 31–42 41

For node (c) the slip state is more complicated as the

contact with the indenter tip occurs only after more than
35% of the indentation process. This means that other
loading components also influence the initial rotations of
the crystal, leading to a more complicated slip system activ-
ity, as shown in Fig. 15c. Nevertheless, one can observe
that the initial tendency of a positive crystal rotation still
prevails (Fig. 14c). Although the same change of the com-
pression axis rotation occurs, due to the high crystal rota-
tions before the contact with the tip is established, this is
sufficient only to decelerate the rotation of the crystal,
not to reverse it.
Fig. 16. Inverse pole figure revealing the rotation of the compression axis
during indentation at the selected node positions a, b and c. The black
cross represents the [11] direction. The schematical image on the right- 6. Conclusions
hand side shows a simplification which describes the changes in the single
crystal kinematics during indentation in terms of a Sachs loading case The main results and conclusions are:
under simple compression tangential to the local contact interface.

– A physically based constitutive crystal plasticity finite

element analysis reveals better agreement with the nan-
oindentation experiments than a phenomenological
viscoplastic material model.
– The complex rotation pattern induced during conical
indentation which is characterized by several rotation
zones is well predicted when using a physically based
constitutive crystal plasticity finite element model. While
the crystal rotates in a certain direction in the zone near
the indenter axis, the neighboring zone shows the oppo-
site rotation.
– The change of the crystal rotation directions can be
explained by the different active slip systems due to the
geometry of the indenter causing a gradual change in
the loading axis during indentation.
– Although the model succeeded in predicting the pile-up
out-of-plane pattern and the rotation pattern under the
indent, it slightly overestimates the crystal rotations.
– For the small deformation scales discussed here there
might be an effect of the free surfaces created by the
FIB technique. Further experimental study in that direc-
tion needs to be done in order to estimate the deviation
in the measurements with respect to the material condi-
Fig. 17. The active slip systems in the indentation process with a conical tion without milling.
indenter. The black cross represents the [11] direction. The curved lines
connect the slip normal and slip direction of a specific slip system. See also
magnification in Fig. 16.
References

[1] Larson BC; personal communication.

compression axis moves away from the ð1 1 1Þ pole [2] McLaughlin KK, Stelmashenko NA, Lloyd SJ, Vandeperre LJ, Clegg
(Fig. 16). This results in a decreased slip activity of slip sys- WJ. Mater Res Soc Symp Proc 2005;841:3–8.
[3] Lloyd SJ, Castellero A, Giuliani F, Long Y, McLaughlin KK,
tem pair (3) on the one hand and on the other hand extends Molina-Aldareguia JM, Stelmashenko NA, Vandeperre LJ, Clegg
the activity of the slip system pair (1). This explains the WJ. Proc Roy Soc A 2005;461:2521–43.
higher value of the initial positive lattice rotation illus- [4] Zaafarani N, Raabe D, Singh RN, Roters F, Zaefferer S. Acta Mater
trated in Fig. 14b compared with node (a) in Fig. 14a. At 2006;54:1863–76.
the end of the indentation the two opposite rotations com- [5] Brown LM. Mat Sci Forum 2007;550:105–17.
[6] Bouvier S, Needleman A. Modell Simul Sci Eng 2006;14:1105–25.
pensate each other, resulting in an almost zero total rota- [7] Kysar JW; http://www.sem.org/NM-kysar.pdf.
tion of the crystal. It is expected that by further [8] Kiener D, Pippan R, Motz C, Kreuzer H. Acta Mater
indenting this position, the rotation-free zone will move 2006;54:2801–11.
outwards, further away from the center of the indent. [9] Ma A, Roters F. Acta Mater 2004;52:3603–12.
42 N. Zaafarani et al. / Acta Materialia 56 (2008) 31–42

[10] Ma A, Roters F, Raabe D. Acta Mater 2006;54:2169–79. [22] Zehetbauer M. Acta Metall et Mat 1993;41:589–99.
[11] Wang Y, Raabe D, Klüber C, Roters F. Acta Mater 2004;52:2229–38. [23] Estrin Y, Tóth LS, Molinari A, Bréchet Y. Acta Mater
[12] Liu Y, Wang B, Yoshino M, Roy S, Lu H, Komanduri R. J Mech 1998;46:5509–22.
Phys Solid 2005;53:2718–41. [24] Roters F. Realisierung eines Mehrebenenkonzeptes in der Plasti-
[13] Casals O, Ocenasek J, Alcala J. Acta Mater 2007;55:55–68. zitätsmodellierung. 1999 Ph.D Thesis, RWTH Aachen [in
[14] Zaefferer S, Kuo JC, Zhao Z, Winning M, Raabe D. Acta Mater German].
2003;51:4719–35. [25] Roters F, Raabe D, Gottstein G. Acta Mater 2000;48:4181–9.
[15] Ma A, Roters F, Raabe D. Acta Mater 2006;54:2181–94. [26] Arsenlis A, Parks DM. J Mech Phys Solid 2002;50:1979–2009.
[16] Raabe D, Sachtleber M, Zhao Z, Roters F, Zaefferer S. Acta Mater [27] Granato AV, Lücke K, Schlipf J, Teutonico L. J Appl Phys
2001;49:3433–41. 1964:2732–45.
[17] Asaro RJ, Needleman A. Acta Metall 1985;33:923–53. [28] MSC. MARC user’s manual 2005, User Subroutines and Special
[18] Wu TY, Bassani JL. Proc Roy Soc A 1991;435:21–41. Routines, vol. D.
[19] Kocks UF. ASME J Eng Mater Tech 1976;98:76–83. [29] Konrad J, Zaefferer S, Raabe D. Acta Mater 2006;54:1369–80.
[20] Estrin Y, Mecking H. Acta Metall 1984;32:57–70. [30] Hosford WF. The mechanics of crystals and textured polycrys-
[21] Mughrabi H. Mat Sci Eng 1987;85:15–31. tals. Oxford: Oxford University Press; 1993.